You are on page 1of 34

`

DESIGN COMPARISONS
FOR CASE STUDIES USING
SAP2000
SAFE NODAL METHOD
AND SAFE INTERNAL METHOD (WOOD-ARMER)
FOR THIN AND THICK PLATES

Jan. 2008
TABLE OF CONTENTS

I REPORT SCOPE .................................................................................................................... 2


II THICK PLATE EXAMPLE ................................................................................................. 3
II-1- MODEL DESCRIPTION ................................................................................................... 3
II-2- MOMENTS IN X DIRECTION ......................................................................................... 4
II-3- MOMENTS IN Y DIRECTION ....................................................................................... 12
III THIN PLATE EXAMPLE ................................................................................................. 20
III-1- MODEL DESCRIPTION ................................................................................................ 20
III-2- MOMENTS IN X DIRECTION ..................................................................................... 21
III-3- MOMENTS IN Y DIRECTION ..................................................................................... 26
IV CONCLUSION ................................................................................................................... 30
APPENDIX A .............................................................................................................................. 31
APPENDIX B............................................................................................................................... 32

Page 1
I REPORT SCOPE

We have been lately notifying unjustified reinforcement when using SAFE models for raft
foundations supported on soil. This situation was mainly detected in case of irregular geometry
of models, unsymmetrical grids of columns, shear walls near the edge of the raft, etc…
The above pushed us to contact the “CSI Technical Support” (refer to Appendix A) and we were
advised to use the internal method (Wood-Armer) along with thick plate analysis in case of point
loads (Raft foundations, Transfer slabs…refer to Appendix B).

In what follows, two shell examples (raft foundation and suspended slab) will be the subject of a
study to compare between the analysis made by SAP2000 and the analysis made by SAFE using
both the nodal method and the internal method (Wood-Armer). The results of this study will be
adopted as a base for all horizontal shell designs.

Page 2
II THICK PLATE EXAMPLE

II-1- MODEL DESCRIPTION

For this example, a raft foundation on soil, supporting two corewalls near the edge and
rectangular grids of walls and columns, was modeled as a thick plate on SAP2000 and on SAFE
V8.0.8.

Fig. 1: Raft Model Geometry on SAFE

Page 3
Fig. 2: Raft Model Geometry on SAP2000

II-2- MOMENTS IN X DIRECTION

If we are to compare the moments in X direction between the SAP2000 model and the SAFE
model, the difference under the right corewall is about 0.4% (59.5t.m for SAP2000 and 59.3t.m
for SAFE) and between the two corewalls about 0.3% (33.96t.m for SAP2000 and 34.06t.m for
SAFE) and directly under a column about 1.3% (148.48t.m for SAP2000 and 150.40t.m for
SAFE). The above differences are truly minimal and can be neglected.

Page 4
Fig. 3: X Moment on SAP2000 Model

Page 5
Fig. 4: X Moment on SAFE Model

Page 6
Fig. 5: MXY Moment on SAP2000 Model

Page 7
Fig. 6: MXY Moment on SAFE Model

Page 8
Assuming that the moments differences between the SAP2000 model and the SAFE model are
insignificant all over the raft model, the SAFE model will directly calculate the strips moments
(column strips and middle strips), using an average value along the width of the strip:

Fig. 7: X Strips Moment From SAFE Model

Particularly, the two values specified on Fig.7 will be considered:


a) 70.534t.m on a 2.22m wide strip and a raft thickness of 1.0m => Reinf. should be 27.13cm2
b) 66.70t.m on a 2.95m wide strip and a raft thickness of 1.0m => Reinf. should be 25.61cm2

Page 9
At these particular locations, the approximate averaging of Mxy on the considered strips is:
a) 16t.m + 70.53 = 86.53t.m =>33.33cm2 of reinforcement.
b) 1t.m + 66.70 = 68.7t.m => 26.4cm2 of reinforcement.

