H
O
D
V
W
L
F
L
W
\
P
R
G
X
O
X
V
>
0
3
D
@
0RUWDUFODVV
%ULFNFODVV
&RPSUHVVLRQHODVWLFLW\PRGXOXV(>03D@IRUVROLGFOD\EULFN
PDVRQU\ZLWKOD\HUWKLFNQHVVDQGPP
M75
M100
M125
M150
M200
Figure 2. Latvian building code methodology deIined mean elasticity modulus Ior elastic
masonry approach
'Computational Civil Engineering 2007, International Symposium 45
To understand the heritage masonry there are 3 point brick bending tests done.
Shown in Iigure 3 bending test results highlight main problem oI heritage masonry
material deIinition load bearing resistance result dispersion induced by home
manuIacturing process variability.
Figure 3. Three point brick bending test results
3 point bending test shows diIIerent tensile resistance by crash Iorce. Test also
shows equal elasticity Ior all specimens and brittle material approach oI stone
material. Structural analysis in elastic stage using FEM linear masonry approach is
not acceptable Ior close to compression resistance calculations. In that cause
plasticity theory must be used. Not only brick manuIacturing signiIicantly aIIect
whole masonry material properties but also various stone material usage, block size
and Iorm variations, mortar component variations and joint thickness variations.
Our aim is to give possibility to understand and analyze masonry arch, in Iact, any
combination oI them, i.e., a masonry building. The 3D surIace scanning with
20mm precision is so detailed that the internal structure can be described easily.
Behind plaster layers Iorming the regular Iorms and levels oI masonry wall surIace
a most irregular internal structure is Iound. Homogeneity, isotropy, uniIorm
mechanical properties, etc., all the common assumptions oI modern conventional
structural analysis cannot be made in this case without violence to the most basic
common sense. Taken Irom Riga Technical University building material laboratory
test report Nr 652000, masonry compression resistance, 3 point bending test given
tensile resistance was used Ior deIinition oI masonry property. Used in
computational soItware material behavior is approximate by material property
variations but can give very close deIormed stage structural understanding. Used in
FEM soItware Staad Pro material properties are described as: Prime modulus oI
46 K. Bondars, A. Korjakins
elasticity (E
0
3600 MPa), average modulus oI elasticity (E
mean
818 MPa),
Yong`s modulus (G 1200 MPa), poisson`s ratio (0.2), thermal linear expansion
coeIIicient (
t
0.000005 1/degree), density ( 18 kN/m3), brick compression
resistance (R
1
4.5 MPa), mortar compression resistance (R
2
2.5 MPa), masonry
compression resistance (R 1.1 MPa), masonry centric tensile resistance (R
t
0.05
MPa), masonry shear resistance Ior head joint (R
sq
0.11 MPa), masonry tensile
resistance in bending Ior head joint (R
tb
0.08 MPa), resistance to main tensile
stresses (R
tw
0.08 MPa), coeIIicient oI creep eIIect (v 2.2).
0
8 . 0 o t   + s
sq rea
R
(1)
DeIined in equation 1 by Latvian building code methodology limiting tangential
stresses, where 0.7 Iriction coeIIicient by joint mortar;
0
compression
stresses average value Irom lightest loading with reduction coeIIicient 0.9.
Tangential limiting value in second stage oI calculation is deIined as crack Iorming
value Ior shear Iorces in macro modeling material approach. Macro modeling
approach is simpliIying material description and may normally be used in analytic
FEM structural computing by Rikards 9. The macro modeling oI masonry as a
composite is latest developed material approach by Rots 10. Loureno and Rots
11 macromodeling technique, speciIically Iormulated Ior the analysis oI masonry
constructions, is based on lumping all inelastic phenomena to the joints by means
oI a composite interIace model. This model, stemming Irom plasticity,
comprehends three diIIerent Iailure mechanisms, namely, a straight tension cutoII
Ior mode I Iailure, the Coulomb Iriction model Ior mode II Iailure as well as an
elliptical cap Ior compression and combined shearcompression Iailure. As in the
previous case, Loureno and Rots model requires the values oI the initial axial and
shear stiIInesses K
N
and K
T
as input data.
3.2. Micro modeling approach
Micro modeling oI masonry material is precise stone material interaction
description possibility but also signiIicantly increase computing time and amount.
In early stage oI micro modeling dry joint approach was used. In latest
developments the block and the mortar in the joints are represented by continuum
models, whilst the interIace unitmortar is represented by discontinuous elements.
The Young model, the Poisson coeIIicient and the inelastic properties oI the units
and the mortar are taken into account. Micro modeling is widely investigated and
described by Lourenco and Rots 12 and Iind mortar material and stone material
plasticity, load bearing tensile and compression resistances depending Irom joint
position, Iriction angle a.c. Micro modeling material approach must be used in
second step oI computational investigation to analyze support settlement deIormed
structure situation. For those purposes additional masonry material properties must
by involved.
'Computational Civil Engineering 2007, International Symposium 47
4. MONITORING DATA COLLECTION
AIter World War Two Riga cathedral was closed as church and reorganized to
concert hall. Soviet government put serious amount oI Iunding to upliIt and repair
the structure, installed new ventilation, heating and Iire saIety systems. Latest
Iunding in those systems renovation, crack monitoring and geological survey make
a possibility to install long term monitoring system and make the present research.
4.1. Cracked arch photo Iixation
Since 1959 cracks has been survived in Riga cathedral. Photo Iixation made by
heritage building protection institutions show lack oI cracks in masonry shells. This
is useIul inIormation to understand the new crack Iorming and prolongation oI
existing in long term monitoring.
4.2. Arch masonry building sequence and layer orientation Iixation
From 1960 to 1961 each arch has been sketched and described by J. Stukmanis. In
very careIul way each brick and block sizes, Iorm, orientation and position in
masonry structure. Shown in Iigure 4 arch orientation, rib positions, brick sizes and
Iorms were Iixed in his work. Also description oI damages and cracks Ior each arch
was written totally on 68 pages and 30 sketches done. Collected in State Cultural
Monument Archive materials about structure investigations and reconstruction
process give we better understanding oI existing situation.
Figure 4. CareIul Iixation oI all cross arch shells was done by J. Stukmanis
48 K. Bondars, A. Korjakins
Fixation oI damages aIter World War Two is very useIul inIormation in nowadays
to understand changes since that time. Crack prolongation, widening and new
cracking Iixation oI masonry structure thereIore can be done by present crack
monitoring. AIter crack Iixation plastering oI arch surIace was accomplished and
cracks hided.
4.3. Cathedral scanning by 3D Leica laser scanners
Building scanning gives the possibility oI virtual three dimensional presentations in
tourist web sites. This very detailed scanning by three dimensional Leica laser
scanner give precise geometric surIace deIining Ior computational analyze oI
structure. Made by Kalinka and Reiniks, specialists Irom geodesic company Merko
laser scanning in 2006 used Ior FEM computational analyze. Precision oI 20mm on
internal surIaces and less detailed Ior outside is today used Ior structures. Increase
out coming virtual model precision theoretically gives whole structure computing
possibility. Future development oI computational hardware will provide us with
powerIul method oI masonry structure analyze.
Figure 5. From Iull 3D model made plan and arch marking in plan made by Stukmanis
'Computational Civil Engineering 2007, International Symposium 49
Figure 5 shows building plan section oI model highlights geometrical uniIormity oI
structure in comparing with Slavietis, Seglins and Drugis in 1959 made hand tool
measurements. Scanned Gothic arch geometry also is preIerable in comparing with
Erdmanis 1963 proportionality Iindings shown in Iigure 6. Proportion Iindings are
simpliIying method oI geometric approach to use the hand calculations oI thrust
line. Geometrical inequalities lead to serious diIIerence between existing situation
and real situation in cross section.
Figure 6. Proportionality Iindings by Erdmanis in 1963
By means oI modern computational FEM soItware there is a possibility to analyze
the whole building. We Iind the laser scanning as very Iast geometric modeling Ior
FEM soItware.
4.4. Microclimate monitoring
StaII managed microclimate monitoring Ior internal temperature and humidity was
perIormed Irom January 2002 to June 2004. Temperature oscillation gives
signiIicant inIluence on masonry deIormations and crack movement. In developed
now SOFO type based monitoring temperature external and internal measure is
included.
50 K. Bondars, A. Korjakins
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
2
0
0
3
.
0
1
.
0
2
2
0
0
3
.
0
1
.
1
6
2
0
0
3
.
0
1
.
3
0
2
0
0
3
.
0
2
.
1
3
2
0
0
3
.
0
2
.
2
7
2
0
0
3
.
0
3
.
1
3
2
0
0
3
.
0
3
.
2
7
2
0
0
3
.
0
4
.
1
0
2
0
0
3
.
0
4
.
2
4
2
0
0
3
.
0
5
.
0
8
2
0
0
3
.
0
5
.
2
2
2
0
0
3
.
0
6
.
0
5
2
0
0
3
.
0
6
.
1
9
2
0
0
3
.
0
7
.
0
3
2
0
0
3
.
0
7
.
1
7
2
0
0
3
.
0
7
.
3
1
2
0
0
3
.
0
8
.
1
4
2
0
0
3
.
0
8
.
2
8
2
0
0
3
.
0
9
.
1
1
2
0
0
3
.
0
9
.
2
5
2
0
0
3
.
1
0
.
0
9
2
0
0
3
.
1
0
.
2
3
2
0
0
3
.
1
1
.
0
6
2
0
0
3
.
1
1
.
2
0
2
0
0
3
.
1
2
.
0
4
2
0
0
3
.
1
2
.
1
8
Temp, A RH, A TempLim RHLim
Figure 7. 2003 year internal temperature and humidity monitoring by staII
At Iigure 7 shown staII made monitoring Ior temperature and humidity Ior year
2003. Line Temp,A shows temperature variation during 2003 year period. RH,A
line shows humidity variations along 2003 year period. Temp,Lim and RH,Lim
lines show lower limiting values Iavorable to Iungous Iorming. Temperature
initiated deIormations simulated by computer modeling show signiIicant inIluence
on crack Iorming, prolongation and movements. Cyclic temperature loading in
eight hundred years oI cathedral liIe time aIIect not only internal comIort but also
stress situation in cross arch system seriously changing thrust line position. Solar
radiation eIIect on cathedral external surIace can be deIined as load on FEM
elements also.
4.5. Support condition survey
Support condition and subsoil situation survey in last two years made by CM GIB
Geotechnical Company presented by Celmins and Markvarts show weak soil layer
presence under Iootings. Started in 2005 geotechnical survey is more detailed than
in Iormer Soviet period ever done. Cross section oI piled supported column Iooting
is shown in Iigure 8. Marked as 7```D low density sand layer around the wooden
pile Iooting is main Riga cathedral part unequal deIormation reason.
'Computational Civil Engineering 2007, International Symposium 51
Figure 8 Column Iooting section in building middle span
Also an unsuited condition wooden piling was Iound. Support deIormations along
time counted as main reason oI crack Iorming in masonry arch shell system. This is
object oI present research to Iix the saIe exploitation limits. Extreme support
deIormation oI masonry shell system change Iorce lines and part interacting
stresses in shell section. Cracking changed thrust line position in cross section is
saIe situation deIinition criteria. Present deIormed situation and LBS Konsultants
make support deIormation simulation by Plaxis in 2005 show approximate unequal
support deIormations in 16 cm by various parts oI cathedral. Future support
deIormations predict crack prolongation, plastic hinge Iorming and unequal support
deIormation possibility.
To understand ground water Ilow and level changes Riga Department ordered
ground water control monitoring Irom BaltOstGeo Company. The ground water
table monitoring during 2006 showed approximately 0.4m wooden pile coverage
and less. In situation when wooden pile caps are not covered with ground water
table wood structure degradation is possible. Solutions must be Iound to prevent air
exposure possibility.
4.6. SOFO deIormation monitoring system development
Surveillance des Ouvrages par Fibres (SOFO) optical Iiber sensors monitoring
system received Irom Smartec and would be mounted in Iew Iollowing months to
52 K. Bondars, A. Korjakins
change mechanical tensometer crack monitoring to a long term. Optical sensor
mounting cracks to provide masonry part movement in Iive years is decided to
control saIe exploitation. According with mechanical tensometer monitoring
reports presented Irom 2005 crack oscillation varies 12mm every year depending
on season. The main reason oI optical tensometer usage is computational control oI
measurement data, long lasting liIe period and minimized side Iactor`s inIluence on
measurement`s quality.
Figure 9 SOFO principles oI optic tensometer
The SOFO measuring system gives precision oI 0.02mm thereIore deIining high
level oI monitoring. Used as reading unit Smartec Bee allows control oI 24 optical
sensor units and communicates with registration PC trough telephone line.
Additional thermo sensors Ior external and internal temperature control are
included in measured program. Internal memory oI reading unit gives possibility oI
data downloading by reasonable schedule. Battery support oI reading unit gives
possibility oI nonstop monitoring. Support deIormation caused crack widening is
the way oI saIe criteria exploitation deIinition.
5. STABILITY CALCULATION AND SAFETY CRITERIA
DEFINITION
As discussed beIore critical thrush line position must be Iound on deIormed
structure oI masonry arch shell structure. Laser three dimensional scanning
geometrical models were used to deIine geometric Iorms oI Gothic arch system.
The latest FEM soItware usage in heritage building masonry calculations in two
stages is developed. In Iirst stage shown in Iigure 10 linear elastic material
'Computational Civil Engineering 2007, International Symposium 53
approach is used. Support deIormation loading is applied by Staad Pro soItware. In
second stage oI analyze deIormed structure shape and plastic material approach is
used to deIine stability by thrust line position in section.
Figure 10 First stages analyzed with Staad Pro elastic material approach using
The second stage oI deIormed structure includes the Iixed and measured by
monitoring support settlements, cracked part interaction and subsoil elasticity.
Plaxis computing tool is developed Ior taking into account masonry plasticity and
stiIIness. Each arch part is stabile and saIe till thrust line locate in curved shell
cross section. DeIined by Heyman Iour hinge Iorming as saIety criteria is also used
to Iind the support deIormation limits by numerical computing.
6. CONCLUSIONS
The present research is a small part oI huge monitoring and investigation amount
done in Riga cathedral. All previously collected investigations data and
calculations increase the cathedral structure detailed understanding. Our view about
masonry material computing technologies and possible material assumptions
received Irom available papers are used to prepare FEM model and computational
strategy choice. Laser scanning based surIace geometric data transIer to FEM
computational soItware signiIicantly reduce modeling time. Improving laser
scanned precision and computational devices will give the powerIul tool Ior
54 K. Bondars, A. Korjakins
structural analyze oI heritage buildings. Detailed investigations oI material
properties, exploitation conditions and deIormed situation must be taken into
account. None methodology describing masonry material is possible to include the
side eIIects oI building process. It is necessary to point out the side eIIect oI
building process: signiIicant dispersion oI material properties; geometric and
material variations; shell structure deIormed conditions and various material
interactions. Wind, solar radiation and temperature loading give no big inIluence
on computational results but can be important. Two step computational method
highlight deIormed shape inIluence on thrust line position in cross section oI arch
shell. The monitoring improvement by long term SOFO optical tensometer tool is
the way Ior detailed structure investigation and saIe exploitation system establish
oI Riga cathedral. Methodology developed with curved arch surIaces laser
scanning in reason to use data as geometrical model Ior FEM computing soItware
is our aim oI Iuture improvement.
Acknowledgements
Thanks to Riga cathedral staII Ior huge amount oI Iormer Soviet time investigation
and monitoring data, which is included in this paper. Involved investigator`s permit
give us big acknowledge to understand the cathedral building. Thanks to the Riga
cathedral representatives understanding in the latest technology usage on arch long
term monitoring advantage. Thanks a lot Ior complex monitoring development and
Iunding in heritage building to keep it in good health.
ReIerences
1. Heyman J., The Stone Skeleton: Structural Engineering oI Masonry Architecture, Cambridge
University Press, Cambridge, 1990 pp 45.
2. Heyman J., The science oI structural engineering, Imperial College Press, 1999.
3. Heyman, J. The equilibrium oI shell structures. OxIord: Clarendon Press. 1997
4. Heyman, J., 'The masonry arch, Ellis Horwood Limited, UK. 1982.
5. Heyman J., The saIety oI masonry arches`, Int. Journ. Mech. Sci., 11 (4), 1969, pp. 363385.
6. Hughes, T.G., Analysis and assessment oI twinspan masonry arch bridges, Proc. Instn. Civ.
Engrs., 110, 1995, 373382.
7. Livesley, R. K. Limit analysis oI structures Iormed Irom rigid blocks. Int. Journ. Ior Num. Meth.
in Eng. 12, 1978, p. 18531871.
8. Loureno, P. B. Computational Strategies Ior Masonry Structures. PhD thesis, DelIt University oI
Technology, 1996
9. Rikards, R., and Cate A., Finite element method, Riga Technical University, Riga, pp 78, 2002
10. Rots, J.G.  Numerical simulation oI cracking in structural masonry, Heron, 36(2), 1991, p. 4963.
11. Loureno, P.B., de Borst, R., Rots, J.G., A plane stress soItening plasticity model Ior orthotropic
materials, Int. J. Numer. Meth. Engng., 40, 1997, p. 40334057
12. Loureno, P.B., Rots, J.G., A multisurIace interIace model Ior the analysis oI masonry structures,
J. Engrg. Mech., ASCE, 123(7), 1997, p. 660668
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Application oI MS Excel Solver
in Solving Structural and Management Problems
Jacek Boro
Builaing Engineering Institute, Wroclaw University of Technology, Wroclaw, 50377, Polana
Summary
Solvers, or optimi:ers, we call software tools that help us as users to fina the best
way to allocate scarce resources. The resources may be everything in limitea
supply. The "best" or optimal solution may mean maximi:ing profits, minimi:ing
costs, or achieving the best possible quality. An almost infinite variety of problems
can be tacklea this way, but here are some typical examples. finance ana
investment, manufacturing, aistribution ana networks.
To use the solver you have to buila a moael that specifies. aecision variables,
constraints ana obfective. The Solver will fina values for the aecision variables that
satisfy the constraints while optimi:ing (maximi:ing or minimi:ing) the obfective.
Spreaasheets such as Microsoft (MS) Excel proviae a convenient way to buila
such a moael.
Anyone who has usea a spreaasheet is alreaay familiar with the process. Cells on
a worksheet can hola numbers, labels, or formulas that calculate new values, such
as the obfective of an optimi:ation. Constraints are simply limits (specifiea with
, or ~ relations) on formula cells. Ana the aecision variables are simply
input cells containing numbers. Whether the Solver can fina a globally optimal
solution, a locally optimal solution, or a gooa solution aepenas on the nature of the
mathematical relationship between the variables ana the obfective function ana
constraints (ana the solution algorithm usea)[1{.
In this paper it will be shown how we can use the ExcelSolver (the proauct
aevelopea by Frontline Systems, Inc. for MS Excel) to fina the best (optimal)
solutions in some structural problems, as well tasks connectea with management of
an enterprise. The illustrative examples will be these, which I always use for my
aiaactics at Builaing Engineering Institute. Numerical results for two of sample
problems aemonstrate how versatile (multipurpose) ana useful tool was usea.
KEYWORDS: excelsolver, modeling, optimizers, structural optimization
problems, managing optimization problems
56 J. Boro
1. INTRODUCTION
The Solver being part oI MicrosoIt Excel (developed by Frontline Systems Ior
MicrosoIt) is one oI the widely used optimization tools. It is able to solve small
scale linear programming (LP), smooth nonlinear programming (NLP), and mixed
integer programming (MIP) problems. An upgraded Premium Solver PlatIorm
(PSP) is a tool that extends Iunctionality and speed oI the MSE Solver to handle
industrialscale problems oI over tens oI thousands variables and constraints. PSP
is also challenging global optimization problems using multistart or clustering
methods and nonsmooth problems using methods based on genetic and
evolutionary algorithms (or tabu search).
Table 1. The Characteristic oI the Enhanced Excel Solvers 1.
'Computational Civil Engineering 2007, International Symposium 57
Remark Ior the Table 1: For integer problems B&B reIers to branch and bound and
P&P reIers to preprocessing and probing. For nonlinear problems 'GRG reIers to
the generalized reduced gradient method and 'SQP reIers to sequential quadratic
programming.
2. USING SPREADSHEETS
2.1. Spreadsheets vs. Individual Programming
Spreadsheets such as MicrosoIt Excel provide a convenient way to build such a
model. Anyone who has used a spreadsheet is already Iamiliar with the process:
Cells on a worksheet can hold numbers, labels, or Iormulas that calculate new
values (such as the objective oI an optimization).
Constraints are simply limits (speciIied with , or ~ relations) on Iormula
cells.
The decision variables are simply input cells containing numbers. Frontline's
Premium Solver products provide powerIul tools Ior solving, or optimizing, such
model.
Another way to build an optimization model is to write code in a programming
language such as Visual Basic or C/C. Instead oI spreadsheet cells, variables or
arrays in the program hold the decision variables and calculate the constraints and
objective.
2.2. How do we deIine a model?
Every optimization model consists oI:
decision (design) variables and parameters,
objective Iunction,
constraints.
2.2.1. Decision variables
They usually measure the amounts oI resources, such as money, to be allocated to
some purpose, or the level oI some activity, such as the number oI products to be
manuIactured, the number oI pounds or gallons oI a chemical to be blended, etc. In
engineering problems decision variables are dimension oI the structures or cross
sections.
58 J. Boro
2.2.2. Obfective function
AIter deIining the decision variables, the next step is to deIine the objective, which
is normally some Iunction that depends on the variables. We'd be Iinished at this
point, iI the model did not require any constraints (but in engineering problems it is
rather impossible). For example, in a curveIitting application, the objective is to
minimize the sum oI squared diIIerences between each actual data value, or
observation, and the corresponding predicted value. This sum has a minimum
value oI zero, which occurs only when the actual and predicted values are all
identical. II we asked a solver to minimize this objective Iunction, we would not
need any constraints.
In most models, however, constraints play a key role in determining what values
can be assumed by the decision variables, and what sort oI objective value can be
attained.
2.2.3. Constraints
Constraints reIlect realworld limits on production capacity, market demand,
available Iunds, and so on. To deIine a constraint, you Iirst compute a value based
on the decision variables. Then you place a limit (, or ~) on this computed
value.
General Constraints. For example, iI A1:A5 contains the percentage oI Iunds to
be invested in each oI 5 stocks, you might use B1 to calculate SUM(A1:A5), and
then deIine a constraint B1 1 to say that the percentages allocated must sum up to
100.
Bounds on Variables. OI course, you can also place a limit directly on a decision
variable, such as A1 100. Upper and lower bounds on the variables are
eIIiciently handled by most optimizers and are very useIul in many problems.
Policy Constraints. Some constraints are determined by policies that you or your
organization may set. For example, in an investment portIolio optimization, you
might have a limit on the maximum percentage oI Iunds to be invested in any one
stock, or one industry group.
Physical Constraints. Many constraints are determined by the physical nature oI
the problem. For example, iI your decision variables measure the number oI
products oI diIIerent types that you plan to manuIacture, producing a negative
number oI products would make no sense. This type oI nonnegativity constraint is
very common. Although it may be obvious to you, constraints such as A1 ~ 0
must be stated explicitly, because the solver has no other way to know that
negative values are disallowed.
As another example oI a physically determined constraint, suppose you are
modeling product shipments in and out oI a warehouse over time. You'll probably
'Computational Civil Engineering 2007, International Symposium 59
need a balance constraint, which speciIies that, in each time period, the beginning
inventory plus the products received minus the products shipped out equals the
ending inventory (and hence the beginning inventory Ior the next period).
Integer Constraints. Advanced optimization soItware also allows you to speciIy
constraints that require decision variables to assume only integer (whole number)
values at the solution. II you are scheduling a Ileet oI trucks, Ior example, a
solution that called Ior a Iraction oI a truck to travel a certain route would not be
useIul. Integer constraints normally can be applied only to decision variables, not
to quantities calculated Irom them. A particularly useIul type oI integer constraint
speciIies that a variable must have an integer value with a lower bound oI 0, and
upper bound oI 1. This Iorces the variable to be either 0 or 1 (nothing in between)
at the Iinal solution. Hence, it can be used to model "yes/no" decisions. For
example, you might use a 01 or binary integer variable to represent a decision on
whether to lease a new machine. Your model might then calculate a Iixed lease
cost per month, but also a lower cost per item processed with the machine, iI it is
used. A solver can help determine whether leasing the machine will yield higher or
lower proIits.
2.2.4. Interpreting solutions
A solution (set oI values Ior the decision variables) Ior which all oI the constraints
in the Solver model are satisIied is called a feasible solution. Most solution
algorithms Iirst try to Iind a Ieasible solution, and then try to improve it by Iinding
another Ieasible solution that increases the value oI the objective Iunction (when
maximizing, or decreases it when minimizing). An optimal solution is a Ieasible
solution where the objective Iunction reaches a maximum (or minimum) value.
A globally optimal solution is one where there are no other Ieasible solutions with
better objective Iunction values. A locally optimal solution is one where there are
no other Ieasible solutions "in the vicinity" with better objective Iunction values
(we can picture this as a point at the top oI a "peak" or at the bottom oI a "valley"
which may be Iormed by the objective Iunction and/or the constraints). The Solver
is designed to Iind optimal solutions (ideally the global optimum) but this is not
always possible. In many cases, though, we may be happy to Iind a good solution
(one that is better than the solution we are using now). Whether the Solver can Iind
a globally optimal solution, a locally optimal solution, or a good solution depends
on the nature oI the mathematical relationship between the variables and the
objective Iunction and constraints (and the solution algorithm used).
2.2.5. What makes a moael hara to solve?
Solver models can be easy or hard to solve. "Hard" models may require a great deal
oI CPU time and randomaccess memory (RAM) to solve (iI they can be solved at
60 J. Boro
all). The good news is that, with today's very Iast PCs and advanced optimization
soItware Irom Frontline Systems, a very broad range oI models can be solved.
Three major Iactors interact to determine how diIIicult it will be to Iind an optimal
solution to a solver model:
the mathematical relationships between the objective and constraints, and the
decision variables
the size of the model (number oI decision variables and constraints) and its
sparsity
the use of integer variables  memory and solution time may rise exponentially as
you add more integer variables
2.3. Using the Solver dialogs
Assuming that we have the standard Excel Solver or a Premium Solver product
installed, the next step is to create a worksheet where the Iormulas Ior the objective
Iunction and the constraints are calculated. In the worksheet below, we have
reserved cells C15, D15, E15 and F15 to hold our decision variables x
1
, x
2
, x
3
and
x
4
: amount oI advertisments through TV, radio, direct mail and newspapers.
Fig.1. Example 1: Advertising
To add the constraints, we click on the Add button, select cells C16:F16 in the Cell
ReIerence edit box (the leIt hand side), and select cells C11:F11 in the Constraint
edit box (the right hand side); the deIault relation is OK (Fig.2.)
'Computational Civil Engineering 2007, International Symposium 61
Fig.2. Adding the constraints in the edit box (polish version oI MS OIIice)
We choose the Add button again (either Irom the dialog above, or Irom the main
Solver Parameters dialog) to deIine the nonnegativity constraint on the decision
variables.
When we've completely entered the problem, the Solver Parameters dialog looks
like this (Fig.3.) This dialog appears Ior the Standard Solver PlatIorm.
Fig.3. The Solver Parameters Dialog Box (polish version oI MS OIIice)
2.4. Finding and using the solution
To Iind the solution, we should click on the Solve button. AIter a moment, the
Solver returns the optimal solution: $9000 in cell C15, $3000 in D15, $0 in E15
and $1875 in cell F15. This means that we should advertise 18 times in TV, 15
times in radio and newspaper, to make a total, minimal cost $13875 (shown at cell
G15). We shouldn`t invest in ads through direct mail. The message "Solver Iound a
solution" appears in the Solver Results dialog, as shown here (Fig.4.). We now
click on "Answer" in the Reports list box to produce an Answer Report, and click
OK to keep the optimal solution values in cells C15:F15.
62 J. Boro
Fig.4. Solver results dialog box
AIter a moment, the Solver creates another worksheet containing an Answer
Report, like the one below, and inserts it to the leIt oI the problem worksheet in the
Excel workbook. (Fig.5.)
Fig.5. Answer report worksheet
3. ILLUSTRATIVE EXAMPLES
3.1. Design (structural) problem optimization oI the welded crosssection
Our target is to optimize a doubleT, bisymmetric crosssection oI a welded, steel
beam, connected with a stiII plate made oI armed concrete (Fig.6.). We allow only
'Computational Civil Engineering 2007, International Symposium 63
one loading condition: bending moment M. We consider 4 design variables t, h, b
and g. Objective Iunction F
c
going to be minimized:
F t + 2 b g) (1)
c
(0,125 R
e
+ s) (h
The worksheet prepared to solve the problem is presented below (Fig. 6.).
67((/%($0237,0,=$7,21
,QSXWGDWD
0 5000 kNm (bending moment)
l
0
= 4,000 m. (beam span)
Strenght constraints (simple bending, by Polish Standard PN90/B03200)
therefore
Considering conection of our beam with stiff plate gives: 
L
= 1 and o
p
= 1,05.
Design (geometric) constraints:
t >= 0,008
0< h <= 0,85
g > 0
6*g<= b <= 20*g prevents stretching out of the belts during welding
cost F
c
= 1663,082 61688,97
design b = 0,387 1,000 starting
variables g = 0,850 1,000 values
h = 0,064 1,000
t = 0,008 0,100
R
e
= ( 2,25E+05 2,15E+05 kPa
plasticity border
J(b,g,h,t) = 0,011 m
4
W
z
(b,g,h,t) = 0,022
m
3
s = 0,7
f
d
= 205 195 MPa strenght of the steel
H
p
= 1,05
steel class: 6W6
62/9(5
W
z
>= M
z
/(o
p
*
L
* f
d
)=
M <= M
R
* 
L
, where M
R
= o
p
* W
z
* f
d ,
Tutaj wstaw wykres momentw zginajcych belki
%HQGLQJPRPHQW0[
0,0
1000,0
2000,0
3000,0
4000,0
5000,0
6000,0
0
,
0
0
,
4
0
,
8
1
,
2
1
,
6
2
,
0
2
,
4
2
,
8
3
,
2
3
,
6
4
,
0
FRRUGLQDWHIURPGRP
0
Y
D
O
X
H
V
L
Q
N
1
P
0
[
Fig.6. Optimization oI the welded steel beam crosssection (4 design variables)
64 J. Boro
3.2. Management oI an enterprise problem  purchasing & transportation
Problem. A concrete manuIacturer has 4 diIIerent plants that all require a certain
amount oI cement. There are 5 diIIerent companies (cement Iactories) where the
cement can be bought. Where should the concrete manuIacturer buy the cement
and how much should it buy, to minimize its' cost and shipping?
Solution. The variables are the amounts oI cement to be bought Irom each
company Ior each plant. On worksheet Purchasing & Transportation Model these
are given the name AmountsoIcementtobuy: cells C15:G18 (see Fig.7.). The
constraints are simple and straightIorward:
Amounts_to_buy (C15:G18) > 0 via the Assume NonNegative option (2)
Total_amounts_to_buy (H15:H18) > Demand (I15:I18) (3)
Total_sold(C19:G19) < Supply(C20:G20) (4)
The objective is to minimize the cost. This is deIined as Totalcost on the
worksheet (cell H24).
Fig.7. Purchasing & Transportation Model (concrete manuIacturing)
Even though this model is very simple, it is one oI the most used models in the
industry. It routinely saves many companies thousands or even millions a year.
'Computational Civil Engineering 2007, International Symposium 65
4. CONCLUSIONS
The optimizer developed by Frontline Systems, Inc. under simple name 'Solver is
multipurpose and Iriendlyinuse tool also Ior engineering and management
problems. ThereIore I decided to include it successIully in my didactic program
beginning Irom the third semester at Faculty oI Civil Engineering.
However in real world optimization problems can be both simple and 'cheap or
complicated and 'expensive, so the types oI mathematical relationships in a model
(Ior example, linear or nonlinear, and especially convex or nonconvex) determine
how hard it is to solve, and the conIidence we can have that the solution is truly
optimal.
These relationships also have a direct bearing on the maximum size oI models that
can be realistically solved. A Iew advanced solvers can break down a problem into
linear, smooth nonlinear and nonsmooth parts and apply the most appropriate
method to each part, but in general, we should try to keep the mathematical
relationships in a model as simple (i.e. close to linear) as possible 2.
ReIerences
1. Fylstra, D. et al., Design ana Use of the Microsoft Excel Solver, Copyright 1998, Institute Ior
Operation Research and the Management Sciences, 00922102/98/2805/0029.
2. Nenov, I., Fylstra, D., Interval Methoas for Acceleratea Global Search in the Microsoft Excel
Solver, Reliable Computing 9: 143159, 2003, Copyright 2003, Kluver Academic Publishers.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Underground structure in discontinuous rock mass
Jii Bostik
Institute of Geotechnics, Brno University of Technology, Brno, 602 00, C:ech republic
Summary
The results of parametric analysis on influence of aiscontinuities orientation on the
response of rock mass auring ariving an unaergrouna structure in the airection of
aiscontinuity planes are presentea in the contribution. Three typical alternatives
were consiaerea. the unaergrouna structure in rock mass with onesiaea aip of
aiscontinuities, the unaergrouna structure situatea in synclinal bena ana the
unaergrouna structure situatea in anticlinal bena. In all three cases, the variability
of aiscontinuities aip angle in the range of 0  90
o
was consiaerea.
Mathematical moaels were usea for the analysis by Finite Element Methoa using
software Plaxis. Mathematical moaels were planar plane strain. Rock Jointea
moael was usea to simulate mechanical behaviour of the rock environment. The
unaergrouna structure is ariven in full face with no lining.
All the analysea alternatives are comparea to one another consiaering the vertical
aisplacement at the top of the unaergrouna structure ana maximum vertical
surface aisplacement. From the results presentea, it can be concluaea that from the
point of view of vertical aisplacement at the top of the unaergrouna structure,
situating the unaergrouna structure in the axis of the synclinal bena, if the aip of
aiscontinuities is not very steep, is the least favourable. On the other hana, taking
into account the maximum vertical surface aisplacement, situating the
unaergrouna structure, in relation to the aiscontinuities orientation, seems to be of
no importance.
KEYWORDS: underground structure, discontinuities, mathematical simulation.
1. INTRODUCTION
The rock environment, in which the underground structures are build and which is
an inseparable part oI them, is usually weakened by several discontinuities. These
discontinuities are created either during the rock Iormation itselI (e.g. stratiIication
oI sedimentary rocks) or in rock which have already been Iormed (e.g. joints, Iault
planes).
'Computational Civil Engineering 2007, International Symposium 67
In case oI sedimentary rocks, the stratiIication, which is developed by lithological
changes oI deposited material, is the most signiIicant structural property.
Occurrence oI the discontinuities in rock mass described above has an impact to its
behaviour during driving oI underground structures. For instance, in 1 the authors
presented that in case oI the stratiIied rocks their placing, thickness and dip oI the
layers and strength oI the rock need to be considered. The angle between tunnel
axis and layers direction, which are generally oI a various aslope, contorted or
Iolded, is a very important Ieature as well.
Driving oI a tunnel in the direction oI layers 1 is less convenient, especially when
the rock is nonhomogenous. The tunnel can be subjected to onesided pressures
when the layers are skewed (Fig. 1a). When the tunnel is situated in the axis oI
anticline (Fig. 1c), the dip oI layers need to be taken into account. Such a situating
the tunnel is very inconvenient in case the layers are steeply skewed as the crown
oI the tunnel was loaded by Iull weight oI the loose layers. Situating the tunnel in
synclinal bend (Fig. 1b) is also considered to be very inconvenient the rock
would pressure on the tunnel Irom either side and huge water inIlow would occur iI
permeable layers were presented. Situating the tunnel path in the direction oI the
vertical or steeply skewed layers is very unsuitable due to large weight oI
overburden aIIecting the crown oI the tunnel.
c b a
Figure 1. Situating tunnel in stratiIied rock in the direction oI the layers 1
2. PARAMETRIC STUDY
In order to analyse an inIluence oI discontinuity planes orientation on the response
oI rock mass during driving an underground structure, the Iollowing model
example was considered. This was solved by Finite Element Method using
soItware Plaxis 2.
68 J. Bostik
2.1. Geometric arrangement
A linear underground structure (collector) oI horseshoe shape, 3.3 m wide, 4.05 m
high and sectional area oI 12.2 m
2
(Fig. 2) is considered. Top oI the underground
structure is 13.2 m under the horizontal surIace, i.e. in depth that equals Iour times
width oI the underground structure.
Figure 2. Schematic arrangement oI underground structure
2.2. Geology conditions
Rock environment is created by one kind oI stratiIied rock. It is assumed, that
single layers are oI the same width (D), which is relatively small as compared to
transverse dimensions oI the underground structure (B), as schematically illustrated
in Fig. 3. Further, it is supposed the discontinuities are not Iilled.
Figure 3. Discontinuities 1, Underground structure 2
Mechanical behaviour oI rock environment in the mathematic model is
approximated by the Jointed Rock model. This model is an anisotropic elastic
perIectly plastic model, especially meant to simulate the behaviour oI stratiIied
and jointed rock layers and its more detailed description can be Iound in 2.
'Computational Civil Engineering 2007, International Symposium 69
Within the analysis, one set oI discontinuities (stratiIication) is used and elastic
behaviour oI rock environment is Iully isotropic, i.e.
1 2
E E = and
1 2
v v = , (1)
) 1 (
2
1
1 1 2
v + = E G , (2)
where
1 1
,v E stand Ior Young`s modulus and Poisson`s ratio oI the rock as a
continuum respectively,
2 2 2
, , G E v stand Ior Young`s modulus, Poisson`s ratio
and Shear modulus in the discontinuity plane direction respectively.
In the discontinuity plane direction it is assumed that shear stresses are limited
according to Coulomb`s criterion. Parameters oI strength in the direction oI
stratiIication are as Iollows: cohesion (c), Iriction angle (), dilatancy angle (),
tensile strength (T  Strength).
Throughout the calculation, the material parameters Irom Table 1 were used.
Table 1. Material parameters used
kN/m3
E
1
kPa
v
1

