Академический Документы
Профессиональный Документы
Культура Документы
ADVANCED MODELLING
CeAS
C ENTR O D I ANALI SI STR UTTU RALE S. R . L.
viale Giustiniano 10
20129 Milano
www.ceas.it
PARATIEPLUS 2014
June 2014
Ce.A.S. s.r.l.
Advanced Modelling
TABLE OF CONTENTS
FOREWORD............................................................................................................................................................... 4
1
SCOPE .......................................................................................................................................................... 32
PROBLEM DESCRIPTION ................................................................................................................................ 32
APPROACHING THIS PROBLEM WITH PARATIE ............................................................................................ 32
RESULT DISCUSSION ..................................................................................................................................... 34
CONCLUSIONS .............................................................................................................................................. 39
ACKNOWLEDGMENT .................................................................................................................................... 39
REFERENCES ................................................................................................................................................ 40
COMMENTS .................................................................................................................................................. 46
REFERENCES ................................................................................................................................................ 47
Ce.A.S. s.r.l.
PARATIEPLUS 2014
June 2014
Foreword
This report includes PARATIE PLUS analysis of some benchmark problems selected among experimental
or numerical studies of flexible retaining structures.
The scope of this report is to check the modelling capabilities of PARATIE PLUS and to discuss some
suggested approaches to practical problems, with respect to available structural and geotechnical
information.
In this respect, the presentation of each problem is limited to the discussion of the methods and
assumptions to be used, as well as the assessment of the obtained results with respect to the expected
behaviour. No precise indication to how use the program is included. It is therefore assumed that the
reader is familiar with the procedures included in PARATIE PLUS, as well as with the assumptions at the
base of the modelling approaches implemented in the program.
Ce.A.S. s.r.l.
Advanced Modelling
Given the high permeability of the surrounding sandy soils (i.e., in the range 10 10 m/s), underwater
excavation was considered the only practical construction method because dewatering would affect a
large area, have a significant environmental impact, produce significant settlements, and potentially cause
damage to historical buildings.
-3
-4
The excavation pit is supported by a 1.21.5 m thick, reinforced-concrete diaphragm wall that extends
around the perimeter of the site, measuring approximately 300 m long and 25 m wide. The wall panels
extend to depths ranging from 25 to 31 m, corresponding to toe embedments of 6.87.8 m below formation
level.
The wall is supported by a single row of prestressed tieback anchors located 23 m below the ground
surface with spacing ranging between 1.0 and 1.5 m. These are installed with dip angles ranging from 25
35 and 8 m fixed (i.e., grouted) anchor lengths into the deeper soil formations (free lengths range from
2640 m). Each tieback typically has eight or nine strands of a high-strength steel (i.e., grade 270 ksi)
tendon.
After installing the diaphragm wall and tieback anchors, excavation was performed underwater by using a
pontoon-mounted crane to an average final formation grade 20.2 m below the initial ground surface.
Before dewatering, the base of the excavation was sealed by a 1.5 m thick underwater concrete slab
supported by an array of tension piles.
The numerical finite element models by Nikolinakou et al. (2011) consider some instrumented cross
sections and are limited to the simulation of the construction stages up to the final underwater excavation
stage. No further construction stages including dewatering stage were analysed.
Among the available sections, wall MQ3 is considered, corresponding with the one whose experimental
behaviour was best reproduced by the numerical models.
Model data is briefly outlined in the following table and in Figure 1-1.
Ce.A.S. s.r.l.
June 2014
PARATIEPLUS 2014
Excavation depth D
D=21.40 m
Wall height
H=28.70 m
Wall thickness
T=1.50 m
Tieback data
Figure 1-1
A critical state friction angle in triaxial compression equal to 31 was also determined, as well as an
estimate of peak friction angles at different initial densities (e0) and different confining pressures.
Initial in situ conditions such as void ratio values at different depths were assessed by means of Dynamic
Probing (DPH) Tests, performed in conjunction with the boreholes.
A correlation between DPH results (which is essentially a dynamic penetration test) and relative density
was used, following the recommended procedure by DIN 4094-3 standard, and the initial void ration was
back-figured, using the usual equation
Dr =
e max e0
with e max e min = e 0.20 according to lab tests.
e max e min
Ce.A.S. s.r.l.
Advanced Modelling
The estimated void ratio profile is shown in the next figure, in which the assumed average initial values
are also reported.
S0
e0 0.6
Dr < 20%
S1
e0 0.5
Dr 50%
S2
e0 0.42
Dr > 80%
At rest coefficient
dry weight
saturated weight
'p
K0
kN/m
kN/m
S0
31
0.5
16.5
20
S1
34
17.1
20.5
S2
37.5
18.3
21.5
It should be remarked that such characterization is closely related to the features of the constitutive soil
model adopted in the cited reference, a model which incorporates several aspects that are usually not
considered in ordinary soil models normally used in engineering analyses in the practice.
