Modeling Guidelines for Nonlinear Analysis

Gregory G. Deierlein Helmut Krawinkler Stanford University

LATBSDC May 9, 2008

Guidelines for NL Analysis

Part I General Modeling Guidelines

Overview of Modeling Issues Energy Dissipation and Damping Deterioration and Modeling Options

Part II Properties of Structural Components

Structural Steel: Beams and Columns Reinforced Concrete: Beam-Columns and Joints

NL Analysis Considerations for Design

Accurate Assessment of Building Performance

Structural Performance and Limit States Non-structural and Contents Performance Limit State: limited yielding/cracking & permanent deformations Modeling: effective (elastic) stiffness Limit State: onset of severe degradation Modeling: post-yield response and degradation (how much?) contribution of gravity systems to lateral response nonstructural components certain energy dissipation, degradation, and failure modes

Evaluation of Serviceability Limit State

Evaluation of Safety (MCE) Limit State

Non-Modeled Effects

Key Considerations for Idealized Models

How faithfully do models capture significant
behavioral effects

Pre-yield (initial) stiffness Strength and failure modes Post-yield deformations and deterioration Relationship to codified standards for strength and deformation capacity Transparent and easy to verify Randomness in input parameters (e.g., fc, Fy) Epistemic model uncertainties

Practicality and reliability for design practice

Ability to Represent Variability in Response

Model Idealizations
Physical Phenomenological

Column Shear

Drift

Axial Load & Post-Peak Response in RC Columns

0.6 fc Ag

0.15 fc Ag

!"# %&'&(")"'* %+%,&-&./"

0"1)1 ,#* 23"4-" 5 6"7".

Example: RC Column Hinge Model

Key Parameters:
strength initial stiffness post-yield stiffness plastic rotation (capping) capacity post-capping slope cyclic deterioration rate

Advantages:
Compatible with design models and standards for stiffness, strength and deformation capacity Easy to calibrate and validate backbone and hysteresis parameters to test data Ability to empirically model complex failure modes and deterioration (e.g., rebar buckling and fracture)

Disadvantages:
Difficult to capture multi-component force interaction (P-M, Mx-My, P-M-V)

Example: RC Shear Wall Fiber Model

Key Parameters:
stress-strain properties of steel and concrete fibers (quasi-uniaxial) quasi-elastic shear stiffness

Advantages:
captures inelastic flexural (P-Mx-My) response out to reasonable deformations Reasonable model for quasi-elastic shear deformations and shear lag

Disadvantages:
Challenges in modeling bond-slip and tension-stiffening degradation due to reinforcing bar buckling and fracture inelastic shear deformations effect of wall openings Difficult to model variability in wall response

Flexurally dominated response

Key Considerations for Model Calibration

Identification and quantification of key behavioral
indices

Model Type Physical versus Phenomenological Stress-strain or force-deformation Modeled versus Un-Modeled Behavior (deterioration) Availability of data or other supporting evidence Unbiased, statistically neutral, mean response prediction Variability in input and response parameters Definition of engineering demand parameters and damage measures Establishment of component limit states (serviceability and safety)

Robust statistical characterization

Acceptance Criteria

UN-MODELED ENERGY DISSIPATION a.k.a. DAMPING

Viscous Damping with NLTH Analysis

Raleigh (proportional) Damping:

Modal Damping:

Explicit Damping Elements

[c]i [c]i configured to represent likely sources of viscous and other incidental damping.

Modeling of Damping in NLTH Analysis

Definition: reduction in dynamic building response due to energy dissipation of structural and nonstructural components of the building, its foundation, and the underlying soil/rock materials Complicating Factors: The interpretation and representation of damping is complicated by relationship of mathematical representation of damping to the physical sources of damping, e.g.,
(1) artificial distinctions between energy dissipation of structural components that is modeled by nonlinear hysteretic response versus equivalent viscous damping, (2) modification of input motions to account for reduction in response due to SSI effects

amplitude dependency (displacement, velocity, acceleration) of damping effects and its effect on building performance for different overall intensity of building response, and (b) different effects for alternative vibration modes.

Physical Sources of Damping

SUPERSTRUCTURE STRUCTURAL:
Primary Structural Components whose nonlinear behavior may be

explicitly modeled in the analysis (e.g., walls, beams, columns, b/c joints) Secondary Structural Components that contribute to response but whose energy absorbing characteristics may not be modeled explicitly (e.g., energy dissipation provided gravity framing, deformations to floor slab at slab-wall connections, etc).