E
2
kPa
v
2

23.7 6x10
4
0.33 6x10
4
0.33
G
2
kPa
c


o


o

TStrength
kPa
2.3x10
4
20 24 0 0
kPa
Three basic alternatives according to Fig. 4 are solved. In the Iirst one (J), one
sided dip oI discontinuities is considered. In the second one (S), the underground
structure is driven in the symmetrical synclinal bend. In the last one (A) the
underground structure is situated in symmetrical anticlinal bend. In all the three
cases above, the various orientation oI the discontinuities, determined by dip angle
o
1
, is involved. Value oI o
1
is chosen Irom the range oI 15  75
o
with 15
o
step.
2.3. Technology
From the point oI view oI driving method, driving oI the underground structure in
Iull  Iace with no use oI lining is considered.
70 J. Bostik
Figure 4. Considered alternatives: J onesided dip oI discontinuities, A symmetrical
anticlinal bend, S symmetrical synclinal bend
2.4. Mathematical model
The planar mathematical model (plane strain) is Iormed by Finite Element Method
using the soItware Plaxis 2. The analysed area is oI a rectangle shape oI the size
oI 35 x 27.7 m. The mesh with 1 336 elements, 2 753 nodes and 4 008 stress points
(Fig. 5) was created by discretisation using sixnode triangular element.
Figure 5. Finite element discretisation
The top boundary oI the mathematical model representing the surIace is Ireely
sliding, while the other model boundaries are tied with standard boundary
conditions: the vertical boundaries restricted on slide in horizontal plane, the
bottom boundary restricted on slide in both horizontal and vertical directions.
'Computational Civil Engineering 2007, International Symposium 71
Solving the mathematical models was perIormed in the two calculation phases
illustrated in Fig. 6:
1. Calculating the primary state oI stress caused, in this case, only by selI
weight oI rock mass . Value oI horizontal and vertical eIIective stresses ratio
(coeIIicient oI lateral earth pressure) K
0
was 0.593.
2. Carrying out the excavation oI the underground structure.
For all the three alternatives J, S and A, Iive models with diIIerent discontinuities
orientation (see section 2.2) were created. Furthermore, cases where set oI the
discontinuities is horizontally and vertically oriented were solved as well. In total,
seventeen models were created.
Figure 6. Calculation phases
3. RESULTS
Vertical displacement at the top oI the underground structure (named as a point I
Iurther in the text) and maximum vertical surIace displacement were selected in
order to compare the results oI the solved alternatives and their variations. Relation
between these values and discontinuities orientation (dip angle) Ior the analysed
cases are graphically illustrated in Fig. 7 and Fig. 8. In these diagrams, the vertical
displacement between calculated values Ior the speciIic discontinuities orientation
are approximated by bisector.
Case J
In the case J (onesided dip oI discontinuities), the maximum vertical displacement
at the top oI the underground structure was Iound Ior dip angle 45
o
and equals
24.0 mm. II the dip angle increases or decreases the value oI vertical displacement
at the top oI the underground structure decreases. The lowest value (19.5 mm) oI
the vertical displacement in the point I was veriIied Ior the case oI horizontally
oriented discontinuities (o
1
0
o
). This value practically corresponds to the value
obtained in the case oI vertically oriented discontinuities (o
1
90
o
).
72 J. Bostik
40
35
30
25
20
15
0 15 30 45 60 75 9
GLSDQJOH>
R
@
Y
H
U
W
L
F
D
O
G
L
V
S
O
D
F
H
P
H
Q
W
S
R
L
Q
W
>
P
P
@
0
J A S
Figure 7. Vertical displacement at the top oI the underground structure
11
10
9
8
7
6
5
0 15 30 45 60 75 9
GLSDQJOH>
R
@
P
D
[
Y
H
U
W
L
F
D
O
V
X
U
I
D
F
H
G
L
V
S
O
D
F
H
P
H
Q
W
>
P
P
@
0
J A S
Figure 8. Maximum vertical surIace displacement
Similar relation can also be Iound Ior maximum value oI the vertical surIace
displacement. The highest value oI 10.3 mm is achieved Ior o
1
60
o
(almost the
same as 10.2 mm Ior o
1
45
o
). Minimum value oI the vertical surIace
displacement oI 5.9 mm was veriIied Ior o
1
0
o
and in Iact corresponds to the
value oI 6.4 mm calculated Ior o
1
90
o
.
Case A
In the case A, the maximum vertical displacement at the top oI the underground
structure was Iound Ior dip angle 60
o
and equals 32.5 mm. The dependency on o
1
'Computational Civil Engineering 2007, International Symposium 73
is oI almost the same as in the case J except the displacements are bigger and the
maximum value is Iound Ior steeper discontinuities dip as presented above.
Variations with o
1
15, 30 and 75
o
when the calculated displacements are slightly
smaller; more speciIically 1.4, 0.3 and 1.6 mm respectively, are an exception.
Maximum vertical surIace displacement reaches the biggest value oI 11 mm Ior o
1
45
o
. For o
1
60
o
, maximum vertical surIace displacement oI 10.3 mm is the
same as in the case J. Nevertheless, this does not apply to vertical displacement oI
a point at the top oI the underground structure, which is Ior o
1
60
o
signiIicantly
bigger than in the case J. The diIIerence equals 8.8 mm, which is approximately
37 .
Case S
In the case oI situating the underground structure in synclinal bend Ior all
alternatives considered, the maximum vertical displacement at top oI the structure
(40.0 mm) is achieved Ior o
1
15
o
. Concurently, Ior dip angle Irom the range oI
30
o
75
o
, vertical displacement oI the point I does not signiIicantly vary, similarly
to the case J.
The maximum vertical surIace displacement (11.0 mm) was Iound Ior o
1
60
o
as
in the case J. The change in relation to the discontinuities orientation is basically oI
simillar nature as in the earlier cases.
Figure
9. Total displacements around the underground structure: a case J (o
1
45
o
), b
case A (o
1
60
o
), c case S (o
1
15
o
)
c b a
A Iield oI displacements in the very vicinity oI shape oI the underground structure
is illustrated by arrows in Fig. 9. Only those alternatives oI analysed cases, where
the biggest vertical displacement at the top oI the underground structure was
calculated, are presented. Here, the deIormation oI the underground structure shape
Ior onesided dip oI discontinuities is asymmetric and the biggest displacement is
approximately localised in the area oI vault support in the direction transverse to
the direction oI discontinuities (Fig. 9a). In the case A, the biggest displacements
74 J. Bostik
occur at the top oI the underground structure and at the sides oI the underground
structure closely below its vault support (Fig. 9b). In the case S, the biggest
displacement was Iound at the top oI the underground structure (Fig. 9c).
4. CONCLUSION
From the results oI the parametric study, it can be concluded that Irom the point oI
view oI vertical deIormation at the top oI the top heading, driving oI the
underground structure in the axis oI synclinal bend, iI the dip oI discontinuities is
not very steep (o
1
15
o
), is the least Iavourable. Further, Ior the case oI situating
the underground structure in the axis oI anticlinal bend, the calculated maximum
vertical displacement at the top oI the underground structure is 19 less than in
the previous case. From this perspective, the most Iavourable is driving oI the
underground structure in case oI onesided askew aslope layers, where maximum
value oI vertical displacement oI the top oI the top heading is 60 oI the value
obtained Ior the underground structure in synclinal bend.
From the perspective oI the maximum vertical surIace displacement, situating the
underground structure, in relation to the discontinuities orientation, seems to be oI
no importance. This observed quantity is practically oI the same value Ior all the
cases analysed.
It also results Irom the analysis, that maximum values oI vertical displacement oI
the top oI the underground structure and vertical surIace displacement are not
reached Ior the same discontinuities orientation, which is signiIicant especially in
the case oI synclinal bend.
Along to the all above it needs to be considered, that the alternatives were analysed
only Ior certain selected values oI dip angle o
1
in the presented analysis. ThereIore,
the maximum values oI the monitored displacements and corresponding values oI
dip angle do not have to be the values oI absolute maximum Ior the case itselI.
Acknowledgements
The contribution was processed within the grant project GA CR 103/07/P323 and
the research project oI MSM0021630519.
ReIerences
1. Zaruba, Q., Mencl, V., Engineering geology, Publisher CSAV, Prague, 1954. (in Czech)
2. Plaxis 2D Version 8, Edited by R. B. J. Brinkgreve, A.A. Balkema Publishers, 2002.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Discrete model Ior the stability oI continuous welded rail
Adam Dosa, ValentinVasile Ungureanu
Department of civil engineering, 'TRANSILJANIA` UNIJERSITY, Braov, Romania
Summary
In this paper a aiscrete moael is aevelopea for the buckling analysis of continuous
welaea rail subfectea to temperature loaa. The moael is basea on a nonlinear
analysis in total lagrangean formulation. The structure consists of beam elements
ana lateral, longituainal ana torsional spring elements. The source of nonlinearity
is aue to the geometric nonlinearity of the rail high axial forces ana also to the
nonlinearity of material type for the lateral ana longituainal resistance of the
ballast ana the torsional resistance of the fasteners. The use of a aisplacement
control algorithm leaas the analysis beyona the critical point ana permits a more
realistic computation of the structural safety. The track moael is encoaea into a
special purpose program which allows a parametric stuay of the influence of
vehicle loaaing, the stiffness properties of the structure ana of the geometric
imperfections on the track stability.
The valiaity of the present moael is verifiea through a series of comparative
analyses with other authors results.
KEYWORDS: Continuous welded rail, Nonlinear stability analysis, Temperature
loading, Structural saIety.
1. INTRODUCTION
The Iirst computational models oI the buckling oI the continuous welded rail
(CWR) were developed at the beginning oI the 1930 years. These models can take
into account the main parameters which control the stability oI CWR like the
horizontal and vertical stiIIness oI the rail, the longitudinal and transversal
resistance oI the rail, the torsional resistance oI the Iasteners, the stresses induced
by the vehicle and temperature loading, the geometry and the misalignment oI the
rail. In the SCFJ model presented in this paper the structure consists oI beam
elements and lateral, longitudinal and torsional spring elements. The beam
elements are modeling the rail and have geometric nonlinear characteristics due to
high compressive thermal stresses. The spring elements are describing the material
nonlinear behavior oI the ballast and the Iasteners.
76 A. Dosa, V. Ungureanu
2. DEVELOPMENT OF THE TRACK MODEL
2.1. The longitudinal ballast behavior
In the SCFJ model the longitudinal resistance oI the ballast is introduced by spring
elements having the, linear or bilinear displacementIorce curves given in Iigure 1.
In the case oI vehicle loading, the bilinear curve is corrected 6 by the equation (1)
taking into account the vertical Iorce Q on each sleeper.
v
c
v L v
c
v
U U , tan Q U U
3
2
> + =  (1)
In the above equation is the reIerence value oI the longitudinal resistance
(without vehicle loading), U is the corrected value oI this resistance and
v
U
c
v L
 is the
angle oI the longitudinal Iriction between the sleeper and the ballast.
Figure 1. Longitudinal resistance versus longitudinal displacement oI the ballast
2.2. The transversal ballast behavior
The transversal resistance oI the ballast is introduced by spring elements having the
displacementIorce curves given in Iigure 2. In both cases the elastoplastic model
includes soItening. This kind oI ballast behavior has been measured Ior
consolidated ballast. In the case oI vehicle loading, the bilinear curve is corrected
6 by the equation (1) taking into account the vertical Iorce Q on each sleeper
using equations (2), (3) or (4).
v
c
v T v
c
v
J J , tan Q J J
3
2
> + =  (2)
(3)
v
c
v r
c
r
J / J J J =
'Computational Civil Engineering 2007, International Symposium 77
Ior tipJ1 :
v r
v
v v
v v
c
r
c
v
c
r
) J J ( J
+
2
2 J = (iI ) (4)
v
v v >
In the above ecuations
v
J is the reIerence value oI the peak transversal resistance
(without vehicle loading), is the corrected value oI this resistance,
c
v
J
T
 is the
angle oI the transversal Iriction between the sleeper and the ballast, is the
reIerence value oI the rezidual transversal resistance (without vehicle loading), and
is the corrected value oI this resistance. In the case oI exponential soItening the
diIIerence J is halI at the middle oI v interval.
r
J
c
r
J
c
r
c
v
J
c
r
c
v
v
Figure 2. Transversal resistance versus transversal displacement oI the ballast
2.3. The torsional stiIIness oI the Iasteners
The resistance oI the Iasteners is introduced by torsional springs having the linear
or trilinear behavior shown in Iigure 3. In the case oI loaded rail this behavior also
can be corrected taking into account the vertical Iorce acting on each sleeper.
Figure 3. The torsional stiIIness oI the Iasteners
78 A. Dosa, V. Ungureanu
2.4. The geometrical and physical characteristics oI the rail
The rail is modeled by beam elements having area oI the cross section A, second
order moment about the vertical and horizontal axes I
:
and I
y
respectively. The
Young modulus and the thermal expansion coeIIicient oI the material are E and
respectively o. In the model the misalignment oI the rail can be described by two
types oI curves: a complete or a halI cosine wave having the total length and the
amplitude o (Iigure 6). The length oI the model is an input oI the program. At the
end oI the model special inIinite boundary elements are introduced equivalent
with the theoretical inIinite rail 6. This elements lead to the reduction oI the
length oI the model and hence the computational eIIort. Further reduction can be
obtained by using the symmetric halI structure.
3. THE NUMERICAL ALGORITHM
Since in a simpliIied manner, the horizontal and vertical behavior are considered
decoupled, the numerical algorithm has two phases.
3.1. The computational model Ior vertical loadings
This model is linear elastic consisting oI a beam on elastic springs. The nodes oI
the structure are considered at the sleepers. Each node has two degrees oI Ireedom:
the vertical translation w and the rotation u
y
. The system oI equations oI
equilibrium is:
F Ka = . (5)
where:
K is the stiIIness matrix oI the structure and results by assembling the stiIIness
matrices k oI the beams and the vertical stiIIness oI the Iasteners.
a is the displacement vector oI the nodes oI the structure.
F is the vector oI Iorces at the nodes oI the structure, which (in this case) results by
assembling the vectors f
0
oI the Iorces on the beams.
The stiIIness matrix k
(4x4)
oI a beam is given by the equation:
B k B k
a T
=
. (6)
Here B
(2 x 4)
is a transIormation vector, which links the vector oI displacements oI
the beam and the reduced vector oI displacements oI the beam. The reduced vector
oI displacements does not contain the rigid body displacements.
'Computational Civil Engineering 2007, International Symposium 79
( )
T
yi i yi i el
a
yi
a
yi a
el
w w
L L
L L
1 1
1
1 1 0 1
0 1 1 1
+ +
+
(
= =
= u u
u
u
a B a (7)
Figure 4. The displacements oI the beam in the vertical plane
k
a
(2 x 2)
is the reduced stiIIness matrix oI the beam.
(
=
4 2
2 4
L
EI
y a
k (8)
II the beam is loaded, the vector f
0
oI equivalent Iorces in the nodes is given by
equations (9).
T
i i i i
i i i i
i i i i
) M T M T (
L ) M M a P ( T L b a P M
L ) M M b P ( T L a b P M
0
1
0
1
0 0
0
1
0
1
2 2 0
1
1
0 2 2 0
+ +
+ + +
+
=
= =
+ + = =
f
(9)
Figure 5. Equivalent nodal Iorces oI the beam
The stiIIness matrices and the load vectors oI the beams are assembled by the
relation (10).
0
f F F k K K + = + =
ina ina ina , ina ina , ina
, . (10)
80 A. Dosa, V. Ungureanu
Here ina is the vector oI the indices oI the displacements oI the current beam.
The stiIIness oI the sleepers is assembled with the help oI the equation (11).
L R
: fna , fna fna , fna
+ = K K (11)
In the above equation fna is the set oI indices oI vertical displacements oI the nodes
fna1, 3, ..., 2nna1. The constraints oI the structure are introduced by setting to
zero the displacements oI the supports. The Iree displacements oI the nodes result
by solving the system oI linear equations:
. (12)
ia ia , ia ia
) ( F K a
1
=
In equation (12) ia is the set oI the Iree displacements oI the structure.
Using the vertical displacements, the vertical Iorce on each sleeper can be
computed by the equation (13)
sleeper :
G L wR Q + = . (13)
The transversal, longitudinal and torsional resistances are corrected taking into
account the Iorces Q on each sleeper using equations (1) to (4).
3.2. The computational model in the horizontal plane
The model is a straight or curved beam on elastic supports with misalignments
(Iigure 6). The nodes oI the structure are considered at the sleepers. At each node
are introduced longitudinal, transversal and rotational spring elements which are
modeling the sleepers. The inIinite boundary elements at the ends oI the model
have equivalent characteristics (Young modulus and thermal expansion coeIIicient)
in order to replace the theoretical inIinite rail 6. The loading oI the model is an
increase oI the temperature in the rail. The characteristics oI the beams and oI the
springs correspond to the two rails oI the track panel. A node has three degrees oI
Ireedom: two linear displacements in the horizontal plane, u and v and the rotation
u
:
around the vertical axis. In the analysis oI the structure the goal is to obtain the
displacementtemperature curve. The problem is solved by a displacement control
based incremental process. The behavior oI the system is determined as a sequence
oI increments oI state parameters (Iorces and displacements). In the current
increment f characterized by the small control displacement ov
cf
, the nonlinear
behavior oI the system can be approximated by a linear relation between the
successive increments oI the state parameters:
f f f f f f
, a K F a a a o o o = + =
+1
. (14)
'Computational Civil Engineering 2007, International Symposium 81
In the above equation a
f
is the displacement vector in the current conIiguration, oa
f
is the increment oI the displacements, oF
f
is the incremental load vector and K
f
is
the incremental (tangent) stiIIness matrix oI the structure.
Figure 6. The model Ior horizontal displacements
By using equations (14), the Iollowing incremental scheme results:
. (15)
f f f f f f
, ) ( a a a F K a o o o + = =
+
1
1
In this paper an improved scheme is used, known as Heun`s or midpoint rule:
. ) (
), (
), (
, ) (
f f f f
f f
f f
f f f f
2 1
1
2 1 1
2 1 2 1
2 1 2 1
1
2 1
2 1
2 1
+
+ +
+ +
+ +
+
+ =
=
=
+ =
F K a a
a F F
a K K
F K a a
o
o o
o
(16)
The incremental load vector oF
f
is not computed. The incremental displacement oa
f
is the result oI a yet unknown increment oI the temperature produced by a known
increment ov
cf
oI the control displacement. For simplicity, in the next equations
indices f oI the current conIiguration are dropped. The displacement control
consists oI loading the system with displacement increments ov
c
in a speciIic node.
As a rule in this paper: the control displacement is taken as the maximum
transversal displacement oI the node on the symmetry axe oI the structure. The
phases oI the computation are the Iollowing:
 It is adopted a base system with the control displacement Iixed at zero.
 This base system is loaded with two load cases:
i) Load 1 is a temperature increase oT 1, which produces displacements oD
(1)
at
the Iree nodes and reaction R
(1)
in the artiIicial support.
ii) Load 2 is a displacement ov
c
oI the artiIicial support, which produces
displacements oa
(2)
at the Iree nodes and reaction R
(2)
in the artiIicial support.
82 A. Dosa, V. Ungureanu
Figure 7. The determination oI the temperature and displacement increments
The base system and the initial system are identical iI the total reaction in the
artiIicial support is zero. R R
(1)
oT R
(2)
0. This equation yields the unknown
variation oT oI temperature and the incremental displacements oa oI the Iree nodes.
, (17)
) ( ) (
R / R T
1 2
= o
. (18)
) ( ) (
T
2 1
a a a o o o o + =
The tangent stiIIness matrix K
f
in the f increment depends on the parameters oI the
system in the current step and results by assembling the stiIIness matrices k
t(6 x 6)
oI
the beams and oI the springs which model the sleepers.
. (19) z z B k k B r r k
T
f
a
G
a T T
t
L / N ) ( L / EA + + + =
In this equation r(cos sin 0 cos sin 0 ),
]( sin cos 0 sin cos 0 ),
(
=
z
z
B
L
1
1 0 0 0 0 0
0 0 0 1 0 0
,
(
=
(
=
4 1
1 4
30 4 2
2 4 L N
,
L
EI
f a
G
:
a
k k ,
'Computational Civil Engineering 2007, International Symposium 83
and N
f
is the axial Iorce in the beam in the fth incremental step: N
f
EA A L
f
/ L
f
,
, L L L
f f 0
= A
2 0 0
1 0
) ( L
i i
x x =
+
,
2
1
) ( L
f
i
f
i f
x x =
+
.
Figure 8. The axial deIormation oI the beam
Matrices and are the material and geometric stiIIness matrices respectively.
They are expressed with the reduced set oI displacements which produce
deIormations and they are not containing the rigid body displacements oI the beam.
This reduced Iorm oI the stiIIness matrices needs less computational eIIort and
speeds up signiIicantly the computation. Equation (19) introduces the nonlinear
eIIect oI the axial Iorce N
a
k
a
G
k
f
. The complete tangent stiIIness matrix in the updated
lagrangean Iormulation used here has two more terms corresponding to the
variation oI the length oI the beam in bending and to the eIIect oI the shear Iorce
1, 2, 3, 4, 5. Since in the current cases the structure is divided in a
suIIicient number oI beams, the errors are very small, when neglecting these two
terms. In a study using the complete tangent stiIIness matrix and equation (19) the
diIIerences between the resulting limit temperatures were only at the IiIth digit.
4. NUMERICAL EXPERIMENTS
The validity oI the present model is veriIied through a series oI comparative
analyses with other author`s results. The numerical example presented here
corresponds with one given in 1. The track length is L24.359/2 m corresponding
to 21 sleepers on the symmetric halI oI the structure. The curvature radius is R400
m. The horizontal misalignment is characterized by a halI wave cosine with a
length 9.144 m and an amplitude o0.0381 m. The rails have the characteristics
oI two AREA 136 rails. The vertical stiIIness oI the ballast elements is R
:
68900
kN/m per meter oI track. The longitudinal stiIIness is R
x
1378 kN/m per meter oI
track. The torsional stiIIness oI the Iasteners is 111.250 kNm/rad per meter oI
84 A. Dosa, V. Ungureanu
track. Laterally the ballast is modeled by the trilinear constitutive behavior given
in Iigure 2. The reIerence values oI the lateral peak resistance and residual
resistance are J
v
17.508 and J
r
9.630 kN per meter oI track. These values are
corrected with the vertical Iorces resulting Irom the vehicle loading. The lateral
displacement at the peak value is v
v
0.00635 m and at the limit value is v
r
0.0381
m. The model is vertically loaded by a vehicle with two bogies represented by Iour
vertical loads oI 293 kN each. The centre spacing between the bogies is 12.85 m.
The spacing between the axles in a bogie is 1.78 m. The centre oI the misalignment
is located in the middle between the bogies. The track is loaded by a temperature
increase Irom zero to a maximal value corresponding to the buckling oI the rail.
The lateral displacement oI the middle node versus the resulting temperature
increase is shown in Iigure 9.
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
0
5
10
15
20
25
30
35
40
45
50
Lateral displacement (m)
T
e
m
p
e
r
a
t
u
r
e
i
n
c
r
e
a
s
e
(
d
e
g
.
C
e
l
s
i
u
s
)
Tmax
Tmin
Figure 9. Lateral displacement versus temperature increase
The curve in the Iigure 9 is characterized by two points: T
max
 the maximum
increase oI temperature Ior which the buckling certainly starts, and T
min
 the
minimum increase oI temperature which occurs in the postbuckling domain. The
values computed by the SCFJ model  T
max
49.5
0
C and T
min
33.3
0
C  are in a
good agreement with those given in 1.
ReIerences
1. Bnu, V. Calculul neliniar al structurilor, Editura tehnic, Bucuresti, 1981.
2. CrisIield, M.A. Nonlinear Finite Element Analysis of Solias ana Structures, Wiley, 1991.
3. Dosa,A., Litr E. Elemente de bar ncovoiat cu precizie mbuntit pentru calculul neliniar si
de stabilitate , Revista Construciilor, 2006.
4. Dosa,A., Popa L. High order beam elements Ior the stability and nonlinear analysis oI Irame
structures, 'Computational Civil Engineering 2006, Iasi, Romnia.
5. Felippa C.A., Nonlinear Finite Element Methoas,
www.colorado.edu/engineering/CAS/courses.d/NFEM.d.
6. Van, M.A., Stability oI Continuous Welded Rail Track, DelIt University Press, 1997.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romania, May 25, 2007
SoilStructure Interaction EIIects oI Damped Structures
Petre Ene
Technical University of Construction, Bucharest, Romania
Summary
In the past, much effort has been aevotea to investigating the seismic behavior of
structures with seismic aampers by neglecting the effects of founaationstructure
interaction. In oraer to ensure the safety of structures, it is necessary to apprehena
the interactive behavior of aampea structures ana unbounaea founaations unaer
intense earthquake grouna motions. This paper presents a rigorous timeaomain
proceaure to aaaress the interaction effects of structures equippea with fluia
viscous aampers ana founaations with an unbounaea meaium. Quantitative results
show that, auring earthquakes, there are significant aifferences between a system
with ana without raaiation aamping. For greater accuracy, raaiation aamping
shoula be properly taken into account. Moreover, the efficiency of fluia viscous
aampers in reaucing seismic aisturbance of a structure is very aepenaent on the
flexibility of the founaation.
1. INTRODUCTION
Seismic dampers have been recently used to enhance seismic perIormance Ior
some tall buildings in CaliIornia (Pong, 1999). For example, an eight story
concrete building located on the campus oI the CaliIornia State University at Los
Angeles was retroIitted by adding viscous dampers (Elhassan et al., 1996). An
eleven story steel building, the new national headquarters Ior the Money Store
located in Sacramento, CaliIornia, was one oI the Iirst buildings in the United
States that use seismic dampers to limit story driIt during earthquakes (Miyamoto
and Scholl, 1998). However, many structural analyses and designs ignore the
behavior oI soilstructure interaction due to the complexity and uncertainty oI
modeling techniques. In the last twenty years, some advances have been made in
this area. Most notable is the development oI procedures based on boundary
integral or boundaryelement methods Ior unbounded media. Nonetheless, a major
need still exists Ior alternative approaches, particularly Ior procedures that could be
implemented within the context oI Iiniteelement analyses. These procedures
would be more Iamiliar to structural analyses, and would allow greater Ilexibility
in terms oI the geometrical and material characterization oI the unbounded
medium.
In this paper, a new timedomain procedure, based on the Iinite element method,
the semianalytical solution and the inIinitesimal Iiniteelement cell method, is
86 P. Ene
presented to address the interactive behavior oI structures equipped with Iluid
viscous dampers and Ioundation with an unbounded region (Fig.1a). To yield more
accurate results, a time domain procedure has been developed to simulate the
radiation Ieatures oI an inIinite domain.
Fig. 1a TenStory Builaing Equippea with Fluia Jiscous Dampers ana Founaation
Fig. 1b Finite Element Mesh for Builaing Equippea with Fluia Jiscous Damper
ana Founaation
2. TIMEDOMAIN SUBSTRUCTURE METHOD
The substructure method is adopted Ior the study. The entire system is divided into
three substructures including the building, Ioundation and Iluid dampers.
Furthermore, the soil medium is also divided into two regions: the near and the Iar
Iield oI the soil. The Iar Iield oI the soil extends to inIinity.
Equations oI Motion oI Building and Near Field oI Foundation
As shown in Figs. 1a and 1b, the structure, Iluid damper and the near Iield oI the
soil can be descretized by the Iinite element method. The equation oI motion oI the
structure and the near Iield oI the Ioundation is given as:
( ) ( ) ( ) ( ) ( )
g
M U t C U t K U t M B u t Q t + + = +
(1)
'Computational Civil Engineering 2007, International Symposium 87
M, C, K symmetrical mass, damping, and stiIIness matrices, respectively, Ior the
building and the near Iield oI the Ioundation;
( ) ( ) ( ) , , U t U t U t
vectors oI nodal displacements, velocities, and accelerations
relative to the ground, respectively;
ground motion;
B displacement transIormation matrix;
And Q (t) vector oI dynamic Iorces resulting Irom the Iar Iield oI the Ioundation.
Analytical Model Ior Fluid Viscous Damper
The Iluid damper exhibits viscoelastic Iluid behavior over a large Irequency. The
simplest model to simulate the mechanical behavior oI the Iluid viscous damper is
the Maxwell model (Bird et al., 1987) given as:
( ) ( ) ( )
0
P t P t C U t + =
(2)
Where 't is the relaxation time, is the damping constant at zero Irequency, P (t)
is the Iorce acting on the Iluid damper, is the damper position velocity.
Another Iinite element Iormulation model Ior Iluid dampers was adopted in this
study (Pong et al., 1994). Using the virtual work principle, the equivalent modal
Iorces F (t) can be obtained as:
(3)
Where the matrix is the added damping result Irom Iluid dampers:
(4)
System Equations for Far Fiela of Founaation
As shown in Fig. 1a, the Ioundation medium is divided into two regions: the near
and Iar Iields oI the Ioundation. The Iar Iield oI the Ioundation extends to inIinity.
Adopting the similarity theorem, as shown in Fig. 2, Song and WolI developed the
consistent inIinitesimal Iiniteelement cell method accounting Ior the radiation
damping Ior the unbounded medium in the time domain (Song and WolI, 1995;
1996).
88 P. Ene
Fig.2 Funaamental Concept of Consistent infinitesimal FiniteElement Cell
Methoa
The governing equation oI the Iar Iield oI Ioundation can be expressed as:
(5)
Where the interaction is Iorce at the interIace oI the near and Iar Iields oI the
unbounded medium and is the acceleration unitimpulse response matrix in
the time domain. The relation oI the vector oI dynamic Iorces Irom the Iar Iield
to those acting on the near Iield is given as:
(6)
As shown in Fig.3, owing to similarity, the coordinates oI the nodes on the exterior
boundary can be expressed by those oI the nodes on the interior boundary and the
dimensionless cell width
(7)
(8)
The nodes on the interior boundary are identiIied by subscripts 1, 2, 3 and the
nodes on the exterior boundary are identiIied by subscripts 4, 5, 6, respectively.
The cell width can be expressed as and
(9)
(10)
'Computational Civil Engineering 2007, International Symposium 89
Fig.3 Twoaimensional Finiteelement Cell ana Parent Element on Descreti:ea
Interface
Note that the subscript i and e portray the interior boundary and the exterior
boundary, respectively. denote the coordinates oI the nodes on the
exterior boundary. indicate the coordinates oI the nodes on the interior
boundary. The shape Iunction matrix oI the Iiniteelement is:
(11)
, (12)
Where , N the shape Iunction vector in the direction.
ThereIore, the stiIIness matrix oI the Iinite element equals:
, (13)
The Jacobean matrix oI a particular element equals:
(14)
Where (15)
The determinant oI the Jacobean matrix is given as:
(16)
where is aenotea as (17)
The strainnodal displacement matrix is
90 P. Ene
(18)
, (19)
Ana (20)
, (21)
the shape functions at noaal point k. The final form of the consistent
infinitesimal finiteelement cell equation in the time aomain is given as.
(22)
Where (23)
(24)
(25)
(26)
'Computational Civil Engineering 2007, International Symposium 91
Fig.4a Comparison oI Relative Displacements between Points A and B while
Elastic Modulus oI Near Field is and Elastic Modulus oI Near Field is
(27)
(28)
mass aensity. After aetermining from eq. (22), the acceleration unit
impulse response matrix may be written as.
(29)
The entire history oI time is divided into n equal intervals . It is assumed that
matrix and the acceleration is piecewise constant over each time step:
, (30)
With the aid oI eq. (30), oI eq. (5) can be expressed as:
(31)
Where (32)
(33)
TimeDomain Finite Element Method Ior Structure Equipped with Fluid Viscous
Dampers and Unbounded Media
Substituting eq. (6) and eq. (31) into eq. (1) results in
(34)
92 P. Ene
Fig.4b Comparison oI Shear Forces at Point B while Elastic Modulus oI Far Field
is and Elastic Modulus oI Near Field is
is an addedload matrix representing the previous time eIIect. Applying the new
mark method to eq.(34), at the N
th
time step, we have
(35)
Where (36)
(37)
Where
.
Application
A ten story steel building Irame was used Ior this application. The building span
width is 9,20m. The typical story height is 3,70m, except the Iirst story height is
4,60m. Point A is designated at the rooI, while Point B is designated at the base.
The material used Ior the building has Poissons ratio and elastic modulus equal to
0,3 and 200 GPa, respectively. The weight Ior each Iloor is 4460,60N/m. The
radius oI the near Iield oI the soil below the building is around the 249m. The
Poisson`s ratio oI the stratum is 0,30 and the weight density is 1,80KN/m
3
.
It was assumed that the Iloor were in their own plane. The damping constant C
0
oI
Iluid viscous damper is 525,59KNs/m and the whole system was subjected to 1940
El Centro earthquake ground motion. Fig. 4a shows the shear Iorce responses at
point B when the building equipped with Iluid viscous dampers during
earthquakes. In these Iigures, the boldsolid line is the seismic response oI the
building on rigid Ioundation (without soil) and the dotted line is the response oI the
'Computational Civil Engineering 2007, International Symposium 93
building on the Ilexible Ioundation with radiation damping during earthquakes.
With reIerence to Fig. 4a and 4b, they indicate that the seismic responses oI the
building have remarkable diIIerences on diIIerent Ioundations.
Fig. 5a Comparison of Max. Relative Displacements between Points A ana B while
Elastic Moaulus of Far Fiela is ana Different Moaulus of Near Fiela
are Appliea
Fig. 5a Comparison of Max. Shear Forces at Point B while Elastic Moaulus of Far
Fiela is ana Different Moaulus of Near Fiela are Appliea
When the elastic modules oI the near Iield oI Ioundation are diIIerent, the
comparison oI the maximum relative displacement responses at points A and B oI
the building equipped with Iluid viscous dampers on the unbounded soil are shown
in Fig. 5a. In addition, Fig. 5b shows the comparisons oI the maximum shear Iorce
responses at point B when the building equipped with Iluid viscous dampers on the
unbounded soil. These two Iigures show that signiIicant diIIerences exist between
the responses oI the building on diIIerent Ioundations during earthquakes. They
also show that the whole system becomes more Ilexible due to the interaction
eIIects oI the Ioundation and the eIIectiveness oI the Iluid viscous damper in
reducing seismic responses oI the building is considerably downgraded.
94 P. Ene
Conclusions
Seismic dampers have been recognized as an eIIective tool in improving seismic
resistance Ior both new and existing buildings. Recently, they have been used
extensively to achieve enhanced structural perIormance Ior the seismic
rehabilitation oI tall buildings in CaliIornia. Nevertheless, soilstructure interaction
is a very complicated behavior and it requires timeconsuming eIIort in structural
analyses and designs. To ensure the saIety oI a structure, it necessitates considering
the interaction behavior oI the damped structure under intense ground motions. It is
prudent to consider the eIIects oI radiation damping and the Ilexibility oI the soil
media when designing tall buildings with seismic dampers since taller buildings
tend to be much more Ilexible than shorter buildings.
This study demonstrates diIIerences between a damped structure with and without
consideration oI the soil media. The structural behavior oI a building can be quite
diIIerent iI the soilstructure interaction is ignored. Quantitative results show that
remarkable diIIerences exist in the seismic behavior oI the system with and without
a Ilexible Ioundation. The role oI radiation damping becomes extremely important
during an earthquake when supplemental damping is added to the structure.
ThereIore, the design oI a tall structure with supplemental damping in an
earthquake one area should consider the eIIects oI soilstructure interaction to
ensure optimal seismic perIormance.
ReIerences
1. Elhassan, R.M., Arminak, A., 1996 Design and Analysis oI a Seismic RetroIit oI a Tall
Concrete Building using Supplemental Viscous Dampers;
2. Miyamoto, H.K., Scholl, R., 1998 Modern Steel Construction;
3. Pong, W.S., 1999 PerIormance  Based Design Procedures Ior Buildings with supplemental
Dampers;
4. Pong, W.S., Tsai, C.S., 1994 Seismic Study oI Building Frames with added Energyabsorbing
Devices;
5. Song, C., WolI, J.P., 1995 Consistent InIinitesimal FiniteElement Cell Method: Out oI Plane
Motion;
6. Song, C., WolI, J.P., 1996 Consistent InIinitesimal FiniteElement Cell Method: Three
Dimensional Vector Wave Equation.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Numerical Modeling oI Fiber ReinIorced Concrete (FRC).
Fiber orientation angle 0
0
, Iiber length l
I
50mm
Liviu Gherman
1
, Petru Mihai
2
, Nicolae Florea
3
, Constantin Gavriloaia
4
, and
Ioan Paul Vod
5
1
Faculty of Civil Engineering, BMTO, Tehnical University 'Gh. Asachi`, Iasi, Romania
2
Faculty of Civil Engineering, BMTO,Tehnical University 'Gh. Asachi`, Iasi, Romania
3
Faculty of Civil Engineering, BMTO, Tehnical University 'Gh. Asachi`, Iasi, Romania
4
Faculty of Civil Engineering, BMTO, Tehnical University 'Gh. Asachi`, Iasi, Romania
5
Faculty of Civil Engineering, BMTO, Tehnical University 'Gh. Asachi`, Iasi, Romania
Summary
The paper presents a new approach on FRC stuay, which has similar
characteristics to a multiphase composite material. Numerical moaeling is
appliea, using a finite element methoa (FEM) program, to stuay the behavior of
linear structural elements, maae from steel fiber reinforcea concrete (SFRC).
The results stress out the favorable influence of the fibers on the bearing capacity
ana especially on the auctility of the loaaea members. Lack of aesign stanaaras
ana practice coaes on this material is bringing this fiela of stuay unaer
researchers consiaeration.
KEYWORDS: numerical modeling, steel Iiber reinIorced concrete, inIluence oI the
steel Iibers.
96 L.Gherman, P.Mihai, N.Florea, Ctin.Gavriloaia, I.P.Vod
1. INTRODUCTION
Enhancement oI SFRC properties is more evident Ior hardened concrete than Ior
Iresh concrete. It is inIluenced by the Iiber characteristics: length and geometry,
reinIorcement percentage, orientation and adherence to the concrete matrix.
For this new type oI material, the workability is decreasing, the concrete having
low to very low plasticity. II a high reinIorced percentage is used, or Iibers having
length higher than the critical one, the workability decreases dramatically, the
Iibers having the tendency to bundle. This shortcoming can be overcome with
supplementary execution costs.
Studying the inIluence oI the Iibers on the concrete behavior, as well as the wide
possibilities Ior using this material makes it necessary that it is studied using
numerical modeling.
2. NUMERICAL MODELING USING FEM
2.1. Used Finite Element Types
Any loaded body with a given shape must be analyzed, in order to establish the
proper Iinite elements to be used Ior meshing. Usually, the choice oI the Iinite
element is determined by the geometry oI the body and by the number oI
independent spatial coordinates necessary to describe the system. Some types oI
Iinite elements, onedimensional or bidimensional, are presented in Figures 1, 2.
When the geometry, properties and other parameters oI the material (stresses,
displacements, pressure, temperature) can be described using only one spatial
coordinate, the onedimensional Iinite element can be used (Figure 1). Even though
this element has a transversal crosssection, it can be idealized as a line.
In some cases, the crosssectional area oI this element may vary along its length.
node
2
node
1
node
2
node
1
Figure 1 Onedimensional Iinite element
When the geometry and details oI the studied problem can be expressed Iunction oI
two independent spatial coordinates, bidimensional Iinite elements are used
(Figure 2), the basic type having triangular shape.
'Computational Civil Engineering 2007, International Symposium 97
3
3 3
4 5 1 4 2 2 2 1 1
Figura 2 Bidimensional triangular Iinite element
2.2. Finite Element Analysis Steps
Most oI the Iinite element analysis steps are perIormed successIully by the FEM
program, so their succession is highly simpliIied:
The preparatory stage, consisting in the precise meshing oI the element, using
a CAD program, aIter which it was exported to a FEM program.
DeIinition oI the physical and mechanical characteristics oI the materials, and
oI the plain stress analysis manner in the nonlinear domain.
Analysis and interpretation oI the results.
2.2.1. The meshing was perIormed in two variants.
The corresponding Iree body diagrams are also presented (Figure 3, 4).
a.) plain concrete beam, where triangular Iinite elements were used,
b.) SFRC beam, where triangular elements were usea for the concrete ana one
aimensional finite elements were usea for the steel fibers. The fiber orientation
angle was consiaerea 0
0
.
3.)
2.)
1.)
Figure 3 Meshing
3.1 Iiber orientation angle
0
0 o = , 3.2 plain concrete beam, 3.3 SFRC beam
98 L.Gherman, P.Mihai, N.Florea, Ctin.Gavriloaia, I.P.Vod
2.2.2. Definition of the physical ana mechanical characteristics of the materials
ana of the plain stress analysis manner in the nonlinear aomain
The studied model has the crosssectional area A 108cm
2
(12x9cm) and the
length 150cm, being simply supported. The distance between the supports is
120cm. It is acted by two concentrated Iorces, 40cm apart Irom each other and
Irom the supports, so that on the middle third the bending moment is constant
(Figure 4). The concrete class is C20/25, with Young modulus E 3000N/mm
2
.
The disperse Iibers (the Iiber orientation angle being considered 0
0
), have the
diameter 1,05mm, the crosssectional area A u ~1mm
2
, Young modulus E
210000N/mm
2
, tensile strength R
ti
200N/mm
2
m , and lengthl m 50
f
= .
Q
M
P P
Figure 4 Free body diagram oI the beam
2.2.3. Analysis ana interpretation of the results.
AIter applying the loading steps, there were obtained the loaddisplacements
curves presented in Figures 5.1 and 5.2, and the stress maps corresponding to the
loading steps (Figures 6 and 7).
1.) Plain concrete beam
P

