problem 6 003
TRANSCRIPT
-
8/12/2019 Problem 6 003
1/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 1
EXAMPLE 6-003
LINK GAP ELEMENT
PROBLEM DESCRIPTION
This example uses a single-bay, single-story rigid frame to test the gap link
element. This link element carries compression loads only; it has zero stiffness
when subjected to tension. The gap element is paced at the bottom of the right-hand column in the frame. The frame is then loaded with a gravity load P (10kips) at the center of the beam. Once the full load P is applied, a lateral load V
(20 kips) is applied, pushing the frame from right to left. The compression loadin the gap element after the full load P has been applied and the uplift at the gapafter the full load V has been applied are compared with independent hand
calculated results.
The model is created in the XZ plane. Only the Ux, Uzand Rydegrees of freedom
are active for the analysis. The gap element is modeled as a single-joint link
element at joint 2. This means that one end of the gap element is connected to theground and the other end is connected to joint 2. The gap element is oriented
such that its positive local 1 axis is parallel to the positive global Z axis. This is
the default orientation of single joint link elements. Only U1degree of freedomproperties are defined for the gap element.
All three frame elements have identical properties. The compression stiffness ofthe gap element is chosen to be approximately 100 times the axial stiffness of the
frame element directly above it, frame element 2. The initial gap element
opening is zero inches.
The loading is applied as follows. First the full load P is applied. Then, while theload P is maintained, the full load V is applied. The analysis is run several times
using different types of analysis cases. Nonlinear static, nonlinear modal time
history and nonlinear direct time history analysis cases are used. See the
subsequent section titled Summary of Analysis Cases for more information.
Two different models are used in this example. In Model A the linear effectivestiffness of the gap element is set equal to zero. In Model B the linear effective
stiffness of the gap element is set equal to the nonlinear stiffness of the gap
element.
-
8/12/2019 Problem 6 003
2/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 2
The gap linear effective stiffness is only used for the linear analysis cases, whichin this example are analysis cases P, V and MODAL. The gap linear effective
stiffness is not used in the other analysis cases and thus has no direct effect on
them; however, it does indirectly affect thenonlinear modal time history cases named
NLMHIST1 and NLMHIST2 because those
cases are solved using the modes from theanalysis case named MODAL.
The lumped joint masses shown in the tableto the right are used for the modal and
modal time history analyses.
GEOMETRY, PROPERTIES AND LOADING
1
P = 10 k
Z
Y
X
Single-joint link element (gap)with compression stiffnessdefined for the U1 degree of
freedom
2
3 4
1 2
3
V = 20 k
144"
144"
Loading
First apply full load P; thenapply full load V
Frame Material Properties
E = 29,900 k/in2
= 0.3G= 11,500 k/in2
Frame Section Properties
A = 10 in2
I = 100 in4
Av = 2 in2 (shear area)
Link Properties (Gap U1 DOF)
Linear Ke (Model A) = 0 k/inLinear Ke (Model B) = 200,000 k/inLinear Ce = 0 k-sec/inNonlinear K = 200,000 k/in
Nonlinear Open = 0 in
Active Degrees of Freedom
Ux, Uz, R
Joint Mass kip-s2/in
Joint DOF Ux DOF Uz
2 0 0.001
3 0.3 0.1
4 0.3 0.1
-
8/12/2019 Problem 6 003
3/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 3
SUMMARY OF ANALYSIS CASES
The following table summarizes the analysis cases that are used in this example.
Analysis Case Description
PA linear static analysis case with a 10 kip gravity load applied in the
negative global Z direction at the center of frame element 3.
VA linear static analysis case with a 20 kip lateral load applied in thenegative global X direction at joint 2.
MODALA ritz-type modal analysis case with starting loads defined for load P, load
V and the link element.
NLMHIST1
A nonlinear modal time history starting from zero initial conditions and
using a ramp load to apply the load P.
20 second ramp load rise time.
99.9% modal damping for all modes.
20 output steps and a 2 second output time step (40 seconds ofoutput total).
Default values for all nonlinear parameters (tolerances).
NLMHIST2
A nonlinear modal time history starting from the conditions at the end ofNLMHIST1 and using a ramp load to apply the load V.
40 second ramp load rise time.
20 output steps and a 4 second output time step (80 seconds of
output total).
99.9% modal damping for all modes.
