What If Fires in Car Parks Developed Quicker Than We Think?
By Wojciech Węgrzyński, PhD FSE
View full PDF here
I was always fascinated by car park fire safety. The general layouts, technical solutions, and fuel loads are usually the same, so the car parks seem very similar (and boring). However, the architecture details and their ventilation features lead to the emergence of overly complex flow patterns, unpredictable fire development and smoke spread. This makes each car park a truly unique challenge in delivering safety to its users. We have already touched on this remarkable diversity in the past, focusing on jet-fan systems' performance [1].
In recent years Europe has seen a series of large fires (Liverpool, UK (2017); Cork, Ireland (2018); Stavanger, Norway (2019); Warsaw, Poland (2020)) that brought the car park fire safety in the focal point of the public discussion over safety once again. Furthermore, we approach the unknown – new fuel vehicles present challenges that we have not faced in the past [2]. Some countries have already banned electric vehicles (EVs) from the enclosed car parks in the absence of knowledge. But will that still be a viable strategy once the EV's become prevalent? Instead of banning, we need to learn how to manage and mitigate the risk. It will be an even greater challenge for countries without a sprinkler culture, like Poland.
Fire safety of car parks is commonly designed with the use of performance-based design methods, and in the heart of every such analysis lies the design fire. The description of a design fire seems straightforward for offices or warehouses - we usually assume some quadratic function of growth in time, often described as the "αt²" fire. We let the fire develop until it consumes all of the fuel, is bounded by the available oxygen or extinguished. In car parks, however, the development of a single vehicle's fire does not reassemble such a simple growth function, Figure 1. Looking at the results of multiple fire experiments around the world, you may notice that the size of the vehicle fire changes significantly after certain events – breaking the windows of the passenger compartment, fuel tank failure or in the case of EV's - the battery failure. Such events happen at different times, and we cannot exclude they will happen very early in the fire, as seen in viral videos of EV fires. Consequently, the growth of the HRR can be rapid, and the fire can shortly reach the peak HRR of a particular event.
Figure 1. HRR measured in the TNO fire experiment [3], arrows indicate the events in the experiment that were followed by a sudden change in the HRR of the fire
It does not necessarily mean that the HRR will grow to a very large value, especially in sprinklered car parks. In fact, in our research, we focused on small fires, with their peak HRR's between 0.25 - 1.50 MW, which can be directly compared to the design fires commonly used today. Our assumption was to model fires that grow from zero to peak HRR within 30 seconds. The reference design scenario was the TNO fire described by the Dutch standard NEN 6098 [4]. We have compared the total amount of heat (and smoke) released from all the scenarios within the first minutes of the fire, representing the evacuation phase. Our fire with peak HRR = 750 kW releases the same amount of heat within the first 4 minutes as the reference design fire. Fires with peak HRR's of 1 000 - 1 500 kW emit more heat than the reference, while 250 - 500 kW emit less heat than the reference (Figure. 2).
Figure 2. The design fires used in this study – the evolution of Heat Release Rate (HRR in kW) and the value of Total Heat Released (THR in MJ)
Once we have defined the fires, we have placed them within our multiparametric car-park fire simulator, the assumptions of which is given in more details in [5]. The tool we have created automatically generates, executes and post-processes CFD simulations of car parks with the FDS model [6], based on the input height and definition of smoke control systems. The investigated car park was 40 x 60 m, and had the height of 2.40 m; 2.70 m; 3.00 m; 3.30 m or 3.60 m. We have investigated a car park with duct ventilation systems (3 variants), jet-fan systems (3 variants), an open car park and one without any systems at all. Overall, we have performed 480 CFD simulations in the project, and the results of 320 which are relevant to the research question of this paper are discussed here. We have focused on analysis of the visibility in smoke in the first 5 minutes of the simulation, which represents the evacuation period. We were looking into how quickly car park fills with smoke, what is the average visibility inside and on what % of the car park the tenability criteria are not achieved after five minutes.
First, we have looked at how visibility changes in time along a line across the car park for three fires. An example of this analysis for a 2.70 m high car park without ventilation systems is shown in Figure 3. You can observe that the loss of visibility (red colour) occurs as soon as between 50 – 100 s, and for the 1000 kW fire, over half of the analysed space is filled with smoke after just 2 minutes. This is highly unsatisfactory, as in such a short time, the car park systems (automatic smoke control, dry pipe sprinklers) would not be able to react to fire. We have found that this result was highly dependent on the height of the car park. In Figure 4 you can see the snapshot of the visibility in the car park after 3 minutes, for car parks with the height of 2.40 m, 3.00 m and 3.60 m, for 1 000 kW fire. While conditions in the smallest car park are untenable, safe evacuation is possible in the same scenario in a high car park (3.60 m). Overall, the results are not great, but are they worse than if analysed with a traditional design fire?