K
N

mm
2.) SFRC beam
P

K
N

mm
Figure 5 Loaddisplacement curve (P) Ior:
5.1. the plain concrete beam, 5.2. the SFRC beam
'Computational Civil Engineering 2007, International Symposium 99
1,1
0,86
0,57
0,29
0
0,29
0,57
0,86
1,1
Max 1.365 at Node 3014
Min 1.502 at Node 627
X
Y
Z
Min 2.893 at Node 627
Max 1.760 at Node 3341
2,8
2,3
1,9
1,4
0,93
0,47
0
0,47
0,93
1,4
X
Y
Z
Min 4.865 at Node 3095
Max 1.740 at Node 1066
4,6
4
3,3
2,6
2
1,3
0,66
0
0,66
1,3
X
Y
Z
Min 6.511 at Node 3095
Max 1.742 at Node 3669
5,8
5
4,1
3,3
2,5
1,7
0,83
0
0,83
1,7
CONTOURSOFSX
STRESS
RESULTSFLE= 1
ncrement 10 Load Factor =4.05000
LOADCASE = 10
X
Y
Z
Min 6.999 at Node 3091
Max 1.737 at Node 3689
7
6,1
5,2
4,4
3,5
2,6
1,7
0,87
0
0,87
CONTOURSOFSX
STRESS
RESULTSFLE= 1
ncrement 11 Load Factor =4.12500
LOADCASE = 11
X
Y
Z
1,4
Figure 7 Stresses maps Ior the plain concrete beam.
100 L.Gherman, P.Mihai, N.Florea, Ctin.Gavriloaia, I.P.Vod
CONTOURSOFS
STRESS
Min 0.9148E04 a
Max 1.347 at Nod
0
0,0842
0,168
0,253
0,337
0,421
0,505
0,589
0,674
0,758
0,842
0,926
1,01
1,09
1,18
1,26
X
Y
Z
Min 0.7766E03 a
Max 4.609 at Nod
0
0,288
0,576
0,864
1,15
1,44
1,73
2,02
2,3
2,59
2,88
3,17
3,46
3,74
4,03
4,32
CONTOURSOFS
STRESS
X
Y
Z
Min 0.1030E02 a
Max 8.334 at Nod
0
0,521
1,04
1,56
2,08
2,6
3,12
3,65
4,17
4,69
5,21
5,73
6,25
6,77
7,29
7,81
CONTOURSOFS
STRESS
X
Y
Z
Min 0.1278E02 a
Max 12.28 at Nod
0
0,767
1,53
2,3
3,07
3,84
4,6
5,37
6,14
6,91
7,67
8,44
9,21
9,98
10,7
11,5
CONTOURSOFS
STRESS
X
Y
Z
Min 0.7169E03 a
Max 3.882 at Nod
0
0,243
0,485
0,728
0,97
1,21
1,46
1,7
1,94
2,18
2,43
2,67
2,91
3,15
3,4
3,64
CONTOURSOFS
STRESS
X
Y
Z
Figure 8 Stresses maps Ior the SFRC beam
'Computational Civil Engineering 2007, International Symposium 101
There are presented the load values Ior each loading steps and the corresponding
displacements, Ior the plain concrete beam (Table 1) and Ior the SFRC beam,
respectively (Table 2). The displacement decrease, Ior each loading step, Ior the
SFRC beam compared to the plain concrete one, is presented in Table 1, column 3.
Notes and comments on the behavior oI the SFRC beam under loading are
presented in Table 3.
Table 1
Loading step Load
magnitude
KN
Displacement
mm
Displacements decrease Ior the
SFRC beam, compared to the
plain concrete beam mm
Comments
0 1 2 3 4
1 1,5 0,119 0,002
2 1,8 0,143 0,003
3 2,1 0,169 0,004
4 2,4 0,2 0,005
5 2,7 0,241 0,008
6 3 0,296 0,014
7 3,3 0,369 0,024
8 3,6 0,483 0,062
9 3,9 0,639 0,105
10 4,05 0,729 0,056
11 4,125 0,825 0,074 Beam Iailure
Table 2
Loading step Load magnitude KN Displacement
mm
Comments
0 1 2 3
1 1,5 0,117
2 1,8 0,14
3 2,1 0,165
4 2,4 0,195
5 2,7 0,233
6 3 0,282
7 3,3 0,345
8 3,6 0,42
9 3,9 0,534
10 4,2 0,673
11 4,35 0,751 The Iirst Iissure occurs
12 4,5 0,86 The Iissure opens
13 4,8 1,231 Inbetween steps 12 and 13,
the Iibers take over the tensile
stresses
14 5,1 1,371
15 5,175 1,48
16 5,3 1,6
38 6,598 4,127
Inbetween steps 14 and 38,
the concrete and the Iibers
work together to take over the
tensile stresses
102 L.Gherman, P.Mihai, N.Florea, Ctin.Gavriloaia, I.P.Vod
Table 3
Bearing capacity increase Ior
the SFRC beam, compared to
the plain concrete beam
Displacements oI the SFRC
beam, compared to the plain
concrete beam
Bearing capacity increase until
the Iirst Iissure occurs
5,45 The displacement oI the SFRC
beam is 9.1 lower
(0,751 compared to 0,825)
Bearing capacity increase until
concrete Iailure
16,36
4,125)
The displacement oI the SFRC
beam is 67,02 higher
(1,231 compared to 0,825)
Bearing capacity increase until
Iibers yielding
23,64
4,125)
The displacement oI the SFRC
beam is 166,18 higher
(1,371 compared to 0,825)
Bearing capacity increase until
Iibers Iailure
62,25
,125)
The displacement oI the SFRC
beam is 500,24 higher
(4,127 compared to 0,825)
(4,35 compared to 4,125)
(4,8 compared to
(5,1 compared to
(6,598 compared to 4
3. CONCLUSIONS
Compared to the plain concrete, the SFRC behavior, on elastic and elastoplastic
domains, is characterized by:
1. elastic domain an increase oI the bearing capacity:
 until the Iirst Iissure, oI 5.45, and the displacement is 9.1 lower;
 until the concrete Iailure, oI 16.36, and the deIormation is 67.02 higher,
leading to the increase oI the area under the Iorcedisplacement curve.
2. plastic domain  an increase oI the bearing capacity:
 until the Iibers yielding, oI 23.64, and the displacement is 166.18 higher;
 until the Iibers strain hardening and Iailure, oI 23.64, and the deIormation is
500.24 higher.
ReIerences
1. Chiaia, B., Fantilli, A.P., Valini, P., Kalamaras, G. Fiber reinIorced concrete Ior massive
structures the case oI craviale tunnel, FIB, Italy, Napoli, 2006.
2. Falkner, H. and Henke, V. Steel Iibre reinIorced concrete, Irom research to standards. Concrete
Structures Annual Technical Journal oI the Hungarian Group oI Iib, 2005; 6: 3946. Proceedings oI
the 2nd Congress Session 2 June 58, 2006 Naples, Italy Design and construction
3. Rilem TC 162TDF, o c design method Iinal recommendation. Materials and structures, 36,
560567, 2003,
4. Barla, G., Barpi, F., Bertolino, C. and Chiaia, B. A note on the design oI IibrereinIorced shotcrete
linings Ior underground support. Computational Modeling oI Concrete Structures EUROC 2003,
Eds. Bicanic, N., de Borst, R., Mang, H. and Meschke, G., A.A. Balkema Publishers, Lisse, 2003:
627634.
5. Balaguru, P.N. and Shah, S.P. Fiber  ReinIorced Cement Composites. McGrawHill Inc, New
York, 1992.
6. Cuteanu, E., Marinov, R.,  Metoda elementelor Iinite n proiectarea structurilor, Ed. Facla,
Timisoara, 1980.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
A new look into Iinite element templates.
Wojciech Gilewski
1
1
Institute of Structural Mechanics, Warsaw University of Technology, Warsaw, Polana
Summary
The present paper is aeaicatea to evaluation ana new methoa of construction the
finite element templates. A template is an algebraic form of element matrices,
which contains free parameters. Setting the parameters to specific values proauces
element instances. Two templates are analy:ea. Bernoulli ana Timoshenko beam.
The number of free parameters is aiscussea by a general methoa.
KEYWORDS: Iinite element template, consistency, ellipticity
1. INTRODUCTION
Beams, plates and shells are widely considered in engineering applications.
However the corresponding discretization procedures are not yet suIIiciently
reliable. It is diIIicult to obtain an element that is optimal. In a Iormulation we
should aim to satisIy: ellipticity, consistency and the inIsup condition.
Ellipticity ensures that the Iinite element model is solvable and physically means
there are no spurious zero energy modes. This condition can easily be veriIied by
studying the zero eigenvalues and corresponding eigenvectors oI the stiIIness
matrix oI a single unsupported Iinite element. Consistency is related to the
convergence. The Iinite element solution must converge to the solution oI a
mathematical problem the element size h is close to zero. The bilinear Iorms used
in the Iinite element discretization must approach the exact bilinear Iorms oI the
mathematical model as h approaches zero. The infsup condition ensures optimal
convergence in bendingdominated problems and is not a subject oI this paper.
One oI the interesting concepts are Iinite element templates proposed by Felippa
1,2. The objective oI this paper is to evaluate Iinite element templates, proposed
by Felippa, with the use oI the energydiIIerence criterion (to check iI the template
is consistent) and the spectral analysis (to check iI the template is elliptical). Two
beam templates are discussed: Ior Bernoulli theory and Ior Timoshenko theory. A
new, general method Ior development oI Iinite element templates is proposed.
104 W. Gilewski
2. TEMPLATES
A Iinite element template is an algebraic Iorm Ior element matrices, which contains
Iree parameters (Felippa 1,2). Setting those parameters to speciIic values
produces element instances. The set oI Iree parameters is called the template
signature. Borrowing the terminology Irom biogenetics, the signature may be
viewed as an 'element DNA that uniquely characterizes it as an individual entity.
Elements derived by diIIerent techniques thet share the same signature are called
clones. The template should IulIill the Iollowing conditions:
consistency (the individual element test is passed),
stability (correct rank and nonnegativity conditions),
parametrization (Iree parameters)
invariance (the element is observer invariant).
The element stiIIness matrix derived through the template approach is based on the
Iundamental decomposition
) ( ) (
f h i b
 o K K K + = ,
where and K are the basis and higherorder stiIIness matrices respectively.
b
K
h
f i
 o , are Iree parameters. These two matrices play diIIerent and complementary
roles. The basic stiIIness takes care oI consistency and elementtypemixing.
The higher order stiIIness is a stabilization term that provides the correct rank
and may be adjusted Ior accuracy.
b
K
h
K
3. BERNOULLI BEAM TEMPLATE
Let us consider a Iinite element template Ior Bernoulli beam 1 oI twonoded
element oI the length with natural d.o.I. L { }
k k i i e
w w   , , , = q .
(
(
(
(
+
(
(
(
(
= + =
2 2
2 2
3
2 2
2 4 2 4
2 2
2 4 2 4
1 0 1 0
0 0 0 0
1 0 1 0
0 0 0 0
L L L L
L L
L L L L
L L
L
EJ
L
EJ
Template
Bernoulli

h b
K K K
3 = For  the well known beam element with Herimite`s polynomial shape
Iunctions. The template depends on one Iree parameter  . The energydiIIerence
procedure (described in details in 3,5, and used Ior evaluation oI beam and plate
Iinite elements in 4,5) can be used to check iI the template satisIies the
consistency condition.
'Computational Civil Engineering 2007, International Symposium 105
The strain energy density oI a beam is expected to be:
2
2
2
~
2


.

\

=
ax
w a
EJ E
s .
The equivalent density oI element strain energy is a quadratic Iorm
e
Template
Bernoulli
T
e
Template ES
s
L
E q K q
1 ~
2 =
.
The nodal displacements can be expressed by average displacements and its
derivatives in the midpoint oI the element with the use oI Taylor series expansion
...,
2
) (
) ( 6
1
2
) (
) ( 2
1
2
) ( ) (
3
3
3
2
2
2
+ 
.

\

A
A

.

\

A
A
+
A
A
=
L
x
x
w L
x
x
w L
x
x
w
x w w
i
...,
2
) (
) ( 6
1
2
) (
) ( 2
1
2
) ( ) (
3
3
3
2
2
2
+ 
.

\

A
A
+ 
.

\

A
A
+
A
A
+ =
L
x
x
w L
x
x
w L
x
x
w
x w w
k
...,
2
) (
) ( 6
1
2
) (
) ( 2
1
2
) (
) (
) (
4
4
4
2
3
3
2
2
+ 
.

\

A
A

.

\

A
A
+
A
A
A
A
=
L
x
x
w L
x
x
w L
x
x
w
x
x
w
i

...
2
) (
) ( 6
1
2
) (
) ( 2
1
2
) (
) (
) (
4
4
4
2
3
3
2
2
+ 
.

\

A
A

.

\

A
A
+
A
A
A
A
=
L
x
x
w L
x
x
w L
x
x
w
x
x
w
k

AIter collecting the expressions with respect to we have L
). ( 0
12
1
) ( 36
~
2
4
4
4
2
2
2
3
3
2
2
2
2
L
x
w
x
w
x
w
EJ L
x
w
EJ E
Template ES
s
+
(
(
A
A
A
A
+


.

\

A
A
+


.

\

A
A
=

In the limit case the Iollowing relation is valid 0 L
s
Template ES
s
L
E E
~ ~
lim
0
=
.
This is a prooI that the element template satisIies the consistency requirement Ior
any  . The basis stiIIness matrix is responsible Ior the Iirst term oI the strain
energy. Formally the
b
K
 parameter is Iree Irom zero to inIinity.
106 W. Gilewski
The element template should also elliptical. To check this condition it is necessary
to calculate the eigenvalues and eigenvectors oI the template. Two zero
eigenvalues related to rigid body motions are expected. The two other eigenvalues
are to be positive and related to the deIormed element. The results are the
Iollowing
L
EJ
L
EJ
 10 , 2 , 0 , 0
4 3 2 1
= = = =
,
{ }
{ }
{ }
{ }. 1 , 2 , 1 , 2
, 1 , 0 , 1 , 0
, 0 , 1 , 0 , 1
, 1 , 0 , 1 ,
1
3
2
1
L L
L
=
=
=
=
w
w
w
w
It is seen that the element is elliptical iI 0 >  . II 0  the Iourth eigenvalue
tends to be zero with the eigenvector that describes deIormed element.
Since the development oI the element template is rather complicated, a couple oI
questions arises aIter reading the texts oI Felippa 1,2:
How many Iree parameters exist in the template ?
Are the Iree parameters exist in both matrices (basis and higher order) ? etc.
Let us examine some more general matrices to answer the questions.
(
(
(
(
+
(
(
(
(
=
144
2
34 24
2
14
34 33 23 13
24
2
23 22
2
12
14 13 12 11
3
44 24
24 22
2 2
2 4 2 4
2 2
2 4 2 4
0 0
0 0 0 0
0 0
0 0 0 0
b L Lb b L Lb
Lb b Lb b
b L Lb b L Lb
Lb b Lb b
L
EJ
a a
a a
L
EJ
Extenaea
Bernoulli
K
The basis matrix is responsible Ior the Iirst term oI the strain energy, so it
depends only on rotations. There are 3 Iormally independent parameters in
this matrix. The second matrix Iormally depends on 10 parameters.
Following the procedure described above one can receive the Iollowing
strain energy density oI the element:
'Computational Civil Engineering 2007, International Symposium 107
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( ) L
x
w
b b b b b b b b b b
EJ
x
w
w b b b b
EJ
x
w
x
w
b b b b b b b b b b
EJ
x
w
x
w
a a EJ
L
x
w
w b b b b b b EJ
L
x
w
x
w
b b b b b b b b EJ
L
x
w
a a a EJ
L
x
w
w b b b b b b b EJ
L
x
w
b b b b b b b b b b EJ
L
x
w
w b b b b b b EJ
L
w b b b EJ
L
E
Extenaea ES
s
0
4 4 8 4 4 4 2 4
16
12
3 4 6 4 3 4 2 4
12
1
3 3 3 3
2
1
2 3 2 3
2
1
2
1
2 2 2 2 2
1
2 2 2 2 2 2
1
4
2
4 ~
2
2
2
2
44 34 33 24 23 22 14 13 12 11
4
4
34 14 23 12
3
3
44 34 33 24 23 22 14 13 12 11
2
2
44 22
3
3
34 33 23 14 12 11
2
2
44 34 33 23 22 14 12 11
2
44 24 22
2
2
2
34 33 23 14 13 12 11
2
2
44 34 33 24 23 22 14 13 12 11
2
34 33 23 14 12 11
3
2
33 13 11
4
+
+


.

\

A
A
+ + + + + +
+
A
A
+ + +
+
A
A
A
A
+ + + + + +
+
A
A
A
A
+ +
+
A
A
+ + + +
+
A
A
A
A
+ + + + +
+ 
.

\

A
A
+ +
+
A
A
+ + + +
+ 
.

\

A
A
+ + + + + +
+
A
A
+ + + +
+ + =
To IulIill the consistency condition the Iollowing equations are to be satisIied:
.
(
(
(
=
(
(
(
(
(
(
4
0
0
1 2 1
1 0 1
1 2 1
44
24
22
a
a
a
108 W. Gilewski
44
34
24
14
33
23
13
22
12
11
4
0
3
0
2
0
1
0
0
4 8 4 1 4 2 4 4 1
1 0 1 0 1 0 0 1 0
4 6 4 1 4 2 3 4 1
3 0 3 1 3 0 0 3 1
3 0 1 1 1 0 2 3 1
2 0 2 1 2 2 0 2 1
2 0 2 1 2 2 1 2 1
1 2 1 1 1 0 0 1 1
0 0 0 1 0 2 0 0 1
b
b
b
b
b
b
b
b
b
b
(
(
(
(
(
(
(
(
(
(
(
(
=
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
There are 3 equations Ior 3 parameters 'a. We have
1
44 24 22
= = = a a a
.
It is seen that the basis matrix oI Bernoulli beam template is independent on the
Iree parameters.
There are 9 equations Ior 10 parameters 'b. Thus
. ,
44 34 24 14 33 23 13 22 12 11 44
  = = = = = = = = = = = b b b b b b b b b b b
3. TIMOSHENKO BEAM TEMPLATE
Let us consider a twonoded element oI the length with natural d.o.I. L
{
k k i i e
w w }   , , , = q . Timoshenko beam template proposed by Felippa 2 is more
complex than Ior Bernoulli beam and depends on 3 parameters:
(
(
(
(
+
(
(
(
(
=
= + =
2 2
2 2
3
2 2
2 4 2 4
2 2
2 4 2 4
3
1 0 1 0
0 0 0 0
1 0 1 0
0 0 0 0
L L L L
L L
L L L L
L L
L
EJ
L
EJ
Template
Bernoulli

o
h b
K K K
.
'Computational Civil Engineering 2007, International Symposium 109
Following a similar procedure like Ior the Bernoulli beam we have:
Density oI strain energy
2 2
~
2 
.

\

+ 
.

\

=
ax
aw
H
ax
a
EJ E
s


Nodal parameters
...,
2
) (
) ( 6
1
2
) (
) ( 2
1
2
) ( ) (
3
3
3
2
2
2
+ 
.

\

A
A

.

\

A
A
+
A
A
=
L
x
x
w L
x
x
w L
x
x
w
x w w
i
...,
2
) (
) ( 6
1
2
) (
) ( 2
1
2
) ( ) (
3
3
3
2
2
2
+ 
.

\

A
A
+ 
.

\

A
A
+
A
A
+ =
L
x
x
w L
x
x
w L
x
x
w
x w w
k
...,
2
) (
) ( 6
1
2
) (
) ( 2
1
2
) ( ) (
4
3
3
2
2
2
+ 
.

\

A
A

.

\

A
A
+
A
A
=
L
x
x
L
x
x
L
x
x
x
i
  
 
...
2
) (
) ( 6
1
2
) (
) ( 2
1
2
) ( ) (
4
3
3
2
2
2
+ 
.

\

A
A
+ 
.

\

A
A
+
A
A
+ =
L
x
x
L
x
x
L
x
x
x
k
  
 
.
Density oI the template strain energy
) ( 0 3 3
12 ~
2
2
3
3
2
2
2
2
2
L
x
w
x
w
EJ
x
w
x
EJ
x
w
L
EJ
x
EJ E
Template
s
+ 
.

\

A
A
A
A

.

\

A
A
A
A
+
+ 
.

\

A
A
+ 
.

\

A
A
=
  



 
o
To IulIill the condition oI consistency it is necessary to take 1 = o and  in the
Iollowing Iorm
( ) ... 1
12
2
+ =
EJ
HL

with any .
110 W. Gilewski
Let us propose more general Iorm oI the stiIIness decomposition:
(
(
(
(
+
(
(
(
(
=
144
2
34 24
2
14
34 33 23 13
24
2
23 22
2
12
14 13 12 11
44 24
24 22
2 2
2 4 2 4
2 2
2 4 2 4
0 0
0 0 0 0
0 0
0 0 0 0
b L Lb b L Lb
Lb b Lb b
b L Lb b L Lb
Lb b Lb b
L
H
a a
a a
L
EJ
Extenaea
Bernoulli
K
Density oI the extended template strain energy
( ) ( )
( ) ( )
( ) ( )
( ) ( )
( ) ( ) ) ( 0 3 6
4
3 3 3 3
2
3 3 3 3 3 6 3
4
3 3 3 3 3
3 6 2
4
2
~
2
33 13 11 34 14 23 12
34 14 23 12
2
44 24 22
33 11 34 14 23 12
2
2
33 13 11
2
44 24 22
44 22
2
2
44 24 22
L b b b
H
x
w
b b b b
H
x
w b b b b H b b b
H
x
w
L
w
b b H
L
w
b b b b H
L
w
b b b H
x
a a a
EJ
x L
a a EJ
L
a a a EJ E
Extenaea
s
+ + + +
A
A
+
+
A
A
+ + + + + +
+
A
A
+ + + +
+ + + 
.

