A Risk-Based Approach to Jet Fan Optimisation

Wind forces can impact a tunnel ventilation system's ability to move smoke in a fire event. How much wind force should the system be designed for?

Introduction 

In addition to providing adequate air quality and maintaining temperatures within acceptable limits, tunnel ventilation systems need to be designed to move smoke in the event of a fire with a ‘good’ level of confidence. Wind and other meteorological forces can negatively affect the performance of the ventilation system, but for how much wind force should the system be designed? 

WSP | Parsons Brinckerhoff’s U.K. tunnel ventilation team is working on a large railway project with a number of tunnels, so it was important to answer this question confidently and with a solid basis. Risk analysis was used as a tool to help the decision-making process. 

This article focuses on the optimisation of a range of rail tunnels that would utilise jet fans to provide smoke control in the form of longitudinal ventilation. Longitudinal ventilation prevents smoke from back-layering, providing a tenable evacuation environment upstream of the fire. In total, seven tunnels with lengths ranging from approximately 500 metres to 3 kilometres were analysed. 

The optimisation was carried out after an initial design phase where the tunnels were found to be sensitive to atmospheric wind. At that stage of the design, questions still remained as to whether the wind force that was being designed for was reasonable. We based the design on a 1 percent probability of exceedance in any year, but should it be 10 percent, 1 percent, 0.1 percent, or something different? 

The design included an assumption that the jet fan nearest the fire was inoperable. In the emerging design, approximately one jet fan per portal was required to overcome the wind forces, two jet fans were required to control the smoke, and one standby/redundant jet fan was provided to handle other random failures. A further question arose as to the probability of both a high wind and a failed jet fan. Was the investment in the redundant jet fan warranted? A quantitative risk analysis was therefore undertaken to understand the acceptability of this risk of removing the redundant jet fan. 

The combination of a fire in the tunnel, a high wind force, and a failure of one of the required jet fans might lead to the back-layering of smoke within the tunnel. Back-layering occurs when the ventilation flow rate is not high enough to meet ‘critical velocity’¹ (CV). The critical velocity will depend on factors such as the fire heat release rate and tunnel gradient. The consequences associated with providing less than critical velocity required evaluation. 

Methodology 

An event tree was generated to consider the probability of various scenarios (see Figure 1). Each branch or scenario of the event tree had an overall predicted event frequency and consequence assigned. This was subsequently used to estimate risk. 

Figure 1 – Event tree (dotted arrows represent uncompleted branches of the tree, only one complete branch is fully shown)

A tunnel fire frequency rate was estimated through interpretation of statistical data from the U.K.’s Railway Safety and Standards Board. Various probabilities were then assigned to each scenario. 

Each event path required an evaluation of consequences to passengers. The consequence analysis was broken down into two constituent parts: 

• Bulk-flow simulations were undertaken using the Subway Environment Simulation (SES)² software for three representative tunnels. This provided information about the tunnel air flow rate for every different configuration of ventilation mode, train location, fire heat release rate, ventilation direction, and wind force that was tested. From this, average percentage of critical velocity was determined for each combination of tunnel, ventilation mode, and wind condition. 
• A 3-D analysis was then performed on a characteristic short tunnel section using the Fire Dynamics Simulator (FDS) software. The evacuation model was enacted within the software which allowed the coincident location of the smoke and passengers to be predicted. These simulated the evacuation of 1,100 passengers within the tunnel with different fire heat release rates and air flow rates. Predicted effects or consequences to passengers during the evacuations were recorded based on the Fractional Effective Dose (FED) method, but adjusted for these simulations to also account for the effects of irritant gasses. The simulations were undertaken for different airflow rates to allow the outcomes to be mapped to the SES simulations. 

Results 

The results of the consequence analysis can be seen in Figure 2. 

Figure 2 – Predicted relationship between percent of critical velocity achieved and total passenger fatalities and weighted injuries for different fire magnitudes.

It is evident that for the larger fires simulated there is always a base equivalent fatality rate of approximately 55 persons. This represents the inherent consequence involved with longitudinal ventilation systems; there is a risk that passengers may be located downstream of the fire location. To minimise passenger numbers downstream of the fire, the ventilation direction is decided by the fire location. To model a condition where an “average” number of passengers were downstream of the fire, the fire was set to be a quarter of the length down the train. As the percentage of critical velocity achieved reduces, the back-layering of the smoke advances. This process is illustrated in Figure 3. The jump in the predicted number of equivalent fatalities from 55 to 250 as seen in Figure 2 was due to the back layering of smoke past an upstream passenger exit (illustrated by scenario C in Figure 3). 

Figure 3 – Illustration of evacuation scenarios: a) CV is achieved, b) CV not achieved but no additional passenger fatalities (60-100% CV), c) CV not achieved and resulted in additional passenger fatalities (<60% CV).

Risk analysis 

The U.K. rail industry has acceptance criteria for the probability of injury for individuals as well as methods to evaluate the so called ‘societal risk’ that can occur for low-frequency high-consequence events such as tunnel fires. 

The risk to an individual passenger was predicted to be 1 in 240,000,000—orders of magnitude lower than the broadly acceptable limit of 1 in 1,000,000 in the U.K. 

Societal risk was evaluated using frequency/severity (FN) curves where the value plotted on the y-axis is the cumulative frequency of experiencing N (passenger fatalities). These are assessed graphically and were compared to the current national railway risk profile of the U.K. railway (see Figure 4). The FN risk should be below this line. 

Figure 4 – Breakdown of societal risk profile of three representative tunnels.

‘System failure’ points on the bottom right of the FN graph represented scenarios where the ventilation system had suffered complete failure. The points were slightly higher than the baseline risk of the U.K. railway. The majority of this risk was predicted to be due to the calculated human error in operating the ventilation system correctly in the event of an incident. Figure 5 shows an FN plot where the element of human error has been removed. This suggests a strong benefit in providing a fully automatic control system. 

Figure 5 – Societal risk with the removal of human error from the operation of the ventilation systems.

Conclusions 

It was concluded that the fire hazard could be managed so far as is reasonably practicable with the proposed ventilation approach of using the spare jet fan to also overcome wind forces. There was no strong case for adding further jet fans to reduce the risk. Eighteen (18) jet fans were eliminated from the ventilation system design, potentially saving many millions of pounds. 

It was also concluded that if efforts were made to reduce the human factor from the operation of the ventilation system, the societal risks attributed to the higher consequence events could be significantly reduced. 

¹Critical velocity – the air flow required to prevent smoke from moving upstream of the fire location.

²Subway Environmental Design Handbook. Volume II. Subway Environment Simulation Computer Program (SES). Part 1. Prepared by Parsons Brinckerhoff as part of a joint venture for the U.S. Department of Transportation, in 1975.