Despite this observation, in the assessment of the resistance parameters considered in the simplified
PARATIE PLUS soil model, we'll us the parameter above as well.
Therefore, as for the assessment of the active and passive thrust coefficients, we'll consider the peak
friction angles above, and wall to soil friction angle = 0.7 'p. Passive coefficients will be computed
Ce.A.S. s.r.l.
June 2014
PARATIEPLUS 2014
according to Lancellotta (2007), including a null seismic acceleration. It must be reminded that in a
PARATIE PLUS analysis, as well as in an analysis by means of any other non linear spring method, the
selection of appropriate (realistic) thrust coefficients KA and KP represents a very important aspect, which
is even more important than the determination of appropriate friction angles.
The selection of elastic properties of the soil in Nikolinakou et al. (2011) is performed in the light of the key
features of their soil model which requires the characterization of soil deformation at very small strains, in
terms of the small strain shear modulus Gmax. Then, their constitutive model is so conceived to adapt the
soil stiffness, so as to model the observed non linear stress-strain behaviour of actual soils.
In the very simplified PARATIE PLUS model, however, a realistic secant modulus with respect to the
expected soil deformations must be defined. The assessment of PARATIE PLUS muduli is briefly outlined
in the following.
The Gmax profiles considered by Nikolinakou et al. (2011) are reported in Figure 1-3(see the cited paper for
more details). In the same figure, an alternative distribution is also included, which will be used in
PARATIE PLUS, according to the following equation
G max
'
= 140 h
pa
0.5
[MPa ]
With 'h = lateral stress and pa = atmospheric pressure (98 kPa). This kind of correlation is quite in
agreement with usual experimental findings in similar soils. However, it should be noted that such fitting is
likely to be an overestimate of actual values, at least in the shallow layers (see discussion in the cited
reference)
'
E 0.2 2.50 140 h
pa
0.5
'
= 70 h
pa
0.5
[MPa ]
'
E vc 35 h
pa
0.5
[MPa ]
Ce.A.S. s.r.l.
Advanced Modelling
In PARATIE PLUS, such values are used to determine, at each stage, the stiffness of the non-linear
springs, by dividing Evc or Eur by an appropriate length representing the assumed width of the soil which is
affected by wall deformation, at either side of the wall. Such length is automatically computed by
PARATIE PLUS according to Becci & Nova (1987).
K 0 = K 0NC OCR where K 0NC = 0.5 and OCR is set equal to 2 for layers S1 and S2.
Ce.A.S. s.r.l.
PARATIEPLUS 2014
June 2014
Figure 1-5: results at anchor installation (lateral displacements, bending moments and anchor forces)
forces)
Figure 1-6: results at final excavation stage (lateral displacements, bending moments and anchor
forces)
At final excavation stage, the maximum inward movement of the wall is about 20 mm. The ground anchor
force displays a 3.7% increase with respect to initial preload. This observation shows that the ground
anchor system was very well tuned. Maximum sagging moment is around 2300 kNm/m, thus requiring a
bending reinforcement As equal to more or less 0.50.6% of the gross wall area (on the external face,
neglecting any long term issue such as cracking control).
10
Ce.A.S. s.r.l.
Advanced Modelling
Figure 1-7: lateral effective stresses at final excavation stage and thrust summary
In Figure 1-7, we recognize that active lateral conditions at driving side are reached along almost all the
wall height. The active pressures computed by PARATIE PLUS do not account, in this case, for any
arching effects.
The passive resistance at the toe is sufficient to ensure safe conditions, as the ratio of the available
passive thrust to the calculated thrust is 1970/935 = 2.1, which is usually acceptable for temporary
situations like this.
1.5 Discussion
In order to compare the obtained results with the experimental and predicted ones, in Figure 1-8 we
superimpose PARATIE PLUS results on to the results presented by Nikolinakou et al. (2011).
Ce.A.S. s.r.l.
11
PARATIEPLUS 2014
June 2014
Of course, the results available by a 2D finite element analysis with an appropriate constitutive model for
soils are much more complete: for example the settlements at the ground level and the heave inside the
excavation cannot be directly captured by a PARATIE PLUS analysis. Moreover, a general failure
mechanism, like the one in Figure 1-9 cannot be investigated.
1.6 Conclusion
Conclusion
A PARATIE PLUS model of a very well documented real benchmark, for which an advanced finite element
simulation is also available, is presented. Limited to the case considered in this example, PARATIE PLUS
results are very close to the published ones, based on an estimate of soil parameters conducted
according to usual methods normally considered in the practice.
In the scientific study from which such example is retrieved, some other similar cases are presented and
discussed. PARATIE PLUS Users are encouraged to attempt a PARATIE PLUS model also for those
other examples. That paper can be obtained in https://cee.mit.edu/whittle
1.7 References
Becci, B. & Nova,R. (1987). "Un metodo di calcolo automatico per il progetto di paratie, Rivista Italiana di
Geotecnica",1,33-47 (in Italian)
Lancellotta R., (2007) Lower-bound approach for seismic passive earth resistance, Gotechnique, Vol.