\

A
A
+ + +
+
A
A
+ + + =





  
To IulIill the consistency requirement 3 equations Ior 3 coeIIicients 'a and 7
equations Ior 10 coeIIicients 'b are to be satisIied:
(
(
(
=
(
(
(
(
(
(
4
0
0
1 2 1
1 0 1
1 2 1
44
24
22
a
a
a
This matrix equation is equivalent to the conditions received Ior the Bernoulli
beam template in the previous chapter.
'Computational Civil Engineering 2007, International Symposium 111
(
(
(
(
(
(
(
(
(
+
(
(
(
(
(
(
(
(
(
(
(
(
=
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
4
2
0
4
0
0
0
0 0 0
0 3 0
0 3 0
3 0 6
0 0 0
0 3 0
0 0 0
0 3 0 6 0 0 1
3 0 3 0 0 3 0
3 0 3 0 0 3 0
0 0 0 0 3 0 0
0 3 0 0 0 0 1
3 0 3 0 0 3 0
0 3 0 6 0 0 1
44
34
24
34
33
23
13
22
12
11
b
b
b
b
b
b
b
b
b
b
One can receive
1
44 24 22
= = = a a a
.
There are 3 independent constants in the group 'b. II we put
3 44 2 34 1 24
, , C b C b C b = = =
the other coeIIicients are the Iollowing:
( ) ( )
( ) . ,
3
1
, 1
3
1
,
3
1
, 3 6 4
3
1
, 1
3
1
, 1
2 34 33 2 23 13
3 1 22 2 12 11
C b b C b b
C C b C b b
= = + = =
= + = =
The derivation above is a prooI that there are 3 independent coeIIicients in the
Timoshenko beam template.
4. CONCLUSIONS
The Iinite element templates, proposed by Felippa 1,2, are algebraic Iorms that
contains some Iree parameters. The Iinite element templates Ior Bernoulli beam
and Timoshenko beam are examined in the paper, Irom the point oI view oI the
consistency condition. A new way oI template construction is proposed. It is
conIirmed that there is 1 Iree parameter Ior Bernoulli beam template and 3 Iree
parameters Ior Timoshenko beam template. The new method oI creation the
template can be extended Ior 2D and 3D problems.
112 W. Gilewski
ReIerences
1. Felippa C.A., A template tutorial I: panels, Iamilies, clones, winners and losers.
Report No. CUCAS0303, University oI Colorado, 2003
2. Felippa C.A., The amusing history oI shear Ilexible beam elements.
Report No. CUCAS0501, University oI Colorado, 2005
3. Gilewski W., Correctness oI plate bending Iinite element with physical shape
Iunctions, Finite Element News, 3, 1993, pp.2934
4. Gilewski W., Evaluation oI Iinite elements.
16
th
Int.Conf. on Computer Methoas in Mechanics CMM2005, Garstecki A., Mochancki B.,
Szczygiol N. Eds., Czstochowa 2005
5. Gilewski W., On the Criteria for Evaluation of Finite Elements From Timoshenko Beam to
HenckyBolle Plate (in Polish), OW Politechniki Warszawskiej, Warsaw 2005
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Evaluation oI the inelastic demand oI structures subjected to
multiple ground motions
Mihail Iancovici
1,2
, Georgiana Ionic
1
1
Department of Mechanics, Statics ana Dynamics of Structures, Technical University of Civil
Engineering (UTCB), Bucharest, 020396, Romania
2
National Center for Seismic Risk Reauction (NCSRR), Bucharest, 021652, Romania
Summary
In the current seismic aesign format, the key issue in establishing realistic seismic
loaas is the behavior factor. It accounts for all the aissipative mechanisms that a
structural system may aevelop unaer a strong grouna motion, however not clearly
enough statea yet. It corresponas to the performance level associatea to the
ultimate limit state (i.e. life safety), relatea to a 100 years mean return interval of
earthquake grouna motion with a prescribea peak acceleration of grouna.
The paper investigates the effect of repeatea Jrancea strong grouna motions on the
behavior factors ana the relatea parameters that accounts for cyclic structural
aeterioration aue to inelastic response. A large number of integratea analyses,
nonlinear response analyses ana energy balancebasea analyses were carriea out
ana estimates were maae on the behavior factors for inelastic SDOF systems
controllea by flexure with stiffness aegraaation. The correlation between behavior
factors ana aamage level are investigatea, using the Bo:orgnia ana Bertero
(2001), improvea aamage inaex. It is shown that multiple grouna motion of
Jrancea type for Bucharest, may leaa to an important increase of force ana arift
aemana of structures that usually is not taken into account.
KEYWORDS: multiple earthquake ground motions, behavior Iactor, hysteretic
energy, damage index, artiIicial accelerogram
114 M.Iancovici, G.Ionic
1. INTRODUCTION
Romanian territory and neighboring countries are repeatedly exposed to medium to
high intensity earthquake ground motions generated Irom the same source, located
in Vrancea region. Bucharest is one oI the most exposed cities to damage to
buildings and human losses as well. It is thereIore obviously needed to explore the
eIIects oI repeated Vrancea strong ground motion and the related implications that
may improve the seismic design oI new buildings and the evaluation procedures oI
the existing ones. In the current Romanian seismic design Iormat, based on strength
principles, the key issue in establishing realistic seismic loads that account Ior the
actual inelastic response, is the Iorce reduction Iactor/behavior Iactor, namely q.
Basically it accounts Ior all the dissipative mechanisms that a structural system
may develop under a strong ground motion, however not clearly enough stated yet.
It is recognized however that the complexity oI inelastic behavior phenomena
cannot be reproduced through a single parameter that is intended to Iully describe
the actual structural response. q Iactor is primarily related to the structural inelastic
response (ductility and cumulative eIIects oI repeated cycles oI inelastic
deIormations) and contains some oI the ground motions properties.
In most oI the seismic design codes and Romanian as well, q Iactor primarily
addresses to the selected structural type and includes the eIIect oI inelastic behavior
and the overstrength eIIect. It does not directly account Ior the inIluence oI strong
motion duration nor Ior the hysteretic behavior oI the structural elements. It
corresponds to the ultimate limit state perIormance level (i.e. liIe saIety), related to
a 100 years mean return interval oI a prescribed peak ground acceleration (PGA).
For repeated earthquake ground motions however, there is no clear evidence on
how this important Iactor might be interpreted and used in analyses.
The purpose oI the paper is to study the eIIect oI repeated strong ground motions oI
Vrancea type, on the behavior Iactors oI buildings located in Bucharest; we study
the variability oI q Iactor and related parameters on structural vibration period and
ductility, and on ground motion parameters as well.
2. BEHAVIOR FACTORS FOR SINGLE INPUT GROUND
MOTIONS; CYCLING LOADING EFFECT
Currently it is usual to estimate the actual Iorce demand by dividing the base shear
Iorce that corresponds to a Iully elastic response by the behavior Iactor.
Early studies revealed the Iact that the equal displacement assumption and the
equal energy assumption provide a Iairly good estimation oI the Iorce reduction
Iactors at long and short periods, respectively. These developments accounts Ior
'Computational Civil Engineering 2007, International Symposium 115
ductility properties only. A study by Newmark and Hall (1973) using 10 ground
motions records oI 1940 El Centro earthquake, proposed reduction Iactors that
include the eIIect oI both, ground motion and structural properties. Nassar and
Krawinkler (1991), Miranda and Bertero (1994), Watanabe and Kawashima (2002)
have been conducted studies on the Iorce reduction Iactors that give Iairly good
estimates on q Iactors especially Ior routine buildings. Ordan et al., 1998 and
Arroyo et al., 2003 Iound that the value oI q strongly depends on ductility and
natural vibration period, and is signiIicantly inIluenced by the soil type.
Inelastic behavior depends on many parameters, associated to the excitation and the
structural system. In order to uniIormly grasp the eIIect oI multiple earthquake
ground motions, and Ior the sake oI clarity we used spectral representations oI
SDOF systems response, having 5 damping ratio, and bilinear restoring Iorce
characteristics, with stiIIness degradation and 10 postyielding stiIIness ratio; all
systems have equal displacement ductility. For Ilexible structures however, SDOF
models are expected to reproduce with some degree oI inaccuracy the actual
response. For routine buildings we expect however realistic estimates.
By deIinition, the behavior Iactor is given by
) , , (
) , (
T F
T F
q
nl y
el el
= (1)
where F
el
and F
y
are the maximum linear and nonlinear base shear Iorce
respectively, is the displacement ductility Iactor,
el
and
nl
are the damping ratios
in the linear and nonlinear behavior range respectively, and T is the vibration
period oI the model. By simplicity, usually
nl
is taken same as
el
.
We Iirst compared the Iormulations oI Newmark and Hall (1973) and Miranda and
Bertero (1994), with q Iactors obtained Irom analysis Ior low and high
displacement ductility systems, using the NS component accelerograms recorded
during 1977, 1986 and 1990 at the INCERC Bucharest station (Iig.1).
0
0.5
1
1.5
2
2.5
3
0 1 2 3 4 5
7V
T
Newmark&Hall
Miranda&Bertero
analysis
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5
7V
T
Newmark&Hall
Miranda&Bertero
analysis
6
2
116 M.Iancovici, G.Ionic
0
0.5
1
1.5
2
2.5
3
0 1 2 3 4 5
7V
T
Newmark&Hall
Miranda&Bertero
analysis
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5
7V
T
Newmark&Hall
Miranda&Bertero
analysis
6
2
0
0.5
1
1.5
2
2.5
3
3.5
0 1 2 3 4 5
7V
T
Newmark&Hall
Miranda&Bertero
analysis
0
1
2
3
4
5
6
7
8
9
0 1 2 3 4 5
7V
T
Newmark&Hall
Miranda&Bertero
analysis
6
2
Figure 1. q Iactors Ior VN77NS, VN86NS, VN90NS records (INCERC Bucharest station)
From the above plots can be observed that q tends to one as T approaches zero.
Proposed relationships Iairly estimate the analysis results, especially Ior short
medium vibration period and Ior lowmedium displacement ductility; Ior Ilexible
structures with high ductility, the results are grossly overestimated, especially in
the case oI 1977 strong ground motion. On the other hand, a large variability oI q
in terms oI T and can be observed. The variability oI spectral response was more
detailed investigated, using a number oI 15 accelerograms oI 1986 Vrancea
earthquake, recorded in Bucharest (Iig.2).
0
0.1
0.2
0.3
0.4
0.5
0.6
0 1 2 3 4 5
7V
&
2
9
6
$
duct1 duct2 duct3
duct4 duct5 duct6
0
0.1
0.2
0.3
0.4
0.5
0 1 2 3 4 5
7V
&
2
9
T
duct2 duct3 duct4
duct5 duct6
Figure 2. COV oI SA and q Iactors Ior 1986 EQGM (15 records, Bucharest)
'Computational Civil Engineering 2007, International Symposium 117
The results are showing a larger variability corresponding to high ductility in small
vibration period region. As Ior the q Iactor, the scattering is much more
pronounced; the variability is lower Ior low ductility structures.
From the deIinition, one oI the major disadvantage oI q Iactor is that does not
account Ior the eIIect oI hysteretic demand, as a powerIul damage indicator (Uang
et al., 1990; Iancovici, 2005). The mass normalized hysteretic energy is given by
(2)
)
=
t
k
a x : t EH
0
2
) ( ) ( ) 1 ( ) ( t t t e 
where,
k
is the ratio oI pre and postyielding stiIIness, is the natural circular
Irequency and :(t) is the nonlinear (hysteretic) displacement. For the sake oI
investigating the patterns oI q Iactors and hysteretic energies, we plotted their mean
plus one standard deviation values on the same graph (Iig.3).
0
0.5
1
1.5
2
2.5
3
3.5
4
0.1 0.8 1.5 2.2 2.9 3.6 4.3 5
7V
T
0
0.02
0.04
0.06
0.08
0.1
0.12
(
+
P
A
V
A
q
EH
0
2
4
6
8
10
12
14
16
18
0.1 0.8 1.5 2.2 2.9 3.6 4.3 5
7V
T
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
(
+
P
A
V
A
q EH
2 6
Figure 3. q Iactor and hysteretic energy spectral representations (15 records, Bucharest)
As suggested, the variation pattern diIIers considerable; generally q Iactor could
not correctly reproduce the hysteretic energy distribution over the whole vibration
periods range. The coeIIicients oI variation Ior the input energy that the structure
will receive and the hysteretic energy that the structure will absorb are plotted in
Iig. 4.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 1 2 3 4 5
7V
&
2
9
(
,
duct2 duct3 duct4
duct5 duct6
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 1 2 3 4 5
7V
&
2
9
(
+
duct2 duct3 duct4
duct5 duct6
Figure 4 . COV oI input and hysteretic energies 1986 EQGM (15 records, Bucharest)
118 M.Iancovici, G.Ionic
A high coeIIicient oI variation can be observed Ior both parameters. For short
natural periods range the variation is higher Ior low ductility; Ior Ilexible systems,
the ductility eIIect on COV nearly vanishes.
3. MULTIPLE GROUND MOTIONS EFFECT ON STRUCTURAL
RESPONSE
The multiple input ground motion eIIect was introduced by a set oI two and three
accelerograms respectively. The eIIect oI longer duration motions was removed
Irom analyses by considering 40 seconds relaxation time intervals between
excitations. We chose Ior our purpose again the NS component accelerograms
recorded at the same site, INCERC Bucharest station during 1977, 1986 and 1990
earthquakes and the generated pulses are shown below.
0 5 10 15 20 25 30 35 40 45
2
0
2
t,s
g
a
1
,
m
/
s
2
0 20 40 60 80 100 120 140
2
0
2
t,s
g
a
2
,
m
/
s
2
0 50 100 150 200 250
2
0
2
t,s
g
a
3
,
m
/
s
2
Figure 5. Multiple input motions oI 1977, 1986 and 1990 EQGM, NS components
(INCERC Bucharest station)
The spectral representation corresponding to low and high displacement ductility
show that there is no sensitivity on q Iactors Ior the considered multiple input
motions.
0
0.5
1
1.5
2
2.5
3
3.
0 1 2 3 4 5
7V
T
sm1
sm12
sm123
0
1
2
3
4
5
6
7
8
0 1 2 3 4 5
7V
T
sm1
sm12
sm123
6
5
2
Figure 6. q Iactors Ior 1977, 1986 and 1990 EQGM, NS components (INCERC Bucharest
station)
'Computational Civil Engineering 2007, International Symposium 119
Taking the VN77NS record as reIerence and representing the ratios oI the
corresponding hysteretic energies (Iig. 7), it can be observed that the variation is
pronounced Ior SM12 and has almost doubled values Ior SM123, especially in the
case oI lowmedium vibration periods. This Iact is not reproduced by the behavior
Iactors ratios.
1
1.2
1.4
1.6
0 1 2 3 4 5
7V
(
+
(
+
(
+
o +
= (3)
120 M.Iancovici, G.Ionic
where is the displacement ductility,
e
is the ratio oI maximum elastic portion oI
deIormation to yield displacement,
mon
is the monotonic displacement ductility
capacity, EH
mon
is the hysteretic energy capacity under monotonic load and 0
1
1.0 is a constant. The associated damage index ratios are represented in Iig. 8.
1.00
1.05
1.10
1.15
1.20
1.25
1.30
0 1 2 3 4 5
7V
'
,
'
,
'
,
J
4
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 1 2 3 4 5
I+]
Q
R
U
P
3
6
'
+
]
Bucharest
PGA0.24g
Bucharest
PGA 0.24g
Bucharest
PGA0.20g
Figure 9. Code spectra compatible accelerograms and corresponding normalized PSD
Iunctions
Typical simulated accelerograms are presented below (Iig.10).
0 5 10 15 20 25 30 35 40
5
0
5
t,s
g
a
1
,
m
/
s
2
0 20 40 60 80 100 120
5
0
5
t,s
g
a
2
,
m
/
s
2
0 20 40 60 80 100 120 140 160 180 200
5
0
5
t,s
g
a
3
,
m
/
s
2
Figure 10. Typical code spectra compatible simulated accelerograms
By analyses, means plus one standard deviation values oI q Iactors were obtained
and are represented in Iig.11.
0
0.5
1
1.5
2
2.5
3
3.5
4
0 1 2 3 4 5
7V
T
sm1
sm12
sm123
0
2
4
6
8
10
12
14
0 1 2 3 4 5
7V
T
sm1
sm12
sm123
2
6
Figure 11. q Iactors (means plus one standard deviation), artiIicial accelerograms
122 M.Iancovici, G.Ionic
The results are showing a general decreasing trend oI the q Iactor; a slight decrease
oI q Iactor corresponding to SM12 and a more pronounced decrease in the case oI
SM123, especially in the case oI Ilexible structures.
0.9
1.0
1.1
0 1 2 3 4 5
7V
T
= (2)
where:
1
,
2
are the unit weight oI the two layers involved, a coeIIicient
experimentally set, E
1
and E
2
are the corresponding moduli oI the materials within
the two layers, h1 the thickness oI the layer Ior which the equivalent thickness is
established. For the present case
2 1
~ , 1 = o and 5 . 2 = n
1
400 =
2
/ cm
. The granular
material is considered to be a ballast with the Iollowing characteristics:
and a modulus . The
thickness oI the cushion is set initially to 1.00m, a Ioundation depth oI 1.50m and a
deIormation modulus oI the soil underneath .
8 6 , / 5 . 21 5 . 20
3
= =
opt a
w m kN
2
/ cm aaN E
2
100 = aaN E
'Computational Civil Engineering 2007, International Symposium 139
Figure 3. The method oI the equivalent stratum
The bearing capacity oI the Ioundation soil is set as p
conv
350kPa, the pressure
delivered by the Iooting p 250kPa, the distribution coeIIicient 0.55 and
consequently, the pressure at the base oI the cushion p
0
137.50kPa. The resulted
equivalent thickness is set as h
equivalent
1.50m. This case is similar with the ones
mentioned in literature, where the ration between deIormation modulus oI the soil
is smaller 1.82.5 the cushion modulus, 150200daN/cm
2
, the ground can be
considered homogeneous.
3. CONCLUSIONS
Based on the analysis perIormed on the Ioundation solutions, the Ioundation
system is recommended to consist oI Ilexible continuous Iooting Ior the external
shear walls and a Ioundation beam network oI the columns inside, all perIormed on
a compacted ballast cushion. The thickness is Iinally set at 2.0m to accommodate
poor soil conditions such as unconsolidated earth Iillings, mud and loose sands, a
nonuniIorm soil layering which is speciIic to swampy areas, and to ensure the
presence oI a drainage layer.
ReIerences
1. Mihu, A., Silion, T., Ciubotaru, V., Boi, N., Musat, V., Grecu, V., Beche, V., Srbu, G., Folosirea
pernelor de balast la Iundarea unor construcii de cldiri pe pmnturi puternic compresibile si
saturate cu ap, Revista nr. 1 Construcii, INCERC Bucuresti, 1976. (in Romanian)
2. Boti, N., Silion, T., Raileanu, P., Cijevschi, Maria, Musat, V., Grecu, V., Stanciu, A., Nicuta,
Ana, Criterii de dimensionare a pernelor de pamant, Sesiunea Stiinific Jubiliar ,40 ae ani ae
Invmant Superior ae Construcii la Iai`, 1981. (in Romanian)
3. Pop, V., Popa, A., Maniu, I., Popa, A., Fundaii de supraIa pe pern de balast, Rotaprint IP Cluf
Napoca, 1978. (in Romanian)
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
The Expertise oI OIIshore Structures
Ludovic G. Kopenetz, FerdinandZsongor Gobesz
Department of Structural Mechanics, Faculty of Civil Engineering ana Builaing Services, Technical
University of ClufNapoca, ClufNapoca, 400020, Romania
Summary
The problem of energy in tight liaison with ecology is representing one of the
greatest problems of humanity. Consiaering an exponential broaaen of marine
proauction in the following years, the international organi:ations have imposea the
aaoption of some highly strict internal ana international prescriptions in oraer to
avoia environmental acciaents (pollution). In this context, the structural safety of
the offshoretype complex plants ana equipments is quite actually.
The structural expertise must cover, besiae the vastness of interfering ranaom
variables, allimportant aspects in oraer to avoia the pollution of the pelagic
environment.
The present paper wishes to bring a humble contribution to the clarification of the
aspects which are proaucing sensitive alterations of structural safety (quality of
material, global ana local stability, fatigue, corrosion ana erosion etc.), using
aavancea ana lifetime orientea analysis concepts.
KEYWORDS: oIIshore structure analysis saIety stability assessment
1. INTRODUCTION
The behavior in situ oI the steel oIIshore structures (Iigure 1) and the ecological
aspects are tightly bonded to structural saIety. In this way, the structural expertise
must answer to the problems oI structural saIety 1, 2, 3. At its turn, the
concept oI structural saIety is bundled to the design, construction and operation oI
oIIshore structures, considering the extent oI interIering random variables (wind
waves, earthquake, construction Iaults, corrosion, Iatigue etc.).
The international standards concerning the conservancy oI the planetary ocean
(covering approximately 70 oI the Earth`s surIace) are very severe 4. In this
context the structural expertise oI the oIIshore structures which are located in the
area oI the coastal shelI oI the Black Sea has a peculiar importance, considering the
Iact that these structures were conceived and erected at the level oI knowledge oI
the years 19701980 5.
'Computational Civil Engineering 2007, International Symposium 141
Figure 1. OIIshore structure types.
It is known that during the years oI 1990 a series oI such structures collapsed
worldwide, such as the Sleipner platIorm made Irom reinIorced concrete) in the
North Sea (Iigure 2, 3), an oIIshore structure made Irom steel in the area oI Brazil`s
territorial waters etc.
During construction Perspective view Section views
Figure 2. Illustrations oI the Sleipner platIorm.
The Iollowing Iactors underlay the occurrence oI such accidents:
 misconceived structural analysis aspects (even using the most advanced
computing programs and supercomputers);
 corrosion due to severe operating conditions in marine environment;
 imperIections during construction and assembling;
 great number oI parameters which were hard to control;
142 L. G. Kopenetz, F. Zs. Gobesz
 post critical structural behavior;
 mismatches between labtest results and the real situation.
Figure 3.
In this paper the authors square up structural analysis aspects which are required
Ior structural expertise in order to assess the level oI saIety oI these kinds oI
structures in situ.
2. LOAD EVALUATION ASPECTS 6, 7, 8
The permanent and nonpersistent actions will be considered both by their lasting
time and their time variation.
'Computational Civil Engineering 2007, International Symposium 143
The evaluation aspects will relay upon a sternly hydro mechanical analysis (wave
theory, hydrostatics and hydrodynamics). In case oI Iloating oIIshore structures,
the Archimedean modiIication oI pushing Iorces will be considered
Model experiments and the simulation oI some regular loadings on such structures
have a distinctly relevance:
 sea rolling waves generated by intermittent wind having 150 175 km/hour
traveling speed,
 seismic waves (tsunami).
3. STRUCTURAL ANALYSIS ASPECTS AND PROBLEMS
CONCERNING FATIGUE TEST 9, 10, 11, 12, 17
Stresses in oIIshore structures are highly dependent Irom global stiIIness (jackets
coupling manner, cable inclination and tension etc.).
A dissimilar eIIect on stresses is coming Irom the vibration oI these structures 13.
The eIIects oI vibrations are originated Irom the Irequency oI the acceleration as
well as Irom the reversing speed oI the acceleration. These vibration eIIects are
acting not only on the bearing structure, but also on the operating crew (there are
persons to whom 0.01g acceleration at a 1Hz Irequency causes sensitivity).
The stress ranges are determined upon an elastic analysis oI the structure Irom
loadings caused by waves, wind etc. 14, 15, 16. The action oI waves has a
dynamic character, similar to seismic action, but having a quasipermanent
persistence. These loads are determined Irom the hydrodynamic analysis
considering the FROUDEKRYLOV components, the incidental hydrodynamic
mass etc.
The coupling between stress status and corrosion is well known. There is a
slackness oI the corrosion in case oI compressive stress, and an accelerated
corrosion in case oI tensile stress.
In the same way there is a connection between Iatigue loading and corrosion speed.
The methodologies used Ior Iatigue examination are based on the principle oI the
linear cumulating oI damages (MINER), based on the prescriptions Irom:
DNV Det Norske Veritas (Norway);
GL the German Lloyd (Germany);
ECCS the European Convention Ior Constructional Steelwork,
API the American Petroleum Institute (USA).
In order to trail in site these kinds oI structures, the authors are advising the
permanent monitoring oI the essential joints oI the structure and the comparison oI
the behavior with a synthetically WHLER curve.
144 L. G. Kopenetz, F. Zs. Gobesz
At the point when the loading spectrum is approaching to the curve, intervention
can be instantly made.
This procedure allows displaying the contribution oI corrosion on the level oI the
stress spectrum. The special tensomatic markers which are able to withstand
marine environment are already on the market at a reasonable cost, the problem
being resumed to the mounting and to the procurement oI proper soItware.
The quantiIication or numerical computation oI the stress spectrums can be done
through the RESERVOIR method.
4. CONCLUSIONS
The recent collapses oI some oIIshore platIorms, although these were made using
highly advanced technology, shows that structural expertise and examination works
are indispensable Ior the existing platIorms.
The emergence oI any damage and the modiIication oI the threshold Ior structural
collapse impose a reasonable limitation oI hazards. Lowering the risk to a certain
level is done usually with uncommon Iinancial exertion. Considering risk
limitation actions, the expertise and eventually monitoring oI these structures have
an incommensurably lower cost than other measures, adopted usually on
probabilistic bases.
Uncontrolled movements due to wind, ice loading and
combination oI wind and ice
Collapsing movements due to earthquakes
Destructive movements due to waves
OFFSHORE MOVEMENTS
Local instability
Explosions
Earthquake
Fire
internal
external
COLLAPSE TYPES ON
OFFSHORE STRUCTURES
The large changes in conIiguration (geometric
nonlinearity)
The drag loading and the dependency oI the load on the
position and orientation oI the structure
The nonlinear loaddeIormation behavior
NONLINEARITIES OF
OFFSHORE STRUCTURES
Figure 4.
'Computational Civil Engineering 2007, International Symposium 145
In case oI steel oIIshore structures, the actions which are causing sensitive
modiIications oI the structural saIety can be grouped in the Iollowing classes
(Iigure 4):
 problems concerning base material quality (coldIlow, diIIerent brittle Iracture
types, lamellar Iracture);
 problems concerning global and local stability (curling oI thin walls and
plates etc.);
 Iatigue and repeated stress;
 creeping;
 problems concerning deIormations induced by static and dynamic loads;
 corrosion and erosion problems.
The actual prescriptions need improvements in the way oI extending the computing
concepts related to the so called 'structural and 'speciIic imperIections.
The Iabrication and maintenance costs oI the oIIshore structures made Irom steel
are huge, due to the aspects mentioned in this paper. In such conditions o possible
way and with spectacular results appears to be the using oI reinIorced concrete by
the accomplishment oI the oIIshore constructions. Alternative structures in this
direction are presented in Iigures 5, 6 and 7, having the decisive advantage oI
corrosion withstanding, relatively simple erection ashore, simple anchoring in
comparison with the special ones Irom steel structures.
Figure 5.
146 L. G. Kopenetz, F. Zs. Gobesz
Figure 6.
Figure 7.
'Computational Civil Engineering 2007, International Symposium 147
The tackling oI Iatigue problems is advisable to be done through permanent
monitoring and by comparing the stress spectrum with a synthetically WHLER
curve.
Structural engineers dealing with such structures are advised to accept only those
results which are conIirmed by a simpliIied, but not shallow structural analysis.
ReIerences
1. Clauss, G., Lehmann, E., stergaard, C., Offshore Structures, Springer Verlag, LondonBerlin
New York, 1992.
2. Ctrig, A., Kopenetz, L., Alexa, P., Analysis Problems oI Tubular OIIshore Structures,
Proceeaings of the Seventh International Symposium on Tubular Structures, A. A. Balkema,
Rotterdam, BrookIield, 1996.
3. Kopenetz, L., Ctrig, A., Practical Structural Dynamics oI Marine Cables, Oviaius University
Annals, Series. Civil Engineering, vol. 1, No. 6, 2004.
4. Kopenetz, L., Ctrig, A., Comportarea 'in situ a structurilor din oel de tip OIIshore, Revista
Construciilor, Nr. 1, 2005.
5. CECMTC6, Recommanaations pour la verification a la fatigue aes structures en acier, CTICM
Centre technique et industriel pour la construction mtallique, SaintRmyles Chevreuse, France,
1987.
6. Brebbia, C. A., Walker, S., Dynamic Analysis of Offshore Structures, NekinesButterworths,
LondonBoston, 1981.
7. Kinsman, B., Wina Waves, their Generation ana Propagation on the Ocean Surface, Prentice
Hall, 1965.
8. Hauptmanns, U., Werner, W., Engineering Risks, Springer Verlag, BerlinLondonNew York,
1991.
9. Haibach, E., Betriebsfestigkeit, VDIVerlag, DsseldorI, 1989.
10. EUROCODE3, Design of Steel Structures, 1993.
11. STAS 10108/01978, Design of Steel Structures (in romanian), 1978.
12. Dalban, C., et alii, Construcii metalice, EDP, Bucuresti, 1983.
13. Kopenetz, L., Ctrig, A., Probleme de analiz structural pentru consolidarea podurilor,
Simpo:ion Reabilitarea Drumurilor i Poausilor, ClujNapoca, 2000.
14. British StandardBS5400, Part 10, Steel, Concrete and Composite Bridges, Coae of Practice for
Fatigue.
15. RP2A, Recommenaea Practice for Planning, Designing ana Constructing Fixea Offshore
Platforms, Washington, American Petroleum Institute, 19871989.
16. APIRP2A, LRFD, its consequences for aaoption to North Sea Offshore aesign practice,
Advanced Mechanics & Engineering Ltd, 1991.
17. EUROCODE1, Basis of Design ana Actions on Structures, 1991.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Scenario and impact oI global warming in building energy
perIormance simulation
Petr Kotek
1,2
, Jordan
1
1
Department of Microenvironmental ana Builaing Services Engineering,
CTU Technical University in Prague, C:ech Republic
2
contact.
kotek.petrcentrum.c:
Summary
Builaing performance simulation (BPS) tools are more ana more wiaely usea in
practice ana aevelopers are trying to facilitate ana anticipate the settings of input
parameter for orainary users. BPS tools are usea to aetermine the energy
consumption, inaoor environment ana to aiagnose heating ana cooling loaa for
relevant HJAC systems in the builaing unaer investigation. Every simulation
moael is still unaer a big uncertainty in input setup aata in case of using BPS tool
before the builaing is built. An uncertainty ana sensitivity analysis (UASA) was
performea in our latest research by using MonteCarlo simulation (with Latin
Hypercube sampling). The aim of this research was to estimate the most influential
input parameter for builaing envelop ana other bounaary conaition.
From UASA was assumea the most influential input parameters such as outsiae
temperature ana sun raaiation (OT, SR) which we (i.e aesigner) can not change or
optimi:e. Ana moreover, these outsiae climate aata are aifferent every year. It
means that if we use some reference climate aatabase we will put a big uncertainty
in simulation moael ana outputs are with a big aifferent uncertainty every year.
This paper is focusea in impact of aifferent OT, SR auring the year ana mainly in
point of view. 'WHAT IF IN FUTURE`. The climate is lunatic every year ana we
can see ana feel that the temperature is slowly rising all over the worla. Nowaaays
the question of global warming is more ana more aiscussea. Accoraing
Intergovernmental panel on climate change (IPCC) we performea a stuay how the
energy consumption will change in preaictea years accoraing to IPCC scenario.
Two case stuaies were chosen for aifferent scenarios of climate change First ana
secona cases are one real aaministration builaing in C:ech Republic, Prague. One
is performea without thermal insulation on the wall construction ana secona with
ana is shown their appropriate sensitivity in point of view of climate change
KEYWORDS: climate change, IPCC scenario, sensitivity analysis, Monte Carlo
simulation, Latin Hypercube sampling
'Computational Civil Engineering 2007, International Symposium 149
1. INTRODUCTION
Uncertainty and sensitivity analysis by using MonteCarlo (MCA) simulation is a
powerIul method Ior estimation the most inIluential parameters and Ior clariIying
how big the uncertainty in results is. The simple crude MCA simulation technique
is shown on Figure 1. Inputs X
i
can be Ior example thickness, density and
conductivity oI thermal insulation, transmittance oI transparent construction,
emissivity oI opaque envelope or some inputs Ior HVAC components etc. 2
thermal
model
new input file
for simulation
with all n
inputs
changed
input x1
input x2
input xn
time
o
u
t
p
u
t
v
a
l
u
e
Monte Carlo Analysis (MCA)
p
r
o
b
a
b
i
l
i
t
y
d
e
n
s
i
t
y
f
u
n
c
t
i
o
n
(
p
d
f
)
repeat for all simulations from 1 to E
results with their
uncertainty
Figure 1. basic principle oI MonteCarlo (MCA) simulation.
Many oI them can be Iuture optimize by using some optimization algorithm but
some oI these like outside dry bulb temperature and sun radiation could not. These
not 'predictable inputs can bring huge uncertainty in results and they can inIluent
heating and cooling load Ior HVAC equipment and change heating and cooling
energy consumption every year. Whence it Iollows that iI we use BPS tool Ior
assessment energy consumption we have to present range oI values where the real
energy consumption per year probably will be. It means, that iI we use some 'old
climate database with average temperature Ior 50 or 30 years, we have to analyze
these values, compare with measured values Ior 5 years backwards Ior example
and perIormed new ones.
From Figure 2. we can see the diIIerence between 30 years observation in Prague
and actual measure temperature. From the chart we can also see, that the long term
observation is little bit 'colder than nowadays temperature. From this we can say,
that the distribution oI the temperature is not normal during the year but more Beta
distribution (with o2 , 5). It is obvious that iI we use the 'red values Ior the
building perIormance simulation we will have diIIerent output values instead oI
using the 'blue one.
150 P. Kotek, F. Jordan
Figure 2. diIIerence between long term observation and actual temperature in Prague
From the Figure 2. we can Ieel about, that the climate is wormer and wormer than
long term observation. Also this winter 2007 a lot oI people save money Ior paying
heating bills because the temperature was so high.
And moreover in 2003, 2004, 2005, 2006 was the annual average temperature
higher than observation among years 19712000 in Czech Republic (CR). Nobody
knows iI our planer will rapidly warmingup but all over the world we can see, that
rising temperature changing a lot oI thing just now.
In 1988 United Nations Organization (OSN) established Intergovernmental Panel
on Climate Change (IPCC) by World Meteorological Organization (WMO) and by
the United Nations Environment Programme (UNEP). IPCC reports are widely
cited in almost any debate related to climate change.
The IPCC published a Iirst assessment report in 1990, a supplementary report in
1992, a second assessment report (SAR) in 1995, and a third assessment report
(TAR) in 2001. A Iourth assessment report (AR4) is currently under way.
The IPCC has made a series oI reports and scenarios related to climate change and
these it can be Iind out in TAR. Statements oI the IPCC or inIormation Irom the
TAR are oIten used as a reIerence showing a scientiIic consensus on the subject oI
global warming. An increasing body oI observations gives a collective picture oI a
warming world and other changes in the climate system (The global average
surface temperature has increasea over the 20th century by about 0.6C)
The TAR estimate Ior the climate sensitivity is 1.5 to 4.5 C over the period 1990
to 2100. Each scenario then has a range oI possible outcomes associated with it.
The most optimistic outcome assumes an aggressive campaign to reduce CO
2
emissions, while the most pessimistic is a "business as usual" scenario. The more
realistic scenarios Iall in between. 6
'Computational Civil Engineering 2007, International Symposium 151
Figure 3. The 140 years observation oI global temperature (IPCC)
Figure 4. The 1000 years climate model observation oI global temperature (IPCC)
According the IPCC long term observation (Figure 3., Figure 4.) we can see, that
the Earth is really warmingup but maybe this is the normal 'behavioroI our
planet and it happened many times in the past. But maybe is it true and every single
human on our planet making our planet warmer by driving car Ior example. This
paper is based on work oI 2500 scientist all over the world in 120 countries and we
used the IPCC outcomes Irom their climate model as an input Ior our investigation.
Investigation in point oI view: How the heating and cooling energy consumptions
will change over the period 1990 to 2100.
152 P. Kotek, F. Jordan
2. IPCC SCENARIOS
From TAR outcomes we used the scenarios
Global mean temperature change (C) (Figure 5a.)
Radiative solar Iorcing (W/m
2
) (Figure 5b.)
a)
b)
Figure 5. The IPCC scenarios a) temperature b) solar radiation
For creating new climate database it can be used crude MCA (via analysis in
Figure 2. and prediction in Figure 5). But the pure MonteCarlo simulation cannot
be applied Ior timeconsuming problems such as dynamic BPS tool, as it requires a
large number oI simulations. A small number oI simulations can be used Ior the
acceptable accuracy oI statistical characteristics oI response using the stratiIied
Latin Hypercube sampling technique LHS 1. For our aim is enough to create
climate database based on scenario B1 (lower bound) and A1Fl (upper bound) with
uncertainty shown in Figure 2. The IPCC report also includes one inconsiderable
'Computational Civil Engineering 2007, International Symposium 153
Iact. Fact, that annual temperature is rising and also every year occur more days
with extreme temperature in winter and summer. This Iact can be seen on Figure 6.
Figure 6. Globe mean temperature change and variance in Iuture
3. PROGRESSION
For Czech Republic exist two *.epw climate databases Ior Prague and Ior Ostrava
city (see Figure 7.). These *.epw Iiles can be download Irom EnergyPlus website
Iree oI charge. 8 and used Ior every king oI BPS tool (ESPr, Trnsys, DOE2,
Design Builder, IESVE~, etc).
154 P. Kotek, F. Jordan
Figure 7. Czech Republic and marked two cities with available climate database
Sometimes these two climate databases are used Ior the whole Czech Republic and
hereby bring big uncertainty in results. But this has not been considered in this
paper. The case study (administration building) is in the capital city Prague and we
will modiIy the downloaded climate Iile CZEPragueIWEC.epw with Weather
Statistics and Conversions (WSC) tool developed by EnergyPlus. 8
The statistical values Ior both climate databases are shown in Table 1.
Table 1. Statistical report oI CZEPragueIWEC.epw
GPS / elevation 5044' N, 1429' E/ 364 m
Annual mean temperature 7.9 C
Days in heating period 245
Average temperature during heating period 4.09 C
min / max temperature 15.2 / 30.5 C
Hours with temperature ~ 0C /year 7078 hours (295 days)
Hours with temperature _ 0C /year 1682 hours (70 days)
By using EnergyPlus WSC tool the Iiles were converted to the *.csv and exported
to the MS Excel modiIied Ior two scenarios (step 10years) and uncertainty made
by SimLab 7. These new weather Iiles (12 Ior each scenario) were converted
back to the *.epw Iile with WCS tool and used Ior simulations.
Table 2. Statistical report oI CZEPragueIWEC.epw in 2100 according IPCC
scenario B1 in 2100 scenario A1Fl in 2100
Annual mean temperature 10.4 C 12.7 C
Days in heating period 204 180
Average temperature during
heating period
3.86 C 4.22 C
min / max temperature 16.3 / 36.7 C 17.8 / 42.6 C
Hours with temperature ~ 0C 7393 hours (308 days) 7596 hours (317 days)
Hours with temperature _ 0C 1367 hours (57 days) 1164 hours (48 days)
'Computational Civil Engineering 2007, International Symposium 155
0
0.01
0.02
0.03
0.04
0.05
0.06
30 20 10 0 10 20 30 40 50
epw_Prague_1995
PCC_scenario A1Fl_2100
Figure 8. Globe mean temperature change and variance according IPCC Ior Prague
For this prototype we used IES VE~ 5 which incorporates ApacheSim, a
dynamic thermal simulation tool based on Iirstprinciples mathematical modeling
oI building heat transIer processes. It has been tested using ASHREA Standard 140
4 and qualiIies as a Dynamic Model in the CIBSE system oI model classiIication
3.
4. CASE STUDIES  ADMINISTRATION BUILDING
For the real administration building in Prague was perIormed two possible
situations.
Adm.case A without thermal insulation like many old building in CR
Adm.case B with 8cm insulation on the wall.
Another construction and setting were used the same Ior both cases.
Figure 9. Photo oI administration building and energy model in IES VE~
The material prosperities Ior both cases can be Iound in Appendix 1.
156 P. Kotek, F. Jordan
Table 3. Basic proportions and settings oI the building
Width x length x hign 12.4 x 15.3 x 8.1 m
Volume 1537 m
3
Area oI glazing 18
Set point Ior heating  day / night, weekend 20C / 15C
Set point Ior cooling  day / night, weekend 24C / 32C
Working hours 6  18 hours
For these cases (adm.case A, adm.case B) were made new 48 'IPCC weather Iiles
and Ior each one made 20 'uncertainty Iiles according probability distribution.
The automatic simulation were perIormed during approximately 4 hours on Intel
Core2 CPU T5500, 1.66 GHz, 1GB RAM.
5. RESULTS
Results shown in Figure 10 represents both cases in point oI view energy load Ior
heating (red curve) and cooling (blue curve) and annual energy consumption.
A
d
m
.
c
a
s
e
A
0
5
10
15
20
25
30
1
9
9
5
2
0
1
5
2
0
3
5
2
0
5
5
2
0
7
5
2
0
9
5
0
10
20
30
40
50
60
1
9
9
5
2
0
1
5
2
0
3
5
2
0
5
5
2
0
7
5
2
0
9
5
A
d
m
.
c
a
s
e
B
0
5
10
15
20
25
30
1
9
9
5
2
0
1
5
2
0
3
5
2
0
5
5
2
0
7
5
2
0
9
5
0
10
20
30
40
50
60
1
9
9
5
2
0
1
5
2
0
3
5
2
0
5
5
2
0
7
5
2
0
9
5
Energy demands (kW) Energy consumption per year (MWh)
Heating Cooling
Fiure 10. Impact oI global warming according scenario IPCC  the uncertainty in results on
the basic shown in Figure 2 represents the black boundary.
'Computational Civil Engineering 2007, International Symposium 157
From results in Figure 10 can be seen, that iI the temperature and solar radiative
Iorcing rise up, than heating load will stay almost the same. The annual heating
will little bit Iall down because oI warmer winter. Obviously the huge impact will
be in cooling demand and consumption. II we compare the results Ior non insulated
and insulated building it is nice to see, that cooling demand will rise up Irom now
in 2100 about 7010 (without insulation) and 409 (with insulation) and
consumption will be higher about 21821 (without ins.) and only 8212
(with ins.) This big diIIerence is because oI easier accumulation in the brick wall
without insulation.
6. CONCLUSIONS
The aim oI this paper is to show how big the impact oI global warming can be in
total annual energy consumption per year. The Iact, that we will need more power
Ior cooling bring together with higher temperature also higher pollution Irom
cooling sources and the ozone hole will be bigger and bigger because oI
greenhouse gases. It is not obvious, that thermal insulation can also cut down the
cooling consumption and it is time to do it Ior every single building not only
because oI reducing heating demand.
Acknowledgements
The authors thank Ior support CTU research aim No. CEZ MSM 6840770003
ReIerences
1. McKay M.D., Conover W.J. and Beckman R.J., A Comparison oI Three Methods Ior Selecting
Values oI Input Variables in the Analysis oI Output Irom a Computer Code, Technometrics Jol.
21. 239245., 1979.
2. Kotek, P., Jordan, F., Kabele, K., Hensen, J., Technique of uncertainty ana sensitivity analysis for
thermal ana HJAC calculations, IBPSA 2007, Beijing, Sep/2007. (not published yet)
3. Crawley, B.D., Hand, W.J., Contrasting the capabilities oI building energy perIormance
simulation programs, United States Department oI Energy, University oI Strathclyde, Wisconsin,
2005.
4. ANSI/ASHRAE Standard 1402001, Standard Method oI Test Ior the Evaluation oI Building
Energy Analysis Computer Programs.
5. IES Virtual Environment~ 5.1. Building perIormance simulation tool,
http://www.iesve.com/content/ , last visited May 2007
6. IPCCThe ScientiIic Basis, Published by the press syndicate oI the university oI cambridge Ior the
Intergovernmental Panel on Climate Change (IPCC), www.cambridge.org ISBN 0521 01495 6,
2001
7. SIMLAB Version 2.2. 2004. Simulation Environment Ior uncertainty and sensitivity analysis,
developed by the Joint Research Centre oI the European commission. http://simlab.jrc.cec.eu.int
8. EnergyPlus website: http://gundog.lbl.gov/
158 P. Kotek, F. Jordan
Appendix 1  Thermal prosperities oI administration building
Adm.case A  without thermal insulation
Wall Construction (outside to inside)
Layer
Conductivity
W.m
1
.K
1