Default values for all nonlinear parameters (tolerances), except inModel B the Force Convergence Tolerance is reduced from the
default 1E-05 to 1E-11.
NLSTAT1A nonlinear static case starting from zero initial conditions and applyingthe load P. This analysis case uses default nonlinear parameters
(tolerances).
NLSTAT2A nonlinear static case starting from the conditions at the end of NLSTAT1and applying the load V. This analysis case uses default nonlinear
parameters (tolerances).
-
8/12/2019 Problem 6 003
4/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 4
Analysis Case Description
NLSTAT3A nonlinear static case starting from the conditions at the end of the directintegration time history NLDHIST1 and applying the load V. This analysis
case uses default nonlinear parameters (tolerances).
NLDHIST1
A nonlinear direct integration time history starting from zero initial
conditions and using a ramp load to apply the load P.
20 second ramp load rise time.
400 output steps and a 0.1 second output time step (40 seconds of
output total). Mass and stiffness proportional damping is used with damping set
to 99.9% at periods of 1 and 3.4 seconds.
Default values for all nonlinear parameters (tolerances) except forthe Maximum Iterations per Substep and the Iteration Convergence
Tolerance. The Maximum Iterations per Substep item is increased
from the default 10 to 100. The Iteration Convergence Tolerance isreduced from the default 1E-04 to 1E-06.
NLDHIST2
A nonlinear direct integration time history starting from the conditions at
the end of NLDHIST1 and using a ramp load to apply the load V.
40 second ramp load rise time. 800 output steps and a 0.1 second output time step (80 seconds of
output total).
Mass and stiffness proportional damping is used with damping set
to 99.9% at periods of 1 and 3.4 seconds.
Default values for all nonlinear parameters (tolerances), except forthe Maximum Iterations per Substep and the Iteration Convergence
Tolerance. The Maximum Iterations per Substep item is increased
from the default 10 to 100. The Iteration Convergence Tolerance isreduced from the default 1E-04 to 1E-06.
NLDHIST3 A nonlinear direct integration time history that is the same as NLDHIST2,except it starts from the conditions at the end of nonlinear static caseNLSTAT1 instead of nonlinear direct integration time history NLDHIST1.
-
8/12/2019 Problem 6 003
5/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 5
The ramp times for the analysis cases are selected using the rule of thumb thatthe ramp loading time should be approximately 10 times the period of interest.
The period of interest is assumed to be the period of the first mode. The
following table shows the first mode period obtained for the two models.
Model
Gap Effective Stiffness for
Modal (Linear) Analysis
k/inMode 1 Period
sec
A 0 3.42
B 200,000 1.90
The load P is applied first and it causes the gap element to always be in
compression. Thus the 1.90 sec period, which is the period when the gap stiffness
is 200,000 k/in, is taken as the period of interest for application of the load P. The
ramp rise time for this load is chosen as 20 seconds, which is approximately 10times the period of interest.
The load V is applied after load P and it causes the gap element to eventually
uplift. Thus the 3.42 sec period, which is the period when the gap stiffness is
0 k/in, is taken as the period of interest for application of the load V. The ramprise time for this load is chosen as 40 seconds, which is approximately 10 times
the period of interest.
Analysis case NLMHIST2 in Model B requires 200 output steps (800 seconds
total) instead of the 20 output steps (80 seconds total) required in Model A. This
occurs because in Model B the modes are calculated using the gap effectivestiffness of 200,000 k/in and the 99.9% damping is applied to those modes.
When the gap opens and the gap effective stiffness becomes zero, the period
lengthens. Because the modal damping coefficient is held constant, the percent ofcritical damping increases in proportion to the period lengthening, thus over-
damping the system. This over-damping that initiates when the gap opens is thereason that it takes so long for analysis to reach its final value in case
NLMHIST2 for Model B.
Analysis case NLMHIST2 in Model B also requires a force convergencetolerance of 1E-11 rather than the default 1E-05. The following equation, which
is equation of motion used to solve nonlinear modal time history analyses, helps
explain the rationale for the change in the convergence tolerance.
-
8/12/2019 Problem 6 003
6/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 6
KLu(t) + KNu(t) + Cu(t) +Mu(t) + rN(t) = r(t) + KNu(t). ..
KLu(t) + KNu(t) + Cu(t) +Mu(t) + rN(t) = r(t) + KNu(t). ..