Considering the reinforcement computed by the SAFE nodal method (Fig.8), the following
results were found:
a) 102.9cm2 of reinforcement.
b) 23.5cm2 of reinforcement.

Fig. 8: X Strips Reinforcement From SAFE Model Using Nodal Method

Page 10
Considering the reinforcement computed by the SAFE internal method (Fig.9), the following
results were found:
a) 33.3cm2 of reinforcement.
b) 26.6cm2 of reinforcement.

Fig. 9: X Strips Reinforcement From SAFE Model Using Internal Method (Wood-Armer)

Page 11
Considering the above results, the use of the combination of moments in Wood-Armer SAFE
method (Mxx + Mxy) is converging with the hand calculations using SAP2000 results. The
nodal method in this case is giving “excessively conservative” reinforcement under the corewalls
(Refer to CSI technical support reply in Appendix A).

II-3- MOMENTS IN Y DIRECTION

Comparing the moments in the Y direction between the SAP2000 model and the SAFE model,
the difference will remain in the same margins as the X Direction.

Page 12
Fig. 10: Y Moment on SAP2000 Model

Page 13
Fig. 11: Y Moment on SAFE Model

Page 14
Fig. 12: MXY Moment on SAP2000 Model (Fig.5)

Page 15
Fig. 13: MXY Moment on SAFE Model (Fig.6)

Page 16
Assuming that the moments differences between the SAP2000 model and the SAFE model are
insignificant all over the raft model, the SAFE model will directly calculate the strips moments
(column strips and middle strips), using an average value along the width of the strip:

Fig. 14: Y Strips Moment From SAFE Model

Page 17
Considering the reinforcement computed by the SAFE nodal method (Fig.15) and the SAFE
internal method (Wood-Armer) (Fig.16), the following results were found:

Fig. 15: Y Strips Reinforcement From SAFE Model Using Nodal Method

Page 18
Fig. 16: Y Strips Reinforcement From SAFE Model Using Internal Method (Wood-Armer)

Considering the above results, the use of the combination of moments in Wood-Armer SAFE
method (Mxx + Mxy) is converging with hand calculations using the SAP2000 values, but the
difference between the two methods is due to additional Mxy only.

Page 19
III THIN PLATE EXAMPLE

III-1- MODEL DESCRIPTION

For this example, a solid slab, supported on walls and columns, was modeled as a thin plate
using SAP2000 and using SAFE V8.0.8.

Fig. 17: Slab Model on SAFE

Page 20
Fig. 18: Slab Model on SAP2000

III-2- MOMENTS IN X DIRECTION

Comparing the moments in X direction between the SAP2000 model and the SAFE model, the
difference at the column on axes B-3 is about 1.1% (32.4t.m for SAP2000 and 32.05t.m for
SAFE) and between the two columns C-3 and D-3 about 0.02% (6.478t.m for SAP2000 and
6.479t.m for SAFE) and at the tip of the wall on axis 4 about 0.3% (14.496t.m for SAP2000 and
14.537t.m for SAFE). The above differences are minimal and can be neglected.

Page 21
Fig. 19: X Moment on SAP2000 Model

Fig. 20: X Moment on SAFE Model

Page 22
Fig. 21: MXY Moment on SAP2000 Model

Fig. 22: MXY Moment on SAFE Model

Page 23
Assuming that the moments differences between the SAP2000 model and the SAFE model are
insignificant all over the raft model, the SAFE model will directly calculate the strips moments
(column strips and middle strips), using an average value along the width of the strip:

Fig. 23: X Strips Moment From SAFE Model

Particularly, the two values on Fig.23 will be considered:


a) 14.326t.m on a 3.0m wide strip and a slab thickness of 0.3m => Reinf. should be 19.96cm2
b) 45.768t.m on a 3.0m wide strip and a slab thickness of 0.3m => Reinf. should be 50.53cm2

At these particular locations, the approximate averaging of Mxy on the considered strips is:
a) 4.5t.m + 14.33 = 18.83t.m =>26.4cm2 of reinforcement.
b) 5t.m + 45.77 = 50.77t.m => 56.3cm2 of reinforcement.