Thickness
m
Capacity
J.kg
1
.K
1

Density
kg.m
3

ENISO
UValue
W.m
2
.K
1

Outside plaster 0,190 0,0125 1000 800
Brickwork 0,800 0,450 900 1700
Inside plaster 0,350 0,010 1000 1000
1,21
Adm.case B  with thermal inslutaliton
Wall Construction (outside to inside)
Layer
Conductivity
W.m
1
.K
1

Thickness
m
Capacity
J.kg
1
.K
1

Density
kg.m
3

ENISO
UValue
W.m
2
.K
1

Outside plaster 0,190 0,0125 1000 800
Polystyrene 0,044 0,080 1270 20
Brickwork 0,800 0,450 900 1700
Inside plaster 0,350 0,010 1000 1000
0,38
Other construction
ENISO
UValue
W.m2.K1
RooI 0,28
Ground Floor 0,9
Floor 0,84
Window 1,7
Glass door 1,7
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Settlement prediction oI a pile group based on design charts
Irina Lungu
1
, Nicolae Boi
2
and Anghel Stanciu
3
1
Department of Roaas ana Founaations, Technical University, Iai, 700050, Romania
2
Department of Roaas ana Founaations, Technical University, Iai, 700050, Romania
3
Department of Roaas ana Founaations, Technical University, Iai, 700050, Romania
Summary
The rehabilitation of transportation infrastructure is an ongoing process that
involves an increasea complexity of briages ana retaining structures with
corresponaing complex requirements to comply with auring service.
When the necessary bearing capacity is not proviaea by means of a shallow
founaation, a aeep founaation, i.e. a piles group is consiaerea appropriate for both
bearing ana aeformation requests.
Pile founaations represent current founaation solutions for these constructions,
given the poor soil conaitions often encounterea on the construction site. Their
aesign is basea on the soil limit states both aeformation ana bearing capacity.
Settlement restrictions that are incluaea with the aeformation limit state require an
accurate evaluation of the settlement preaiction regaraing the pile group.
Depenaing on the purpose of using piles for the founaation system, one of the
following three cases will be the most suitable in a practical aesign. (a) the piles
are supposea to carry the full loaa, (b) the piles will act as settlement reaucers for
the raft which is a less important bearing element of the founaation system, the
piles are requirea to have a factor of point ana shaft capacity safety somewhat
greater than unity, (c) the piles are proviaea as settlement reaucers but they will
operate at a shaft factor of safety of unity, ana piles of aifferent length or stiffness
are usea . The raft is the principal bearing element of the founaation.
Sometimes, simple settlement calculations seem to be aaequate for practical
purposes proviaea that the appropriate insitu soil properties have been carriea
out with engineering fuagment. The errors introaucea by the simple methoas are
usually small comparea with those that can occur auring sampling ana testing.
The laboratory ana fiela tests on single piles ana pile groups as well as a careful
monitoring activity of builaings with pile founaation have known a tremenaous
evolution, implementing all this in aata base systems usea in the computer fiela ana
aavancea numerical methoas for engineering purposes.
KEYWORDS: pile group, settlement, design charts, pilesoil interaction, single
pile, Finite Element Method, Boundary Element Method.
160 I. Lungu, N. Boi, A. Stanciu
1. INTRODUCTION
There is a strong intention to modiIy the actual capacity based design to a
settlement based design when Ioundation systems on piles are concerned
1,2,5. In pursuing this issue there are the Iollowing arguments:
bearing capacity oI the single pile is very sensitive to the pile installation
procedure while its stiIIness is less aIIected;
bearing capacity oI an individual pile in a pile group may signiIicantly vary
when piles have diIIerent dimensions and spacings, or when due to soil
variation, the installation procedure may introduce some diIIerences among
them, while pile group behavior under loading intend to make this diIIerences
less signiIicant;
the connecting raIt may deliver a signiIicant part oI the load to the subsoil by
direct contact and in the same time, its inIluence on the Ioundation settlements
may be less important 5,39;
when piles act as settlement reducers the most important Iactor that decide on
the necessary number oI piles is the diIIerential settlement.
2. SETTLEMENT PREDICTION OF A PILE GROUP BASED ON
THE SINGLE PILE BEHAVIOR
Considering that the single pile rigidity was established by load tests on single
piles, the stiIIness oI a pile group may be quantiIied through the use oI an
eIIiciency, q, the inverse oI the group settlement ratio, R
s
, deIined as:
1
1
nk
k
R
p
s
= = q (1)
with k
1
 pilehead stiIIness oI a single pile; k
p
 stiIIness oI the pile group (in terms
oI average settlement).
To a Iirst approximation, the group eIIiciency was taken as a simple power law oI
the number oI piles in the group, as: , thereIore the group stiIIness
becomes: 3.
e
n
~ q
1
1
k n k
e
p
~
The exponent, e, generally lies in the range oI 0.30.5 Ior primarily Iriction piles,
rising to 0.6 or higher Ior end bearing piles. Fleming presented a set oI design
charts (Iigure 1) in terms oI a base value ,e
1
, and Iour correction Iactors c
1
to c
4
, as
indicated in the equation:
) ( ) ( ) / ( ) / ( ) (
4 3 2 1 1
v c c a s c G E c a l e e
p
= (2)
'Computational Civil Engineering 2007, International Symposium 161
Figure 1. Design chart to evaluate the pile eIIiciency
The simple use oI eIIiciency Iactors does not provide any indication oI the
increased proportion oI load carried by the pile bases in a group oI piles  and thus
the greater importance oI underlying soil strata compared with the case oI a single
pile 5.
162 I. Lungu, N. Boi, A. Stanciu
Poulos 2 pointed out that since the base response is generally more nonlinear
than the shaIt response, this may lead to lower group eIIiciency than using linear
elasticity. But ignoring nonlinear eIIects will overestimate the interaction between
piles, which will compensate Ior overestimation oI the base stiIIness.
3. COMBINATION OF BOUNDARY ELEMENT WITH FINITE
ELEMENT APPROACH
Hain and Lee 4 developed a method Ior pile raIt settlement prediction which
included a Iinite element approach Ior the raIt and a boundary element approach
using Mindlin`s solution Ior the pile group interaction (so called raIt on a pile
reinIorced continuum).
The model considers a variable stiIIness raIt oI any geometry supported by a
random group oI identical piles and a soil mass idealized as an elastic continuum.
Consideration is given Ior a deep homogeneous mass, a modiIied linear elastic
analysis Ior a layer oI Iinite thickness, a Iailure oI individual piles at load less than
the total group capacity by applying an excess load cutoII procedure.
Four types oI interactions are included in this approach and represented in Iigure 2.
Figure 2. PilesoilraIt interactions basic representations
'Computational Civil Engineering 2007, International Symposium 163
Imposing the compatibility and equilibrium conditions between the soilpile group
system and the raIt a set oI stiIIness equations can be Iormed. Two assumptions are
involved:
only vertical Iorces are transmitted Irom the raIt to the pile heads;
each pile occupies the whole oI the 'constant pressure area around a
particular node.
These assumptions can not be appropriate to be applied Ior the analysis oI piled
isolated Iootings where lateral Iorces and overturning moments are the major
proportion oI the applied loads.
The pile compressibility is also considered and as a conclusion, the author stated
that the percentage oI load suported by the pile group diminishes as the pile
compressibility increases.
A decrease in the length oI the piles had the similar eIIect on the load distribution
in the Ioundation system. The Iinal results with all the inIluences assessed at the
begining are presented in Iigure 3 as design charts.
Figure 3. Design charts with parameters inIluencing the pile group behavior under vertical
loads
164 I. Lungu, N. Boi, A. Stanciu
The method is considered the Iirst attempt in considering a sound interaction in a
piled raIt Ioundation with the limitations oI a linear elastic behavior oI the soil, a
regular pile arrangement: the raIt and piles with the same length and geometry 5.
4. CONCLUSIONS
The stiIIness oI a single pile becomes more important in design practice not only
by its value but Ior learning the motivation oI a suggested procedure in the
available literature and thus to be able to make a sound engineering judgement
each time a new case comes up.
Design charts proved to be useIul tools to set comparative values at work in order
to select the Ioundation elements in the best combination to Iace even the most
pessimistic hypothesis oI soilIoundation interaction. A successIul and economic
design is reached only when the interaction eIIects are taken into account.
ReIerences
1. Van Impe, W.F., Developments in pile aesign, DFI ConIerence, Stressa, 1991
2. Poulos, H.G., Founaation economy via pilearaft systems, Piletalk International, K. Lumpur,
Malayesia, pp 97106, 1991
3. Fleming, W.G.K., A new method Ior single pile settlement prediction and analysis,
Geotechnique, vol 42, no 3, pp 411425, 1992
4. Hain, S.J. and Lee, I.K., The analysis oI Ilexible raItpile systems, Geotechnique, vol 28, no 1,
pp 6583, 1978
5. Lungu, I., Tez de doctorat, Contribuii privina conlucrarea intre terenului ae funaare, funaaie
i structura ae re:isten, Iasi, 1997
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
The importance oI controlling subsystem in inIormation systems
Ior building industry
Peter Mesaros
1
, Frantisek Mesaros
2
1
Department of Management Sciences, Faculty of Business Economics, University of Economics,
Kosice, 04130, Slovakia
2
Research Institute of Builaing Informatics Lta., Kosice, 04001, Slovakia
Summary
The paper presents various aspects ana opportunities of using information systems
ana information technologies in builaing inaustry consiaering their impact on
competitiveness. Especially we focus on new subsystem of controlling ana its
integration in corporate information system. It is a contribution to the research
profect Nr. 2330256/07 of University of Economics in Bratislava, namea
Knowleage Management as Assumption of Successful Development of Slovak
Enterprises.
KEYWORDS: inIormation systems, inIormation technologies, controlling system,
building industry
1. INTRODUCTION
An increasingly rapid change and development in the business environment oI
building Iirms have been noted in recent years. The growth oI domestic as well as
Ioreign competition creates a constant pressure on intensive exploitation oI the
latest inIormation technology, innovation oI products and increase in the quality oI
construction perIormance. This appears to be the way oI acquiring a competitive
advantage and achieving the building Iirm prosperity. Important Iactors oI
prosperity are mainly composed oI:
proIit which is not Iormed at the expense oI the Iuture gains,
shortterm liquidity and longterm capacity oI Iunding,
economy oI the whole Iirm with an accent on economization and quality oI
the construction perIormance,
eIIectiveness in using relevant inIormation and inIormation technologies Ior
its processing in order to achieve competitiveness.
Managers in the building Iirms, even despite the Iact that various programme
systems have already been introduced in the corporate inIormation systems, report
on the lack oI relevant current and exact inIormation related to the corresponding
dates and suitable Iorm oI decision making in an eIIicient and eIIective way. The
current corporate inIormation systems in the building industry are not capable oI
166 P. Mesaros, F. Mesaros
providing the inIormation that is needed by managers. In connection with this, it is
particularly the inIormation required Ior:
comparing the plan and reality as to the ongoing conduct oI the building i.e.
the inIormation necessary Ior the ongoing management oI proIit and costs oI
the construction,
making the deviations analysis as compared to the planned state,
rapid conduct oI calculation with providing the data Ior the needs oI Iurther
measures to eliminate the deviations Irom the plan.
The current inIormation systems do not assure a rapid indication oI problems and
analysis oI their causes with the proposal Ior their elimination (Carnick, 1999).
Managers can Iind this inIormation in a huge amount oI data concerned with the
plan, accounting, marketing, various reports, etc. They select only certain
inIormation, make their own analysis in order to make the optimum decisions.
However, this activity is rather timeconsuming and not so eIIective, as a result oI
this process making a relevant directing decision appears to be Irequently rather
subjective and intuitive. Similar associations are also mentioned by various other
authors (M. Kozlovska, 2002). In recent years a large business has grown up
around the subject oI inIormation systems and inIormation technologies and its
implementation in business practice oI all industries oI national economies. ICT
companies and management consultants have developed methods and techniques Ior
identiIying and agreeing investment in inIormation systems that support business
strategy and create new strategic options and build the competitiveness. InIormation
technology has changed the nature oI competition and has created new
opportunities Ior using inIormation services. Using a technology to reduce the cost
oI building company activities or to add value to existing products and services can
provide a competitive advantage.
2. INFORMATION SYSTEMS FOR BUILDING INDUSTRY
2.1. ERP systems and its exploitation in building companies
According to Wikipedia deIinition, Enterprise Resource Planning systems (ERPs)
integrate all data and processes oI an organization into a uniIied system. A typical
ERP system will use multiple components oI computer soItware and hardware to
achieve the integration. A key ingredient oI most ERP systems is the use oI a
uniIied database to store data Ior the various system modules.
Enterprise Resource Planning is a term originally derived Irom manuIacturing
resource planning (MRP II) that Iollowed material requirements planning (MRP).
MRP evolved into ERP when routings became major part oI the soItware
'Computational Civil Engineering 2007, International Symposium 167
architecture and a company's capacity planning activity also became a part oI the
standard soItware activity. ERP systems typically handle the manuIacturing,
logistics, distribution, inventory, shipping, invoicing, and accounting Ior a
company. Enterprise Resource Planning or ERP soItware can aid in the control oI
many business activities, like sales, marketing, delivery, billing, production,
inventory management, quality management, and human resource management.
ERPs are crossIunctional and enterprise wide. All Iunctional departments that are
involved in operations or production are integrated in one system. In addition to
manuIacturing, warehousing, logistics, and InIormation Technology, this would
include accounting, human resources, marketing, and strategic management.
Implementing ERP soItware is typically not an internal skill, so even smaller
projects are more cost eIIective iI specialist ERP implementation consultants are
employed. The length oI time to implement an ERP system depends on the size oI
the business, the scope oI the change and willingness oI the customer to take
ownership Ior the project.
2.2. Controlling subsystem new component in ERPs Ior building industry
Special applications
In our view, controlling presents a speciIic Iorm oI work with inIormation, its
gathering, processing, evaluating and providing Ior the needs oI managerial
perIormance and decisionmaking. From the point oI view oI time, two main
directions have been Iormed, namely the strategic and operational controlling. The
strategic controlling is aimed at direction oI a longterm detecting oI the business
company potentials. Externally, it analyzes the threats and opportunities. The
operational controlling directs its activities within the given proIit potential. Its aim
is the optimisation oI subject, time and value parameters oI the corporate activities
aimed at achieving the prosperity and proIit in the process oI building production
conduct.
2.2.1. Assumptions of Controlling Subsystem Implementation
In the controlling subsystem the existing databases are used that have been Iormed
within the calculationplanning subsystem (ex ante inIormation) and the database
oI personal and economic inIormation subsystem (ex post inIormation).
A simpliIied connection oI the given subsystems and their groups is illustrated in
Fig.1 and Fig. 2.
The data are taken Irom the group oI tasks Proauction calculation that presents the
proposed economic and capacity expression oI technical, technological, material
and organizational perIormance oI the object construction or just a part oI it.
A whole range oI data can be Iound here, namely the limit ones, the norm/standard/
and planned data Ior the whole construction, an object, or a part oI it. Continuously
168 P. Mesaros, F. Mesaros
once a month, these data are taken Irom the group oI tasks Proauction billing Ior
the concretely undertaken volume in the given construction, object, part, or an
order.
Calculationplanning
Subsystem
1. standarbasis
budget
2. production calculation
3. production billing
transportation
4. budgeted indexes
5. time plan
Subsystem of social and
economic information
1. wages
2. human resources
3. accounting
4. transportation
5. material and technical
assurance
Subsystem
OPERATIONAL
CONTROLLING
Figure 1. Operational controlling as a subsystem oI the inIormation system
The production billing uses the same standard basis as the production calculation.
The production billing (PB) is calculated as a multiplication oI the data oI the real
vector oI production (Q sk) and the data Irom the standard basis (SB).
SB Qsk PB  = (1)
The production billing can be processed Irom:
organizational points oI view Ior orders, or objects when the order includes
more objects (constructions, centres, centre oI costs, division, plant),
the point oI view oI time in a month, exceptionally even a shorter time period
(summarized quarterly, annually, Ior the whole order).
The area of drawing the building costs: The tool Ior perIorming this activity is
accounting. Even smaller enterprises should have an independent area oI the so
called (managerial) accounting oI costs. The accounting oI costs enables to use
such procedures and methods oI accounting that are not allowed Ior Iinancial
accounting. The objective oI this procedure is to enable to deIine the respective
volume oI variable costs and to Iind the socalled contribution proIit (a contribution
to cover the Iixed costs Irom the proIit production) at every calculation unit, centre,
etc.
'Computational Civil Engineering 2007, International Symposium 169
The essential diIIerence between adding the costs to the Iinancial accounting and
accounting oI costs consists in the Iact that:
Iinancial accounting is managed by the accounting classes Ior the needs oI
making a record on the proIit and losses. The socalled balance sheet and
thereIore the main criterion here, is a strict evidence oI documents;
accounting oI costs is more governed by the principle oI causality. The basic
kinds are classiIied with an accent and in dependence on certain decisions so
that they could directly enable the respective management measures.
The data on the reality are mainly taken Irom the group oI tasks oI intracorporate
accounting in the structure deIined within the calculationplanning subsystem in
connection with the organizational structure oI the given building Iirm. For
example:
EC  economic centre
CC  costs centre
EA  economic activity
CU  calculation unit/entity (construction object)
ThereIore, it is inevitable to create a code oI numbers oI the actual suppliers in the
structure  a plant, EC, CC, EA.
Determination
of quantities
Detection of
deviations
Analysis of
deviations
Proposal of
measures
CONTROLLING AND ITS IMPLEMENTATION
Plan of costs Comparing plan
 reality
Reasons of
deviations
Recommendatios
for management
Figure 2. Content oI activities and procedural steps in the controlling subsystem
Within the controlling subsystem even the links to the material and technical
supply Ior the purposes oI continuous comparing the planned material need with
the real material use, not only in the Iinancial but also in the material expressing,
are deIined as well.
When selecting the Iollowed indexes the individual building Iirms themselves
determine the type oI indexes, as well as the extent and degree oI details. However,
170 P. Mesaros, F. Mesaros
their extent is given by the principle oI eIIectiveness oI the inIormation system. It
is required that the costs Ior their provision are lower than the eIIects provided by
the detected deviations and accepted decisions.
2.2.2. Aavantages of implementing ana exploiting the controlling subsystem
The advantages arise Irom technically calculated and justiIied norms oI
manuIacturing and technical character oI unit costs elaborated by the Department
oI manuIacturing and technical preparation oI production. Thus, controlling
contributes to the prosperity oI a building Iirm,
creates presumptions Ior acquiring a higher quality oI building work
help achieve rapid outcomes oI costs drawing in the course oI construction
(once a month at the minimum), i.e. not aIter the completion oI construction,
serves to common checkup oI drawing the individual costs Ior the
construction,
enables to carry out a detailed analysis oI deviations, provides the possibility
oI analysing the causes oI their origin and points to the divisions responsible
Ior their origin,
has a psychological signiIicance since it upgrades the motivation oI workers
towards savings (as this method enables a preventive checkup oI costs
drawing, it also motivates the workers towards higher responsibility Ior their
own economic behaviour),
creates pressure towards the perIect arrangement oI organizational relations
among divisions, sticking to the technical, technological and organizational
conditions oI the construction perIormance,
enables to evaluate the eIIectiveness oI production divisions activities as well
as the activities oI divisions oI preproduction and production preparation
(Travnik, 1998), e.g. incomplete project documentation oI the construction
with a poor quality, on the basis oI which the technical preparation oI the
construction is elaborated. The production calculation may thus cause rather
great deviations Irom the norms that are not due to production but are
generally ascribed to production,
enables to analyse continuously the objectivity oI understanding the right
overheads Ior the respective construction job,
provides important data Ior the analysis oI the secondary budgeted costs that
are oIten incorrectly accounted within the construction overhead costs.
In the context oI the above said it is highly recommended to set up a position oI the
controller who, on the basis oI his knowledge and practical experience, could
analyse the detected deviations between the plan and reality. He thus assures the
working oI inIormation Ilows, i.e. provides objective inIormation Ior decision
making on the basis oI real processes taking place in the construction.
'Computational Civil Engineering 2007, International Symposium 171
Within his work capacity, the controller assures the Iollowing tasks, in particular:
 prepares supporting documents Ior planning and decisionmaking,
 processes the methodological materials, regulations Ior processing the
normative basis oI the enterprise, production calculations, budgets,
production billing, budgeted indexes, supporting papers Ior a Iollowup oI
production cost indexes in the intracorporate accounting,
 Iollows the legal acts in the Iield oI pricing, accounting, calculations,
statistics, taxes, etc., and tries to implement them into the corporate
inIormation system as soon as possible,
 provides methodological counselling in the given areas Ior the enterprise
workers.
The setting oI the controller tasks provides only a Iramework oI his activities.
ThereIore, it is rather necessary to take into account the speciIic Ieatures oI every
building Iirm as well.
3. CONCLUSIONS
Application oI the controlling subsystem appears to be a relatively simple and
transparently conceptual Iact. It assures the complex approach and presents an
eIIective Iorm oI acquiring, processing and interpreting the inIormation on the
costs and proIit, which support decisionmaking and Iormulating the strategy by
managers. It gives presumptions Ior increasing the eIIiciency oI a building Iirm and
its prosperity according to the objective indexes. It enables to act successIully in
the competitive environment as well as to respond to constantly increasing
requirements oI investors to the quality oI building production.
ReIerences
1. Carnick, S., Why the top management workers need executive inIormation systems? Acta
Oeconomica Cassoviensia N3, Faculty oI Business Economics Kosice oI the University oI
Economics in Bratislava, 1999.
2. Kozlovska, M., Management oI purchasing, Eurostav, vol. 8, 2002.
3. Mesaros, F., Mesaros, P., InIormation Systems Strategy in Building in Industry, Economics ana
Business Management, vol. 1, 2003.
4. Travnik, I., et al., Directing the builaing work value, Slovak Technical University in Bratislava,
Bratislava, 1998.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Analysis oI Water Consumption in Residential Building
Roman Musil
CTU in Prague, Faculty of Civil Engineering, Department of Microenvironmental ana Builaing
Services Engineering, Thakurova 7, 169 34 Prague 6,email. roman.musilfsv.cvut.c:
Abstract
This paper consiaer about measurement consumption of cola water in resiaential
builaings. In the year 2005 ran over extensive monitoring cola water consumption
in 62 resiaential builaing in Pilsen implement by waterworks Pilsen with
cooperation aepartment of microenvironmental ana builaing services systems in
Prague. The paper consiaers about analysis this measurement ana comparsion
with C:ech norm ana prescriptions. There will be evaluate aaily ana hourly
maximal coefficient cola water takeoff ana aaily ana hourly cola water
consumption profiles for resiaential builaings.
Keywords
monitoring, residential building, water consumption
1. INTRODUCTION
Need and consumption oI cold water in residential building is magnitude which is
not initial only Ior dimension supply water systems but it shows liIe style and user
consumer habits. This value is evolving and changing with society development.
Impact on user behaviour is in relation prices development. The second aspect is
technical building Iacilities development in hot water preparation which enables
using Ilexible Iacilities reacting on variable water needs.
2. WATER NEED
Water need is supposed takeoII while consumption is the real takeoII water
amount Ior certain time period. Water consumption assessment is input value
which is covered with suIIicient equipment capacity Ior water supplying. Problem
is when consumption is Iunction oI time and we have to observe it in diIIerent time
scale. ThereIore water need assessment isn`t possible simpliIy to one value but
time period has to be written by more parameters which are:
'Computational Civil Engineering 2007, International Symposium 173
water need during the chosen period (speciIic water need)
water takeoII during time period distribution
Basic statement is speciIic water need qp. This value is given in dependence on
kind oI operation and activity in designed building. It is water need per basic
consumption unit (person, bed, eat,.) and it is in liter/(unit*day). Probably water
losses Irom source to consumer are included in speciIic water need. Probably water
losses are supposed about 20. State speciIic water need Ior residential buildings
according to valid Czech public notice number 428/2001 sb. are in table 1.
Table 1. SpeciIic water need m
3
/person/year according to valid Czech notice 428/2001 Ior
residential buildings.
3. WATER NEED ASSESSMENT
We calculate with subsequent values Ior assessment oI water need whichever
object:
1) Daily average water need
Q 1
P
q . n  l/d
Where:
q speciIic water need (liter/unit/day)
n number oI unit
3.1 Daily maximum water need
Facilities Ior water takeoII Irom source, capacity oI water and water pipelines Ior
supplying water to watersupply tank are designed on maximal water need.
Q 2
m
Q
P
. k
d
 l/d
Where:
174 R. Musil
Q
p
average daily water need
k
d
daily inequality coeIIicient
Table 2. TakeoII daily inequality coeIIicient CSN 736701
3.2 Hourly maximum water need
We dimension Iacilities Ior supply water to consumers on this value. Running oI
vary hourly need is use Ior dimension oI pumping device which supply water
directly to consumers (Ior example pumping station Ior increasing water pressure).
Hourly water need is one value Irom component Ior assessment oI watersupply
tank value.
Q 3
h, max
Q
m
. k
h, max
. z  l/hod
3.3. Hourly minimum water need
Q Q
m
. k
h, min
. z  l/hod 4
h, min
Where: Qm daily maximum water need l/day
kh, max hourly maximum inequality coeIIicient
kh, min hourly minimum inequality coeIIicient
z takeoII water period residential building a hotels z 24 hours
administrative buildings z 10 hours
Balance values Q
m
and Q
h
serve to accumulate reservoir and automatic pumping
station designing which are components oI internal watersupply. Determination oI
daily and hourly inequality coeIIicient is very complex Ior residential and civil
buildings. These coeIIicients depend on various Iactors. Water need during day is
unequal and depends on building operation character, on yearly season,
temperature change, Ilat and house Iurniture oI building services water systems, on
social and age oI user groups and their behavior. Water takeoII varies during week
too.
'Computational Civil Engineering 2007, International Symposium 175
In 2005 year went ahead wide monitoring water consumption in 62 residential
buildings in Pilsen practiced by the waterworks Pilsen in cooperation with
Department oI Microenvironmental and Building Services Systems, CTU in
Prague. Monitoring on cold water was done in 3 week period by water meter Iixed
on building Ioot with hourly period saving measured values. These values are
primarily determined Ior processing stochastic loads proIiles and mathematical
urban space modelling. Final values dont include hot water which is supplied to
building by centralized heating system. Evaluate measurement results are in next
graphs.
&ROG:DWHU&RQVXPSWLRQ
0.00
1.00
2.00
3.00
4.00
5.00
6.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
WLPHKRXU
O
L
W
U
H
S
H
U
V
R
Monday Tuesday Wednesday Thursday
Friday Saturday Sunday average
Figure 1. Daily cold water consumption during week
Table 3. Comparsion oI hourly inequality coeIIicient in several days with average value
176 R. Musil
&ROG:DWHU&RQVXPSWLRQ
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
WLPHKRXU
O
L
W
U
H
S
H
U
V
R
Monday Tuesday Wednesday Thursday
Friday Saturday Sunday average
Figure 2. Maximal cold water consumption during week envelope curve 2.
'DLO\:DWHU&RQVXPSWLRQ
0.00
20.00
40.00
60.00
80.00
100.00
120.00
M
o
n
d
a
y
T
u
e
s
d
a
y
W
e
d
n
e
s
d
a
y
T
h
u
r
s
d
a
y
F
r
i
d
a
y
S
a
t
u
r
d
a
y
S
u
n
d
a
y
7LPHK
&
R
O
G
:
D
W
H
U
S
H
U
S
H
U
V
R
Q
D
Q
G
G
Average daily water consumption
Maximum daily water consumptionenvelope spline
Figure 3. Total daily cold water consumption
'Computational Civil Engineering 2007, International Symposium 177
$YHUDJH&ROG:DWHU&RQVXPSWLRQ5HJDUGLQJWR%XLOGLQJ
2FFXSDQF\
0
10
20
30
40
50
60
70
80
90
M
o
n
d
a
y
T
u
e
s
d
a
y
W
e
d
n
e
s
d
a
y
T
h
u
r
s
d
a
y
F
r
i
d
a
y
S
a
t
u
r
d
a
y
S
u
n
d
a
y
$
Y
H
U
D
J
H
F
R
Q
V
X
P
S
W
L
R
Q
O
L
W
U
H
S
H
U
V
R
Q
40  50 inhabitants 50  75 inhabitants
100  300 inhabitants 300  400 inhabitants
400  500 inhabitants above 500 inhabitants
average
Figure 4. Total daily cold water consumption divided according to number oI occupant in
the building
$YHUDJHKRXUO\WDNHRIILQHTXDOLW\FRHIILFLHQWIRUUHVLGHQWLDOEXLOGLQJV
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
M
o
n
d
a
y
T
u
e
s
d
a
y
W
e
d
n
e
s
d
a
y
T
h
u
r
s
d
a
y
F
r
i
d
a
y
S
a
t
u
r
d
a
y
S
u
n
d
a
y
kmax for 4050 inhabitants kmin for 4050 inhabitants
kmax for 5075 inhabitants kmin for 5075 inhabitants
kmin for 75100 inhabitants kmax for 100300 inhabitants
kmin for 100300 inhabitants kmax for 300400 inhabitants
kmin for 300400 inhabitants kmax for 400500 inhabitants
kmin for 400500 inhabitants kmax for more than 500 inhabitants
kmin for more than 500 inhabitants
Figure 5. Hourly inequality coeIIicient according to Ilat occupancy
178 R. Musil
4. CONCLUSION
From measurement results result that total yearly water consumption Irom all 62
measurement is 56 liter/person/day (Fig. 3 and Fig 4.). Average water consumption
in same object type is Ior us important at total water need evaluation and is very
depended on concrete measured residential buildings, its inhabitants and technical
equipment. For building services systems designing average value doesn`t have
predicative value. As well as in other technical areas Ior example structural
mechanics Ior beam calculation we have several load states which are partly
covered and results is maximum load envelope curve. It is similar with water
consumption, in this case we have 62 load states (done measurement) where
envelope curve copy maximal takeoII values in separate hours (Fig. 2). Average
maximal takeoII is 90 liter/person/day. From equation 2 is coeIIicient inequality
1,6 which in comparable with Czech norm value ( Ior 1000 inhabitant is k
d
1,5).
From Fig. 3 is possible to see that average daily water consumption during week
has minimum diIIerence. DiIIerences are in time period oI takeoII (Fig. 1 and
Fig.2) where is possible to see extreme in noon and in the evening. Greatest take
oII peak is in Saturday (noon) and in Sunday (noon and evening). In Saturday
evening is surprisingly lowest Irom the whole week. Measurement results shows
also number oI inhabitants inIluence on average cold water consumption in object
(Fig. 4). Average cold water consumption in object with lower number oI
inhabitant is bigger than by bigger occupancy oI building. The next important
indicator is hourly inequality coeIIicient (Fig.5) which is in most oI groups
expressively other than CSN 736701 values, how is shown in consequent table.
Table 4. Comparsion oI hourly inequality coeIIicient in several days with average value
Values were calculated Irom measured dates always Ior every day in week and
were averaged Ior compare with Czech norm values. Table shows that values given
by calculation Irom measured cold water consumption in residential buildings are
extensively diIIerent. Residential building with smaller number oI inhabitants has
bigger diIIerences (as Iar as about 55) and other way round with more people in
object is diIIerence smaller. Smaller daily inequality coeIIicient values mean lower
hourly takeoII water diIIerences which is inIluenced by Iact that was measured
only cold water and there are no taken in peak hot water takeoIIs. In table 4, we
can see percentage comparison calculated k
hmax
Ior individual days with average
hourly maximal coeIIicient during week. Table shows, that weekly average value
'Computational Civil Engineering 2007, International Symposium 179
oI hourly inequality coeIIicient is with regard to small daily value diIIerences
suIIicient.
Acknowledgement
This paper would not be created without excellent cooperation and technical
Iacilities oI Waterworks Pilsen, a.s which ensured measurement and date
collection. Paper was created with support oI investigative intention CEZ MSM
6840770003.
ReIerences
1. Kadlecova M., Notices to new Czech norm oI hot water design, Topenastvi Instalace 3/99
2. Kabele K. and collective, Energy and Environmental building systems, CTU 2005
3. Czech notice oI Agriculture Ministry 428/2001
4. Vrana J. Water and Sewerage in Home and Flat, Grada publishing 2005
5. Musil V.and collective, Technical Buildings Facilities I
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Improving Civil Engineering Physics TeachingLearning with
Mathematica 5.1
Irina Radinschi
1
, Brindusa Ciobanu
2
1
Department of Physics, 'Gh. Asachi` Technical University, Iasi, 700050, Romania
2
Department of Physics, 'Gh. Asachi` Technical University, Iasi, 700050, Romania
Summary
In this paper we present some important examples of Mathematica 5.1 techniques
for stuaying physical concepts ana phenomena ana for plotting a general class of
physics phenomena. We point out that a computer algebra system like Mathematica
is very useful for physics stuaies. We have a longterm experience in using this
program in physics teaching ana learning, as in approachea fielas as mathematical
physics ana computational physics.
Mathematica 5.1 is aescribea through some examples of plotting waves
interference, constructive interference ana aestructive interference. These
examples highlight the power ana versatility of Mathematica ana inaicate its
application to a much wiaer range of problems in computational physics.
Mathematica 5.1 assists the stuaents for solving aifferential equations,
aifferentiating, integrating, making sums, finaing roots, plotting ana proaucing
animations. We emphasi:e the graphical capability of Mathematica ana use its
power to go beyona finaing solutions ana performing length calculations ana bring
the stuaies alive with animations, ana other graphical tools. This is possible
because Mathematica 5.1 proviaes a gooa environment for computation, a high
level programming language, text, graphics, ana animation. Further, GrTensorM
which is embeaaea in Mathematica is a package which allows us carrying out
length calculations ana evaluating the components of some tensorial ana
pseuaotensorial quantities.
For these reasons we implement Mathematica 5.1 in our course of physics, for
exploring, teaching, ana applying powerful mathematical methoas in physics. This
computer algebra system allows the stuaents to aevelop their ability to learn
physics, to solve aifficult problems with a symbolic computing engine, ana to
review the basic mathematics. We introauce Mathematica program into our
courses for stuaents who mafor in technical fielas because is a reliable tool for
teaching ana learning physics in a moaern way.
KEYWORDS: Mathematica 5.1, engineering physics teachinglearning,computer
algebra system, physics, physical concepts and phenomena.
'Computational Civil Engineering 2007, International Symposium 181
1. INTRODUCTION
It is wellknown that nowadays, physics studies are not allowed without using
adequate soItware. Mathematica is a powerIul mathematical soItware system Ior
teachers and students. In the recent years, Mathematica has changed the way
physics is taught and teached. Mathematica 5.1 gives the possibility to visualize
and display physics concepts and phenomena, to perIorm length calculations and to
generate numerical and graphical solutions to physics problems. It also provides
modern techniques and Ior plotting 2D and 3D having many options Ior graph
drawing.
Our experience in computational physics 16, 1014 and particularly in using
Mathematica 5.1, make us to decide that a better way Ior engineering physics
teachinglearning is to incorporate this computer algebra system in our course oI
physics. Mathematica 5.1 is a very powerIul and useIul general purpose program
78 which can Iind solutions to algebraic equations, it can do calculus, it can
evaluate equations numerically and it can plot and generate animations and sounds.
Mathematica 5.1 also covers essential physical situations and phenomena. This
allows the students leaving the length calculations to Mathematica and pay more
attention to the physics studies. Mathematica has also a powerIul online help built
in, which can be used even by the new users 78.
Mathematica is a reliable tool oI choice, in engineering analysis and modeling, and
in technical education in universities. Mathematica 5.1 combines powerIul
computing soItware with a convenient user interIace. Mathematica's Ieatures
include symbolic and highperIormance numeric computation, 2D and 3D data
visualization, broad programming capabilities, and onestep creation oI web
documents 9.
2. PRESENTATION OF MATHEMATICA 5.1 APPLICATIONS
2.1. Applications oI Mathematica 5.1 in physics
Mathematica 5.1 includes advanced programming methods Irom modern computer
science as well as adding a host oI new ideas oI its own 9.
Mathematica 5.1 applications in physics studies allow the students to test their
understanding oI the concepts and phenomena, and applying them to many
physical situations. It also allows solving rapidly physics problems without
perIorming length calculations, and plotting and producing animations. We show
how Mathematica 5.1 will be applied to take the students beyond the limitations oI
182 I. Radinschi, B. Ciobanu
traditional instruction. We present some Mathematica 5.1 applications in physics
which are implemented in our courses.
2.2.1. Mathematica 5.1 commanas for stuaying the wave interference. Plotting 2D
interference.
The wave interIerence is an interesting topic, and with the graphical and animation
capability oI Mathematica, we can improve our physics courses. Once the students
are able to handle the equations which describe the wave interIerence, they also
discover results on their own by varying the amplitudes, the values oI pulsation,
initial phase and time, and generating constructive and destructive interIerence.
Mathematica 5.1 is also a reliable tool Ior graph drawing. It has many options Ior
the color oI the curves and Ior the color oI the background. These can be set by
using the option Graphics`Colors` and choosing Irom there the colors. The builtin
Mathematica color directives RGBColor, Hue, and CMYKColor handle the most
common systems. This package gives color directives using other standard
systems. For axes we use the option AxesLabel and we plot the physical quantities
on Zaxis, Yaxis and Xaxis.
The Mathematica 5.1 commands Ior plotting 2D the wave interIerence are given
below. We give some examples Ior constructive and destructive interIerence.
Graphics
ln1:Plot0.2*SinPi/4*tPi/3,0.2*SinPi/4*tPi/3},
t,1,10},PlotStyleHue0.1},Hue0.2}},BackgroundRGBColor0.94174`,
1.`,1.`,AxesLabelt, }
2 4 6 8 10
t
0.2
0.1
0.1
0.2
Figure1. Plotting two waves
Out1 Graphics
'Computational Civil Engineering 2007, International Symposium 183
ln2:Plot0.2*SinPi/4*tPi/3,0.4*SinPi/4*tPi/7,0.2*SinPi/4*tPi/30.4*
SinPi/4*tPi/7},t,1,10},PlotStyleHue0.1},Hue0.2,
Hue0.3}},BackgroundRGBColor0.94174`,1.`,1.`, AxesLabelt, }
2 4 6 8 10
t
0.6
0.4
0.2
0.2
0.4
0.6
Figure 2. Plotting 2D interIerence
Out2 Graphics
In this case, the resultant wave Iunction is evaluated using the command
ln3: TrigExpand0.2*SinPi/4*tPi/30.4*SinPi/4*tPi/7
Out30.346759*CosPi/4*t0.460388*SinPi/4*t
Out3 
4
t
Sin 0.460388 
4
t
Cos + 0.346759
ln4:Plot0.2*SinPi/4*t,0.4*SinPi/4*t,0.2*SinPi/4*t0.4*SinPi/4*t},t,0.
1,50},PlotStyleHue0.1},Hue0.2},Hue0.7}}, BackgroundRGBColor
0.94174`,1.`,1.`,AxesLabelt,}
10 20 30 40 50
t
0.6
0.4
0.2
0.2
0.4
0.6
Figure 3. Plotting constructive interIerence
184 I. Radinschi, B. Ciobanu
Out4 Graphics
The resultant wave Iunction is computed with the command
ln5: TrigExpand0.2*SinPi/4*t0.4*SinPi/4*t
Out50.6*SinPi/4*t
Out5 
4
t
Sin 0.6
ln6:Plot0.2*SinPi/4*tPi/3,0.2*SinPi/4*tPi/3, 0.2*SinPi/4*tPi/3
0.2*SinPi/4*tPi/3}, t,1,10}, PlotStyleHue0.1}, Hue0.2}},
BackgroundRGBColor0.94174`,1.`,1.`, AxesLabelt, }
2 4 6 8 10
t
0.2
0.1
0.1
0.2
Figure 4. Plotting destructive interIerence
Out6 Graphics
ln7:Plot0.2*SinPi/4*t,0.2*SinPi/4*tPi,0.2*SinPi/4*t0.2*SinPi/4*tPi
},t,0.1,50},PlotStyleHue0.1},Hue0.2},Hue0.7}},
BackgroundRGBColor 0.94174`,1.`,1.`,AxesLabelt, }
10 20 30 40 50
t
0.2
0.1
0.1
0.2
Figure 5. Plotting destructive interIerence
'Computational Civil Engineering 2007, International Symposium 185
Out7 Graphics
2.2.2. Mathematica 5.1 commanas for plotting 3D wave interference
We give an example oI plotting 3D the interIerence. The Mathematica 5.1
commands Ior plotting 3D waves interIerence are presented.
ln8:p1Plot3D0.2*Sin*t,t,1,10},,Pi/7,Pi/4},BackgroundRGBColor
0.94174`,1.`,1.`,AxesLabelt,,}
2
4
6
8
10
t
0.5
0.6
0.7
0.2
0.1
0
0.1
0.2
2
4
6
8
t
Figure 6. Plotting 3D wave Iunction
Out8 SurIaceGraphics
ln9:p2Plot3D0.2*Sin*tPi,t,1,10},,Pi/7,Pi/4},BackgroundRGBColo
r 0.94174`,1.`,1.`,AxesLabelt,,}
2
4
6
8
10
t
0.5
0.6
0.7
0.2
0.1
0
0.1
0.2
2
4
6
8
t
Figure 7. Plotting 3D wave Iunction
186 I. Radinschi, B. Ciobanu
Out9 SurIaceGraphics
ln10: SurIaceGraphics
Showp1,p2},AxesTrue
0
2.5
5
7.5
10
t
0.5
0.6
0.7
0.2
0.1
0
0.1
0.2
0
2.5
5
7.5
t
Figure 8. Plotting 3D destructive waves interIerence
Out10 SurIaceGraphics
The sum oI two wave Iunctions can also be plotted using the commands given
below.
ln11:p3Plot3D0.2*Sin*t0.2*Sin*tPi,t,1,10},,Pi/7,Pi/4},Backgro
undRGBColor 0.94174`,1.`,1.`,AxesLabelt,,}
2
4
6
8
10
t
0.5
0.6
0.7
110
16
0
110
16
2
4
6
8
t
Figure 9. 3D graph Ior destructive waves interIerence
'Computational Civil Engineering 2007, International Symposium 187
Out11 SurIaceGraphics
The resultant wave Iunction Ior destructive waves interIerence is computed using
the command
ln12: TrigExpand0.2*Sin*t0.4*Sin*t
Out110*Sin*t
For plotting constructive interIerence we use the next commands.
ln13 p3Plot3D0.2*Sin*tPi/3,t,1,10},,
Pi/4,Pi/4},BackgroundRGBColor 0.94174`,1.`,1.`,AxesLabelt,,}
2
4
6
8
10
t
0.5
0
0.5
0.2
0.1
0
0.1
0.2
2
4
6
8
t
Figure 10. Plotting 3D wave Iunction
Out13 SurIaceGraphics
ln14 p4Plot3D0.2*Sin*tPi/4,t,1,10},,
Pi/4,Pi/4},BackgroundRGBColor 0.94174`,1.`,1.`,AxesLabelt,,}
188 I. Radinschi, B. Ciobanu
2
4
6
8
10
t
0.5
0
0.5
0.2
0.1
0
0.1
0.2
2
4
6
8
t
Figure 11. Plotting 3D wave Iunction
Out14 SurIaceGraphics
The resultant wave Iunction is computed using the command
ln15 TrigExpandSin*tPi/30.2*Sin*tPi/4
Out150.0317837*Cos*t0.241421*Sin*t
ln16 Plot3D0.0317837*Cos*t0.241421*Sin*t,t,1,10},,
Pi/4,Pi/4},BackgroundRGBColor 0.94174`,1.`,1.`,AxesLabelt,,}
2
4
6
8
10
t
0.5
0
0.5
0.2
0
0.2
2
4
6
8
t
Figure 12. Plotting 3D constructive waves interIerence
Out16 SurIaceGraphics
'Computational Civil Engineering 2007, International Symposium 189
Plot3D includes a setting Ior the PlotPoints and one Ior the Mesh option. For the
graph Irom the Figure 12 we use these options and we obtain
ln17 Plot3D0.0317837*Cos*t0.241421*Sin*t,t,1,10},,
Pi/4,Pi/4}, BackgroundRGBColor 0.94174`,1.`,1.`, PlotPoints40,
MeshFalse, AxesLabelt,,}
2
4
6
8
10
t
0.5
0
0.5
0.4
0.2
0
0.2
0.4
2
4
6
8
t
Figure 13. Plotting 3D constructive waves interIerence
Out17 SurIaceGraphics
3. CONCLUSIONS
The latest version Mathematica 5.1 raises the level in mathematical soItware,
providing more than 50 new Iunctions, toolkits, and perIormance improvements in
a powerIul program 1516. The Mathematica 5.1 version adds a host oI new
capabilities, especially Ior working with largescale, diverse types oI data. New
innovative algorithms are introduced to deliver unmatched perIormance Ior all
steps in the data handling process, importing, analyzing, manipulating, or plotting.
We have introduced Mathematica 5.1 as a teaching and learning tool in our course
oI physics Ior civil engineering. Mathematica 5.1 is a powerIul program and has
many advantages. It provides a good environment Ior computation, a highlevel
programming language, text, graphics, and animation. We assume that the students
are already Iamiliar with Mathematica, but we also oIIer a basic training in this
computer algebra system Ior beginers.
The cases oI Plot and Plot3D wave interIerences was illustrated with some
examples and commands Ior plotting 2D and 3D waves interIerence, and also
190 I. Radinschi, B. Ciobanu
options like PlotStyle, BackgroundRGBColor, AxesLabel, PlotPoints, and Mesh
have been added Ior improving the graph drawing.
Our main goals are to implement Mathematica 5.1 and Maple 9.5 programs to
assist students in physics learning and Ior solving problems. We want our students
to make progresses and we also want them to make these progresses rapidly. On
the other hand, we`ll give them the possibility to choose between Mathematica 5.1
and Maple 9.5, or to work with both programs.
ReIerences
1. Ciobanu B., Radinschi I., One Computational Algorithm Ior Physics Modeling, Proceeaings of 5
th
International Conference on Electromechanical ana Power Systems, SIELMEN, Chisinau, Rep.
Moldova, October 68, 2005, pp. 222225.
2. Radinschi I., Ciobanu B., Testing Physics, Junimea Publishing House, Iasi, Romania, 2006.
3. Radinschi I., Ciobanu B., Physics for Engineers, Junimea Publishing House, Iasi, Romania, 2006.
4. Ciobanu B., Radinschi I., A Sequential Programming Algorithm a Tool Ior Study the Standing
Waves, Proceeaings of International Symposium IEEI 2006, Iasi, Romania, Bulletin oI IP Iasi,
Vol. LII(LVI), Fasc.5, pp. 129134.
5. Radinschi I., Scripcariu L., Ciobanu B., Frunza M. D., Online TeachingwhileQuizzing Test,
Proceeaings of the 2
na
National Conference on Appliea Physics, June 910, Galati, Romania,
2006, in press Romanian Journal oI Physics
6. Ciobanu B., Radinschi I., Teaching and Learning Physics Ior Engineers by Aid Computers,
Proceeaings of International Symposium IEEI 2006, Iasi, Romania, Bulletin oI IP Iasi, Vol.
LII(LVI), Fasc.5, pp. 135140.
7. Mathematica Ior Scientists and Engineers: Using Mathematica to do Science, by Richard Gass
(Prentice Hall).
8. www.wolIram.com.
9. http://www.pugh.co.uk.