Figure 3. Space-time plots presenting the visibility measurement results along a line across the car park (the x-axis) in the function of time (y-axis). This is a 2D representation of how a parameter (visibility) changes in the function of time and space—results for 500 kW, 1 000 kW and 1 500 kW scenarios in 2.70 m car park.
Figure 4. Visibility at the height of 1.80 m above the floor, for a car park with duct ventilation system (17 m³/s) and the height of 2.40 m, 3.00 m and 3.60 m and 1 000 kW fire scenario
To verify that, we took a snapshot of visibility in the car park from the CFD simulation with the reference design fire (TNO) and our rapidly-growing fires. Then we have subtracted their results, Figure 5. The figure shows the difference between these simulations' results after five minutes for a car park with 2.70 m height and a 34 m³/s duct ventilation system. The blue colour indicates that the reference simulation had worse results, and the red indicates that rapidly growing fire leads to worse outcomes. It can be noticed that for the first two fires, the outcomes of the reference case are worse, and for the peak HRR = 750 kW the outcomes are the closest between simulations. What you cannot see on the static figure is the evolution of the changes – looking at transient animations, you can notice that the conditions for rapid fires are worse, but the untenable conditions are observed much earlier than for the traditional design. Especially for larger fires (> 750 kW), the outcomes of rapidly growing fires were significantly worse. As of today, we do not have reliable data on the sizes of EV fires after a battery failure in an early phase of the fire. Considering the findings of this research, we can predict that if such fires are larger than 750 kW, they are probably more challenging than the design fires used today.
Figure 5. The difference between reference and rapid fire simulations after 300 s, the blue colour indicates lower (worse) visibility in the reference case, and red indicates lower (worse) visibility in the rapidly growing fire case. Simulations in a 2.70 m height car park, with a duct ventilation system (34 m³/s)
In the final part of our analysis, we verified how modern smoke control systems change the fires' outcomes. For this, we did investigate the % of the car park with untenable visibility (less than 10 m), Figure 6. First, we have again noticed that the change of height had a profound impact on the results. Furthermore, for the smallest car parks (2.40 m height) and our larger fires (> 750 kW) the characteristics of the ventilation system did not really matter – the car park did always end up filled with smoke. In taller car parks, we have observed significant improvement in ventilated car parks' conditions over the unventilated case. Furthermore, the differences in performance between particular smoke control systems were also substantial. This confirms that smoke control is essential for the provision of safety in car parks, but is not a miraculous cure - you cannot mitigate the car park's low height by oversizing the smoke control system.
Figure 6. The % of the car park filled with smoke exceeding the tenable value of visibility (here chosen as 10 m) for various car park heights, systems and fire growth scenarios
So, what if fires in car parks developed quicker than we think?
It seems that a rapidly growing fire scenario is more challenging than the design fires currently in use. Such scenarios may represent an EV fire after a battery failure or a rapidly growing fire of an internal combustion vehicle after a fuel tank rupture. As a consequence of such a rapid growth of the fire, the car park may fill with smoke quickly, even before smoke control or dry-pipe sprinkler systems can react. The outcomes of such fires in low-height (in our case – 2.40 m) car parks were unsatisfactory. Our research has shown that the best mitigation strategy for rapidly growing fires is to provide sufficient height of the car-parks (2.70 m+), allowing adequate space for the smoke layer and providing longer escape time. This strategy should be used primarily for car parks where many people can be present simultaneously (e.g. malls, sports arenas).
Wojciech Węgrzyński is with the Building Research Institute (ITB)
REFERENCES:
[1] W. Węgrzyński, JET-FAN SYSTEMS IN CAR PARKS DESIGN METHODS: AN OVERVIEW AND ASSESSMENT OF PERFORMANCE, SFPE Eur. (2018). https://www.sfpe.org/publications/sfpeeuropedigital/sfpeeurope9/issue9feature3
[2] H. Boehmer, M. Klassen, S. Olenick, Fire Protection Research Foundation report: "Modern Vehicle Hazards in Parking Garages & Vehicle Carriers," 2020. http://www.nfpa.org//-/media/Files/News-and-Research/Fire-statistics-and-reports/Building-and-life-safety/RFModernVehicleHazards-in-ParkingGarages.pdf.
[3] N. van Oerle, A. Lemaire, P. van de Leur, Effectiveness of Forced Ventilation in Closed Car Parks, in: TNO Rep. No. 1999-CVB-RR1442, 1999.
[4] NEN 6098:2010, Ontw. nl Rookbeheersingssystemen voor mechanisch geventileerde parkeergarages, (2010).
[5] W. Węgrzyński, Multiparametric CFD Analyses To Understand The Key Variables In Car Park Smoke Control, in: Fire Evacuation Model. Tech. Conf., 2020. https://www.femtc.com/events/2020/d2-10-wegrzynski/
[6] K. McGrattan, S. Hostikka, R. McDermott, J. Floyd, C. Weinschenk, K. Overholt, Fire Dynamics Simulator User’s Guide, Sixth Edition, 2017. doi:10.6028/NIST.SP.1019.