Performance of Deep Beams with Topologically Optimized Reinforcements: Experimental Verification and Comparison

Abstract

Conventionally designed reinforced concrete (RC) deep beams often fail in a nonductile shear failure mode. A deep beam may embrace significant ductility through a careful layout optimization of the reinforcements, but this optimization intention still needs effective experimental validation. This paper presents such an experimental validation for two evolutionary optimization methods the authors developed recently: a single-objective evolutionary structural optimization (SOESO) aiming to achieve uniform stress in steel rebars and a biobjective counterpart (BOESO) with a secondary objective to create a smoother convergence by reviving some deleted reinforcement elements. Two specimens designed using the two optimization methods were statically tested. The loading-displacement curves of both specimens exhibited an apparent postyield plateau, confirming that both optimization methods can design ductile deep beams. In comparison, the BOESO beam outperformed the SOESO beam in the sense that the stresses of all rebars in the BOESO beam were closer to the yielding stress at failure, while most oblique rebars in the SOESO beam did not yield and the longitudinal rebars at different rows were uneven. Furthermore, the BOESO-designed reinforcement layout meets the design requirement of the bottle-shaped strut according to the strut-and-tie model (STM). The experiment results recommend using the BOESO method as a reference for future RC deep beam design.