10. Radinschi I., Frunza M. D., Ciobanu B., Online Virtual Model Ior Testing the Knowledge,
Proceeaings of INTED 2007, March 79, Valencia, Spain.
11. Radinschi I., Ciobanu B., Implementation oI Computational Methods in Physics Learning,
Proceeaings of 4
th
International Symposium Computational Civil Engineering 2006, Iasi,
Romania, 2006, pp. 251257.
12. Frunza M. D., Radinschi I., A Virtual Model Ior Testing Mathematics, Physics and Chemistry,
Proceeaings of International Symposium for Design ana Technology of Electronic Packages,
SIITME, 2006, p. 154, Iasi, Romania.
13. Ciobanu B., Radinschi I., The Electrical Properties oI sample Investigation by Computational
Programs, Proceeaings of 5
th
International Conference on Electromechanical ana Power Systems,
SIELMEN, October 68, 2005Chisinau, Rep. Moldova, 2. p. 218221.
14. Ciobanu B., Radinschi I., One computational program Ior predictions oI properties oI the yarns,
Proceeaings of International Symposium Present ana Perspective in Textile Engineering,
November 1012, 2005, Iasi, Romania, p. 563569.
15. www.gtpcc.org
16. www.pcmag.com; www.uscb2b.com; www.cambridge.org; www.uc.edu.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Physics with Maple 9.5
Irina Radinschi
1
, Brindusa Ciobanu
1
, Mircea Daniel Frunza
2
1
Department of Physics, 'Gh. Asachi` Technical University, Iasi, 700050, Romania
2
Faculty of Electronics ana Communications, 'Gh. Asachi` Technical University, Iasi, Romania
Summary
In this paper we sketch how computer algebra systems like Maple 9.5 can be usea
to helping the stuaents connecting physical concepts ana phenomena with powerful
mathematical formalisms ana graphical representations. We aemonstrate that a
computer algebra system like Maple is very useful for physics stuaies. The
experience in this area aemonstratea that computational physics, together with
traaitional theoretical ana experimental physics, is a reliable tool which helps
stuaents to explore a wiae variety of phenomena ana give them a aeeper
unaerstanaing of these topics.
Maple 9.5 proviaes a gooa environment for computation, a highlevel
programming language, text, graphics, ana animation. Maple 9.5 is also capable
of, ana usea for, highlevel computation by stuaents ana this making it an iaeal
computer assistea teaching ana learning tool. It is also a comprehensive
environment for exploring, teaching, ana applying mathematical methoas in
physics. Using Maple the stuaents will be unconstrainea by traaitional
mathematical ana physical limitations, ana they will also be capable of a better
unaerstanaing of computational methoas. Maple 9.5 for physics helps the stuaents
to aevelop their ability to learn physics, to solve aifficult problems with a symbolic
computing engine, ana to review the basic mathematics. Maple 9.5 assists the
stuaents for solving aifferential equations, aifferentiating, integrating, making
sums, finaing roots, plotting ana proaucing animations. In this work are presentea
some applications of Maple 9.5 in physics, which help the stuaents to briage the
gap between mathematical formalism ana physical phenomena.
Step by step we want to introauce Maple program into our courses for those who
plan to mafor in technical fielas. The main features of our course are the use of
computer program like Maple for teaching ana learning physics, ana for solving
physics problems. The stuaents can also use the computers outsiae class for further
instruction ana problem solving. In this way we want to take our stuaents beyona
the limitations of traaitional instruction ana make them to become familiar with
highlevel computer algebra systems.
KEYWORDS: Maple 9.5, computer algebra system, physics, physical concepts
and phenomena.
192 I. Radinschi, B. Ciobanu, M.D. Frunza
1. INTRODUCTION
Analyses oI current developments in physics teaching and learning have shown the
importance oI implementation oI highlevel computer algebra systems like Maple
7 in our courses. Maple is a mathematical application package that supports
symbolic and numeric computation and graphics. Maple 9.5 helps students
connecting physical concepts and phenomena with powerIul mathematical
Iormalisms and graphical representations. The extensive mathematical assistance,
symbolic manipulations, computational power and graphical abilities oI programs
like Maple 1114 can greatly help students to explore physical topics and
experiment with ideas without perIorming cumbersome calculations.
We have a longterm experience in computational physics 16, 810 and
particularly in using Maple 9.5 program, with applications in many Iields oI
physics. Our course will be designed to incorporate the use oI Maple Ior solving
diIIerential equations, diIIerentiating, integrating, making sums, Iinding roots,
plotting and producing animations.
Students use Maple V soItware to achieve a better conceptual understanding oI the
material oI the course while still gaining a good knowledge oI the methods oI
problemsolving. Further, this computational method allows a rapid learning oI
physics and is also a powerIul tool Ior solving problems.
2. PRESENTATION OF MAPLE 9.5 APPLICATIONS
2.1. Applications oI Maple 9.5 in physics
Maple 9.5 applications in physics gives students the opportunity to show what they
have learned by testing their understanding oI the concepts and phenomena, and
applying them to real situations. Further, it allows solving rapidly physics problems
without perIorming length calculations, plotting and producing animations. We
present some Maple 9.5 applications in physics which are implemented in our
courses.
2.2.1. Evaluation of speea ana aisplacement
~ ode1 : m*diII(x(t),t)4*t`2/3;
oae1 : m
a
a t
x t ( )
4
2
3
t
'Computational Civil Engineering 2007, International Symposium 193
~ ans1 : dsolve(ode1);
ans1 : x t ( )
4 t
3
9 m
C1
and
~ diII(4*t`3/(9*m),t);
4 t
2
3 m
2.2.2. Plotting the speea ana aisplacement
The commands oI Maple 9.5 Ior plotting 2d, Ior producing animations and Ior
plotting 3d Ior the speed are:
a) plotting 2d
~with(plots);
~plot(4*t`2/3,t0...10);
Figure 1. Graph 2d oI the speed
b) producing animations
~with(plots);
~animate(plot,A*(4*t`2/3),t0...10, A5...5);
194 I. Radinschi, B. Ciobanu, M.D. Frunza
Figure 2. Animation 2d graph oI the speed
c) plotting 3d
~with(plots);
~plot3d(4*t`2/(3*m),m0.1...1,t0...10);
Figure 3. Plot 3d oI the speed
'Computational Civil Engineering 2007, International Symposium 195
The commands oI Maple 9.5 Ior plotting 2d, Ior producing animations and Ior
plotting 3d (in this case we consider that the integrating constant is C
1
0) Ior the
displacement are
a) plotting 2d
~with(plots);
~plot(4*t`3/9,t0...10);
Figure 4. Plot 2d oI the displacement
b) producing animations
~with(plots);
~animate(plot,A*(4*t`3/9),t0...10, A5...5);
196 I. Radinschi, B. Ciobanu, M.D. Frunza
Figure 5. Animation 2d graph oI the displacement
c) plotting 3d
~with(plots);
~plot3d(4*t`3/(9*m),m0.1...1,t0...10);
Figure 6. Plot 3d oI the displacement
'Computational Civil Engineering 2007, International Symposium 197
2.2.3. Plotting the wave propagation
There are considered three wave Iunctions. The commands oI Maple 9.5 Ior
plotting the waves propagation in 2d and 3d cases are given below.
a) plotting 2d
~with(plots);
~I:4*sin(Pi/4*tPi/7);
~h:2*cos(Pi/4*tPi/3) ;
~g:0.7*sin(2*Pi/3*tPi/7);
~ plot(I(t),g(t),h(t)},t0...10);
Figure 7. Plot 2d oI the wave propagation
b) plotting 3d
~with(plots);
~I:4*sin(omega*tPi/7);
~g:0.7*sin(omega*Pi/3*tPi/7);
~h:2*cos(Pi/4*tPi/3) ;
~plot3d(I,g,h},omega0...Pi/4,t0...10);
198 I. Radinschi, B. Ciobanu, M.D. Frunza
Figure 8. Plot 3d oI the wave propagation
2.2.4. Plotting the kinetic ana potential energy
The commands oI Maple 9.5 Ior plotting the kinetic energy and the potential
energy (with respect oI SI units) are given bellow
~with(plots);
~plot3d(m*v`2/2,m0.1...1,v0...10);
Figure 9. Plot 3d oI the kinetic energy
~plot3d(k*x`2/2,k0.1...10,x1...100);
'Computational Civil Engineering 2007, International Symposium 199
Figure 10. Plot 3d oI the potential energy
We gave some examples which demonstrate the useIulness oI Maple 9.5 Ior
teaching and learning physics.
Maple 9.5 can also be used as a powerIul tool Ior solving physics problems. AIter
the students will become Iamiliar to most oI Maple commands they will be capable
to solve more complicated problems.
3. CONCLUSIONS
This paper points out some advantages oI using Maple 7, 1114 in studying
physics. We presented some examples Irom our course to demonstrate the
useIulness oI this highlevel computer algebra system. We used Maple 9.5 Ior
evaluating the speed and displacement in a particular case, Ior plotting 2d and 3d oI
the speed and displacement, and Ior producing animations. The program was also
used Ior plotting the wave propagation oI three waves in 2d and 3d cases, and Ior
plotting the kinetic energy and the potential energy in 3d case.
The main goals are to allow students the ability to learn physics, to perIorm length
calculations, to make graphical representations and introduce animations, and also
to solve physics problems using Maple 9.5. The applications will be introduced
step by step, Iirstly students have to learn how to handle with derivatives and
integrals, solving equations and making graphs. At a higher level, they will be able
to solve challenging and physically realistic problems.
200 I. Radinschi, B. Ciobanu, M.D. Frunza
ReIerences
1. Ciobanu B., Radinschi, I., One Computational Algorithm Ior Physics Modeling, Proceeaings of
5
th
International Conference on Electromechanical ana Power Systems, SIELMEN, Chisinau,
Rep. Moldova, October 68, 2005, pp. 222225.
2. Radinschi I., Ciobanu B., Testing Physics, Junimea Publishing House, Iasi, Romania, 2006.
3. Radinschi I., Ciobanu B., Physics for Engineers, Junimea Publishing House, Iasi, Romania, 2006.
4. Ciobanu B., Radinschi I., A Sequential Programming Algorithm a Tool Ior Study the Standing
Waves, Proceeaings of International Symposium IEEI 2006, Iasi, Romania, Bulletin oI IP Iasi,
Vol. LII(LVI), Fasc.5, pp. 129134
5. Ciobanu B., Radinschi I., Teaching and Learning Physics Ior Engineers by Aid Computers,
Proceeaings of International Symposium IEEI 2006, Iasi, Romania, Bulletin oI IP Iasi, Vol.
LII(LVI), Fasc.5, pp. 135140.
6. Radinschi I., Scripcariu L., Ciobanu B., Frunza M. D., Online TeachingwhileQuizzing Test,
Proceeaings of the 2
na
National Conference on Appliea Physics, June 910, Galati, Romania,
2006, in press Romanian Journal oI Physics
7. http://grtensor.org; http://www.maplesoIt.com/maplebooks.html.
8. Radinschi I., Frunza M. D., Ciobanu B., Online Virtual Model Ior Testing the Knowledge,
Proceeaings of INTED 2007, March 79, Valencia, Spain.
9. Radinschi I., Ciobanu B., Implementation oI Computational Methods in Physics Learning,
Proceeaings of 4
th
International Symposium Computational Civil Engineering 2006, Iasi,
Romania, 2006, pp. 251257.
10. Frunza M. D., Radinschi I., A Virtual Model Ior Testing Mathematics, Physics and Chemistry,
Proceeaings of International Symposium for Design ana Technology of Electronic Packages,
SIITME, 2006, p. 154, Iasi, Romania.
11. http://www.wIu.edu/physics/cel/maple.html
12. http://www.maplesoIt.com/applications/
13. wwwteaching.physics.ox.ac.uk
14. www.adsabs.harvard.edu
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Physics Studies Computational Methods, a Strong Connection
Irina Radinschi
1
, Brindusa Ciobanu
2
1
Department of Physics, 'Gh. Asachi` Technical University, Iasi, 700050, Romania,
raainschiyahoo.com
2
Department of Physics, 'Gh. Asachi` Technical University, Iasi, 700050, Romania,
bciobanu2003yahoo.com
Summary
The aim of this paper is to point out the importance of the connection between
theoretical physics ana computational methoas. At national ana international level,
consiaerable efforts have been maae for implementing the computational methoas
in the stuay of physics. This work is focusea on the use of Mathematica ana Maple
programs for physics stuaies. Because our applications are part of the algebraic
computations in physics, new calculation programs in Mathematica ana Maple are
conceivea. We also usea methoas of animatea graphs for illustrating the physics
phenomena. The calculations are performea with Mathematica ana Maple
programs. In some particular cases of our applications, these programs have
attachea the GrTensor packages. Many graphs were also maae using the
Mathematica ana Maple programs. This goal will be reachea easily establishing
gooa collaborations with specialists in the fiela from our country ana from abroaa.
On the other hanas, unaergraauate physics implies to work with stuaents ana in
this way they have the possibility to enlarge the circle of collaborations ana gather
a lot of information in the fiela ana in aafacent areas. Platforms as Mathematica
ana Maple have some aavantages as flexibility ana speea, ana more aavancea
graphical facilities. GrTensorII which is embeaaea in Maple is a package which
allows us carrying out length calculations ana evaluating the components of some
tensorial ana pseuaotensorial quantities. This is because it has the facility for
aefining new tensors. Furthermore, it allows the simplication of large terms. The
auration of the calculations is shorter comparea to that which usea other software
such as REDUCE. GrTensorM runs unaer Mathematica. GrTensorM has a large
number of preaefinea obfects, with automatic inaex generation ana facilitates the
creation of new obfects with a compact intuitive interactive aefinition facility. An
important step is the aevelopment of the existing methoas, ana in this light we
intena to enlarge the areas of stuay by using the Mathematica ana Maple programs
to more classes of applications. In this way a powerful connection between physics
stuaies ana computational methoas is establishea.
KEYWORDS: physics studies, computational methods, Mathematica, Maple,
GrTensorII, GrTensorM.
202 I. Radinschi, B. Ciobanu
1. INTRODUCTION
The connection between the physics studies and the computational methods is one
oI the most important issues at national and international level. The mathematical
methods implemented in our physics course represent the main way oI instruction
and provide the skills needed Ior advanced studies in physics. Mathematica and
Maple have the potential to strongly impact the way problem solving methods are
taught in undergraduate physics and the curriculum oI such courses. In order to
perIorm length calculations, evaluate the components oI some tensorial and
pseudotensorial quantities and make graphs we use powerIul programs like
Mathematica and Maple. In physics we want to use more animated graphs Ior
illustrating the phenomena. Nowadays, physics studies are not allowed without
using adequate soItware. Because the use oI good soItware is very important, this
is the reason why specialists in the Iield Irom our country and Irom abroad have
been consulted. Further, special attention was paid Ior involving our students in
this interactive study: physical phenomenacomputational methods.
We consider that this work can be extended to many physical situations and a lot oI
inIormation in the Iield can be achieved. Because the study oI physics requires in
many cases making graphs and carrying out calculations which can be perIormed
using computer algebra methods, we strongly recommend programs like
Mathematica and Maple.
We have a longterm experience in computational physics 13, 57 and
particularly in using these programs, with applications in many Iields oI physics.
Our course will be designed to incorporate the use oI the Mathematica and Maple
programs as the primary means oI teaching mathematical methods Ior solving
physics problems. Further, we want to understand how this course impacts the
students` understanding oI the course material, and to what extent Mathematica and
Maple can be reliable problem solving tool Ior students at a similar level in their
study oI physics. This is an important point, because one oI the main problems that
would arise is how will be the course instruction aIIected by the use oI these two
computer programs and what are the strengths and weaknesses oI Mathematica and
Maple as teaching and learning tools.
These studies imply to work with students, and in this way they have the possibility
to enlarge the circle oI collaborations and gather a lot oI inIormation in the Iield
and in adjacent areas. Further, the computational methods allow a rapid learning oI
physics and our viewpoint is they can succesIully replace the traditional methods.
We hope that our students will Ieel conIident and they will be able to handle with
the Mathematica and Maple programs, in order to easily and rapidly solve and
explore most problems they encounter in their physics studies, and enrich their
knowledge.
'Computational Civil Engineering 2007, International Symposium 203
2. CONNECTION BETWEEN PHYSICS STUDIES AND
COMPUTATIONAL METHODS
2.1. Presentation oI Mathematica and Maple programs
As we pointed out in Introduction, the study oI physics requires adequate soItware.
In recent years computer algebra systems have become increasingly powerIul tools
in teaching physics. The extensive mathematical assistance, symbolic
manipulations, computational power and graphical abilities oI programs like
Mathematica and Maple 4 can greatly help students to explore physical topics and
experiment with ideas without perIorming cumbersome calculations. We consider
that we are able to obtain reliable results in teaching and learning physics, and this
is due to the good connection between the physics studies and the methods oI
computer calculation and graphic simulation. The applications which are obtained
Ior many physical situations also allow to propose some useIul suggestions Ior
Iuture studies. Concerning our courses and seminars we want to perIorm the
calculations with Mathematica and Maple. Many graphs will also be made using
the Mathematica and Maple programs. Special attention will be paid to animated
graphs Ior illustrating the physics phenomena. In the whole process oI teaching and
learning physics we will take into account that many oI our students are novice
users oI these programs, and will choose the adequate level oI applications.
PlatIorms as Mathematica and Maple have some advantages as Ilexibility and
speed, and more advanced graphical Iacilities. GrTensorII is a package which
allows calculating the components oI tensorial and pseudotensorial quantities, and
is embedded in Maple. This is because it has the Iacility Ior deIining new tensors.
Furthermore, it allows the simpliIication oI large terms. GrTensorM runs under
Mathematica. GrTensorM has a large number oI predeIined objects, with automatic
index generation and Iacilitates the creation oI new objects with a compact intuitive
interactive deIinition Iacility. We have a longterm experience in using these
programs in physics teaching and learning, as in approached Iields as mathematical
physics and computational physics.
2.2. Advantages in using Mathematica and Maple Ior teaching and learning
physics
We sketch the advantages oI these powerIul programs in teaching and learning
physics. It is obviously that Mathematica and Maple program have to satisIy some
requirements which indicated them as powerIul tools. These requirements are:
1) to be a good mathematical tool Ior solving equations, carrying out length
calculations, and writing tensorial and pseudotensorial quantities,
204 I. Radinschi, B. Ciobanu
2) Mathematica and Maple computer algebra systems have to allow the students to
easily and quickly change the mathematical models and the parameters and then
generate new answers,
3) giving the possibility to be applied Ior a large number oI applications and
yielding meaningIul results,
4) Iacilitating to make predictions and analogies between diIIerent physical
situation,
5) giving the possibility to make graphs and animated graphs in order to illustrate
the physical phenomena,
6) having the entire documentation electronically searchable.
These Ieatures make Mathematica and Maple ideal Ior use in upperlevel physics
courses Ior students completing assignments.
3. APPLICATIONS OF MATHEMATICA AND MAPLE
3.1. Maple 9.5 applications
We present some examples Irom a typical calculus course. Further, some graphs
are inserted. The topics that we have chosen are solving diIIerential equations,
evaluating the Iirst and second derivative oI some physical quantities and making
graphs and animated graphs. The Iollowing examples, solutions and graphs
demonstrate such cases.
3.1.1. Maple commanas for simple harmonic oscillations ana aampea harmonic
oscillations
Firstly, we consider the simple harmonic oscillations and damped harmonic
oscillations and we give the Maple sequences Ior dsolve solve ordinary
diIIerential equations (ODEs). The Iirst example is using the command
dsolve(ode1) which yields the roots oI ode1. In the second case we use the
command dsolve(ode2) which yields the roots oI ode2.
~ ode1 : diII(x(t),t,t)omega`2*x(t);
oae1 . (aiff(x(t),(t, 2)))omega2*x(t)
~ ans1 : dsolve(ode1);
ans1 : x t ( ) C1 sin
w
t ( ) C2 cos
w
t ( )
'Computational Civil Engineering 2007, International Symposium 205
and
~ ode2 : diII(x(t),t,t)2*gamma*diII(x(t),t)omega`2*x(t);
ode2 : (diII(x(t),(t, 2)))2*gamma*(diII(x(t), t))omega`2*x(t)
~ ans2 : dsolve(ode2);
ans2 : x(t) C1*exp((gamma(gamma`2omega`2)`(1/2))*t)C2*exp((
gamma(gamma`2omega`2)`(1/2))*t)
3.1.2. Maple commanas for evaluating the first ana secona aerivative of
aisplacement for the simple harmonic oscillations
Secondly, we give some examples oI evaluating the Iirst and second derivative oI
displacement Ior the simple harmonic oscillations. This will give the velocity and
acceleration Ior the simple harmonic oscillations.
~ diII(A*sin(omega*tphi),t);
A cos
w
t
f
( )
w
~ diII(A*sin(omega*tphi),t,t);
A
sin
w
t
f
( )
w
2
3.1.3. Maple commanas for plotting 2a ana 3a for simple harmonic oscillations
Thirdly, we make the graphs Ior simple harmonic oscillations. We make a 2
dimensional graph and a 3 dimensional graph.
For the 2 dimensional case we plot the expression Ior the displacement with the
amplitude A2, the pulsation Pi/4 and the initial phase Pi/3 and t0.10, and
we obtain the graph Irom the Figure 1.
Figure1. Simple harmonic oscillations, the 2d graph
206 I. Radinschi, B. Ciobanu
~ with(plots);
~ plot(2*sin(Pi/4*tPi/3),t0...10);
For the 3 dimensional case we plot the expression Ior the displacement with the
amplitude A0.1.1, the pulsation Pi/4, the initial phase Pi/3, and t0.10.
We give the sequences bellow and in the Figure 2 we plot the displacement Ior
diIIerent values oI the amplitude and time.
~ plot3d(A*sin(Pi/4*tPi/3),A0.1...1,t0...10);
Figure 2. Simple harmonic oscillations, the 3d graph
For the graphs the selection animate or animate3d can be selected.
3.2. Mathematica 5.1 applications
We give some examples connected to the topics oI solving diIIerential equations,
evaluating the Iirst and second derivative oI some physical quantities and making
graphs and animated graphs.
3.2.1. Mathematica commanas for simple harmonic oscillations ana aampea
harmonic oscillations
We present the simple harmonic oscillations and damped harmonic oscillations and
we give the Mathematica sequences Ior DSolve. We obtain
'Computational Civil Engineering 2007, International Symposium 207
ln1 DSolvex''t `2 xt, x, t
Out1 xFunctiont},C1 Cos tC2 Sin t}}
and
ln2 DSolvex''t 2 x't `2 xt, x, t
Out2 xFunctiont},  2   1 
) ( ) (
2 2 2 2
C e C
t t e o o e o o +
+ e
3.2.2. Mathematica commanas for evaluating the speea ana acceleration for the
simple harmonic oscillations
In this case the Iirst and second derivatives oI displacement Ior simple harmonic
oscillations are computed. They will give the velocity and acceleration Ior the
simple harmonic oscillations.
ln1 DA*Sin*t,t
Out1A Cos*t
ln2 DA**Cos*t
Out3 A `2 Sin*t
3.2.3. Mathematica commanas for plotting 2a ana 3a for simple harmonic
oscillations
c) We plot the displacement Ior the simple harmonic oscillations. We make a 2
dimensional graph and a 3dimensional graph, and use the same values Ior the
amplitude, pulsation, initial phase and time parameters as in the case oI Maple
program.
For the 2dimensional case we obtain the graph Irom the Figure 3.
ln1 Plot2*SinPi/4*tPi/3,t,1,10}, PlotStyleHue.1, BackgroundRGBColor
0.94174`,1.`,1.`, AxesLabelt, }
Out1Graphics
208 I. Radinschi, B. Ciobanu
2 4 6 8 10
t
2
1
1
2
Figure 3. Simple harmonic oscillations, the 2d graph
There are many options Ior the color oI the curve and Ior the color oI the
background. These can be set by using the option Graphics`Colors` and choosing
Irom there the colors. The builtin Mathematica color directives RGBColor, Hue,
and CMYKColor handle the most common systems. This package gives color
directives using other standard systems. For axes we use the option AxesLabel and
we plot the wave Iunction on Yaxis against time t on Xaxis.
The 3dimensional case is illustrated in Figure 4. We plot the expression Ior the
wave Iunction using the same option Ior RGBColor as in the 2d case. For axes we
use the option AxesLabel and we plot the wave Iunction on Zaxis, against
amplitude A on Xaxis and time t on Yaxis.
ln2 Plot3DA*SinPi/4*tPi/3, A,0.1,1}, t,1,10}, BackgroundRGBColor
0.94174`,1.`,1.`, AxesLabelA,t,}
Out2 SurIaceGraphics
0.2
0.4
0.6
0.8
1
A
2
4
6
8
10
t
1
0.5
0
0.5
1
0.2
0.4
0.6
0.8
A
Figure 4. Simple harmonic oscillations, the 3d graph
'Computational Civil Engineering 2007, International Symposium 209
ln3 ShowI,MeshFalse
Out3 SurIaceGraphics
0.2
0.4
0.6
0.8
1
2
4
6
8
10
1
0.5
0
0.5
1
0.2
0.4
0.6
0.8
Figure 5. Simple harmonic oscillations, the 3d graph
Further, the SelectionAnimate or Sound Generation can be selected. Mathematica
is a powerIul computational tool which provides a large spectrum oI choices Ior
SelectionAnimate and Ior Sound Generation. It also provides simulations oI
relevant concepts.
In some particular cases oI our applications, Mathematica and Maple programs
have attached the GrTensor packages. These are useIul Ior evaluating the
components oI some tensorial and pseudotensorial quantities. Important
calculations can be carried out with GrTensorII and GrTensorM Ior studies
concerning the general relativity. These packages allow evaluating the determinant
oI a metric, the covariant and contravariant components oI the metric tensor, the
components oI the Einstein and Ricci tensor, and also to deIine new tensors and
pseudotensors.
We also strongly encouraged the students to use either Mathematica or Maple to
work on the homework problems.
4. CONCLUSIONS
In this paper, we examined the connection between physics studies and Computer
Algebra Systems. Teaching and learning undergraduate physics using Mathematica
5.1 and Maple 9.5 is described through examples. Both Mathematica 5.1 and
Maple 9.5 are very useIul Ior learning, teaching, and carrying out research in
physics.
210 I. Radinschi, B. Ciobanu
We presented some examples Irom our course to demonstrate the useIulness oI
these computer programs. We have chosen the cases oI simple harmonic
oscillations and damped harmonic oscillations. Mathematica and Maple are used
Ior solving the equation oI motion in these two cases, Ior evaluating the velocity
and acceleration in the case oI simple harmonic oscillations and Ior plotting the
displacement in both cases. We work with Mathematica 5.1 and Maple 9.5. For
evaluating tensorial and pseudotensorial quantities the GrTensorII and GrTensorM
packages are attached to Mathematica and Maple programs, respectively.
We will implement Mathematica and Maple as teaching and learning tools in our
course oI physics Ior civil engineering taking into account the advantages oI these
programs.
ReIerences
1. Ciobanu B., Radinschi I., One Computational Algorithm Ior Physics Modeling, Proceeaings of 5
th
International Conference on Electromechanical ana Power Systems, SIELMEN, Chisinau, Rep.
Moldova, October 68, 2005, pp. 222225.
2. Radinschi I., Ciobanu B., Implementation oI Computational Methods in Physics Learning,
Proceeaings of 4
th
International Symposium Computational Civil Engineering 2006, Iasi,
Romania, 2006, pp. 251257.
3. Frunza M. D., Radinschi I., A Virtual Model Ior Testing Mathematics, Physics and Chemistry,
Proceeaings of International Symposium for Design ana Technology of Electronic Packages,
SIITME, 2006, p. 154, Iasi, Romania.
4. http://grtensor.org; http://www.maplesoIt.com/maplebooks.html.
5. Ciobanu B., Radinschi I., Teaching and Learning Physics Ior Engineers by Aid Computers,
Proceeaings of International Symposium IEEI 2006, Iasi, Romania, Bulletin oI IP Iasi, Vol.
LII(LVI), Fasc.5, pp. 135140.
6. Radinschi I., Scripcariu L., Ciobanu B., Frunza M. D., Online TeachingwhileQuizzing Test,
Proceeaings of the 2
na
National Conference on Appliea Physics, June 910, Galati, Romania,
2006, in press Romanian Journal oI Physics
7. Ciobanu B., Radinschi I., A Sequential Programming Algorithm a Tool Ior Study the Standing
Waves, Proceeaings of International Symposium IEEI 2006, Iasi, Romania, Bulletin oI IP Iasi,
Vol. LII(LVI), Fasc.5, pp. 129134.
8. Radinschi I., Frunza M. D., Ciobanu B., Online Virtual Model Ior Testing the Knowledge,
Proceeaings of INTED 2007, March 79, Valencia, Spain.
9. Radinschi I., Ciobanu B., Testing Physics, Junimea Publishing House, Iasi, Romania, 2006.
10. Radinschi I., Ciobanu B., Physics for Engineers, Junimea Publishing House, Iasi, Romania, 2006.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Computer Algebra System Ior EnergyMomentum Localization
Irina Radinschi
1
, Brindusa Ciobanu
1
1
Department of Physics, 'Gh. Asachi` Technical University, Iasi, 700050, Romania,
raainschiyahoo.com
2
Department of Physics, 'Gh. Asachi` Technical University, Iasi, 700050, Romania,
bciobanu2003yahoo.com
Summary
The aim of this paper is to sketch how computer algebra systems like Maple 9.5
with the GrTensorII package is usea for performing calculations for energy
momentum locali:ation. This computer algebra platform Maple ana GrTensorII
allows aefining new obfects like tensorial ana pseuaotensorial quantities, ana for
this purpose there are createa proceaures with GrTensorII 1.50 version. As the
software Maple allows the use of complicatea graphical methoas, they were usea
to make the graphical representations. Maple has some aavantages as flexibility
ana speea, ana also aavancea graphical facilities as animation. GrTensorII which
is embeaaea in Maple is a package which allows us carrying out length
calculations ana evaluating the components of some tensorial ana pseuaotensorial
quantities. This is because it has the facility for aefining new tensors ana
pseuaotensors. Furthermore, it allows the simplification of large terms ana gives
the possibility to evaluate the energy ana momentum aensity components for a
given metric. The auration of the calculations is shorter comparea to that which
usea other software such as REDUCE. At international level, consiaerable efforts
have been aone to fina a generally acceptea expression for the energymomentum
aensity. This incluaes working with superenergy tensors, energymomentum
complexes, quasilocal expressions ana teleparallel quantities. Concerning the
aevelopment of the existing theories ana methoas, we mention the enlargement of
the areas by applying the energymomentum complexes ana their teleparallel
versions in the case of aifferent 2, 3 ana 31 aimensional spacetimes. Special
attention has been paia to the connections between the results ana furthermore to
the connections with the teleparallel theory of gravitation. In this context, we
present some programs elaboratea in GrTensorII which have been usea for
evaluating the energy ana momentum components for the Einstein, LanaauLifshit:
ana Moller energymomentum complexes ana make an investigation of the
connection between physics stuaies ana computational methoas. We point out
several important results that the energymomentum complexes have yiela for
aifferent geometries ana some scientific perspective for future research in this
area. Further, special attention was also paia for involving our master stuaents in
this interactive stuay. physical phenomenacomputational methoas.
KEYWORDS: computer algebra system, Maple, GrTensorII, energymomentum
localization.
212 I. Radinschi, B. Ciobanu
1. INTRODUCTION
It is wellknown that a computer algebra system like Maple is very useIul Ior
physics studies, and GrTensorII package 1 can be used Ior perIorming length
calculations and evaluating the components oI some tensorial and pseudotensorial
quantities. Further, GrTensorII is a reliable tool Ior evaluating the energy and
momentum components oI the energymomentum complexes oI Einstein 2,
LandauLiIshitz 3 and Moller 4 and gives a deeper understanding oI these
prescriptions.
The diIIerent attempts at constructing an energymomentum density don't give a
generally accepted expression. The 'center oI general relativity is one oI the most
interesting and challenging ideas in modern science, the one that gravity is the
geometry oI curved Iourdimensional spacetime. In conclusion, gravity is
geometry. For localizing the energy there have been used many deIinitions,
including superenergytensors, the energymomentum complexes, the quasilocal
quantities, and the teleparallel theory oI gravitation. The energymomentum
complexes deIinitions include the Einstein 2, LandauLiIshitz 3, and Moller 4
prescriptions. These prescriptions have been criticized because oI their drawback,
they are coordinate dependent. Except the Moller energymomentum complex
which allows evaluating the energy distribution in any coordinate system, the other
prescriptions give meaningIul results only iI the calculations are restricted to quasi
Cartesian coordinates. There are doubts that these pseudotensorial deIinitions yield
good results Ior a given spacetime. In this light, some remarks are needed: 1) the
energymomentum complexes (ELL) are noncovariant, coordinate dependent
expressions and yield acceptable results only the calculations are carried out in
quasiCartesian coordinates, 2) only the Moller energymomentum complex
enables one to calculate the energy in any coordinate system and 3) this agrees with
the equivalence principle which states that gravity cannot be detected at a point.
Chang, Nester and Chen 5 showed that the energymomentum complexes are
actually quasilocal and legitimate expressions Ior the energymomentum. They
concluded that there exist a direct relationship between energymomentum
complexes and quasilocal expressions because every energymomentum
complexes is associated with a legitimate Hamiltonian boundary term. In the last
decades the issue oI the energymomentum localization by applying the energy
momentum complexes was reopened and many interesting results have been
obtained 6, which demonstrate that these prescriptions are powerIul concepts Ior
energymomentum localization. All these considerations point out the signiIicance
oI these prescriptions and stress the useIulness oI Maple program and GrTensorII
package which allow calculating the energy and momentum components Ior a
given spacetime, and also making graphical representations.
'Computational Civil Engineering 2007, International Symposium 213
2. GRTENSORII PACKAGE AND ENERGYMOMENTUM
COMPLEXES
2.1. Energymomentum complexes
2.2.1. General comments
As we pointed out in Introduction, the lack oI the pseudotensorial prescriptions is
connected to the dependence oI the energymomentum complexes on the
coordinate systems. Only the Moller prescription allows evaluating the energy and
momentum in any coordinate system. The other energymomentum complexes oI
Einstein and LandauLiIshitz yield reliable results iI the calculations are carried out
in quasiCartesian coordinates. In this context, another important problem appears.
According to recent studies, although the energymomentum complexes behave
under general coordinate transIormations like nontensorial objects, the local
conservation laws obtained by them hold in all coordinate systems. The Einstein
and LandauLiIshitz energymomentum complexes give the same energy
distribution as the Penrose deIinition Ior a general nonstatic spherically symmetric
metric oI the KerrSchild class 7. In the case oI the most general nonstatic
spherically symmetric metric these deIinitions disagree concerning the expression
Ior the energy distribution 7.
2.2.2. Maple ana GrTensorII platform for the Einstein, LanaauLifshit: ana
Moller prescriptions
The GrTensorII programs Ior the Einstein, LandauLiIshitz and Moller energy
momentum complexes are given below. Further, examples Ior a class oI solutions
belonging to the string theory are presented 810.
GrTensorII Ior the Einstein energymomentum complex:
~readlib(grii):
~grtensor();
~ grdeI(`Hk`m`l`n}:(detg)*(g`k`m}*g`l`n}g`l`m}*g`k`n})`):
Created deIinition Ior H(dn,up,up,up)
~ grdeI(`H1i`k`l}:1/sqrt(detg)*gi m}*Hk`m`l`n,n}`):
Created a deIinition Ior H(dn,up,up,up,pdn)
Created deIinition Ior H1(dn,up,up)
214 I. Radinschi, B. Ciobanu
GrTensorII Ior the LandauLiIshitz energymomentum complex:
~readlib(grii):
~grtensor();
~ grdeI(`S3`m`j`n`k}:detg*(g`m`n}*g`j`k}g`m`k}*g`j`n})`):
Created deIinition Ior S3(up,up,up,up)
~ grdeI(`S4`m`n`k}:S3`m`j`n`k,j}`):
Created a deIinition Ior S3(up,up,up,up,pdn)
Created deIinition Ior S4(up,up,up)
GrTensorII Ior the Moller energymomentum complex:
~ grdeI(`Mi`k`l}:sqrt(detg)*(gi m,n}gi n,m})*g`k`n}*g`l`m}`):
Created deIinition Ior M(dn,up,up)
The Iirst two commands in these procedures are Ior starting the GrTensorII
package. The command grdeI is used Ior deIining the required energymomentum
complexes.
The Einstein, LandauLiIshitz and Moller energymomentum complexes are used
Ior evaluating the energy and momentum components Ior the spacetime which
describes a magnetic stringy solution. The metric is used in the initial Iorm Ior the
Moller prescription and is transIormed in quasiCartesian coordinates Ior the
Einstein and LandauLiIshitz prescriptions. The Iollowing results are obtained:
~ qload(kar1);
Calculating ds Ior kar1 ... Done. (0.000000 sec.)
~ grcalc(M(dn,up,up)):
Calculating g(dn,dn,pdn) Ior kar1 ... Done. (1.000000 sec.)
Calculating detg Ior kar1 ... Done. (0.000000 sec.)
Calculating g(up,up) Ior kar1 ... Done. (0.000000 sec.)
Calculating detg Ior kar1 ... Done. (0.000000 sec.)
Calculating M(dn,up,up) Ior kar1 ... Done. (0.000000 sec.)
~ grdisplay(M(dn,up,up));
'Computational Civil Engineering 2007, International Symposium 215
~ qload(kar);
Calculating ds Ior kar ... Done. (0.000000 sec.)
~ grcalc(H1(dn,up,up)):
Calculating g(dn,dn,pdn) Ior kar ... Done. (1.000000 sec.)
Calculating detg Ior kar ... Done. (0.000000 sec.)
Calculating g(up,up) Ior kar ... Done. (0.000000 sec.)
Calculating detg Ior kar ... Done. (0.000000 sec.)
Calculating H1(dn,up,up) Ior kar ... Done. (0.000000 sec.)
~ grdisplay(H1(dn,up,up));
216 I. Radinschi, B. Ciobanu
~ qload(kar);
Calculating ds Ior kar ... Done. (0.000000 sec.)
~ grcalc(S3(up,up,up,up,pdn),S4(up,up,up)):
Calculating g(dn,dn,pdn) Ior kar ... Done. (1.000000 sec.)
Calculating detg Ior kar ... Done. (0.000000 sec.)
Calculating g(up,up) Ior kar ... Done. (0.000000 sec.)
Calculating detg Ior kar ... Done. (0.000000 sec.)
Calculating S3(up,up,up,up,pdn) Ior kar ... Done. (0.000000 sec.)
Calculating S4(up,up,up) Ior kar ... Done. (0.000000 sec.)
~ grdisplay(S4(up,up,up));
'Computational Civil Engineering 2007, International Symposium 217
The command qload in these procedures is Ior loading the metric, grcalc is
designated Ior calculating the energy and momentum components Ior the
aIorementioned energymomentum complexes, and the command grdisplay is used
Ior displaying the results. The energy components oI the Einstein, LandauLiIshitz
and Moller energymomentum complexes are involved in calculating the
expression Ior energy in the case oI the aIorementioned magnetic stringy solution.
For plotting the expressions Ior energy obtained in the Einstein, LandauLiIshitz
and Moller prescriptions on the same graph the next Maple 9.5 commands are
used. We denote with I, g and h the expressions Ior energy in the LandauLiIshitz,
Einstein and Moller prescriptions, respectively. In all expressions we consider the
case M1.
~ with(plots);
~ I:r/2*(r*Q`22*r2*Q`2)/((r2)*(rQ`2));
~ g:1/2*(r*Q`22*r2*Q`2)/(rQ`2);
~ h:r/2*(2Q`2)/(rQ`2);
~ plot3d(I,g,h},r0...100,Q0.1...1);
Figure 1. Energy on Zaxis plotted against r on Xaxis and Q on Yaxis
218 I. Radinschi, B. Ciobanu
3. CONCLUSIONS
In the last decade the problem oI the energymomentum complexes has been re
opened and acceptable results have been obtained 6 Ior diIIerent spacetimes.
Many authors used diIIerent energymomentum prescriptions to investigate the
energy distributions Ior a given geometry. They demonstrated that in many cases
several energymomentum complexes yield the same and acceptable result. Chang,
Nester and Chen 5 showed that the energymomentum complexes are actually
quasilocal and legitimate expressions Ior the energymomentum. Their idea
supports the energymomentum complexes and the role which these are playing in
energymomentum localization. Furthermore, important studies have been done
about the new idea oI quasilocal approach Ior energymomentum complexes 11
12 and a large class oI new pseudotensors connected to the positivity in small
regions have been constructed 13. In this light, the quasilocal quantities are
associated with a closed 2surIace (L. B. Szabados, 12 and
http://relativity.livingreviews.org/Articles/lrr20044/). The Hamiltonian boundary
term determines the quasilocal quantities Ior Iinite regions and the special quasi
local energymomentum boundary term expressions correspond each oI them to a
physically distinct and geometrically clear boundary condition 14.
Maple and GrTensorII procedures are used Ior evaluating the energy and
momentum components oI the Einstein, LandauLiIshitz and Moller energy
momentum complexes. Some oI our results are presented, and the metric under
consideration describes a magnetic stringy solution. GrTensorII yields the energy
and momentum components which are involved in the expression Ior energy and
momentum densities. The expressions Ior energy in the Einstein, LandauLiIshitz
and Moller prescriptions are plotted on the same graph using Maple 9.5 commands
Ior 3d graphs. The computer algebra platIorm Maple 9.5 and GrTensorII package is
used Ior perIorming the calculations and graphical representations, and in this way
a powerIul connection between physics studies and computational methods is
established. The young master students play an important role in this activity and
their tasks are clearly deIined. They are mostly needed to clariIy the problems
related to theoretical physics and computational and mathematics demands. Their
contributions are meant to be strictly related to their dissertation papers. They also
have the possibility to enlarge the circle oI collaborations abroad and gather a lot oI
inIormation in the Iield and the adjacent topics.
ReIerences
1. http://grtensor.org; http://www.maplesoIt.com/maplebooks.html.
2. Einstein, A., Preuss. Akaa. Wiss. Berlin 47, 778, 1915; Addendumibid. 47, 799, 1915; Trautman,
A. in Gravitation: an Introauction to Current Research, ed. L. Witten, Wiley, New York, p. 169,
1962.
3. Landau, L. D., LiIshitz, E. M. The Classical Theory of Fielas, Pergamon Press, p. 280, 1987.
'Computational Civil Engineering 2007, International Symposium 219
4. Moller, C. Ann. Phys. (NY) 4, 347, 1958.
5. Chang, ChiaChen, Nester, J. M., Chen, ChiangMei, Phys. Rev. Lett. 83, 1897, 1999.
6. Aguirregabiria, J. M., Chamorro, A., Virbhadra, K. S., Gen. Rel. Grav. 28, 1393, 1996; Vagenas,
E. C., Moa. Phys. Lett. A21, 1947, 2006; ShariI M., Fatima, Tasnim, Astrophys. Space Sci. 302,
217, 2006; Gad, Ragab M., Gen. Rel. Grav. 38, 417, 2006; Grammenos, Th., Moa. Phys. Lett.
A20, 1741, 2005; Radinschi, I., Moa. Phys. Lett. A15, Nos. 11&12, 803, 2000; Radinschi I.,
Grammenos, Th., Int. J. Moa. Phys. A21, 2853, 2006; Aydogdu, O., Salti, M., Astrophys. Space
Sci. 302, 61, 2006.
7. Virbhadra, K. S., Phys. Rev. D60, 104041, 1999.
8. Radinschi, I,. Rom. J. Phys. Vol.50, Number 12, 57, 2005.
9. Radinschi I., Ciobanu B., International ConIerence FiIty Years oI Romanian Astrophysics,
Bucharest, September 2730, AIP, 2006
10. Radinschi, I., Yang, IChing, On the Energy of Stringy Black Holes, New Developments in String
Theory Research, ed. Susan A. Grece, Nova Science Publishers, Inc New York, U.S.A., p. 17,
2006.
11. Chang, ChiaChen, Nester, J. M., Class. Quant. Grav. 16, 1279, 1999.
12. Chang, ChiaChen, Nester, J. M., Chen, Grav. Cosmol. 6, 257, 2000; Szabados, L. B., Living.
Rev. Relativity 7, 4, 2004.
13. So, Lau Loi, Nester, James M., Chen, Hsin, grqc/0605150, to appear in Proceedings oI the 7th
International ConIerence on Gravitation and Astrophysics; Deser, S., Franklin, J.S., Seminara, D.,
Class. Quant. Grav. 16, 2815, 1999.
14. Nester, James M., Class. Quant. Grav. 21, S261, 2004; Chen, ChiangMei, Nester, J. M., Tung,
RohSuan, Phys. Rev. D72, 104020, 2005.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
The state oI stresses in RC beams with installation holes located
in compressed zone
Przemyslaw Siwiec
1
, Sebastian To
1
1
Institute of Builaing Engineering, Wroclaw University of Technology
Summary
This paper presents a question of influence of installation holes locatea in the
compressea :one of the reinforcea concrete beam on functioning. In oraer to
research the actual functioning of the element a numerical moael basea on Finite
Element Methoa has been createa.
Two inaepenaent analyses were performea in which linear ana nonlinear
constitutive relations for concrete ana for steel have been usea. The results
acquirea have been comparea ana aiscussea.
KEYWORDS: RC beam, nonlinear FEM analysis, nonlinear material, numerical
model, numerical analysis.
1. INTRODUCTION
Designing process oI reinIorced concrete structures (RCS) is based on current state
oI knowledge. Codes oI Practice (CP) in many countries are based on this
knowledge. It can be considered that iI an element is designed according to
requirements oI the CP and it IulIills requirements oI border states (in both phases:
service and mount) as well as construction requirements, there is certain probability
that it will be saIe and will properly carry adequate loads.
The problem occurs when there is need to speciIy the state oI saIety oI reinIorced
concrete element not designed in accordance with requirements oI the CP and
especially when it doesn`t IulIill construction requirements. Then it is necessary to
perIorm a complete analysis oI stresses and strains according to rules oI mechanics.
And here a problem arises: according to rules oI a linear analysis it is not possible
to speciIy correctly the state oI stresses in reinIorced concrete elements in the II
(2nd) phase. So there is a need to make the analysis based on physical and
geometrical nonlinearities in reinIorced concrete.
'Computational Civil Engineering 2007, International Symposium 221
2. NONLINEAR MODEL
Fundamental physical stressstrains relations based on assumption oI linear elastic
relation come Irom the Hook`s law. The structure deIormations are not
proportional to the loads and this irregularity in material oI a structure. This kind oI
material is called plastic.
From the point oI view oI theory oI elasticity, material turns to undetermined state
aIter exceeding limit value oI stresses (criterion oI plasticity) deIined by
constitutive relationships between stresses and strains. There are many criteria oI
plasticity in complex state oI stresses 1,2,3. The most popular are: Trasca and
Von Mises (MHV) yield criterion Ior describing materials such as steel and
aluminum; and Coulomb and DrucknerPrager yield criterion Ior describing such
materials as concrete, rocks, soils and sands. In literature, Ior describing plasticity
in reinIorced concrete the most popular are Von Mises criterion (MHV) Ior steel
and DrucknerPrager criterion Ior concrete under compression. The most
divergences among researchers oI the problem have been caused by concrete under
tension. The solutions used 4.5.6.7 are mainly connected with adopted calculation
techniques. In this paper the Finite Element Method as a basic analytical tool has
been used. Within a Iramework oI this method the smeared crack approach Ior
modeling concrete under tension has been used.
Basic assumption oI theory oI plasticity in FEM Iormulation is decomposition oI
strains to elastic and plastic parts:
(1)
p e
c c c + =
Such a model is called elastoplastic. The most important consequence oI adopting
this material model relation between total stresses Ior certain time t and total strains
Ior time t, is additional parameter a Iunction oI stresses and strain history. This
Iunction is taken into account by internal parameter oI state , governed by a
speciIic evolution law. Properties oI elastoplastic material are determined by
assumptions:
the elastic relation between total stresses and strains is given by:
(2)
e
c o D =
the yield condition determines the state oI stresses when the plastic Ilow is
initiated. This condition is determined by vector Iunction oI stresses and by
the parameter oI state M, according to relationship:
0 ) , ( = k o f (3)
the Ilow rule speciIies the plastic strain rate vector as a Iunction oI the state
oI stresses in relationship:
222 F. Lastname, S. Author, T. Author
_
=
c
c
=
n
f
f
f
p
g
1
o
c
(4)
Ior n Iunctions oI plastic potential g
f
which can also be considered as a Iunction oI
the stress vector and state parameter g
f
(U,M).
The plastic multiplier is determined Irom KuhnTucker conditions:
f
(5)
=
>
s
0
0
0
f
f
f
f

