Modeling SCB frames using beam-column elements
Vesna Terzic UC Berkeley
January 2013
Agenda
Different modeling approaches of SCBFs Line-element model of SCBF
3 different models of gusset plate connections will be considered and demonstrated on an example
Comparison of seismic responses of a SCBF considering different gusset plate connection models Sensitivity of the model to geometric imperfection of the brace and the number of elements used to model the brace Consideration of further simplifications of the model (demonstrated on an example) Conclusions and summary Q & A with web participants
Introduction
Special Concentrically Braced Frames (SCBF) are commonly used as the seismic resisting system in buildings. During large seismic events they may experience buckling of the braces. Inelastic deformation of the braces place inelastic deformation demands on beams, columns and connections.
Modeling approaches for SCBF
Continuum models (shell or brick elements)
Accurate Computationally expensive
Line-element models (beam-column elements and zero-length elements)
Simple = Computation time significantly reduced Accurate simulation of global behavior Reasonable predictions of many local behaviors
OpenSees elements used in Line-element models
Braces, beams and columns can be modeled with force-based (FB) fiber beam-column elements. Rigidity of the gusset, gussetto-beam, and gusset-to-column connections can be modeled with rigid elastic elements. Beam-column connections of shear tab type can be modeled with zero-length rotational spring model (Liu & AstanehAsl, 2004)
OpenSees elements used in Line-element models
Gusset plates (GP) connection can be modeled in two ways: 1. Force-based fiber elements (Uriz & Mahin, 2008) 2. Rotational hinge (Hsiao et al., 2012)
FBE
Lavg=(L1+L2+L3)/3
Analytical Predictions
Uritz & Mahin, 2008 Hsiao et al., 2012
Example
One story-one bay SCBF with chevron configuration of braces Beams: W27x84 Columns: W14x176 360 in. Braces:HSS10x10x0.625 Gusset plate: tapered plate with t=1.375 in Beam-column connections are shear tab connections (not designed for the purpose of this example)
180 in.
Buckling of HSS braces
Hsiao et al. 2013
OpenSees model 3D model
Y
Shear-tab connections are modeled as pins
3 31 321 322 421 431 3101+#El 4 422 432 3201+#El 522 521 5 51
Braces are modeled: with 10 FB elements. Corotational geometric transformation. Quadratic out-of-plane imperfection (Leff/1000) Gusset plate connection modeled in following ways: 1. FB element 2. Out-of-plane rotational spring 3. Pin
12
3101 13 1
3201 23 2
21
Nonlinear FB elements Rigid elastic elements Nodes Pinned connection
OpenSees model
All nonlinear elements are modeled using Steel02 wrapped with Fatigue material 3 integration points (IP) are used for braces and beams and 4 IPs are used for columns Nonlinear rotational spring is modeled using zero-length element and Steel02 material assigned to it. All rigid elements are modeled with elastic beamcolumn elements with 10 times bigger A and I than that of the corresponding element.
OpenSees model - loads
Loads: Gravity Ground motion with its two components (horizontal and vertical)
1.5
Ground motion [g]
1 0.5 0 0.5 1 1.5 0 5 10 15 Time [sec] 20 25 30
Seismic performance of SCBF with different gusset plate connections
Base shear vs. displacement
1500 1000 Spring FBE
1500 1000
Force [kip]
Spring FBE pin
Force [kip]
500 0 500 1000 1500 1 0.5 0 0.5 Disp. [in] 1
500 0 500 1000 1500 1 0.5 0 0.5 Disp. [in] 1
Note: period is ~ the same for all three types of models: T=0.156 sec
Seismic performance of SCBF with different gusset plate connections
Story drift
0.6
Story Drift [%]
0.4 0.2 0 0.2 0.4 0.6 0
0.6
Spring FBE
10
15 Time [sec]
20
25
30
Story Drift [%]
0.4 0.2 0 0.2 0.4 0.6 0 5 10 15 Time [sec] 20 25
Spring FBE pin
30
Seismic performance of SCBF with different gusset plate connections
Floor acceleration
Floor acceleration [g]
1.5 1 0.5 0 0.5 1 1.5 0 5 10 15 Time [sec] 20 25 30 FBE Spring
Floor acceleration [g]
1.5 1 0.5 0 0.5 1 1.5 0 5 10 15 Time [sec] 20 25
FBE Spring pin
30
Seismic performance of SCBF with different gusset plate connections
Axial force deformation at the middle of the left brace
left brace 1000 Spring FBE 1000 Spring FBE pin left brace
Axial Force [kip]
500
Axial Force [kip]
500
500
500
1000 0.02 0.01 0 Axial Strain [in/in] 0.01
1000 0.02 0.01 0 Axial Strain [in/in] 0.01
Seismic performance of SCBF with different gusset plate connections
Stress-strain of a fiber at the midd cross-section of the left brace
60 left brace 60 left brace
40
40
Stress [ksi]
Stress [ksi]
20
20
20
20
40 Spring FBE 60 0.04 0.02 0 Strain [in/in] 0.02
40
Spring FBE pin 0.02 0 Strain [in/in] 0.02
60 0.04
Summary
GP connections modeled with either FBE or rotational spring provide similar global and local responses of the system. FBE element is simpler to model (input information are t, Ww) than rotational spring (input information are t, Ww and Lavg) Pinned GP connection results in great loss of accuracy and is not recommended for estimating a seismic performance of SCBF under large earthquakes that can induce the buckling of the braces.
