Workshops and Conferences

List of workshops and conferences that I have presented, co-authored, or involved in. Click the title for more details, slides, papers, web-app, and other resources.

Academic Conferences

Beyond Statistical Fitting: Improving Reliability Analysis through Physically Consistent Bridge Traffic Load Model

Existing literature on bridge traffic load distributions can predict loading beyond what is physically impossible. For example, it is possible to predict load larger than a truck fully loaded with Osmium, the densest stable element there is. How could this be? This presentation proposes a method to prevent impossible loading from being predicted. We urge that future research properly consider the physical nature of the problem being modelled, and goes beyond just chasing some goodness-of-fit, predictive power, or other metrics when proposing a statistical model.

Presented in ICOSSAR25 (2025), Los Angeles, USA.

Phenomenological Probabilistic Modelling of Highway Bridge Traffic Loading

This work presents recent advances in probabilistic modelling that adopt a phenomenological, generative approach to highway bridge traffic loading. First, a hierarchical Bayesian model is introduced to use traffic load data from multiple bridges within a network to inform the statistical modelling. Second, a Bayesian model is proposed to incorporate engineering and physical constraints into probabilistic traffic load models. These developments demonstrate how generative modelling can offer deeper insights into traffic loading processes while improving predictive accuracy. This shift leads to more robust, interpretable probabilistic models, improving bridge design, assessment, and network-wide risk management.

Presented in ICOSSAR25 (2025), Los Angeles, USA.

Insightful Modelling: What Colour is Your Box?

The last few years have seen a tremendous increase in the use of artificial intelligence and machine learning (AI/ML) approaches in structural engineering research. However, the intrinsic opacity of these “black box” models presents significant challenges in developing new insights, as they often obscure the underlying mechanisms. This lecture contrasts black box models with “white box” approaches, such as Bayesian hierarchical modeling, which prioritize transparency and interpretability by requiring explicit hypotheses. This contrast is illustrated through a case study on highway bridge traffic load modeling, where a Bayesian hierarchical model not only enhances predictive accuracy but also uncovers critical correlations and reduces uncertainties—insights that are difficult, if not impossible, to achieve with black box models. Overall, this talk advocates for the continued use of white box, or even grey box, models in structural engineering research, emphasizing their role in generating new knowledge and advancing the field beyond mere prediction.

Presented in ISRERM 2024 (2024), Hefei, China.

Bayesian hierarchical modelling of bridge traffic loading across a road network

In this work, a modern Bayesian hierarchical model is developed using the generalized extreme value distribution, covering intermediate spans where data is not available at the time of fitting. This paper is mostly a work in progress of a paper we published in 2024. The unique contribution of this paper is in the verification of the proposed hierarchical model by accurately estimating a known true parameter (the traffic upper bound), thereby indirectly verifying the accuracy of the 2024 paper.

Presented in ICASP14 (2023), Dublin, Ireland.

Workshops

Structural Safety and Code Calibration Workshop

This series of workshops presented by Prof. Colin Caprani and Dr. Shihabudin Khan, with guest lecture by the distinguished Prof. Mark Stewart introduces the basic of statistics and probability, its application in reliability analysis, and how safety is achieved in structural engineering via probabilistic assessment and code calibration.

I developed a web-app with Prof. Colin Caprani and Dr. Shihabudin Khan to accompany the workshops. The web-app contains interactive elements to visualize probability distributions, reliability analysis, and the complex task of code calibration. You can access the web app by clicking the link in the title.