Model uncertainties and bias in SHEAR strength predictions of slender stirrup reinforced concrete beams

Abstract

Prior studies point to high model uncertainty in shear prediction. Model factors for various shear prediction models are characterized here, based on a recent and well‐vetted database of shear failures. Characterization includes estimation of main statistical parameters; and importantly also correlation and regression analysis to assess consistency of model factors over ranges of design parameters. The aim of the study is to identify models suitable for use as general probabilistic model (GPM) in future reliability assessments. The Variable Strut Inclination Method (VSIM), shear level of approximation III of the fib Model Code 2010, and ACI 318 displayed high bias, variability and various correlations with shear input parameters which make them unsuitable choices for GPM. The compression chord capacity model and the modified compression field theory displayed low bias and variability with consistent model factors over the ranges of shear design parameters and are thus suitable as GPM. Concerns stem from observed inconsistency in capacity prediction by VSIM and size effect in ACI. Reliability assessment is advised.

Assessment of reliability of EN 1992‐1‐1 variable strut inclination method of shear design provisions for stirrup failure

“…The authors quantified the model uncertainty for six different prediction models of shear capacity (consisting of newly developed shear models) using a critically vetted published database of shear experiments developed by Reineck et al and the results were subsequently used in this contribution. The FORM method was implemented for the performance analysis of the VSIM procedure.…”

“…Olalusi and Viljoen show that the model factors of the MCFT (R2k) and CCC are less biased, with lower uncertainty and less unwanted trends with shear parameters than alternative shear models including the VSIM, shear level of approximation III of the fib Model Code 2010, and ACI 318. This motivates the choice of the MCFT (R2k) and CCC as GPMs for shear reliability assessment in this paper.…”

“…The trend of EC2 VSIM mean values with the amount of shear reinforcement ρ w f ywm led to the investigation of alternative prediction models suitable for establishing the general probabilistic model (GPM) for shear. 20 The principal GPM for the reliability assessment was established as the stochastic representation of capacity predictions from MCFT-based analysis program Response 2000 (MCFT [R2k]) 6 (see Section 3.1). An alternative GPM based on the stochastic representation of capacity predictions from CCC model 8 (see Section 3.1) was established to validate basic investigations conducted using MCFT (R2k)-based GPM.…”

“…

The best estimate shear strength predictions of the operational EN 1992-1-1 (EC2) variable strut inclination method (VSIM) have been shown to underestimate shear capacity at low stirrup quantities. Conversely, the model overestimates shear capacity at high stirrup quantities (Cladera and Mari, 2007;Olalusi and Viljoen, 2019). Sufficient safety performance of the design formulation should be confirmed at high levels of stirrup reinforcement, and over conservatism at low stirrup reinforcement is not ideal.

…”

Probabilistic studies on mechanics‐based shear capacity models for reinforced concrete beams with stirrups

“…A similar study was conducted by Olalusi and Viljoen for determining the modeling error of shear capacity models recommended by different codes of practice. 55 The study 55 brings out the importance of consideration of modeling error in probabilistic description of shear capacity of slender beams.…”