Road tunnels need a robust risk management strategy being critical component of modern transportation network. Change in vehicle types with intrinsic hazard source change due to evolution of power sources (H2, EV etc.), tunnels must evolve into intelligent infrastructures to safeguard lives through robust engineering and proactive risk governance mechanism.

The risk profile within a road tunnel fluctuates instantaneously based on the types and volumes of vehicles present. Traditional design-stage risk assessments, often grounded in conservative assumptions, fall short in managing such dynamic infrastructure. Advancements in high-performance computing and computer vision technology, a real-time tunnel risk estimation can be achieved. This necessitates a robust methodology for continuous risk quantification, enabling proactive monitoring and timely intervention to manage risk.

The primary objective of this research work was to develop dynamic risk estimation framework for road tunnels, forming the basis for a smart tunnel Risk Monitor that evaluates real-time risk based on vehicle types, traffic volume, and hazard potential. Applied to the Bhatan tunnel under simplified traffic assumptions and simulated traffic flow conditions, to derive a correlation between overtaking and severe accident collision probability. The methodology models event progression using an event tree and estimates risk every second over a one-year period. The framework demonstrates scalability across diverse tunnel configurations, vehicle categories, and traffic volumes.

As secondary objective, it provided the method to derive the URCL, the Upper Risk Control Limit and ARTL, the Acceptable Risk Threshold Limit for effective risk management of a tunnel and demonstrated the evaluation using the estimated one-year risk profile for Bhatan tunnel. Further, it recommended administrative actions and restrictions that can be initiated triggered once instantaneous risk reaches the URCL and stopped with the restoration of ARTL. 

Computational fluid dynamics (CFD) simulations were used to estimate peak heat release rates (HRR), aligning closely with results from the Runehamar tunnel fire experiment involving heavy goods vehicles (~200 MW). Simulations were extended to five vehicle categories viz. cars, SUVs, six and ten-wheeler trucks and buses with certain substitute material fire properties like n-heptane as engine oil and mixture of polyvinyl chloride (PVC) & urethane as burnable materials in all vehicles. The observed peak HRR values exceeded significantly for cars, SUV/LMVs (~25 MW vs ~ 5 MW) and bus (~ 200 MW vs ~ 20 MW) those suggested by some of the widely adopted international guidelines. This study therefore proposes revised HRR benchmarks for individual passenger and freight vehicles, intended for use in tunnel design-stage safety and risk assessments.

This study introduced and applied a potential two-vehicle collision scenario weighted methodology to estimate the design fire load or peak heat release rate (HRR) for road tunnels. The approach was implemented for five reference vehicle categories in Bhatan to estimate the design basis peak HRR at 81 MW, offering a refined framework for evaluating fire severity under realistic multi-vehicle conditions.

This study established the groundwork for the development of a Tunnel Risk Monitoring and Management System (TRMMS) for a smart tunnel. Integrating the proposed risk estimation framework with computer vision and deep learning for vehicle classification, hazard assessment, and speed detection can enable intelligent, real-time risk monitoring.