57, No. 3, pp. 319-321
Nikolinakou, M.A., Whittle, A.J., Savidis S., & Schran, U. (2011) Prediction and interpretation of the
performance of a deep excavation in Berlin sand, ASCE Journal of Geotechnical and
Geoenvironmental Engineering, 137(11), 1047-1061.
12
Ce.A.S. s.r.l.
Advanced Modelling
Ce.A.S. s.r.l.
13
June 2014
PARATIEPLUS 2014
= 35
= 0.7
At rest coefficient as well as KA and KP values are automatically computed by PARATIE PLUS according
to the usual methods.
Ground anchors are modelled as linear springs whose stiffness is given by the axial stiffness (EA) of the
steel stem divided by the deformable length, set equal to the free length plus 50% of the grouted length.
In PARATIE PLUS, the hydraulic barrier is modelled by simply assigning a reduced permeability to that
zone: we assume kplug knat/100. Once the plug is prescribed, water pressures at both sides of the walls
are determined based on usual 1-D seepage scheme, which is the routinely adopted procedure by
PARATIE PLUS. It should be noted that in this case no other soil parameter is modified in this case, even
if normally some increase stiffness and resistance is also assigned.
In Figure 2-2 the PARATIE PLUS model layout at final excavation stage is shown: the soil zone where a
modified permeability is assigned is highlighted by a dashed hatch.
14
Ce.A.S. s.r.l.
Advanced Modelling
Figure 2-3: PARATIE PLUS model - deformed shape and bending moments
Ce.A.S. s.r.l.
15
PARATIEPLUS 2014
June 2014
reproduces quite well the observed behaviour, unless in the top part where some reduced anchor stiffness
is highlighted by the measures.
This effect is probably related to a limited upper anchor length whose foundation is disturbed by wall
movements: such aspect may be highlighted just by a continuum model (Itasca (2011)). In this respect,
FLAC results reported in Itasca (2011) and reproduced in Figure 2-6 confirm this observation. In situations
like the present one, the intrinsic limitations of the non-linear spring approach become quite evident.
16
Ce.A.S. s.r.l.
Advanced Modelling
Ce.A.S. s.r.l.
17
June 2014
PARATIEPLUS 2014
-200
-100
100
200
300
0
-5
-10
Z [m]
-15
-20
-25
-30
-35
Figure 2-9: UNCONFINED SEEPAGE SOLUTIONS - excess and total pressure isoiso-curves
18
Ce.A.S. s.r.l.
Advanced Modelling
Figure 2-10:
10: UNCONFINED SEEPAGE SOLUTIONS - flow vectors and finite element mesh
2.4 Conclusion
A PARATIE PLUS model of a real anchored bulkhead is presented. PARATIE PLUS results are fairly in
agreement with both field measures and some other numerical predictions. The most important
discrepancies with published data have been addressed and clearly explained.
The usual 1-D seepage model in PARATIE-PLUS is compared with more advanced 2-D seepage models.
In this case, the obtained results are essentially the same using all the modelling schemes included in
PARATIE PLUS. Limited to examples like current one, the 1-D seepage scheme can be used for practical
design purposes.
2.5 References
Brinkgreve, R.B.J., Engin E., Engin H.K., (2010). Validation of empirical formulas to derive model
parameters for sands. Proc. Numerical Methods in Geotechnical Engineering NUMGE2010,
Trondheim, Norway, June 2-4 , pp. 137142.
Itasca Consulting Group, (2011). Installation of a Triple-Anchored Excavation Wall in Sand, FLAC version
7.0, Example Applications Manual, section 17.
Schweiger, H. F. (2002). Results from numerical benchmark exercises in geotechnics, Proc. 5th European
Conf. Numerical Methods in Geotechnical Engineering, Presses Ponts et chaussees, Paris, 2002, pp
305314.
Ce.A.S. s.r.l.
19
June 2014
PARATIEPLUS 2014
20
Ce.A.S. s.r.l.
Advanced Modelling
By means of PARATIE PLUS, such procedure may be reproduced as follows: we restart analysis
discussed in example 1.
STAGE B
The bottom slab is activated by adding an elastic spring element, whose stiffness (per unit running
excavation out-of-plane length) is given by
Kslab = Ect/(0.5B )
in which Ec is the Young modulus of the concrete, t = slab thickness and B is the excavation
breadth (~ 23 m): thus
Kslab = 3010 1.5 / (0.523) = ~3.910 kN / m / m
6
It should be noted that, if no other change is prescribed to the model, no modifications to previous
stage are expected
STAGE C
A cover slab (1 m equivalent thick slab is assumed) at top elevation is activated, by means of an
elastic spring element whose stiffness is given by
Ktop = 3010 1 / (0.523) = ~2.610 kN / m / m
6
It is also assumed that water level inside excavation is lowered down to EL. -12.50 m, but such
lowering must not modify the hydraulic conditions in the surrounding soil. To do this, the following
three operations must be done
1. the water table and the excavation side must be kept to the original level (-2 m) but the
lining option must be activated;
2. the stabilizing pressures of remaining water inside excavation below El. -12.50 (8.9 m
water height) must be applied as external lateral loadings to the retaining wall;
3. the bottom of the excavation must be stabilized by a surcharge accounting for the weight
of the bottom slab, the weight of the water inside excavation and the reaction of the
tension piles. Note that the sum of such contribution must correspond with u= 194 kPa)
determined above. Therefore the average uplift force to be resisted by piles, in this phase,
is given by
qnet= 194 - 251.50 - 10(7.4) = 194-37.5-74 = 82.5 kPa
However, to PARATIE PLUS, a surcharge q = 194 kPa must be applied at the bottom of
excavation.
