Paul Williams, Norman Disney & Young

Paul Williams

Which of the two highway tunnel ventilation systems - transverse or longitudinal - provides the lowest risk to fire life safety? Paul Williams of Norman Disney & Young in New Zealand describes how the spacing of emergency cross-passage exits along highway tunnels influence the selection of the ventilation system selected and applied, particularly in the context of the overall life safety strategy of the facility.

The key conclusion is that with the appropriate choice of exit spacing a longitudinal ventilation system can be shown to present an equal or lower risk to life safety than if a local smoke extraction system is applied.

The control of smoke generated by a fire in a highway tunnel is managed in different ways, depending on the type of ventilation system adopted and installed. The two types of mechanical ventilation available for highway tunnels comprise either transverse systems with ducting created by false ceilings, or longitudinal systems based on installation of jet fans in the tunnel crown.

In tunnels with local smoke extraction systems, smoke clearance is achieved by opening a number of remotely actuated mechanical dampers that are distributed evenly along the ceiling of the tunnel that creates the high level ducts of transverse ventilation systems (Fig 1). The alternative longitudinal system provides airflow along the tunnel, typically using longitudinal jet fans, in order to clear smoke through the tunnel portals (Fig 2).

Of the two highway tunnel ventilation systems, the one that provides the lowest risk to life safety depends on the individual tunnel design, and can be determined by a number of factors. The inter-relationship between four of these specific factors is considered.
• The functionality of the two different ventilation systems:
• The reliability of the ventilation system in operation
• The frequency of traffic congestion within the tunnel
• The travel distances between emergency exits including the portals and cross passages that link the two tubes

In an ideal world a simple table or flow chart addressing these four factors would result in the correct choices being made. In reality the choice is more complicated and requires a quantitative risk analysis to arrive at optimum ventilation design decisions.

The risk analysis model developed follows a fairly typical approach whereby event trees are constructed to determine the frequency of specific events. The consequence of each event is determined separately and the risk of a single event, or combined events, can then be calculated through the product of the frequency and the consequence. Ultimately, since the risk is a product of the frequency of an event and its consequences, it is given that the risk can be affected by controlling either or both of these parameters.

The best way to demonstrate the capability of the model risk analysis is through an example. The model risk analysis, however, can be used for any tunnel project to determine the impact of the ventilation system design, congestion in the tunnel, and the emergency exit spacing to demonstrate how the same level of risk can be achieved using different combinations of these parameters. Alternatively, if the acceptable level of risk is pre-defined, this model risk analysis can be used to find the balance between all these factors in order to achieve the stated level of risk.

To demonstrate the model, a risk analysis is applied to the new Waterview Connection twin tube highway tunnel in Auckland, New Zealand, which is currently under design and soon to be into construction. The Waterview Connection tunnel is 2.5km long and is expected to carry 90,000 vehicles/day. The two tubes of the tunnel are uni-directional and there are no entry or exit junctions within the tunnel.

Frequency of an event
Vehicle fires within the tunnel are expected to occur as a result of two initial events; either a collision leading to a fire or a technical fault leading to auto-ignition. This provides the starting point for the model risk analysis for an estimated number of fires across a range of maximum heat release rates (Table 1).

Table 1. Distribution of heat release rates as a
percentage of the expected number of fires

Maximum heat release rate Percentage of fires Number of fires/year Approximate fire return period 5MW 90.00% 0.27 4 years 30MW 9.90% 0.03 33 years 50MW 0.09% ˜0.00 3,700 years 100MW 0.01% ˜0.00 33,000 years

In the event of a fire, the chosen ventilation system is expected to operate, but since no system is 100% reliable, this is not guaranteed. This provides the first branch in the model risk analysis event tree and incorporates two factors; first the choice of ventilation system from the two available options, and secondly the reliability of the chosen ventilation system. The second branch in the event tree is related to the volume of traffic, and more specifically, the frequency of traffic congestion.

At the point at which a fire occurs, these variables govern the number of occupants who will be affected and the extent to which they are affected.

Estimating the likely frequency of congestion for a tunnel under construction should be set on a range of frequencies between 0% and 100% of the time. In reality congested traffic for more than 5% of the time (equating to approximately two hours/week day) is unlikely to be acceptable and would highlight potential problems with the overall traffic network or the proposed new highway tunnel design.

Congested traffic has a number of influences on life safety in a tunnel fire and specifically in the context of the choice of ventilation system. The tunnel used in this example is uni-directional and therefore it is assumed that in the event of a fire all vehicles upstream of the incident are prevented from travelling any further, while in uncongested flow, vehicles downstream are able to drive out of the tunnel. As such the ventilation system should be designed to protect occupants upstream of the fire incident until they are able to reach an exit on foot. In operation, both ventilation system options should achieve this performance objective.

Consider now the congested traffic case, whereby vehicles downstream of the fire are unable to drive out of the tunnel. Occupants of these vehicles must also evacuate on foot and as such the ventilation system should also afford these users some protection. Using a ventilation system with local extract means that, in general, occupants within a zone of 200m to 300m downstream of an incident may be affected by smoke. Using a longitudinal system, occupants much further downstream may be affected. While the impact is expected to drop the further away occupants are from the fire source, it is clear that there is the potential for more tunnel users to be at risk.

Once the ventilation system reliability and the congestion frequency are determined, the frequency of a number of different scenarios can be estimated. For example, if the ventilation reliability is 95% and the tunnel is congested 1% of the time, the frequency of ventilation failure in the event of 5MW fire in a congested tunnel is approximately once every 7,400 years.