)
t
t t
2
) ( ) (  ) , (
where *` denotes the complex conjugate, g is the short time window, x(t) is the
signal, t is the time location parameter, f is Irequency and t is time. In the two
dimensional timeIrequency joint representation, the vertical stripes oI the complex
valued STFT coeIIicients STFT(t, f) correspond to the Fourier spectra oI the
windowed signal with the window shiIted to given timest. The main disadvantage
oI linear timeIrequency transIorm is that the time Irequency resolution is limited
to the Heisenberg bound. This is due to the imposition oI local time window g(t). II
this window is more resolved in time, the Irequency resolution suIIers because the
eIIective width oI its Fourier transIorm G(f) increases, and viceversa.
Quadratic (nonlinear) methods present the second Iundamental class oI time
Irequency distributions. Quadratic methods are based upon estimating an
instantaneous power (or energy) spectrum using a bilinear operation on the signal
x(t) itselI. The class oI all quadratic timeIrequency distributions to time shiIts and
IrequencyshiIt is called Cohen`s class. Similarly, the class oI all quadratic time
Irequency distributions covariant to timeshiIt and scales is called the AIIine class.
The intersection oI these two classes contains timeIrequency distributions, like the
WignerVille distribution, that are covariant to all operators.
Cohen 1 generalised the deIinition oI the time Irequency distributions in such a
way as to include a wide variety oI diIIerent distributions. These diIIerent
distributions can be represented in several ways. Cohen`s class deIinition like the
Fourier TransIormation, with respect tot, oI the generalised local correlation
Iunction is most common. With a twodimensional kernel, the bilinear time
Irequency distribution oI the Cohens class is deIined according to equation 2:
( ) ( ) t u
t t
t u
u t t t u t
a at a t x t x e f t C
t f f f t f
x

.

\


.

\

+ =
)))
+ '
2 2
, ,
* 2 2 2
(3)
where x is the signal, t (t) is the time, t is the time location parameter, e is
angular Irequency, u is shiIt Irequency parameter, (u, t) is called the kernel oI the
time Irequency distribution. A distribution C
x
(t, f) Irom Cohen`s class can be
interpreted as the twodimensional Fourier TransIormation oI a weighted version oI
the ambiguity Iunction oI the signal
'Computational Civil Engineering 2007, International Symposium 249
( ) ( ) ( ) u t t u t u
u t t t
a a e e A f t C
t f f f
x x
=
))
2 2
, , , (4)
where A
x
(u, t) is the ambiguity Iunction oI the signal x(t), given by equation:
( ) at e t x t x A
t f

.

\


.

\

+ =
)
u
t t
t u
2 2
,
*
(5)
We note that all integrals run Irom  to . The weighted Iunction (u, t) is called
the kernel. It determines the speciIic properties oI the distribution. The product
A
x
(u, t)(u, t) is known as the characteristic Iunction. Since the ambiguity
Iunction is a bilinear Iunction oI the signal, it exhibits cross components, which, iI
allowed to pass into time Irequency distribution, can reduce autocomponent
resolution, obscure the true signal Ieature, and make interpretation oI the
distribution diIIicult. ThereIore, the kernel is oIten selected to weight the ambiguity
Iunction such that the autocomponents, which are centred at the origin oI the (u, t)
ambiguity plane, are passed, while the crosscomponents, which are located away
Irom origin, are suppressed. This means that the suppression oI crosscomponents
might be understood as the Irequency response oI a twodimensional lowpass
Iilter.
When a low pass kernel is employed, there is a tradeoII between cross
components suppression and autocomponent concentration. Generally, as the
bandpass region oI the kernel is made smaller, the amount oI crosscomponent
suppression increases, but at the expense oI autocomponent concentration.
There is deIinition oI the kernel Ior Rihaczek TransIormation in equation 6
( )
2
,
t u
t u
=
f
e . (6)
Equation 4 can also be rewritten into the Iollowing Iorm 5
(7) ( ) ( ) ( ) u t u t u t a a WJT f t f t C
x
H =
) )
, , ,
where
(8) ( ) ( )
( )
e t u
u t t
a at e f t
t f f
= H
) )
2
, ,
is the twodimensional Fourier transIorm oI the kernel and WVT presents
WignerVille transIorm. Cohen`s class has a simple interpretation as a smoothed
WignerVille distribution 5.
250 J. Smutny, L. Pazdera
3. ANALYSIS OF DYNAMIC PARAMETERS
The model used Ior laboratory measurements and analysis oI dynamic parameters
oI a sample oI rail Iastening is presented below. The rail grid model was
constructed oI concrete sleepers B 91, on which there were Iastened rails oI
construction shape UIC 60 by Ilexible Iastening Vossloh SKL14.
For the testing oI the dynamic properties oI the sample, the method oI measuring
the response to mechanical shock was used. Mechanical shock was stimulated by a
special hammer in the radial direction on the railhead. A part oI this hammer is a
Iorce detector.
The response was measured by accelerometers at diIIerent points oI the rail
structure, on the rail Ioot and sleepers (10 cm Irom Iastening). Figure 1 show the
location oI detectors. From the response time signals Irequency transIer Iunctions
(accelerance) were calculated in order to obtain standardised responses 5.
Figure 1 General view oI the working place
Signals Irom measurements on the rail and sleepers were used Ior the presentation
oI particular analyses in this contribution. The measuring system consisted oI a
Brel and Kjaer PULSE modular analyses Ior recording the vibration parameters
together with B&k cubic acceleration detector and a B&k shock stimulation
hammer (Figure 1).
'Computational Civil Engineering 2007, International Symposium 251
The accelerometers were Iastened to the measured construction by means oI bee
wax. The results were recorded digitally.
The analysis oI the response to mechanical shock was evaluated by means oI the
Iollowing methods and parameters 5:
Time records oI the duration oI impulse response Iunction (in principle
standardized acceleration value)
Frequency analysis with the use oI Irequency response Iunction (according to
equation 1)
TimeIrequency method oI spectral analysis (Ior the transIer Irom time to
time Irequency domain, the algorithm oI Short Time Fourier TransIormation
and Rihaczek TransIormation was used)
Time histories oI the impulse response Iunction, recorded by accelerometers,
located on the rail Ioot, are depicted on the upper graph oI Figure 2. The maximum
positive value oI acceleration oI 300 ms
2
is reached 1 ms Irom the observed
beginning. The maximum negative value oI acceleration oI 300 ms
2
is reached 2
ms Irom the beginning. Damping oI the signal Irom the acceleration 300 ms
2
to
the acceleration lower than 30 ms
2
took 15 ms.
In the leIt graph oI Figure 2 is depicted the amplitude spectrum oI this Irequency
response Iunction calculated according to equation 1. In the graph, six important
Irequencies (0.2 kHz, 0.7 kHz, 1.9 kHz, 2.4 kHz, 3.3 kHz and 3.7 kHz), are visible.
The important values are taken as those which have the damping up to 20 dB Irom
the maximum value oI amplitude spectrum.
Time Irequency amplitude spectrum estimated by application oI Short Time
Fourier TransIormation to the impulse response Iunction is depicted in the middle
graph in Figure 2. As shown on this graph, the time history oI important Irequency
components essentially diIIer.
Frequency component 1.9 kHz reaches the highest values Ior a relatively long time
(compared to other Irequency components). It appears in the signal nearly in its Iull
history, i.e. approximately 40 ms by damping up to 40 dB. The second most
important component is the Irequency 3.3 kHz. This appears in signal up to the
time oI 20 ms Irom the above. Other notable Irequencies 2.4 kHz and 3.7 kHz are
in the signal Ior the time oI 5 ms up to 15 ms.
Similar conclusions are visible Irom the middle graph oI Figure 3, which present
the analysis oI impulse response Iunction on the rail Ioot by the use oI Rihaczek
TransIormation. This transIorm belong to the category oI nonlinear time
Irequency proceedings Irom the Cohen class.
252 J. Smutny, L. Pazdera
Figure 2 Accelerometric detector located on the rail Ioot, time Irequency analysis by
Short Time Fourier TransIormation
Figure 3 Accelerometric detector located on the rail Ioot, time Irequency analysis by
Rihaczek TransIormation
'Computational Civil Engineering 2007, International Symposium 253
Signals (impulse response Iunction) taken by a second transducer, located on the
concrete sleeper, have diIIerent character. From the time record (see upper graph oI
Figure 4) it is apparent that the maximum impulse response Iunction amplitude
acquires lower Irequency values as a result oI the inIluence oI the transIormation oI
the signal through the Iastening oI rail, clip plate, sleeper to the accelerometer and
reaches values oI 50 ms
2
. These values were reached 2 ms Irom the Iirst rise time
Irom 'amplitude pack". Values oI acceleration are considerably lower than those
by the transducer located on the rail Ioot which was located nearer to the source oI
mechanical impulse.
In the leIt graph oI Figure 4 is depicted the amplitude spectrum oI Irequency
response Iunction. The Iorm oI spectrum considerably diIIers Irom the
characteristics measured by the Iirst transducer located on the rail Ioot. The most
important components appear in the lower Irequencies Irom the transducer located
on the rail Ioot: in the interval oI 0.2 kHz up to 2 kHz, there are also more in
number.
Similar conclusions are given by the middle graph oI Figure 4 which presents the
time Irequency amplitude spectrum estimated by the application oI the Short Time
Fourier TransIormation. From this graph it can be seen that time occurrence oI
signiIicant components included in signal is considerably shorter (the longest is
approx. 20 ms Irom the imaginary beginning) than it is Irom the signal Irom
transducer located on the rail Ioot.
Figure 4 Accelerometric detector located on concrete sleeper, time Irequency analysis by
Short Time Fourier TransIormation
254 J. Smutny, L. Pazdera
Figure 5 Accelerometric detector located on concrete sleeper, time Irequency analysis by
Rihaczek TransIormation
Similar conclusions apply to the middle graphs oI Figure 5 which present the
analysis oI signals Irom transducers located on the sleeper by the use oI Rihaczek
TransIormation. The signiIicant Irequency components which are calculated by the
Rihaczek TransIormation (Figure 6) are Irequencies oI 0.2 kHz, 0.7 kHz, 1.9 kHz,
2.7 kHz, 3.2 kHz, 3.4 kHz and 3.7 kHz. The most signiIicant spectrum component
is the Irequency component 0.2 kHz which appears within this spectrum Ior a
relatively long time in relation to the activity oI other components.
On the whole, it is possible to state Irom the middle graphs in Figure 2 to Figure 5
that in contrast to linear methods whose ability to resolve the Irequency elements in
the time region is limited by certain window Iunctions, quadratic methods can
achieve this objective. Higher distinguishing makes more Iavourable localization oI
signiIicant Irequency components in time possible. The quality oI time and
Irequency achieved by measuring the signal response to mechanical shock and
applying by these transIormations is a good choice.
4. CONCLUSIONS
Based on measurements and analyses, it is possible to state that the methods
presented above are very good Ior the measurement oI dynamic parameters oI rail
'Computational Civil Engineering 2007, International Symposium 255
Iastenings. The use oI these methods enables the testing oI new types oI rail
Iastenings and diIIerent types oI rail washers under rails and the opportunity to
optimise the geometric location oI damping elements on rail etc. From the
mathematical means oI signal analysis it is possible to utilise both Short Time
Fourier TransIormation and Rihaczek TransIormation Ior timelocalisation oI the
occurrence oI Irequency elements oI stationary and nonstationary signals.
Based on the experience acquired, it is oI great advantage Ior the analysis oI real
signals to utilise the properly selected time and Irequency sections. This procedure
seems to be more suitable than the spatial arrangement. It is possible Ior more
precise localisation oI time records to separate signiIicant Irequency components or
to depict all important Irequencies. Analysis oI signals, acquired by measurement
and analysis oI response to mechanical shock gives new, more detailed insights to
transition characteristics oI railway and tramway structures. Hence, it grants
valuable knowledge Ior a thorough analysis oI these constructions, which can be
important Ior consequent optimisation oI construction and operational conditions.
Also the Iact that by time Irequency proceedings analysis oI dynamic load oI
railway and tramway constructions provides real data Ior consequent Iormulation
oI mathematical models. From this point oI view, both linear and nonlinear time
Irequency transIormations are applicable. These methods give a Iast and accurate
localisation oI Irequency components included in the measured signal. It is possible
to apply the described method successIully not only on samples oI several
constructions oI railway and tramway superstructure but also directly in the Iield
on real tracks.
Acknowledgements
This research has been supported by the research project 103/07/0183
("The investigation oI dynamic eIIects due to the rail transport by the method oI
quadratic time and Irequency invariant transIormations ") and by research project
MSM 0021630519 ('Progressive reliable and durable loadbearing structural
constructions)
ReIerences
1. Cohen L., TimeIrequency distributions, Proc. IEEE, 1989, vol. 77 no. 7, pp. 941981
2. O'Neill J.C., Quartic Functions Ior TimeFrequency Analysis with Applications to Signal
Adaptive Kernel Design, SPIE  Advanced Signal Processing Algorithms, 1997
3. Poularikas A. D.: The TransIorm and Applications Handbook, IEEE Press, 1996
4. Melcer, J., Kucharova, D. : Mechanical properties oI rubber pads under static and dynamic load,
proceedings oI International ConIerence on Materials Science and Engineering,
BRAMAT 2003, Romania, Brasov, 3/2003, University oI Brasov, 2003, pp. 18314.
5. Smutn J., Pazdera L.: New techniques in analysis oI dynamic parameters rail Iastening, InSight,
The Journal oI The British Institute oI NonDestructive Testing, Vol 46, No 10, October, 2004,
pp. 612615, ISSN 13542575
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
The use oI time Irequency transIormations
in testing structural elements
Jaroslav Smutny
1
, Lubos Pazdera
2
1
Department of Railway Constructions ana Structures, Faculty of Civil Engineering,
University of Technology Brno, Brno, 602 00, C:ech Republic
2
Department of Physics, Faculty of Civil Engineering,
University of Technology Brno, Brno, 602 00, C:ech Republic
Summary
The paper aeals with the nonaestructive testing of structural elements by means of
the acoustic response using the fointea time ana frequency transforms. These
methoas make it possible to locali:e the beginning ana the ena of frequency
components containea in the measurea signal ana in this way, they enable us to
analy:e perfectly the spectrum of the nonstationary noise. In this way, this
mathematical proceaure enables us to aistinguish a gooa specimen from a
aefective one.
KEYWORDS: Time Frequency TransIorm, Wavelet TransIorm, Acoustic
Response Analysis, Ambiguity Function
1. INTRODUCTION
Many times we have already been able to convince ourselves that utilising the
experience and knowledge having their roots even in the Iar past have brought
surprisingly good results. One oI such experience is the knowledge that noise
resulting Irom the shock applied to a structure with cracks (disturbances) is
signiIicantly diIIerent Irom the same subject without cracks. This phenomenon has
been known Ior a long time. As early as in the Middle Ages this phenomenon was
used to detect cracks in ceramic pots aIter their Iiring. The phenomenon mentioned
makes it possible to detect cracks in metallic materials as well. Its very old
application in the railway transport is generally known. However, in the
development and application oI the methods used to detect deIects in structural
elements and materials, this phenomenon has oIten been neglected. The absence oI
the advanced measuring techniques and appropriate mathematical instruments
necessary Ior the evaluation oI measured signals were the main reasons.
There are only the methods oI the timeIrequency analysis in connection with the
classic spectral analysis that make a thorough analysis oI the signals measured with
good possibilities oI classiIying and identiIying possible deIects. The noise excited
'Computational Civil Engineering 2007, International Symposium 257
by Iorce impulse to the tested construction or element is a very interesting
phenomenon since it contains a number oI mutually independent and well
recognisable symptoms according to which it is possible to diIIerentiate materials
with cracks Irom those without cracks 8, 9.
These symptoms are particularly included multispectra oI the measured noise
signals. The composition oI the spectrum oI noise oI each material is given by
several characteristics with their own side elements with respect to time and
Irequency. The spectrum oI the measured noise is changed Ior the duration oI the
process. The symptoms oI cracks are as Iollows:
Changes oI the amplitudes oI particular characteristic Irequency components
Frequency shiIts oI characteristic Irequency components
Appearance oI new Irequency components
Presence oI modiIied spectrum in comparison with a good product
2. THEORY OF TIME FREQUENCY ANALYSIS
InIormation relating to any technical or physical occurrence is represented in the
signal by time changes oI immediate values or physical phenomena described. The
direct evaluation oI the timeamplitude representation oIten appears not to be easy.
That is why there is a practice oI the signal transIorm Irom the time domain to
some diIIerent ones. In some cases, some important pieces oI inIormation Irom the
Irequency area may be obtained. There are a lot oI various transIorms applied Ior
transition between time, Irequency and jointed time Irequency domain. The best
known method is the Fourier transIorm and some oI its modiIications.
The Fourier transIorm and some oI its modiIications are the techniques which are
especially suitable Ior processing stationary signals. These methods can analyse
transient and nonstationary signals as well as in the cases when we are interested
only in the Irequency components contained in the whole time behaviour oI signal.
However, these do not provide us with inIormation on the occurrence oI important
Irequency components in the time Ilow.
One oI the possible way how to analyse the time occurrence relating to the
Irequency components oI transient and nonstationary signals is the use oI the so
called timeIrequency analysis method.
The goal oI this paper is to demonstrate some lessknown methods Ior creating
Iunctions representing the energy oI the signal simultaneously in time and
Irequency domain.
The Short Time Fourier and Wavelet TransIorms represent the examples oI the
linear timeIrequency distributions. The main idea oI the Short Time Fourier
TransIorm (STFT) is to split a nonstationary signal into segments where the signal
258 J. Smutny, L. Pazdera
is considered to be stationary. The Fourier transIorm on each oI these segments is
computed. The STFT is deIined by equation 3
(1) at e t g t x f STFT
t f f
=