Effect of initial imperfection on the results GPC = FBE
Global responses
1500 1000 p=1%Leff p=0.2%Leff p=0.1%Leff p=0.05%Leff
Geometric imperfection 1%Leff 0.2%Leff 0.1%Leff 0.05%Leff
Max. Drift [%] 0.62 0.59 0.54 0.52
Max. Acc. [g] 1.33 1.56 1.59 1.61
Force [kip]
500 0 500 1000 1500 1 0.5 0 0.5 Disp. [in] 1
Note: compression elements usually have constriction tolerance of 0.1%Leff
Effect of initial imperfection on the results GPC = FBE
Local responses
left brace 1000 p=1%Leff p=0.2%Leff p=0.1%Leff p=0.05%Leff
right brace 1000 p=1%Leff p=0.2%Leff p=0.1%Leff p=0.05%Leff
Axial Force [kip]
500
Axial Force [kip]
500
500
500
1000 0.025 0.02 0.015 0.01 0.005 0 Axial Strain [in/in] 0.005
1000 0.025 0.02 0.015 0.01 0.005 0 Axial Strain [in/in] 0.005
Effect of number of FBE used to model the brace GPC = FBE
Global responses
noEle=2 noEle=4 noEle=8 noEle=16
1500 1000
Force [kip]
500 0 500 1000 1500 1 0.5 0 0.5 Disp. [in] 1
Number of elements 2 4 8 16
Max. Drift [%] 0.55 0.54 0.54 0.54
Max. Acc. [g] 1.59 1.59 1.59 1.59
Effect of number of FBE used to model the brace GPC = FBE
Local responses
left brace 1000 noEle=2 noEle=4 noEle=8 noEle=16
Axial Force [kip]
right brace 1000 noEle=2 noEle=4 noEle=8 noEle=16
Axial Force [kip]
500
500
500
500
1000 0.03 0.02 0.01 Axial Strain [in/in] 0
1000 0.03 0.02 0.01 Axial Strain [in/in] 0
To capture failure of the brace it is suggested to use 10-20 elements (Uriz & Mahin 2008)
3D vs. 2D frame GP connection modeled with rotational spring
Global responses
1500 1000
Force [kip]
2D 3D
500 0 500 1000 1500 1 0.5 0 0.5 Disp. [in] 1
Spatial dimension 2D 3D
Max. Drift [%] 0.591 0.586
Max. Acc. [g] 1.569 1.554
3D vs. 2D frame GP connection modeled with rotational spring
Local responses
left brace 1000 2D 3D 1000 2D 3D right brace
Axial Force [kip]
500
Axial Force [kip]
500
500
500
1000 0.025 0.02 0.015 0.01 0.005 0 Axial Strain [in/in] 0.005
1000 0.025 0.02 0.015 0.01 0.005 0 Axial Strain [in/in] 0.005
3D vs. 2D frame GP connection modeled with FBE
Global responses
1500 1000
Force [kip]
2D 3D
500 0 500 1000 1500 1 0.5 0 0.5 Disp. [in] 1
Spatial dimension 2D 3D
Max. Drift [%] 0.58 0.54
Max. Acc. [g] 1.60 1.60
3D vs. 2D frame GP connection modeled with rotational spring
Local responses
left brace 1000 2D 3D 1000 2D 3D right brace
Axial Force [kip]
500
Axial Force [kip]
500
500
500
1000 0.025 0.02 0.015 0.01 0.005 0 Axial Strain [in/in] 0.005
1000 0.025 0.02 0.015 0.01 0.005 0 Axial Strain [in/in] 0.005
Summary and conclusions
GP connections modeled with either FBE or rotational spring provide similar both global and local responses of the system. GP connections should not be modeled as pinned if buckling of the braces is expected. Global and local responses are sensitive to the value of the geometric imperfection at the middle of the brace
AISC specifies construction tolerance of steel elements under compression to Leff/1000 (design documents)
Local response of the system is sensitive to the number of FB elements used to model the brace.
To capture the fracture of the brace it is recommended to use 10-20 elements
3D frame models can be replaced with 2D models without compromising the accuracy of the results (especially in the case of GP connections modeled with rotational springs)
References
1. Patxi Uriz, and Stephen A. Mahin, (2008), Toward EarthquakeResistant Design of Concentrically Braced Steel-Frame Structures, PEER report 2008/08. 2. Po-Chien Hsiao, Dawn E. Lehman, and Charles W. Roeder, (2012), "Improved analytical model for special concentrically braced frames", Journal of Constructional Steel Research 73 (21012) 80-94. 3. Liu J, and Astaneh-Asl A, (2004), Moment-Rotation Parameters for Composite Shear Tab Connections, Journal of structural engineering, ASCE 2004;130(9). 4. Po-Chien Hsiao, Dawn E. Lehman, and Charles W. Roeder, (2013), A model to simulate special concentrically braced frames beyond brace fracture, Earthquake Engng Struct. Dyn. 2013; 42:183200
