Академический Документы
Профессиональный Документы
Культура Документы
BY:
W.P.R Indrajith (168913V)
1.0 Introduction............................................................................................................. 1
Variables:
Radius of the tank R
Height of the tank H
Thickness of the cylinder wall t
Results to be discussed
Hoop stress-resultant N
Vertical stress-resultant Nx
Page 1
2 Hoop stress-resultant ( N ) and Vertical stress-resultant ( Nx )
Let's use membrane hypothesis. Considering the pressure by liquid at which the point which we
consider is P, then the equilibrium of the half of the cylinder is considered and shown below.
Considering a unit height along the cylinder, force equilibrium can be written as follows.
There are two forces acting in the system where water pressure and self weight of the wall. But since
we consider only the water pressure loading case, there is no effective force acting along the vertical
direction of the wall. Hence the vertical stress resultant should be zero.
Nx = 0
Page 2
3 Choosing of variables R, H, t
R 1000 H/R 4
H 4000 t/H 0.025
t 100 t/R 0.1
4.1 Properties & boundary conditions in SAP 2000 for initial analysis
6 2
Modulus of elasticity 34 x 10 kN/m
Member - shell - membrane - 100 mm in thickness
Fixed at X=0 level
Define material
Page 3
Define tank wall section - membrane
Page 4
Extruded view of the generated tank Assigning boundary conditions - Fixed base
Page 5
Application of water pressure using the joint pattern
Page 6
Area local axis in the model
horizontal
dir.
local 2 dir.
vertical dir.
local 1 dir.
local 3 dir.
Red colour arrow shows - Local axis 1 - Results out put F11 - N
White colour arrow shows - Local axis 2 - Results out put F22 - Nx
Indigo colour arrow shows - Local axis 3 - Results out put F12
Page 7
Hoop stress resultant distribution of the tank wall
Water tank with the selected geometric parameters (H=4m, R=1m, t=100mm) is modelled with 4
types of mesh arrangement and Hoop stress resultant along the vertical wall is compared in
different cases.
SAP2000 results along the wall for each mesh arrangement is shown below
All results start to converge well in 100mmx100mm mesh arrangement. Hence let's select
100mmx100mm mesh for the rest of the modelling works.
Page 8
SAP 2000 Hoop stress-resultant contours with each meshing arrangements
Page 9
5 Results comparison
Considering following geometric parameters of water tank, for each case, finite element models
were developed with 3 member formulations as,
- Membrane sections
- Thin shell sections
- Thick shell sections
Geometric parameters
Value (mm) Value
Variable Case 1 Case 2 Case 3 Variable Case 1 Case 2 Case 3
R 1000 2000 4000 H/R 4 2.5 0.625
H 4000 5000 2500 t/H 0.025 0.03 0.05
t 100 150 125 t/R 0.1 0.075 0.03125
Page 10
Case 2 - H=5000mm, R=2000mm, t=150mm
Page 11
Case 3 - H=2500mm, R=4000mm, t=125mm
Page 12
5.2 Results comparison - General
N vs X , Case 1 - H=4000mm
60
50
40
SAP Membrane
N (kN/m)
30
SAP Thin shelll
20 SAP Thick shelll
Manual calc
10
0
0 1000 2000 3000 4000 5000
-10
X (mm)
Nx vs X , Case 1 - H=4000mm
4
3
2
1
0 SAP Membrane
Nx (kN/m)
-3 Manual calc
-4
-5
-6
X (mm)
Page 13
5.2.2 Variation of N and Nx vs X , case 2 ( H = 5000mm, R = 2000mm, t =150mm )
N vs X , Case 2 - H=5000mm
140
120
100
80 SAP Membrane
N (kN/m)
0
0 1000 2000 3000 4000 5000 6000
-20
X (mm)
Nx vs X , Case 2 - H=5000mm
6
4
2
0
0 1000 2000 3000 4000 5000 6000
-2 SAP Membrane
Nx (kN/m)
-8 Manual calc
-10
-12
-14
X (mm)
Page 14
5.2.3 Variation of N and Nx vs X , case 3 ( H = 2500mm, R = 4000mm, t =125mm )
N vs X , Case 3 - H=2500mm
140
120
100
80 SAP Membrane
N (kN/m)
0
0 500 1000 1500 2000 2500 3000
-20
X (mm)
Nx vs X , Case 3 - H=2500mm
10
0 SAP Membrane
Nx (kN/m)
-10
-15
X (mm)
Page 15
5.2.4 Bending moment in X direction vs X , case 1 ( H = 4000mm, R = 1000mm, t =100mm )
Mx vs X , Case 1 - H=4000mm
1.5
1
Mx (kNm/m)
-0.5
X (mm)
Mx vs X , Case 2 - H=5000mm
5
4
3
Mx (kNm/m)
2
SAP Thin shell
1
SAP Thick shell
0
-1 0 1000 2000 3000 4000 5000 6000
-2
X (mm)
Mx vs X , Case 3 - H=2500mm
3
2
Mx (kNm/m)
Page 16
5.3 Results comparison - Varying parameters
Based on the results comparison done in section 5.2, It seems that H/R ratio and thickness of the
element may have an influence to stress resultants. In order to further investigate the behaviour, 9
more SAP models were developed using membrane elements as described below.