We finally assume that the existing ground anchor is still active at this stage.
STAGE D
An intermediate slab (1 m equivalent thick slab is assumed) at EL -12 m is activated, by adding an
elastic support with the same stiffness as for the top support.
Now the water inside excavation is completely removed. To model this, just the stabilizing water
pressures inside excavation at previous stage are removed as well. Therefore both the lining
option and the surcharge q = 194 kPa at the excavation line must be preserved.
The existing ground anchor is still active at this stage.
STAGE E
The ground anchor is finally removed, thus assuming very long term conditions.
Ce.A.S. s.r.l.
21
PARATIEPLUS 2014
June 2014
Relevant results for long term stage E are included in the following plot. We note that the water pressures
in soil elements always depend on the original phreatic level, at both side of the wall.
We cannot compare our results with measured or computed results by others, since no information is
currently available, on our best knowledge.
This problem is just proposed to highlight some quite advanced concepts in PARATIE PLUS such as the
lining option feature.
However, just some very qualitative comments can be done as well. For example, we note that the
bending moment increase at STAGE C is very dramatic, whereas the beneficial effects of the intermediate
support activated at stage D is very minor. Therefore, based on our results, such construction sequence is
apparently poorly optimized and it probably does not correspond with actual design.
Input data for this example are included as final stages in the EXAMPLE discussed in section 1.
22
Ce.A.S. s.r.l.
Advanced Modelling
Hexc
HP
Hs
Hw
sat
Ce.A.S. s.r.l.
23
June 2014
PARATIEPLUS 2014
w Hw
sat H S + plug H P
Fs
where Fs is a safety factory normally ranging between 1.10 and 1.30. According to Eurocode 7, part 1, this
condition may be formally rewritten within the Ultimate Limit State method, by checking the compliance of
the plug with respect to the UPL (uplift) limit state. Including the Italian partial safety factors (see Italian
NADs, issued on March, 2013), the condition above is formally rewritten as:
in which the factor 1.1 multiplying the left-hand term represents the partial safety factor for unfavorable
destabilizing permanent actions (water pressure in this case), whereas the factor 0.9 on right-hand term is
the partial safety factor for stabilizing permanent actions (plug and soil weight).
Beyond the definition of overall or partial safety factors, a most important aspect to carefully consider is
the definition of a realistic value for the soil plug unit weight P. Such value should be carefully selected
according to the preliminary tests which should be normally foreseen in actual projects. As for a very
preliminary assessment, a value equal to about sat in the natural soil may be assumed.
According to such criteria, the depth HP of the jet-grout columns depends on the thickness HS of
unimproved natural soil just below excavation level. In some circumstances, HS is set to zero, since the
jet-grouting columns start just below the dredge line. In this case the jet-grouting plug also offers a very
relevant contribution to retaining wall stability. However no ballasting contribution by natural soil is
considered and the thickness of the plug may be quite large. Assuming that P sat, and that for sandy
soils, sat 2w Eurocode Design equation above, taking HS=0, becomes
H P 1.23 H exc
in which Hexc is the water table elevation with respect to the dredge line level (or top plug elevation) (see
Figure 3-4, left). In many practical cases, it may be more effective to start jet-grouting columns fairly below
the dredge line. In this case, a relevant reduction in jet-grouting volume is obtained. However, some
shortcomings arise as well, as summarized in Figure 3-4, right.
impervious
retaining wall
impervious
retaining wall
internal support
wall deformation
HP
HP
natural soil
Hs
Hexc
wall deformation
Hexc
internal support
jet-grouting plug
jet-grouting plug
Pro
Good contribution to wall
stability
Can be used as subfoundation for long-term
conditions, if possible
contra
Pro
Uneconomical
Cost-effective
contra
Minor contribution to wall toe
stability
Limited visual control
Unuseful for long term
conditions
Extra drilling required
Long walls may be required
24
Ce.A.S. s.r.l.
Advanced Modelling
interface at the wall toe must also be realized with much care. However, some residual water inflow must
be considered as well, and an appropriate pumping system must be provided accordingly.
When the total hydrualic head is very high, both aforementioned solutions may be unfeasible. In such
cases, a different strategy must be considered, as in the real case discussed in section 1.
