COMPUTERS & STRUCTURES INC.

Software Verification
PROGRAM NAME: REVISION NO.:

SAP2000 2

CONCLUSIONS
The conclusions are presented separately for frame, shell, plane, asolid, solid and link elements, as well as for analysis cases in the following subsections. FRAMES The SAP2000 verification and validation example problems for frames all show acceptable, and in many cases exact, comparison with the independent solutions. The accuracy of the SAP2000 results for certain classes of frame examples depends on the discretization of the frame objects. For those classes of examples, as the discretization is refined, the solution becomes more accurate. The table below lists those classes of examples and the verification examples that address them.
CLASSES OF FRAME EXAMPLES WHERE SOLUTION ACCURACY IS DEPENDENT ON OBJECT DISCRETIZATION
Problem Class Buckling analysis Tension stiffening using the P-Delta option available in static nonlinear analysis Static nonlinear analysis of a model with large bending displacements Tension stiffening using P-Delta force assigned to a frame object Approximation of uniform mass Example Problems 1-019 1-016, 1-017

1-029

1-016

1-014, 1-015

AREA ELEMENTS - SHELLS, PLANES AND ASOLIDS In general the SAP2000 verification and validation example problems for shells, planes and asolids show acceptable comparison with the independent solutions. The verification problems highlight several important modeling issues to be noted when using these area elements. Those issues include element meshing and in-plane shear and bending behavior when using irregular-shaped elements. Those items are explained in the following subsections. Meshing of Area Elements It is important to adequately mesh area elements to obtain satisfactory results. The art of creating area element models includes determining what constitutes an adequate mesh.
CONCLUSIONS - 1

COMPUTERS & STRUCTURES INC.

Software Verification
PROGRAM NAME: REVISION NO.:

SAP2000 2

In general, meshes should always be two or more elements wide. Rectangular elements give the best results and the aspect ratio should not be excessive. A tighter mesh may be needed in areas where the stress is high or the stress is changing quickly. When reviewing results, the following process can help determine if the mesh is adequate. Pick a joint in a high stress area that has several different area elements connected to it. Review the stress reported for that joint for each of the area elements. If the stresses are similar, the mesh likely is adequate. Otherwise, additional meshing is required. If you choose to view the stresses graphically when using this process, be sure to turn off the stress averaging feature when displaying the stresses. In-Plane Shear and Bending with Irregular-Shaped Elements As shown in Example 2-002 and Example 3-002, when modeling for in-plane shear and bending, the area element is sensitive to geometric distortions and to aspect ratio. Rectangular- and parallelogram-shaped elements provide good behavior. Triangular elements are not recommended. Trapezoidal-shaped elements should be avoided for use where in-plane shear and bending is significant, if it is possible to use rectangularshaped or parallelogram-shaped elements. Where the use of trapezoidal elements is necessary, the following modeling tips are suggested: 1. Always use a mesh that is two or more elements wide. 2. Minimize the angle between opposite sides of the trapezoid. 3. Use aspect ratios near one to one. 4. Review the results carefully to ascertain stress continuity between elements as explained in the previous subsection. Thin Shell versus Thick Shell The main difference between the thin shell option and the thick shell option is that, unlike the thin shell option, the thick shell option includes the effects of out-of-plane shear deformations in the analysis. For most shell element models, the effect of out-of-plane shear deformations is negligible. Example 2-012 is a problem where the shear deformations are significant, and thus, the thick and thin plate solutions yield different results.

CONCLUSIONS - 2

COMPUTERS & STRUCTURES INC.