Considering the reinforcement computed by the SAFE nodal method (Fig.24) the following
results were found:
a) 15.45 cm2 of reinforcement.
b) 50.534 cm2 of reinforcement.

Page 24
Fig. 24: X Strips Reinforcement From SAFE Model Using Nodal Method

Considering the reinforcement computed by the SAFE internal moment method (Fig.25) the
following results were found:
a) 26.758 cm2 of reinforcement.
b) 55.927 cm2 of reinforcement.

Fig. 25: X Strips Reinforcement From SAFE Model Using Internal Method (Wood-Armer)

Page 25
Considering the above results, the use of the combination of moments in Wood-Armer SAFE
method (Mxx + Mxy) is converging with the hand calculations using SAP2000 values. The
SAFE nodal method in this case will give reinforcement taking into account Mxx only.

III-3- MOMENTS IN Y DIRECTION

Comparing the moments in the Y direction between the SAP2000 model and the SAFE model,
the difference will remain in the same margin as the X Direction, which can also be neglected.

Fig. 26: Y Moment on SAP2000 Model

Page 26
Fig. 27: Y Moment on SAFE Model

Fig. 28: MXY Moment on SAP2000 Model (Fig.21)

Page 27
Fig. 29: MXY Moment on SAFE Model (Fig.22)

Assuming that the moments differences between the SAP2000 model and the SAFE model are
insignificant all over the raft model, the SAFE model will directly calculate the strips moments
(column strips and middle strips), using an average value along the width of the strip:

Fig. 30: Y Strips Moment From SAFE Model

Considering the reinforcement computed by the SAFE nodal method (Fig.31) and the SAFE
internal method (Wood-Armer) (Fig.32), the following results were achieved:

Page 28
Fig. 31: Y Strips Reinforcement From SAFE Model Using Nodal Method

Fig. 32: Y Strips Reinforcement From SAFE Model Using Internal Method (Wood-Armer)

Page 29
Considering the above results, the use of the combination of moments in Wood-Armer SAFE
method (Mxx + Mxy) is converging with hand calculations using SAP2000 values, but the
difference between the two methods is due to additional Mxy only.

IV CONCLUSION

Considering the results found for both thick and thin shell elements using SAFE and SAP2000,
and referring to the response made by CSI Technical support team (Appendix A), it is
recommended to adopt the following analysis methods for each case described below:
a. For shell elements supporting point loads (ex.: raft foundations and transfer slabs…);
Model the shell as thick plate and use the internal moment method for the design
(Wood-Armer).
b. For shell elements supported on columns and walls, and having irregularities in column
grids, slab limits, and geometry; Use the internal moment method to take into
consideration the torsional moment Mxy that shall not be neglected.
c. For slabs supported on columns and walls and having regular column grids, slab limits
and geometry, there is no difference whether to use the nodal method or the internal
moment method in the design, and this is due to the insignificant torsional moments
Mxy in the slab; both methods should give approximately the same results.

Finally, it would be preferable to adopt the internal Wood-Armer method at all cases since it will
always detect any torsional effects that might be neglected using nodal method, leading to under-
design of structures.

Page 30
APPENDIX A

From: CSI Technical Support [mailto:support@csiberkeley.com]


Sent: Thursday, December 13, 2007 7:47 PM
To: Charbel Ghanem; CSI Technical Support
Subject: RE: - (DarRef: GENERAL.1/ Bey-07-129910 EML)

Dear Charbel,

SAFE version 6.46 and later versions allows users to choose between the nodal moments method of
design as documented in the SAFE manual and the Wood-Armer method of design as documented in
Eurocode 2 1992, Section A2.8(3). The Wood-Armer method explicitly accounts for Mxy moments and
some jurisdictions require that. The results are normally close for the two methods except when
concentrated loads are present and the thick plate element has been used. In this case the nodal
moment method gives overly conservative results. The user chooses the method to use in the
Preference Dialog box.