SUPERSTRUCTURE NONSTRUCTURAL
Exterior Cladding a likely source of considerable damping, depending on

the material, method of attachment, expansion joints. Interior Wall Partitions and Finishes (issues materials, method of attachment of finishes to structural walls/braces/columns, method of attachment of partitions to slabs, density of partitions i.e., open office versus partitioned residential) Mech/Electrical - piping, electrical conduit, HVAC risers, elevator rails and cables, stairs, etc.

SUBSTRUCTURE FOUNDATION & SITE

soil-foundation interface (e.g, soil yielding and gapping)) radiation damping

Current/Recent Practice for NLTHA

Assume that energy dissipation at large deformations
is primarily accounted for by hysteretic response

Raleigh (proportional) damping is usually expressed as

a percentage of critical damping (for first few modes) to reflect other sources of un-modeled energy dissipation:

SF AB83 (2007): 5% LA Alternative Procedure (2005) not specified SAC Joint Venture (1995): 2% ATC 63 (2007): 5% LATBDC (1989): Design EQ 5% to 10% Geared to elastic analysis. MCE EQ 7.5% to 12% Too high for NL analysis!!

Recorded Strong Motion Data (US)

REF: Chopra/Goel

Observations: 1.Damping in the range of 2% to 8% of critical 2.Reduction in damping with increasing height Possible Explanations: - proportionally less soil/foundation damping - proportionally less nonstructural damping - smaller excitation (amplitude) in measured earthquakes

Recorded Strong Motion Data (US)

REF: Chopra/Goel

Observations: 1.Modest trend for increase in % Critical Damping with increasing roof drift demand. 2.Interstory drift ratios may vary considerably between cases and may reveal stronger trends

Recorded Motion Data (Japan)

REF: Satake, 2003

Observations: 1.Data from shaking tests, strong motion and a few ambient 2.Damping ranges: - - Steel-Framed Buildings: 0.5% to 3% RC and SRC Buildings: 0.5% to 8%

3.Decreasing damping with building height (suggested trend 1/height) - Satake et al. hypothesize soil/foundation effects as a major factor

Measured Wind-Induced Damping

Fair number of tall building measurements that demonstrate amplitude dependence of damping. However, measured response is limited to very low acceleration and displacement amplitudes with correspondingly small damping values. The plot above is a typical example, where damping during wind events is typically around 0.5% to 1% of critical.

Measured Wind-Induced Damping

Damping ratios under wind-induced (storm) vibrations of tall buildings range from 0.5% to 1.4% of critical

Considerations for Raleigh Damping

1. Target Damping Ratio 2. Specified Period Range 3. Definition of Stiffness 4. Other variations:

- Initial or Tangent Stiffness, Ko or Kt - Basis for Ko - Mass or Stiffness Only - Damping parameters (a, b) vary during analysis

Definition of Stiffness
1. Justification for using Ko
- damping mechanisms do not change during loading - Ko or Kt is not a major factor, aM dominates first mode damping - use of Ko is simpler and more robust

2. Justification for using Kt

- Maintains original damping ratios as the system as the structure softens and velocities increase (avoids over-damping) - safeguards against equilibrium problems with ill-conditioned stiffnesses of yielding/cracking elements - numerical studies suggest better correlation with tests

3. Definition of Basis Stiffness, Ko

- gap elements/materials, such as cracked concrete in fiber-type formulations, can lead to large initial [C] - rigid-plastic hinges (or similar) can lead to local force imbalance

Components to Damping Matrix

[c]i

b[K]
Stiffness

a[M]
Mass Discrete

Damping Recommendations
Damping = a/N 1% - 2% N = # stories a: 70 to 80 steel 80 to 100 RC 130 dual sys.

Target Damping (Percentage of Critical)

2% - 5%

Damping Recommendations (Raleigh)

Period Range: 0.2To to 1.5To Halls method for setting a & b to control error over period range Use Kt or reduced (effective) Ko

Damping: Suggested Future Study

1. Measurements/Evaluation of Damping
- Laboratory shake table tests - Instrumented buildings - Understand how un-modeled energy dissipation changes during inelastic loading?

2. Examine sensitivity of tall building (40+ stories) response to assumed damping models
- damping in higher modes (deformations, internal forces) - Significance at serviceability and safety (MCE) levels

3. Clarify numerical and implementation issues

- Initial versus tangent stiffness - Raleigh vs. Modal vs. Discrete damping - Influence of hysteretic model properties (e.g., degradation in unloading/reloading stiffness)