H = 2500 mm
t = 125 mm R = 4000 mm
R = 4000 mm R = 2000 mm R = 1000 mm t = 100 mm t = 125 mm t= 150 mm
Model 1 Model 2 Model 3 Thin shell Thick shell Thin shell Thick shell Thin shell Thick shell
N vs X , H=2500mm
140.00 H/R 0.625
Manual
120.00
H/R 0.625
100.00 SAP
Manual
60.00
H/R 1.25
40.00 SAP
H/R 2.5
20.00 Manual
0.00 H/R 2.5
0 500 1000 1500 2000 2500 3000 SAP
-20.00
X (mm)
Page 17
5.3.2 Effect of Thin shell and Thick shell in Variation of N, Keeping H and R constant and vary t
N ( kN/m)
t=100 mm ( t/H= 0.04, t/R= 0.025) t=125 mm ( t/H= 0.05, t/R= 0.03) t=150 mm ( t/H= 0.06, t/R= 0.038)
X(mm) SAP Thin shell SAP Thick shell SAP Thin shell SAP Thick shell SAP Thin shell SAP Thick shell
0 -0.06 -0.07 -0.05 -0.06 -0.04 -0.05
100 2.83 3.4 2.23 2.9 1.83 2.6
200 9.70 10.7 7.80 8.9 6.50 7.7
300 18.60 19.7 15.10 16.4 12.70 14.2
400 27.80 28.9 23.00 24.3 19.50 21.1
500 36.50 37.5 30.70 32 26.40 27.1
1000 56.10 56.2 51.50 51.9 47.20 47.8
1500 43.80 43.7 43.50 43.4 42.50 41.8
2000 21.70 21.6 22.90 22.8 23.90 23.7
2500 1.20 1.1 0.20 0.2 1.80 6.3
N vs X , H=2500mm, R=4000mm
120.00
100.00
80.00
20.00
0.00
0 500 1000 1500 2000 2500 3000
-20.00
X (mm)
Page 18
6.0 Observations and discussion
- Values obtained for N and Nx with manual calculations with membrane theory, SAP2000 using
membrane elements, SAP2000 using Thin shell and SAP2000 using Thick shells are not exactly the
same.
- Basically the variation shows near to the base ( X = 0) as shown in the following graph.
- As we goes along the vertical axis ( increasing X), all the results start to converge well with
membrane theory regardless of the element formulation.
N vs X , Case 1 - H=4000mm
60
50
40
SAP Membrane
N (kN/m)
30
SAP Thin shelll
20
SAP Thick shelll
10 Manual calc
0
0 1000 2000 3000 4000 5000
-10
X (mm)
- Situation is quite same in the case of Nx as well as shown below. Although Nx is zero for entire
tank in accordance to the membrane theory, SAP2000 results give a numerical value in all cases,
yet they converges to zero after a certain height of the tank.
Nx vs X , Case 1 - H=4000mm
4
3
2
1
0 SAP Membrane
Nx (kN/m)
-3 Manual calc
-4
-5
-6
X (mm)
Page 19
- The variation of N between Membrane theory and SAP2000 (using membrane elements) shows a
little variation with changing H/R ratios as shown in the graph below. Although the variation is not
that significant, it can be seen that with higher H/R ratios, values converges better than in the
case of lower H/R rations.
N vs X , H=2500mm
140.00 H/R 0.625
Manual
120.00
H/R 0.625
100.00 SAP
Manual
60.00
H/R 1.25
40.00 SAP
H/R 2.5
20.00 Manual
0.00 H/R 2.5
0 500 1000 1500 2000 2500 3000 SAP
-20.00
X (mm)
-
The effect of using thin shell and thick shells did not make a significant impact even with varied
wall thickness with respect to the height and radius of the tank as shown below.
N vs X , H=2500mm, R=4000mm
120.00
100.00
80.00
t/H 0.04, t/R 0.025 Thin shell
t/H 0.04, t/R 0.025 Thick shell
60.00
N (kN/m)
0.00
0 500 1000 1500 2000 2500 3000
-20.00
X (mm)
Page 20
- The effect of using thin shell and thick shells did not make a significant impact for bending
moment Mxx as shown below. Values are almost the same although there is a absolute difference
between values.
Mx vs X , Case 2 - H/R=2.5
5
4
3
Mx (kNm/m)
2
SAP Thin shell
1
SAP Thick shell
0
-1 0 1000 2000 3000 4000 5000 6000
-2
X (mm)
Mx vs X , Case 3 - H/R=0.625
3
2
Mx (kNm/m)
Page 21
7.0 Conclusions
- Different H/R ratios show the same variation, in general, but it can be observed that there is an
improvement for higher H/R ratios.
- As far as hoop stress resultant concern, use of Thin and Thick shells did not show a remarkable
variation of the results, although numerical values show a difference.
- With all the observations, it can be seen that there is a significant effect of the Boundary
conditions also. Specially the much higher value of hoop stress and Vertical stress resultant not
being zero as expected can lead to that conclusion. Hence a separate model was developed with a
different joint restrain to the this effect.
A model was developed with restraining the base only along vertical axis (Z) and the results of N
and Nx follow exactly as per Membrane theory. The results are shown below.
Page 22
- Another important observation was made during the study. During the mesh sensitivity study,
several models were required to develop from a coarse mesh to a fine mesh. In general cases
which we encounter during daily works, meshing of slabs is a common activity which we do with
SAP2000. In this meshing process, the selected area or areas can be meshed as per the required
mesh density. But in this assignment, this general meshing process with SAP2000 for membrane
sections was not worked as expected.
Following figures show the way of unsuccessful work of general meshing lead to incorrect results
F11- Initial 500mmx500mm mesh F11- Incorrect results for 250mmx250mm mesh which
was obtained from 500mmx500mm model
General meshing produced results which are incorrect. Hence reproducing of the model by specifying
meshing intervals at the very beginning produced a correct model. Results out put with the correct
model is shown below for comparrison.
Page 23