Our testing shows that the nodal method and the internal moments method (Wood-Armer) produce
similar reinforcements when the reinforcement is being provided in essentially the principal directions. In
this case the Wood-Armer method gives slightly higher reinforcements as it involves some absolute
terms. On the other hand when the twist governs the Wood-Armer method is more reliable. Whether the
nodal method gives more or less reinforcement is dependent upon how we treat moments of different
signs at the two nodes of the same element, what element (thin or thick) is used, the aspect ratio of the
mesh, etc. Actually, the main reason we put the Wood-Armer method in SAFE, besides that some
codes now require it, is that we were getting very large reinforcements in mats under the columns
when the nodal method was used with thick shell elements.

Regards,
Faisal

From: Charbel Ghanem [mailto:Charbel.Ghanem@dargroup.com]


Sent: Thursday, December 13, 2007 6:11 AM
To: CSI Technical Support
Subject: (DarRef: Not Referenced)

Dear Mr. or Mm,

I have an important question regarding the SAFE software, and I will appreciate it if I get an answer as
soon as possible.
I am doing a SAFE model for a raft foundation on soil support; using a thick plate shell, I am obtaining
reinforcement under a core wall that is more than three times the reinforcement obtained from thin plate
analysis. So please let me know when to use thick plate analysis, knowing that from my experience, the
“thick plate” should be used for shells where shear deformation is important ex. rafts and transfer slabs…,
and is it normal to obtain that much increase in reinforcement under core wall, knowing that the design
ultimate shears and moments are not compatible with the reinforcement results.

Waiting for your reply, please accept my sincere salutations.

Best Regards
Charbel Ghanem

Page 31
APPENDIX B

SAFE v8 - CSiDETAILER
ANALYSIS, DESIGN and DRAFTING of SLAB SYSTEMS
SAFE is a special purpose program that automates the analysis and design of simple to complex
concrete flat plates and foundation systems using powerful object based modeling. The program
can analyze and design slabs or mats of arbitrary shapes and varying thickness, drop panels,
openings, edge beams and discontinuities. Foundations can be combinations of Mats, Strip Footings
or Isolated Spread Footings.

Page 32
Introduction

SAFE is a sophisticated, yet easy to use, special purpose analysis and design program developed
specifically for concrete Slab/Beam, Basemat/Foundation systems. SAFE couples powerful object-
based modeling tools with an intuitive graphical interface, allowing the user to quickly and efficiently
model slabs of regular or arbitrary geometry with openings, drop panels, ribs, edge beams, and slip
joints supported by columns, walls or soil. Design is seamlessly integrated with modeling and
analysis, and provides comprehensive reporting of the required reinforcing calculated by the program
based on the user’s choice of design code. And with the optional CSiDETAILER program, detail
drawings may be produced almost effortlessly for the slabs and beams designed using SAFE.

The analysis is based upon the Finite Element method in a theoretically consistent fashion that properly
accounts for the effects of twisting moments. Meshing is Automated based upon User Specified
Parameters. Foundations are modeled as thick plates on Elastic Foundations, where the
Compression Only Soil Springs are automatically discretized based upon a modulus of subgrade
reaction that is specified for each Foundation Object.

The Software produces reinforcing layouts and evaluates the effects of punching shear around column
supports. Options are available for including cracked properties in the finite element model based upon
the slab reinforcing that is provided.

Also a comprehensive export option from ETABS is available that will automatically create complete
SAFE models of any ETABS floor or foundation for immediate design by SAFE.

Based on the finite element method, this program offers accuracy and flexibility that cannot be
matched by traditional hand calculations or equivalent frame computer programs. Traditional methods
for the analysis of simple slab systems are tedious and time consuming, and are often inapplicable for
complex geometries or loadings. General purpose finite element programs are capable of handling
much more complex models, but are often cumbersome and difficult to use and also produce results
that are not directly usable by the structural engineer.

Page 33

You might also like