In the preceding equation KLis the stiffness of all the linear elements, including
the linear degrees of freedom of the link elements; KN is the linear effective
stiffness matrix for all link element nonlinear degrees of freedom; C is theproportional damping matrix; M is the diagonal mass matrix; r is the vector of
applied loads; and rN is the vector of forces from the nonlinear degrees of
freedom of the link elements that is computed by iteration at time t.
In the preceding equation, if the KN term is very large compared to the other
terms, the relative force convergence tolerance needs to be very small to capturean accurate representation of rN. This is why the 1E-11 force convergence
tolerance is required for analysis case NLMHIST2 in Model B.
The nonlinear direct integration analysis cases need the maximum number of
iterations per substep set to 100 rather than the default 10 and the iterationconvergence tolerance set to 1E-06 rather than the default 1E-04. Reducing the
iteration convergence tolerance to 1E-06 eliminates the slight fluctuations (high
frequency chatter) in the results.
For this particular example, increasing the maximum allowed number of
iterations per substep to 100 decreases the running time of the problem by severalorders of magnitude. Using the default 10 maximum iterations per substep causes
the program to have to cut the time steps from the initially specified 0.1 second
to as low as approximately 1E-07 second to solve the problem. When 100
maximum iterations per substep are allowed, the program does not have toreduce the time step size below 0.1 second to solve the problem.
TECHNICAL FEATURES OF SAP2000 TESTED
Gap element links Force-controlled nonlinear static analysis
Nonlinear modal time history analysis Nonlinear direct time history analysis
Frame point loads
Joint force loads
Joint mass assignments Ramp loading for time histories
-
8/12/2019 Problem 6 003
7/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 7
RESULTS COMPARISON
Independent results are hand calculated using the unit load method described on
page 244 in Cook and Young 1985. The results are presented separately forModel A and Model B
Results for Model A (Gap Ke= 0 k/in)
OutputParameter
AnalysisCase SAP2000 Independent
PercentDifference
NLMHIST1 -4.534 0%
NLSTAT1 -4.534 0%
Link force after
full load P is
applied
kip NLDHIST1 -4.534
-4.534
0%
NLMHIST2 3.917 0%
NLSTAT2 3.917 0%
NLSTAT3 3.917 0%
NLDHIST2 3.917 0%
Link
deformation
after full loadV is applied
inch
NLDHIST3 3.917
3.917
0%
-
8/12/2019 Problem 6 003
8/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 8
Results for Model B (Gap Ke= 200,000 k/in)
OutputParameter
AnalysisCase SAP2000 Independent
PercentDifference
NLMHIST1 -4.534 0%
NLSTAT1 -4.534 0%
Link force afterfull load P is
applied
kip NLDHIST1 -4.534
-4.534
0%
NLMHIST2 3.917 0%
NLSTAT2 3.917 0%
NLSTAT3 3.917 0%
NLDHIST2 3.917 0%
Link
deformationafter full load
V is applied
inch
NLDHIST3 3.917
3.917
0%
COMPUTER FILE: Example 6-003a, Example 6-003b
CONCLUSION
The SAP2000 results show an acceptable comparison with the independentresults both for Model A where the gap linear effective stiffness, Ke, is 0 k/in and
for Model B where the gap linear effective stiffness, Ke, is 200,000 k/in. The
comparison is exact when sufficiently small convergence tolerances are used.
This example illustrates that solution of nonlinear problems using gap elements
can be sensitive to the convergence and iteration tolerances that are used. For thisverification example, it is easy to determine if the tolerances used are sufficient
because the hand calculated results were available. In other situations it is helpful
to run the analysis two or more times using tolerances that are different by anorder of magnitude or more and verify that the results are the same. This is a
good check that the tolerances used are sufficient.
-
8/12/2019 Problem 6 003
9/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 9
HAND CALCULATION
-
8/12/2019 Problem 6 003
10/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 10
-
8/12/2019 Problem 6 003
11/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 11
-
8/12/2019 Problem 6 003
12/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 12
-
8/12/2019 Problem 6 003
13/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 13
-
8/12/2019 Problem 6 003
14/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 14
-
8/12/2019 Problem 6 003
15/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 15
-
8/12/2019 Problem 6 003
16/16
COMPUTERS &
STRUCTURES
INC.
R Software Verification
PROGRAM NAME: SAP2000
REVISION NO.: 0
EXAMPLE 6-003 - 16