Research

Thoughts 

The ultimate goal of any structural analysis is to reliably estimate the global performance under expected loads. To achieve this, one has to gain a reasonable command of the capacity estimations of different elements with various configurations. For instance, column configurations with identical material and reinforcement properties but different axial load ratios can result in entirely different failure modes, capacities and energy dissipations. While reliable capacity estimation is reasonably complex in the case of a new design, it is much more arduous for existing structures, especially when they are not suitably designed per code requirements. 

The required confidence level in response predictions is attained through experiments that help understand the force-resisting behaviour and other member-level performances. Based on this knowledge, analytical models for load and displacement capacities are derived. These are again fed into the global level analyses to produce reliable structural performance estimates. For seismic or any time-dependent dynamic loading, cyclic behaviour is important. In these circumstances, evaluation of future performance is dependent on historical load cycles. Here, the hysteretic behaviour - which carries time-dependent performance and expected energy dissipation - comes into play. Over the past few years, my research has focused on evaluating and recommending suitable analytical models to estimate force-displacement relationships for critical members such as shear walls, pile caps, flat plates, columns, etc. These models were often tested against state-of-the-art design provisions, where they performed exceedingly well in prediction accuracy, especially in capturing the parametric influences. In addition to strength modelling, a reasonable level of confidence was also achieved in modelling the hysteretic behaviour of columns and walls using collated data from experimental results published worldwide. These models are simple and effective to be incorporated into popular commercial design software, thereby assisting design engineers to carry reliable nonlinear dynamic and time-history analyses for complex structures. With emerging material technologies, there is an increasing need to understand the responses with new materials at both member and global levels, as this is essentially starting with a clean slate. More pressing and urgent issues involving retrofit requirements for existing structures must be dealt with subjectively to provide viable solutions. 

On the adoption of ML tools in structural engineering: The generous amount of test data/images available offers new avenues of explorative research by adopting next-generation tools such as computer vision and deep learning to help researchers and practitioners solve design issues with speed and specificity. However, one must be careful in using such data-based tools, for they may give the reader an illusion of expertise, potentially resulting in unscientific conclusions. Balancing data-driven approaches with physics-based learning ensures the effective and informed use of ML tools in structural engineering. Therefore, for optimal utilization of ML/ML tools in structural engineering, hands-on experimental experience, essential scientific rigour, and strong fundamental knowledge of member-level and global-level structural behaviour are essential. 

Size Effect in Reinforced Concrete 

How reliably can we translate findings from reduced-scale laboratory tests to larger structures for design? 

Most laboratory tests are conducted on a reduced scale, from which suggestions are made for much larger real structures. However, in translating the results from small-scale elements to larger counterparts, one has to consider the "Size Effect." The size effect reduces the normalized nominal strength of members compared to their experimental small-scale replicas. In lay terms, it would mean that the strength of smaller lab specimens would be higher than that of their up-scaled replicas in practice. This is important for design, as design equations are based on experimental data heavily comprised of small-scale tests. Thus, quantifying strength reduction with size effect is an important scientific problem. In some cases, the size effect would not significantly impact the design as the lower strength may already have been accounted for through either strength reduction and/or load amplification factors. However, this remains an issue to be considered separately in particularly large-sized specimens. 

To quantify the size effect in two-way shear members, an experimental plan consists of twelve total pile cap specimens and three specimens each at different scales (1x, 2x and 4x). The largest set of four specimens were 1250mm thick; the next set consisted of four pile caps with 625mm thickness, which is half their larger counterparts; and the smallest set of four specimens were further halved in size. Observed results were compared with the size effect provisions from American and Japanese codes, and necessary recommendations were made. A brief visual summary can be found below. 

For more details, please refer to the published manuscript in the ACI Structural Journal. For a free full copy of this paper, send a request here. 

Hysteretic Modelling for Reinforced Concrete Members

Could we take advantage of available test data to predict the complex hysteretic behaviour of RC members? 

In structural analysis, especially in the case of seismic loading, it is essential to model both capacity curves and the hysteretic behaviour of each member carefully. Hysteretic modelling is generally a complex task involving predicting different loading-unloading cycles at different drift levels. Moreover, different parametric configurations must be considered for accurately estimating hysteretic behaviour, just as they are accounted for in capacity predictions. 

Using the hysteretic test data from 113 columns and 108 walls, equations for pivot parameters α and β capturing the influence of key parameters were proposed. Calibration of these equations was carried out to minimize the differences in predicted vs experimental total energy dissipations of hysteretic curves. For this complex optimization problem, a metaheuristic algorithm titled "simulated annealing" was used to escape local minima while arriving at global minimums. Notable hysteresis differences in shear versus flexure, such as more significant pinching in shear, were accurately captured. These results made it possible to carry out accurate nonlinear time history analyses that result in reliable global predictions. 

For more details, please refer to the two manuscripts on columns and walls published in the Journal of Earthquake Engineering and Structural Dynamics. Kindly send a request for receiving these papers here:~column~wall~. 

Shear Capacity Estimation in Reinforced Concrete

Lateral strength of structural walls & punching strength of pile caps and flat plates & shear strength of beam-column joints 

Broadly, member failures can be classified into either flexure or shear. Several engineering assumptions made it possible to provide flexural capacity estimations with reasonably good confidence. Flexure-dominant members are also desired in design for their ductile responses. However, some types of members are inherently shear-dominant, and their failure is brittle. Thus, shear strength estimation is an area of great interest to researchers and designers. More complex tools, such as strut-and-tie models, are commonly adopted. 

Based on the available test data, suitable force transfer mechanisms were derived for structural shear walls with openings (for windows) consistent with observed failure crack patterns. In addition to shear walls, the punching behaviour of pile caps (with nonsymmetric pile configurations) and flat plates (with and without shear reinforcement) were studied to propose models that reliably predict their strength. These analytical models helped provide reliable estimates and advocated for improved design recommendations in state-of-the-art standards for RC design. 

Papers on walls, pile caps, flat plates, and beam-column joints are published in ASCE Structural and ACI Structural Journals. Kindly send a request for receiving these papers here~wall~pile cap~flat plate~beam-column joints. 

Please refer to Publications for a complete list of manuscripts