Software Verification
PROGRAM NAME: REVISION NO.:

SAP2000 2

In most problems where shear deformations are not significant the thin and thick plate options will converge to essentially the same answers. The thick plate option usually requires a finer mesh than the thin plate option to converge. The thick plate results for twisting behavior are more sensitive to aspect ratio and geometric distortions than the thin plate results. This is illustrated in load case 4 in Example 2-002. In general we recommend using the thin plate option, except in instances where out-ofplane shear deformations may be significant. Incompatible Bending Modes Option for Planes and Asolids Models that have bending behavior and do not use the incompatible bending modes option typically require a finer mesh than models using the incompatible bending modes option to obtain the same level of accuracy in the results. We recommend that you always use the incompatible bending modes option when you use plane and asolid elements. SOLIDS In general the SAP2000 verification and validation example problems for solids show acceptable comparison with the independent solutions. It is important to adequately mesh solid elements to obtain satisfactory results. Rectangular- and parallelogram-shaped elements give the best results and the aspect ratio should not be excessive. Trapezoidal-shaped elements should be avoided where possible. Where trapezoidal elements are unavoidable, the difference in angle between opposite sides should be minimized. A tighter mesh may be needed in areas where the stress is high or the stress is changing quickly. Models that have bending behavior and do not use the incompatible bending modes option typically require a finer mesh than models using the incompatible bending modes option to obtain the same level of accuracy in the results. In addition, the models without incompatible bending modes appear to be more sensitive to the element aspect ratio. We recommend that you always use the incompatible bending modes option when you use plane and asolid elements.

CONCLUSIONS - 3

COMPUTERS & STRUCTURES INC.

Software Verification
PROGRAM NAME: REVISION NO.:

SAP2000 2

LINKS In general the SAP2000 verification and validation example problems for links show acceptable comparison with the independent solutions. The verification problems highlight some important modeling issues to note when using link elements. When using nonlinear links in an analysis, it is important to recognize that careful study of the problem is required. Parametric studies of the link properties used in the SAP2000 model are useful. Also, as described in the following subsection entitled Analysis Cases, parametric study of some of the analysis case parameters should be performed to ensure an appropriate solution. As illustrated in example problem 6-007, when damper elements with velocity exponents other than one are used, the results obtained can be sensitive to the behavior of the damper at low velocities. Thus, it is very important to obtain accurate information about the force-velocity characteristics of the dampers and then to adjust the damper properties in SAP2000 to match those characteristics. In particular, the stiffness, k, can be adjusted to modify the low velocity behavior of the isolator. We suggest that when nonlinear velocity exponents are used, parametric studies using different k values should be performed. See example problem 6-007 for more information. ANALYSIS CASES For some types of static nonlinear analyses, the accuracy of the results is dependent on the discretization or meshing used in the model. Examples of this are shown in example problems 1-016, 1-017, 1-029, and 2-019. The accuracy of the time history analysis results can depend on the output sampling time interval. If that time interval is too long, peak responses may not be captured. This is illustrated in example problem 1-022. In general, the accuracy of the results of buckling analysis cases is dependent on the discretization or meshing used in the model. An example of this is shown in example problem 1-019. Nonlinear analyses typically require parametric studies of the convergence tolerances to verify that an appropriately small tolerance has been used. In general, you should assume a tolerance and then run an analysis using that tolerance and another using a smaller tolerance. If the results of the two analyses are not significantly different, the assumed tolerance was acceptable. Otherwise, a smaller tolerance should be tried. Similar to the parametric studies for convergence tolerances, for direct integration time histories, parametric studies should also be performed to confirm that the time step used

CONCLUSIONS - 4

COMPUTERS & STRUCTURES INC.

Software Verification
PROGRAM NAME: REVISION NO.:

SAP2000 2

is sufficiently small to give consistent results. This is described in example problem 6-011. Note that for direct integration time histories, control the size of the time step in the analysis using the Maximum Substep Size parameter, and control the size of output steps reported using the Output Time Step Size parameter. For example, set the Maximum Substep Size parameter to 0.0005 second to force the analysis to use steps no larger than 0.0005 second, and at the same time, set the Output Time Step Size parameter to 0.02 second so that results are reported at a 0.02-second interval.

CONCLUSIONS - 5