3.3
The hydraulic barrier shown in Figure 2-1 (pertaining to a real excavation in Berlin) has been most likely
designed in accordance to the design criteria in Figure 3-4, right. For such case we have:
H w = (32 3) = 29 m
HS = (30 16.80) = 13.20 m
HP = 2 m
sat plug 2 w
therefore
Simplified Eurocode requirement is apparently not fulfilled. By further inspection of actual values, we
should observe what follows.
If we assume that the head loss is just dissipated along the soil plug, the water pressure at the plug
bottom face (point A in Figure 3-3) is
uA = (32-3)10 = 290 kPa
With the dredge line al EL -16.80, the total weight of the soil column above point A is
v,A (32-16.80)20 = 304 kPa > 290 kPa
A safety factor equal to about 1.05 is guaranteed.
Now consider the seepage reduction provided by simplified or more precise seepage analysis discussed
in section 2.3. By inspection of PARATIE PLUS results as outlined also in Figure 2-8, highest value for uA
is 272 kPa. Therefore a safety factor equal to about 1.11 is obtained, in accordance with traditional criteria
used prior to Eurocodes.
According to Eurocode 7 (Italian NAD), we have
0.90 v,A = 273 kPa
(design resistance)
(driving action)
Even in this case, Eurocode requirement is not met. However, if a higher value for hydraulic plug were
considered, according to actual data, a different conclusion may be reached.
Moreover, due to the very deep position of the soil plug, a quite low contribution to wall deformation can
be given by such plug: in this respect, the plug was most likely neglected in wall stability calculation.
In the PARATIE PLUS model, accordingly, the plug was simulated by simply setting the plug permeability
to 1/100 the natural permeability of the surrounding soil. Note that for such analysis it is only important to
reasonably define the relative values among different permeability zones. No soil stiffness of cohesion
improvement was defined at all.
Ce.A.S. s.r.l.
25
June 2014
PARATIEPLUS 2014
3.4
In the construction of the High Speed Railway Viaduct crossing the Po River near Piacenza (Italy), large
diameter bored piles are adopted for pier foundation system (Nova & Becci (2008))
Due to stringent hydraulic requirements, including severe provisions for scour prevention, the foundation
of most piers had to be placed at a quite low elevation with respect to the riverbed level. In order to allow
pier construction, deep excavations have been necessary, dealing with very severe water conditions.
STAGE A
Steel casings were driven down to around 15 m
below the riverbed, corresponding with the position
of foundation piles.
A lateral steel structure was fabricated just below
the top of such steel piles, in order to provide a
template for sheetpile installation and a lateral
supporting system.
STAGE B
AZ36 sheetpiles were driven all around the
foundation area.
26
Ce.A.S. s.r.l.
Advanced Modelling
STAGE C
A steel-concrete deck was fabricated at the top, so
as to create a working area well above the river.
Such offshore temporary islands were connected
to the land by means of temporary walkways.
STAGE D
Foundation Piles (28 2 m dia. drilled shafts) have
been put in place operating on that working deck.
Once all the piles have been completed, some load
tests have been conducted, using the O-Cell
technology.
STAGE E
About 4 m of the soil inside the working area was
dredged and a massive unreinforced concrete plug
was poured underwater. Then the water inside the
cofferdam was pumped off.
STAGE F
An additional collar was welded to each steel
casing to improve long term bond between piles
and concrete plug. An addition 50 cm concrete
layer was then cast in place.
STAGE G
Temporary steel casings projecting above have
been cut off and the concrete pile tops were
regularized.
Now the construction of foundation reinforcement
could start.
STAGE H
At bridge completion, sheetpiles have been
removed.
Most severe conditions for sheetpiles are represented by STAGE E, just before concrete pouring, as well
as quite long term conditions after STAGE F until bridge completion. Since STAGE E was considered
Ce.A.S. s.r.l.
27
June 2014
PARATIEPLUS 2014
quite temporary, less severe water conditions where selected, corresponding with average water level.
For subsequent stages, a more severe condition was assumed.
Two main design aspects will be discussed, namely:
: +51.00 m
: +37.00 m
:4m
: A=855 m
: 23 kN/m
: W = 855 4 23 = 78.66 MN
The weight of the plug is not sufficient to balance U, thefore some load must be resisted by existing piles
Overall action to be resisted by piles
: U-W=75.24
75.24 MN
Since 28 piles per pier are provided, each pile must ensure a tensile resistance equal to about
: N = 75.24 / 28 = 2.69 MN per pile
Such resistance is easily provided by the shaft resistance of the drilled piles. Of course the resistance of
the connection between each pile and the surrounding concrete pluge has been also addressed.
As for geotechnical check of the tension piles, the tension N has been amplified by an appropriate
coefficient, however, not much greater than unity, as the uncertainties are limited the unit weight of the
plug.