Consequence of an event
For each scenario there will be a consequence. In this context the consequence is considered to be the number of fatalities. In general, relatively fewer fatalities are expected as a consequence of smaller fires than with larger sized fires. Similarly to the frequency of each event, the number of fatalities can be analysed based on the following factors:
• The local smoke extraction versus longitudinal ventilation
• The selected ventilation system reliability
• The frequency of traffic congestion

The point of difference between frequency and consequence is that the consequence of an event can be based also on the travel distances between emergency exits. The underlying assumption is that the longer users are within the tunnel, the longer they are exposed potentially to fire and smoke, and hence to a higher level of life safety risk.

With a longitudinal ventilation system and no congestion, there should be a limited number of fatalities in the event of a fire as occupants upstream of the source will be protected by the smoke control system and occupants downstream of the fire will be able to drive clear of the tunnel. There will be similar consequences if a local smoke extract system is installed, with occupants upstream largely protected and occupants downstream able to drive out of the tunnel.

Assume now that the tunnel is uncongested but that the chosen ventilation system has failed. Regardless of the choice of ventilation system, occupants upstream of the incident are now at greater risk as they must evacuate on foot with nothing to prevent smoke spreading in their direction. Occupants downstream, meanwhile, are still able to drive clear of the tunnel and are considered to be at no greater risk than if the ventilation system was working.

The next step is to consider the consequences if the tunnel is congested during a fire event. In a congested scenario, occupants downstream of the fire are no longer able to drive clear of the tunnel. Should there be congestion and a failure of the ventilation system, occupants downstream are at equal risk, irrespective of whether a longitudinal or a local smoke extract system is provided. When the ventilation system does operate, the local smoke extract system is beneficial since, unlike the longitudinal system, it prevents smoke spreading downstream.

The final factor that is considered to influence the expected number of fatalities in the event of a tunnel fire is the distance between emergency exits. At this stage a linear relationship has been assumed between exit spacing and the number of fatalities. In reality, it is unlikely that the exit spacing will linearly affect the number of fatalities and a more rigorous approach, involving a fractional equivalent dose (FED) analysis, might be more optimal. While this is outside the bounds of this risk analysis, the general trend is expected to be such that the number of fatalities reduces proportionally with reduced exit spacing.

The risk to life safety for highway tunnel users is calculated by multiplying the frequency of an event by the consequence of the same event. There are, in fact, an infinite number of scenarios that can be calculated by varying the reliability of the two systems, the frequency of traffic congestion and the spacing or distance between emergency exits.

The results of all these calculations can be used to develop a plot that, for different ventilation reliabilities and congestion frequencies, can be used to determine which ventilation system presents the lowest risk to life safety (Fig 3).

Fig 3. Plot highlighting choice of ventilation system as a function of the frequency of congestion and the system reliability

In area A and area C, the relative reliability of the two ventilation systems and the level of congestion indicate that there is a clear lowest risk solution. This is irrespective of exit spacing. The model risk analysis will always demonstrate that either the longitudinal system or the local smoke extract systems is preferable.

Within the central shaded area (area B), both ventilation systems can be shown to present a similar risk, subject to the appropriate choice of exit spacing.

For the sake of study, assume the following parameters:
• Longitudinal ventilation system reliability - 98%
• Local smoke extract system reliability - 96%
• Frequency of congestion - 2% or approximately 40 minutes per day

By varying the exit spacing, either the longitudinal ventilation system or the local smoke extract system can be demonstrated to result in the lowest risk to life safety.

At 350m exit spacing, the longitudinal ventilation system, when compared to the local smoke extra system, results in an extra 0.35 fatalities per 100 fatalities.

At 100m exit spacing, the longitudinal system, when compared to the local smoke extra system, results in 0.61 fewer fatalities per 100 fatalities.

By decreasing the exit spacing from 350m to 100m, the ventilation system that presents the lowest risk changes from a local smoke extract system to the longitudinal system. Logically, therefore, there is an exit spacing at which the risk to life safety is identical for both systems. In this example the answer is approximately 288m.

Optimising ventilation choices
The risk analysis model described outlines an approach for determining the balance between choice of ventilation system, traffic congestion and exit spacing in order to minimise the risk to life safety.

The key conclusion is that with the appropriate choice of exit spacing a longitudinal ventilation system can be shown to present an equal or lower risk to life safety than if a local smoke extraction system is provided, even assuming a reasonable level of congestion.

Often the cost and time associated with building a new highway tunnel can be reduced significantly if a local smoke extraction system is avoided. This is achieved by eliminating the transverse ventilation ducting; reducing the internal cross-section of the tunnel tubes as a result, and by installing a less complex, longitudinal ventilation system design.

In a number of cases this risk analysis model can show that a ducted system is not necessary to achieve the required level of life safety. As such, this risk analysis model approach to the selection and design of the ventilation system has the potential to improve the financial viability of tunnel projects, as well as reducing the construction risk and lowering life- cycle maintenance requirements.

References

New Zealand awards mega-TBM Waterview highway tunnel undertaking - TunnelTalk, August 2011

Excerpt of a paper presented at the International Symposium on Tunnel Safety and Security, New York, March 2012 Download a copy of the full conference paper

Search

TunnelTalk Articles TunnelCast Videos Archive Feedback Current Diary Dates World News Briefs Career Moves Accolades & Tributes Stay Connected Procurement Job Opportunities Equipment Sales DIRECTORY Marketplace About Us Advertising Contact Us Legal Notices Home Office     Latest Alert Alert Signup Alert