)
t
t t
2
) ( ) (  ) , (
where *` denotes the complex conjugate, g(t) is a short time window, x(t) is
a signal, t is a time location parameter, f is a Irequency and t is a time. In the
two dimensional timeIrequency joint representation, the vertical slice oI the
complex valued STFT coeIIicients STFT(t, f) correspond to the Fourier
spectra oI the windowed signal with the window shiIted by time t. The time
Irequency resolution is limited to the Heisenberg principle, which is the
main disadvantage oI the linear timeIrequency transIorm. The signal
component cannot be presented as a point in the time Irequency space. Only
its position inside the AtAf rectangle region may be determined. This is due
to the imposition oI local time window g(t). II the width oI the window is
increased, Irequency resolution improves but time resolution becomes poor
and viceversa.
The Wavelet transIorm (WT) is a new mathematical tool developed mainly since
the middle oI the 1980s. It is eIIicient Ior the local analysis oI nonstationary and
Iast developing transient signals. Similarly to the STFT, the Wigner distribution,
ambiguity Iunction and the Wavelet transIorm map the signal to the timescale
(Irequency) joint presentation. The temporal aspect oI the signal is preserved. The
WT provides a multiresolution analysis with a dilated window. The high
Irequency analysis is made using a narrow window and the lower Irequency
analysis is made using a wide window. The Wavelet analysis is similar to the
Fourier analysis because it breaks a signal down into its constituent parts Ior the
analysis. Whereas the Fourier transIorm decomposes the signal into a set oI sine
waves oI diIIerent Irequencies, the Wavelet transIorm decomposes the signal into
its 'wavelets, scaled and shiIted versions oI the 'parent Wavelet. The Wavelet
transIorm allows us an outstanding localisation in both the time domain via
translation oI the parental wavelet and in the scale (Irequency) domain via
dilatation. The translation and dilatation operations applied to the parent wavelet
are perIormed to calculate the wavelet coeIIicients representing the correlation
between the wavelet and the localised section oI the signal. The wavelet
coeIIicients are calculated Ior each wavelet segment, giving a timescale Iunction
relating to the wavelets correlation to the signal 6, 7.
at
s
t
t x
s
s WT

.

\

=

)
t
t ) (
1
) , ( (2)
'Computational Civil Engineering 2007, International Symposium 259
where *` denotes the complex conjugate, x(t) is a signal, t is a time, t is a
translation Iactor, s is a scale Iactor (Irequency) and (x)is a wavelet. This
transIorm is called Continuous Wavelet TransIorm because the analysing wavelet
can be used at any scale, and its position can also be shiIted continuously over the
entire time domain oI the signal being analysed.
As the parent wavelet (which is obviously the most suitable especially Ior transient
processes) the Morlet Wavelet and the Mexican Hat Wavelet are oIten used. The
parent Morlet Wavelet is preIerred Ior the analysis oI the vibration measurement.
The Morlet`s basic wavelet Iunction is a multiplication oI the Fourier basis with a
Gaussian window according to equation (3)
2
2
0
t
t f
e e
=
e
(3)
Its real part is a CosineGaussian Iunction and the imaginary part is a Sine
Gaussian Iunction. The Wavelet transIorm is oI a particular interest Ior the analysis
oI nonstationary and transient signals. The Wavelet transIorm provides an
alternative to the classic Short Time Fourier TransIorm or the Gabor transIorm, and
it is more eIIicient than the Short Time Fourier TransIorm.
The WignerVille transIorm 2, 4 is another alternative method Ior a ShortTime
Fourier TransIorm and the Wavelet TransIorm Ior processing both the stationary
and nonstationary signals. The Wigner distribution was proposed by Wigner in
1932 Ior the region oI quantum physics, and about 15 years later, this was modiIied
Ior the region oI the signal analysis by Ville. The WignerVille transIorm is
deIined Ior the time Irequency region by relation
( )
( )
t
t t
t t
a e t x t x f t WJT
f f
x

.

\


.

\

+ =
)
2 *
2 2
, (4)
where * is a complex conjunction, t is a time, t is a shiIt along the time axis, x(t)
is a time representation oI the signal and WJT
x
(t,f) is a jointed time and Irequency
representation oI the signal. In contrast to the linear time Irequency transIorms in
which the resolution is limited by the window Iunction, the WignerVille spectrum
oIIers us an excellent resolution both in the Irequency and time domain. The
calculation is not limited by the Heisenberg principle oI uncertainty which is its
important characteristic since it is a more general transIormation that does not
utilise the weighing Iunction.
Quadratic (nonlinear) methods represent the second Iundamental class oI time
Irequency distributions. The quadratic methods are based upon estimating an
instantaneous power (or energy) spectrum using a bilinear operation on the signal
x(t) itselI. The class oI all quadratic timeIrequency distributions to timeshiIt and
IrequencyshiIt is called the Cohen`s class. Similarly, the class oI all quadratic
timeIrequency distributions covariant to timeshiIt and scale is called the AIIine
260 J. Smutny, L. Pazdera
class. The intersection oI these two classes contains timeIrequency distributions
like the WignerVille distribution that are covariant to all operators 2. Cohen 1
describes the nonlinear time Irequency transIormation (especially shiItinvariant
class) that can be derived Irom the Wigner distribution.
Cohen 1 generalised the deIinition oI the time Irequency distributions in such a
way which includes a wide variety oI diIIerent distributions. These diIIerent
distributions can be represented in several ways. The Cohen`s class deIinition like
the Fourier TransIorm, with respect tot, oI the generalised local correlation
Iunction, is the most common. With a twodimensional kernel, the bilinear time
Irequency distribution oI the Cohens class is deIined according to equation 1, 10
( ) ( ) t u
t t
t u
u t t t u t
a at a t x t x e f t C
t f f f t f
x

.

\


.

\

+ =
)))
+ '
2 2
, ,
* 2 2 2
(5)
where x(t) is a signal, t is a time, t is a time location parameter, e is an angular
Irequency, u is a shiIt Irequency parameter, a Iunction (u, t) is called the kernel
oI the time Irequency distribution. Distribution Cx (t,e, ) Irom the Cohen`s class
can be interpreted as the twodimensional Fourier TransIorm oI a weighted version
oI the ambiguity Iunction oI signal
( ) ( ) ( ) u t t u t u
u t t t
a a e e A f t C
t f f f
x x
=
))
2 2
, , , (6)
where A
x
(u, t) is the ambiguity Iunction oI the signal x(t), given by equation:
( ) at e t x t x A
t f

.

\


.

\

+ =
)
u
t t
t u
2 2
,
*
(7)
Note that all integrals run Irom  to . The weighted Iunction (u, t) is called the
kernel. It determines the speciIic properties oI the distribution. The product A (u,
t) (u, t) is known as the characteristic Iunction. Since Iunction ( ) t u , A
x
represents a bilinear operation Ior processing the signal, the contributions Irom the
socalled cross components are exhibited during its calculation, which
consequently deteriorates the diIIerentiation oI the given transIormation. This
eIIect may be limited by a suitable choice oI the socalled kernel Iunction. Then the
kernel Iunction unambiguously determines the properties oI a given transIormation.
x
ThereIore, the kernel is oIten selected to weight the ambiguity Iunction such that
the autoelements that are centred at the origin oI the (u, t) ambiguity plane are
passed, while the crosselements that are located away Irom origin are suppressed.
This means that the suppression oI crosselements might be understood as the
Irequency response oI a twodimensional lowpass Iilter.
When a lowpass kernel is employed, there is a tradeoII between the cross
elements suppression and the autoelement concentration. Generally, as the band
'Computational Civil Engineering 2007, International Symposium 261
pass region oI the kernel is made smaller, the amount oI the crosselement
suppression increases, however, at the expense oI the autoelement concentration.
Table 1 presents deIinitions oI the kernels Ior various time Irequency distributions
5.
Table 1 DeIinitions oI the kernels Ior various time Irequency distributions
Distribution Kernel Iunction (u, t)
Rihaczek
( )
2
,
t u
t u
=
f
e
MargenauHill ( )

.

\

=
2
t u
t u cos ,
Page
( )
2
,
t u
t u
=
f
e
Equation 6 can also be rewritten into the Iollowing Iorm 5, 10
(8) ( ) ( ) ( ) u t u t u t a a WJT f t f t C
x
H =
) )
, , ,
where
(9) ( ) ( )
( )
e t u
u t t
a at e f t
t f f
= H
) )
2
, ,
is the twodimensional Fourier transIorm oI the kernel and WJT presents the
WignerVille transIorm. Cohen`s class has a simple interpretation as a smoothed
WignerVille distribution.
3. ANALYSIS OF DYNAMIC PARAMETERS
The text to Iollow pursues the determination oI the quality oI structural elements
(in this case the 'hurdis brick) by the method oI acoustic response analysis. The
basis Ior the methodology designed by the author is the analysis oI the response oI
the noise signal to a mechanical impulse, particularly by timeIrequency
procedures. The measured specimen was placed on a special device Iixed to
supports (Figure 1). The mechanical shock was excited by a special pendulum with
a deIined choice oI the shock intensity. The measuring device (made by the
Bruel & Kjaer Company) consisted the PULSE 3360C signal analyser, a
262 J. Smutny, L. Pazdera
microphone and the measuring soItware. The sampling Irequency was 12 kHz. The
electric signal detected and digitally recorded resulted Irom the measurement, and
it was adequate to the amplitude oI the acoustic pressure in the place oI the
microphone location.
AIter completing the analysis, the check measurement and calculations, the
Iollowing methods and parameters were used to the analysis oI the response to the
mechanical shock:
diagram time history oI the acoustic pressure
Irequency analysis with help oI behaviour oI the power spectral density (the
algorithm oI the Iast Fourier transIorm was used)
linear jointed time and Irequency methods oI the spectral analysis
(the algorithms oI the Wavelet transIorm were used)
nonlinear jointed time and Irequency methods oI the spectral analysis
(the algorithms oI the Rihaczek transIorm were used)
The analyzed Iigures are composed oI three diagrams. The upper diagram shows
the time history oI the acoustic pressure. In the leIt diagram, the amplitude
spectrum oI the acoustic response calculated by the direct employing oI the Fourier
transIorm on measured signal is shown. The middle diagram shows the 3D view oI
jointed time and the Irequency spectrum oI the amplitude spectrum oI the acoustic
pressure. The spectra in particular diagrams in Figures 2 to 5 were gradually
calculated by means oI the Wavelet transIorm or more precisely by the Rihaczek
transIorms.
The values oI the acoustic pressure in the decibel scale are depicted in the middle
diagrams in diIIerent colours. It should be said that the maximum value is in black
colour. Figures 2 to 5 present three groups oI diagrams oI the signals measured by
the microphone using the impactecho method used to a deIective product and Ior
the deIectIree one.
Comparing the time records oI a good and a deIective specimen, it may be
considered that the signal coming Irom the good product has a lower damping, i.e.
the specimen sounds Ior a longer time.
'Computational Civil Engineering 2007, International Symposium 263
Figure 1 General view oI the working place
Then the Irequency characteristic shows that in a deIective specimen the distinctive
Irequency components have 'moved in the direction oI the lower Irequency
values and that the spectrum is wider compared with deIectIree specimens. The
comparison oI the course oI the timeIrequency transIorm clearly shows the
diIIerent characters oI the spectra oI the deIective specimens and the deIectIree
specimens.
The text to Iollow will only deal with the diagram oI the timeIrequency
characteristics obtained by means oI diIIerent mathematical procedures.
The Wavelet transIorm will be considered as a basic transIorm. The timeIrequency
curve Ior the deIectIree product (Figure 2) clearly shows two Irequency
components on Irequencies 2.5 kHz and 3.3 kHz. The component on Irequency
2.5 kHz is damped by 20 dB approximately in 30 ms, while the Irequency
component 3.3 kHz is damped in 50 ms. These two Irequency components
compared with other Irequency components are remarkable in their spectra. The
change oI the timeIrequency spectrum oI a deIective specimen (Figure 3) can
easily be distinguished. The distinctive Irequency region, approximately Irom
700 Hz to 3.2 kHz contains more than two important Irequency components e.g.
it is possible to choose the values on 700 Hz, 1.1 kHz, 1.8 kHz, 2.2 kHz, 2.5 kHz,
2.8 kHz and 3.2 kHz their values are less than one order lower than the value oI
the maximum component. The time periods Ior damping oI these Irequency
components by 20 dB are usually shorter than the periods Ior selected components
264 J. Smutny, L. Pazdera
in deIectIree specimens. These periods reach the values oI damping approximately
between 15 ms and 30 ms.
Let us mention that the Wavelet transIorm is one oI the basic and also Iast
procedures Ior the timeIrequency analysis oI signals. However, the accuracy and
appropriateness oI this method depends upon the choice oI the window Iunction.
The application oI the method requires a certain experience gained Ior the
'rational deIinition oI input parameters and also in the interpretation oI its
spectrum.
ThereIore it is oIten more advantageous to use nonlinear time Irequency
transIormations Ior this analysis oI the transient signals. The characteristic Ieature
oI nonlinear transIormations rests in their resultant resolution in time and
Irequency which is not limited by the Heisenberg uncertainty principle. This Iact
includes a high resolution capability in the time Irequency level which
demonstrates itselI by the 'accurate localisation oI important Irequency
components in time.
The Rihaczek transIorms applied to both signals (deIective ones in Figure 4 and
deIectIree ones in Figure 5) show a similar jointed timeIrequency spectrum oI the
standardised acoustic pressure level as in case oI the Wavelet transIorm. The time
Irequency spectrum (the Rihaczek transIorm) shows local Irequency maximums oI
particular Irequency components in time more distinctively (precisely) than it is
with the spectrum calculated by the Wavelet transIorm. This is given especially by
the type oI transIorm where mainly the kernel and the local autocorrelation
Iunction participate in the phenomenon presented.
The timeIrequency spectrum calculated by the Rihaczek transIorm shows sharper
courses oI particular distinguished Irequency components ("slighter lines") than it
is with the spectra calculated by the Wavelet transIorm.
As apparent Irom Figures 2 to 5, the whole process oI response may be roughly
divided into three stages. In the Iirst stage, a very Iast growth oI the amplitudes oI
the key Irequency components to maximum values is apparent. In the second stage,
the descent and extinguishing oI higher Irequency components occur. However, in
both phases oI the response the contents oI the Irequency crosssections remain
approximately the same. In the third stage, the Iading oI the response occurs. This
is usually characterised by the existence oI the lowest Irequency component.
II some material cracks occur, the speed oI the growth and the declination oI
amplitudes oI the spectrum are higher than Ior a deIectIree material. It should,
however, be stated that the speed oI the growth and declination oI amplitudes oI
particular important elements are not the same. The Irequency or the time
Irequency spectrum oI deIective materials is generally much wider, and the so
called clusters oI important Irequency components occur there. This phenomenon
is properly interpreted by particular diagrams in Figures 2 to 5.
'Computational Civil Engineering 2007, International Symposium 265
Figure 2 DeIectIree product  time behaviour signal Irom microphone,
Irequency spectrum and time  Irequency analysis by Wavelet transIorm
Figure 3 DeIective product  time behaviour signal Irom the microphone,
Irequency spectrum and time  Irequency analysis by the Wavelet transIorm
266 J. Smutny, L. Pazdera
Figure 4 DeIectIree product  time behaviour signal Irom the microphone,
Irequency spectrum and time  Irequency analysis by the Rihaczek transIorm
Figure 5 DeIective product  time behaviour signal Irom the microphone,
Irequency spectrum and time  Irequency analysis by the Rihaczek transIorm
'Computational Civil Engineering 2007, International Symposium 267
4. CONCLUSIONS
Based on the measurement and analyses made, it is possible to state that the
experiment checked the possibilities oI using the given methodology Ior the
detection oI structural deIects oI the measured products and materials Irom the
homogeneity and the cracks Iormation points oI view. The analysed parameters
enable us to distinguish a good product Irom a deIective one. The given
methodology may be successIully applied to all ceramic and concrete products. It
can also be established that modern means oI signal analysis, especially the time
Irequency transIorms highly contributed to a quality processing oI the
measurement. These methods provide us with a time localisation oI Irequency
components contained in the measured signal. In this way, these methods oIIer new
possibilities in the experimental analyses oI structural elements and materials. This
possibility is included in both the linear and the nonlinear timeIrequency
procedures. By comparing the obtained results we can clearly observe that when
analysing the responses to the mechanical shock (heavily damped time signals) the
procedures, namely those including into the class oI the Cohen timeIrequency
transIorms, show a very good time localisation oI signiIicant Irequency
components contained in the analysed signal.
In some particularly important cases the depicting oI results oI the computes and
analyses by the time and Irequency sections may be completed. Thus, these provide
us with a proIound support in the analysis oI the
timeIrequency results. This procedure oIten appears as more suitable than e.g. a
separate space arrangement. This makes it possible to locate very precisely the time
behaviour oI particular important Irequency components or to show all important
Irequency components contained in the spectrum in a given time.
In conclusion it may be mentioned that the given methodology can also be
incorporated in the process oI the halIautomated quality control oI products under
in the production line. When using the above methodology together with the
methods oI qualitative analysis or the artiIicial intelligence (Iuzzy and rough sets,
neural networks, genetic algorithms etc.), then the process oI evaluating the quality
oI products can Iully be automated.
Acknowledgements
This research has been supported by the research project 103/07/0183
and by research project MSM 0021630519 ('Progressive reliable and durable
loadbearing structural constructions)
268 J. Smutny, L. Pazdera
ReIerences
1. L. Cohen: TimeIrequency distributions  a review, Proc. IEEE, Vol. 77 No. 7, 941981, 1989
2. Wahl T. J.: Bolton J. S., The Application oI the Wigner Distribution to the IdentiIication oI
StructureBorne Noise Components, Journal oI Sound and Vibration, pp. 101122, 1993, ISBN
0022460X/93/100101
3. Poularikas A. D.: The TransIorm and Applications Handbook, IEEE Press, 1996
4. J.C. O'Neill: Quartic Functions Ior TimeFrequency Analysis with Applications to Signal
Adaptive Kernel Design, SPIE  Advanced Signal Processing Algorithms, 1997
5. Hammond J. K., White P. R.: The Analysis oI NonStationary Signals Using TimeFrequency
Methods, Journal oI Sound and Vibration, pp. 419447, 1996,
and ISBN 0022460 X/93/100101
6. Marasek C., Piotrkowski R., Serrano E., Ruzzante J. E.: Monitoring oI the tool condition with
acoustic emission signal analysis using wavelet packets, Insight  NonDestructive Testing and
Condition Monitoring Vol. 44 No. 12, December 2002, ISSN 13542575
7. Chang Y. F.: Wavelet deconvolution beIore scanning in ultrasonic nondestructive testing,
Insight  NonDestructive Testing and Condition Monitoring Vol. 44 No. 11, November 2002,
ISSN 13542575
8. Young S. Cho: Nondestructive testing oI high strength concrete using spectral analysis oI
surIace Wales, NDT & E International, Volume 36, Issue 4, June 2003, pp. 229235, ISSN 0963
8695
9. Allen G. Davis : The nondestructive impulse response test in North America: 19852001
NDT & E International, Volume 36, Issue 4, 2003, pp. 185193, ISSN 09638695
10. Smutn J., Pazdera L.: New techniques in analysis oI dynamic parameters rail Iastening, InSight,
The Journal oI The British Institute oI NonDestructive Testing, Vol. 46, No 10, October, 2004,
pp. 612615, ISSN 13542575
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Design oI reinIorced soil works Textomur structures based on
the computer program Cartage
Anghel Stanciu
1
, Oana Col
2
and Irina Lungu
3
1
Department of Roaas ana Founaations, Technical University, Iasi, 700050, Romania
2
Department of Roaas ana Founaations, Technical University, Iasi, 700050, Romania
3
Department of Roaas ana Founaations, Technical University, Iasi, 700050, Romania
Summary
Reinforcea soil represents an alternative solution to consoliaate earth massifs ana
perform retaining structures incluaea in the present transportation infrastructure.
One of the aavantages consists in applying this type of structure in poor founaation
soil conaitions, aue to the fact that the inaucea pressures are small ana uniformly
aistributea. Construction costs may also be aecreasea by using the local material
from the construction site as filling material within the reinforcea soil work.
Finally, the resultea platform may also be support for aaaitional construction
works.
Regaraing the construction methoa, the variety of the reinforcements (steel,
geosynthetics, geogrias) or of the facing elements (precast concrete blocks, steel
structures, reversea geogrias or geosynthetics) allows various performing
proceaures to aevelop when builaing reinforcea soil structures.
Textomur system is incluaea in the category of reinforcea soil with geosynthetics,
aevelopea mainly in France. The main aifference between this system ana other
reinforcing techniques consists in builaing up the facing element, by using
prefabricatea steel elements.
The aesign of the reinforcea soil works in Textomur system is implementea via a
computer program Cartage. This program was initiatea in 1985 LCPC
(Laboratoire Central au Ponts et Chaussee), France ana it is basea on the
analytical calculation methoa aevelopea also at LCPC, consiaering the limit
equilibrium.
By running this computer program stresses within the reinforcements can be
evaluatea, the aisplacements of the soil massive, ana the safety coefficients at
anchorage as well, basea on the stability analysis of the massive in case of
cylinarical failure surfaces going through the toe of the reinforcea soil structure.
KEYWORDS: reinIorced soil, retaining structures, geosynthetics, Textomur,
Cartage.
270 A. Stanciu, O. Col, I. Lungu
1. INTRODUCTION
The Iirst modern system oI soil reinIorcing was initiated and developed in the mid
60s by the French engineer Henri Vidal, with the name Terre Armee (reinIorced
soil), using steel reinIorcements. Beginning with 1970, geosynthetics have been
created as an alternative oI the steel reinIorcements. The acceptance oI
geosynthetics in reinIorced soil construction has been triggered by a number oI
Iactors, including aesthetics, reliability, cost, simple construction techniques, good
seismic perIormance, and the ability to tolerate large deIormations without
structural distress.
In Romania, the Iirst attempts oI perIorming retaining walls by reinIorced soil were
made in 1973...1974, when such a construction work was designed and built over
50m length. AIter 1990 many retaining walls have been perIormed as soil
reinIorced with geotextiles and geogrids within road rehabilitation works.
Textomur system is included in the category oI reinIorced soil with geosynthetics,
developed mainly in France. The main diIIerence between this system and other
reinIorcing techniques consists in building up the Iacing element, by using
preIabricated steel elements.
2. TEXTOMUR REINFORCING SYSTEM
Textomur is essentially a reinIorcing technique oI soil using geogrids or
geotextiles. There are many variants oI the Textomur system and the name
corresponds to Textomur system with mineral, vegetal, or neutral Iacing elements.
The Iacing elements oI Textomur have a length oI 4.75m the steel within has a
diameter oI 6 and 8mm. For the Textomur system with mineral Iacing the
Iormworks are galvanized aIter welding and bending at the appropriate angle. The
height oI one layer is 6065cm, being variable depending on the requirements oI
the project. The slope oI the Iormwork may vary between 40 to 90. The
reinIorcement length is established Ior each project but an initial 0.7 oI the total
height is recommended.
2.1. Advantages
Local material Irom the construction site may be used as Iilling material.
Flexibility curves, edges, terraces oI variable length may be perIormed.
Long lasting liIe the liIe span is estimated Ior approximately 120 years.
Easy to built, without a Ioundation.
Construction may be perIormed without additional bracing.
'Computational Civil Engineering 2007, International Symposium 271
2.2. Textomur with mineral Iacing
Textomur with mineral Iacing includes galvanized welded steel Iacing elements
behind which large boulders oI 30100cm are manually placed over a width oI
30cm and along the entire height oI the Iacing (Iigure 1). The boulders are
preIerred to be with plan surIaces and consequently the arrangement will display
less voids.
Figure 1. Cross section oI a Textomur system with mineral Iacing
This type oI Textomur is best applied Ior retaining structures (Iigure 2), the ensure
slope stability, Ior signiIicant height, reinIorced soil works in steps can be
perIormed, with an aesthetic role as well.
Figure 2. Retaining wall oI Textomur with vertical mineral Iacing
272 A. Stanciu, O. Col, I. Lungu
3. COMPUTER PROGRAM  CARTAGE
The computer program Cartage is based on the LCPC calculation method oI the
reinIorced soil 1, 2, considering the limit equilibrium oI the massive. Slope
stability is analyzed based on circular Iailure surIaces through the toe oI the
massive. The program computes the stresses within reinIorcements, considering
each Iailure surIace. The saIety coeIIicient Ior internal stability is set as 1.5. The
maximum stress in each reinIorcement resulted Irom the program is considered the
service stress Ior the corresponding Iailure surIace. The calculation continues until
the saIety requirements are met, altering the reinIorcing scheme or the
reinIorcement type.
3.1. Calculation method oI L.C.P.C.
The method is based on Rankine`s theory to compute the active earth pressure on
the retaining element (Iigure 3). The soil layer horizontally placed between the i1
and i reinIorcements is considered to establish the tension Iorce within the
reinIorcement 3, 4. Regarding the main stress as the vertical one ( H = o
v
),
the tension Iorce in the i reinIorcement would counteract the active earth pressure
acting on the Iacing oI height H A .
H
AH
i1
i
ov H
T=Ka ov AH
oh=Ka ov
Figure 3. Principle oI L.C.P.C. method
Consequently the tension Iorce is given by:
( )
2
i
2
1
H K H K H T
a a
A + A = (1)
but and thus the Iinal relationship is the Iollowing: H i H A =
2
i
2
1
1 H K T
a
A

.

\

+ = (2)
'Computational Civil Engineering 2007, International Symposium 273
3.2. Design hypotheses
The stability analysis is perIormed considering the simultaneous action oI several
pessimistic Iactors the highest height, a horizontal platIorm at the top, the service
load as close as possible towards the Iacing, and ignoring the presence oI the
material in Iront oI the reinIorced structure toe.
A case study is presented as a retaining wall on the road RN 94, Embrun, France.
The less Iavourable case is adopted in the design proIile (Iigure 4), height oI 5,40m
corresponding to 9 levels oI Textomur Iormwork, with an inclination oI 27
0
5.
The presence oI soil over the Iirst Iormwork height was disregarded.
Figure 4. Design proIile Ior Cartage, considering one potential Iailure surIace
The geotechnical parameters oI the local material are: unit weight 
21kN/m
3
;
internal Iriction angle  35
o
; cohesion  c 5kPa.
For Iilling the space between reinIorcements, the local material is considered
cohesionless.
Initially, 5 layers oI geotextile were considered, with strength in tension oI
100kN/m, with 0.60m interdistance, 2 layers with 200kN/m strength at 0,60m
interdistance and 3 layers oI 200kN/m strength at 0,30m interdistance.
The input data Ior Cartage are represented by: geometry oI the massive;
characteristics oI the Iilling soil, Ioundation soil and soil behind the retaining
structure; characteristics oI the reinIorcements; coordinates oI the Iailure circles;
displacement value at the massive top; the presence oI underground water.
The input data are presented in Iigure 5, a veriIication oI their accuracy being
easily done.
274 A. Stanciu, O. Col, I. Lungu
Figure 5. Format oI the input data Ior the computer program Cartage
'Computational Civil Engineering 2007, International Symposium 275
3.3. Results
By running the program the output data are presented in a table consisting oI the
extreme values Ior the saIety coeIIicients according to the Iailure surIaces
considered, Ior the displacement, coeIIicient oI anchorage and maximum stress.
Based on the required values by the designed norms, the necessity oI altering the
reinIorcing scheme is analyzed.
The alteration oI the reinIorcing solution is made according to the diIIerence
between the imposed and obtained values and not in the least, based on the design
experience oI the engineer.
Within the presented case study, the Iinal values are set in a table Ior the Iirst 12
Iailure circles as in Iigure 6 5.
Figure 6. Final values as output data Irom the Cartage program
276 A. Stanciu, O. Col, I. Lungu
LEGEND:
Values between parantheses ( , )  number oI the massive and reinIorcement
with the smallest value oI the coeIIicient oI anchorage (ANCR), respectively
the maximum stress (TENR) Ior each Iailure circle.
DELTA displacement value ;
F0 value oI the saIety coeIIicient with reinIorcing the massive ;
FSOL value oI the saIety coeIIicient aIter reinIorcing the massive.
4. CONCLUSIONS
Using computer programs to design retaining structures made oI reinIorced soil
decreases signiIicantly the design time. In order to perIorm an eIIicient design it is
required to master correctly the behaviour oI the reinIorced soil generally, and
especially the response oI such structures subjected to various external loads.
ReIerences
1. Silion T., P. Rileanu, A. Stanciu  Fundaii  Pmant armat, Institutul Politehnic Iasi, Facultatea
de Construcii, 1980.
2. Cartage manual de utilizare, L.C.P.C, 1985.
3. Schlosser F  La terre armee, Recherches et realisations, Bull. liaison Labo P. et.ch. 62, nov.
dec., 1972.
4. Schlosser F., Long T.N. Dimensionnement ae murs en terre armee, Session de Iormation
permanente, E.N.P.C., 1974.
5. Proiect tehnic si detalii de execuie  ,RN 94  Deviation aEmbrun", Geonove, Frana.
Computational Civil Engineering 2007, nternational Symposium
Iai, Romnia, May 25, 2007
Computer program Ior Pipe Section Column Sizing
Dragos Voiculescu
1
, Daniela Preda
1
1
Steel Structures, Technical University of Civil Enginering, Bucharest, Romania
Summary
The paper presents the Excel Program for Pipe Section Column Si:ing. The
program has three main parts. the first one is the aata input, where the proposea
si:e of the column is set, the secona one is the loaa section, where the maximum
loaas are input from the structural analysis program, such as SAP or ETABS, the
thira part gives the results of the structural checks on the proposea section.
The program is interactive, meaning that if one of the checks is wrong, it can be
correctea by changing the si:e of the Pipe section.
The program is very useful in appliea engineering aesign, as it gives immeaiately
the final section of a specific column.
KEYWORDS: pipe section; column; sizing.
1. INTRODUCTION
The computer program was developed in MicrosoIt Excel in order to be used
without major problems on any machine. It has some Visual Basic Ieatures to make
it very easy to use; also it was realized in such a manner to be printed as calculation
notes in real projects.
2. PROGRAM PRESENTATION
The program has three main parts, interconnected. The Iirst part is the data input,
where the initial geometrical characteristics oI the column are given. This data is
used to compute all sectional characteristics oI the pipe section, which will be used
later in the program. Also the section class is shown. (Fig.1)
Immediately aIter, the pipe steel grade is input. The user may choose in between
the two common steel grades Ior pipes. AIter choosing the steel grade, the Iactored
strength is shown. All strength checks will be reported to this strength.
278 D. Voiculescu, D. Preda
Figure 1.Input data
AIter this data is established, the program demands the input oI the buckling length
Iactors (Fig.2). These ones are computed using another program, according to the
column end supports and beam characteristics. The buckling length Iactors will be
used to compute the buckling coeIIicient which will be used Iurther.
Figure 2.Buckling data input
Then the second part oI the computer program starts, the part where the column
loads are input. These loads are taken Irom a structural analysis program such as
SAP or ETABS. The data is input in a table, and is used in all the strength and
stability checks. (Fig.3)
Loads are given in three hypothesis: The Iirst one is the maximum axial load with
the correspondent bending moments; the second one is the maximum zz bending
moment with the correspondent others and the third one is the maximum yy
bending moment with the correspondent others. All Iurther calculations is reIerred
to these hypothesis.
'Computational Civil Engineering 2007, International Symposium 279
Figure 3.Load input
The third part oI the computer program starts with the strength checks Ior the
proposed pipe section. Here the applied Iormulas are written and the result is given
Ior each hypothesis. (Fig.4)
Figure 4.Strength checks
280 D. Voiculescu, D. Preda
Also the buckling checks are given in this part oI the program (Fig.5).
Figure 5.Buckling checks
'Computational Civil Engineering 2007, International Symposium 281
3. CONCLUSIONS
This program is very useIul when Iast section sizing is demanded. It may be
printed and gives the calculation sheet Ior the speciIic column.
ReIerences
1. STAS 10108/078.