In the calculation of sheet piling the effect of the tension piles has been included by assigning, at the
bottom of the excavation, fictitious surcharge qs at least equal to:
qs = (U-W) / A = (75.24 MN) / 855 m = 88 kPa
3.4.2 Sheetpile
Sheetpile design
The aforementioned staging sequence can be easily modelled with PARATIE PLUS.
The soil at the site, at relevant depths for sheetpile design, is silty sand, for which the following constant
properties can be assumed:
d= 17 kPa
sat= 20 kPa
Evc=7.75 MPa
c'=0
=31
= /3
28
Ce.A.S. s.r.l.
Advanced Modelling
Ce.A.S. s.r.l.
29
PARATIEPLUS 2014
June 2014
zero at the plug top elevation. It's worth observing that the effective vertical stress below the plug is 0 in
both cases.
Maximum unfactored bending moment is about 986 kNm /m, which was deemed compatible with the
assumed sheetpile shape and a S355 material.
It's remarked that some other design aspects should be addressed, including an assessment of minimum
embedment depth for stability. Such check can be easily performed by means of the automatic cutting
procedure offered by PARATIE PLUS.
30
Ce.A.S. s.r.l.
Advanced Modelling
It's finally recommended to make some other experiments so as to get familiar with the seepage features
in PARATIE PLUS: for example a 2D seepage analysis for both approaches is suggested. In both cases,
fixed excess pore pressures must be imposed at both top boundaries as well as on the left side
(representing the riverbed), whereas symmetry conditions must be given along the right vertical side. In
next figure 2D analysis for approach 1 is anticipated.
Boundary conditions
3.5 References
Nova. R, Becci B., 2008, Experimental and Numerical Assessment of Osterberg Load Tests on Large
Bored Piles in Sand, 33rd Annual 11th International Deep Foundations Institute Conference
Proceedings, New York, NY, Oct 15-17, pp 225-233.
Ce.A.S. s.r.l.
31
PARATIEPLUS 2014
June 2014
Scope
Prediction capabilities of the clay model in PARATIE-PLUS in untrained soil modelling are assessed
Figure 4-1: A BRACED WALL IN "BOSTON BLUE CLAY" (Whittle & Hashash (1994))
The construction sequence comprises the following steps: 1) soil is initially excavated to a depth, hu=2.5m
with no internal bracing; 2) the wall is then propped at the surface; 3) subsequent excavation proceeds in
2.5m increments with crosslot supports installed at the current grade level (total excavation depth, H).
The analyses assume that there is minimal migration of pore water throughout the excavation process and
hence, the soil is subject to undrained shearing. Initial pore pressures are hydrostatic with the
groundwater table located at a depth, dw=2.5m (typical of conditions in Boston) and full capillarity in the
overlying clay. The clay is saturated and exhibits normalized engineering properties which are typical of
Boston Blue Clay (BBC). A deep layer of K0-normally consolidated clay (OCR=1) with in situ lateral earth
NC
pressures, K0 ='h0/'v0=0.53 is assumed.
Remaining soil parameters are discussed in the cited reference as well as, for example, in a study by
Maiorano & Aversa (1997) based on the same problem.
In these studies, two-dimensional finite element models and complex critical state constitutive laws are
adopted, like MCC (Modified Cam Clay) and MITMIT-E3:
E3 we note that such model need very different soil
parameters as respect to PARATIE data that will be derived from the previous by some approximations.
4.3
Approaching
Approaching this problem with PARATIE
As usual, an initial at rest recovery step is followed by progressive excavation phases in which rigid struts
are progressively activated which prohibit further wall deformations at each application level.
32
Ce.A.S. s.r.l.
Advanced Modelling
A very crude soil modelling is of course in used, with respect to the complex critical state models in the
cited studies: in PARATIE just a critical state (cv) and a peak friction angle, some thrust parameters ad
elastic moduli are needed.
As for cv, values between 26 and 32.5 will be adopted, whereas the peak angle will be computed based
on then suggested correlation included in the PARATIE PLUS reference manual.
Active and passive thrust coefficients are computed assuming a perfect wall adhesion to soil, according to
the cited reference (/=1).
As for elastic moduli, a relation in the form EVC=RVC(v/pa) is assumed, which is suitable for NC clay,
taking EUR/EVC=constant.
To estimate RVC, from Figure 4-2 the Gsec/ v ratio can be derived, depending on the shear strain level ,
which can be anticipated based on Figure 4-3.
modulus
(after
Corresponding with =12%, Gsec/ v ranges between 1030: therefore RVC values between 2000 and
6000 kPa will be adopted.
Since OCR=1, the analysis can directly start in undrained conditions because PARATIE PLUS assumed
drained initial conditions anyway.
Ce.A.S. s.r.l.
33
PARATIEPLUS 2014
4.4
June 2014
Result discussion
Results by Whittle & Hashash (1994, as shown in the next figures, highlight some differences among
different constitutive model predictions.
Figure 4-4: wall deformation at different excavation depths (after Whittle & Hashash (1994))
In the following, PARATIE PLUS results will be compared with the MIT-E3 ones, which will be assumed as
the exact solution.
The investigated parameter assumptions within reasonable ranges as discussed above are summarized
in the following table.
34
Ce.A.S. s.r.l.
Advanced Modelling
EUR/EVC=5
cv
Rvc=3000
kPa
EUR/EVC=3.75
Rvc=4000
kPa
26
Rvc=3000
kPa
Rvc=4000
kPa
EUR/EVC=3
Rvc=3000
kPa
Rvc=4000
kPa
C1
27
C2
C3
C5
28
C6
29
30
C4
32.5
C7
In the next figures, wall deformed shapes computed by PARATIE PLUS at different excavation stages are
compared with Whittle & Hashash MIT-E3 results.
MIT-E3 vs PARATIE
'cv=26 Evc = 40 'v Eur /E vc=5
-200
-150
-100
-50
0
0
-5
-10
-15
10
-20
depth z [m]
-25
-30
-35
-40
H=
22.5
20
15
-45
Ce.A.S. s.r.l.
35
June 2014
PARATIEPLUS 2014
MIT-E3 vs PARATIE
'cv=27 Evc = 30 'v E ur /Evc=5
-200
-150
-100
-50
0
0
-5
-10
depth z [m]
10
H=
22.5
20
-15
-20
-25
-30
-35
-40
-45
15
MIT-E3 vs PARATIE
'cv=27 Evc = 40 'v E ur /Evc=5
-200
-150
-100
-50
0
0
-5
-10
depth z [m]
10
H=
22.5
20
15
-15
-20
-25
-30
-35
-40
-45
36
Ce.A.S. s.r.l.
Advanced Modelling
MIT-E3 vs PARATIE
'cv=30 Evc = 30 'v E ur /Evc=5
-200
-150
-100
-50
0
0
-5
-10
-15
-20
-25
-30
depth z [m]
10
H=
22.5
20
-35
-40
-45
15
-150
-100
-50
0
0
-5
-10
-15
depth z [m]
10
-20
-25
-30
-35
-40
H=
22.5
20
15
-45
Ce.A.S. s.r.l.
37
June 2014
PARATIEPLUS 2014
MIT-E3 vs PARATIE
'cv=28 Evc = 40 'v E ur /Evc=3
-200
-150
-100
-50
depth z [m]
10
H=
22.5
20
0
-5
-10
-15
-20
-25
-30
-35
-40
-45
15
MIT-E3 vs PARATIE
'cv=32.5 Evc = 40 'v Eur /Evc=3
-200
-150
-100
-50
0
0
-5
depth z [m]
10
H=
22.5
20
15
-10
-15
-20
-25
-30
-35
-40
-45
38
In intermediate steps, (H=10 and 15m), the exact solution is reproduced with best agreement by
C2 and C4 parameters; C6 and C7 choices are also good.
Ce.A.S. s.r.l.
Advanced Modelling
The solution at H=22.5m is somehow similar to the exact one only in C6 and C7 cases, whereas
for other parameter assumptions, the predicted solution is very different.
at final excavation stage, a solution somehow similar to the exact one may be obtained provided
EUR/EVC 3
In summary, C6 and C7 solutions show the best agreement with the reference ones. As far as elastic
behaviour prevails (until H<15m), the Eur modulus plays an important role: in C6 solution, such parameter
is set equal to 120(v) which therefore seems to be the best estimate choice.
In Figure 4-13, maximum bending moments by PARATIE PLUS are compared with the exact ones in
Figure 4-5. PARATIE PLUS predictions are in very good agreement with the exact ones, at least down to
excavation depth H=20m. At H=22.5, PARATIE PLUS predictions agree with reference one, limited to
cases C6 and C7, as already observed for deformations.
MIT-E3 vs PARATIE
2,0
MIT-E3
Whittle&Hashash (1994)
PARATIE C1
1,5
PARATIE C2
PARATIE C3
1,0
PARATIE C4
PARATIE C5
0,5
PARATIE C6
PARATIE C7
0,0
12
16
20
24
Figure 4-13:
13: PARATIE PLUS - MITMIT-E3 comparison: max wall bending moments vs. excavation depth
4.5 Conclusions
In this example, the computed results by the undrained clay model included in PARATIE PLUS are in
agreement with the solutions obtained by much more complex numerical models.
In intermediate analysis steps, predicted wall deformations by PARATIE PLUS are very near to those
shown in the cited reference, adopting consistent compressibility parameters with the ones used in the
comparison studies.
Near to wall failure conditions, PARATIE PLUS solution highly depends on selected Eur/Evc ratio. This is in
agreement with the assumption that such ratio governs the stress path slope toward passive (or active)
limit state, as shown in the Reference Manual.
4.6 Acknowledgment
Acknowledgment
We would like to acknowledge Professor Andrew Whittle and Professor Youssef Hashash for their kind
and valuable information about their work on which this example is taken. Dr. Alessandro Flora is also
acknowledged.
Ce.A.S. s.r.l.
39
PARATIEPLUS 2014
June 2014
4.7 References
Whittle, A.J. & Hashash, Y.M.A (1994) Soil modeling and prediction of deep excavation behavior, Proc.
Intl. Symp. on Pre-Failure Deformation Characteristics of Geo-Materials (IS-Hokkaido), pp. 589-594
Maiorano, R.M.S, Aversa S., (1997) Modellazione numerica di uno scavo in argille normalmente
consolidate: una variazione sul tema, Atti convegno Il modello geotecnico del sottosuolo nella
progettazione delle opere di sostegno e degli scavi, Perugia, pp. 297-311
40
Ce.A.S. s.r.l.
Advanced Modelling
(M)
(S)
(S)
If the NONLINEAR GAP option is not activated, they share also the same lateral
displacement.
If the NONLINEAR GAP option is activated, they share the lateral displacement
only if they are in contact, i.e. if the master node pushes the slave and viceversa
If the force between them is a tensile force, such pair of nodes miss their bond
and behave as separate entities.
PARATIE PLUS automatically determines such conditions, by means of the
numerical procedures explained in the Theory Manual.
The aim of such feature is to simply model two adjacent walls interacting if they
are in touch. A very simple model is now discussed to get familiar with this
feature.
Ce.A.S. s.r.l.
41
June 2014
PARATIEPLUS 2014
EJ
B'
L1
B'
P
yA
X
A
L2
yB
B
NON-LINEAR
GAP INTERFACE
EJ
y A , master =
(P X ) L3
3EJ
= y A , slave =
X L31
3EJ
==>
X=
P L3
L3 + L31
Therefore
y B = y A + A L 2 where A =
X L21
2EJ
Assume
a rectangular concrete section with H=600 mm for both beams:
EJ=539280 kNm/m
P=10 kN/m
L=15 m
42
L1=7 m L2=5.50 m
Ce.A.S. s.r.l.
Advanced Modelling
We obtain
X= 0.90775 P = 9.0775 kN/m
yA=1.9245 mm
A=0.0041 rad
yB=4.1927 mm
Ce.A.S. s.r.l.
43
June 2014
PARATIEPLUS 2014
As for design purposes, a uniform sand is assumed, with a coefficient of permeability k=110 m/s. The
design chart in Figure 6-2 is used.
d=8 m
D=10 m
h=8 m
therefore
b/D=0.8
0.8
d/D=1
We assume two wellpoints (w.p.) lines at each side of the excavation, as shown in the next figure. At each
w.p. one half of the total flow q is assigned.
44
Ce.A.S. s.r.l.
Advanced Modelling
According to the estimate above, at each w.p. an outflow q/2 equal to 310-4 m/s is assigned.
The model is extended by 40 m at each sided as well as below ground level. The results are shown in the
next plots.
Assumption A
-4
Ce.A.S. s.r.l.
45
PARATIEPLUS 2014
June 2014
Assumption B
-4
Results according to ASSUMPTION A are satifactory, because the phreatic line inside excavation is
realistic and fairly below the dredge line.
ASSUMPTION B results show that the assigned flow is excessive, since the water table in the
surrounding soil can move and no flow can enter the system through such boundary, in contrast with
ASSMPTION A. So the requested flow to be discharged is a little bit less than in ASSUMPTION A.
Let try applyng about 90% of the dicharge flow above. We obtain the following results, which are
satisfactory.
-4
-4
Figure 6-6: ASSUMPTION B - results for q/2 = 2.7x10
2.7x10 m/s
6.1 Comments
In this example, just steady state seepage analysis aspects are considered. No retaining wall behaviour is
investigated. In this respect the seepage analysis has been assigned as the first and only analysis stage.
This simple example aims at showing that a realistic seepage analysis set up requires a trial and error
procedure to reasonably match expected results: several hydraulic boundary conditions shold be
assessed, as well as different solution strategies; the assigned discharge flow at well points must also be
tuned iteratively.
46
Ce.A.S. s.r.l.
Advanced Modelling
Once acceptable results are obtained, seepage analysis pressures can be coupled with ordinary
PARATIE analysis of the retaining wall.
Limited to this example, we believe that the predicted discharge flow according to ASSUMPTION B is
more realistic, but the results given by ASSUMPTION A are more conservative, as far as w.p. design is
considered. The water pressures on the walls are quite the same for both approaches.
Finally it is worth noting that the results obtained by the seepage analysis may be slighlty unsymmetric
even if the model is symmetric. This is due to the fact that a slightly unsymmetric finite element mesh may
be automatically generated by PARATIE PLUS.
6.2 References
Lancellotta R. (1987) Geotecnica, 1st ed., Zanichelli, Bologna
Ce.A.S. s.r.l.
47
PARATIEPLUS 2014
June 2014
48
Ce.A.S. s.r.l.
Advanced Modelling
Ce.A.S. s.r.l.
49
PARATIEPLUS 2014
June 2014
50
Ce.A.S. s.r.l.