Open car parks in fire

The behaviour of all structures in fire has come under close scrutiny in recent years, and since the unfortunate event at Luton Airport last year steel framed car parks are receiving particular attention, with different clients asking for different levels of resistance. Most structures are designed for a period of resistance, given in minutes, when subject to a standard ISO fire curve which defines the temperature at a given time. Whilst standardisation allows direct comparisons, the relationship between a ‘standard fire’ and a ‘real fire’ is not at all obvious. In this article Dr Yigit Ozcelik of SCI compares the flexural buckling resistance of columns exposed to fire, considering prescriptive and performance-based fire design approaches.

Introduction

The unfortunate fire incidents at the Luton Airport car park in 2023 and the Kings Dock car park in Liverpool in 2017 have raised fundamental questions around the fire safety of open sided car parks (Figure 1). Whether or not the recent fires show there is a problem with current designs is open to question (did these structures do what they were intended to do or not?), but the fact that vehicles have changed since the current guidance was written is undeniable and it is quite likely that design rules and regulations will be modified in the near future in order to improve the perceived safety of such car parks.

Approved Document B^[1], providing guidance on the fire safety buildings in England, requires the structural frames of open car parks less than 30m in height to have 15 minutes fire resistance. (Note the AD is one way of meeting the requirements of the Building Regulations; however, there may be other ways to comply with the Building Regulations than following the guidance provided by the AD.^[1]) At the time the AD was produced this period was deemed enough to allow evacuation should a fire take hold. However, various stakeholders are currently asking themselves if 15 minutes is adequate when modern vehicles are considered. For example, the IStructE Car Park Design Guide^[2] suggests that such a low fire rating should be used with caution.

Modern vehicles are typically larger than older vehicles and they make greater use of plastic and synthetic materials. As a result, they have a larger fire load, and when plastic fuel tanks rupture there may be a greater tendency for a fire to spread between vehicles. Modern vehicles with alternative power sources such as LPG and lithium-ion batteries have different fire loads when compared to older vehicles^[2]. A recent CROSS report^[3] on fire risks states a 15-minutes fire resistance may not satisfy the functional requirement of the building regulations in multi-storey car parks occupied by modern vehicles. However, whilst a longer period of resistance, remembering this is resistance when exposed to a standardised ‘fictional’ fire, would undoubtedly increase fire capability, before concluding this is the correct way to go, we must remember that carbon and financial costs must also be taken into account in order to achieve the best overall solution. Adding fire protection that is not needed would not be a sensible thing to do.

Most of steel open sections used in the UK have 15 minutes inherent fire resistance^[4]. Hence, currently, fire protection is rarely needed or used for steel members in open car parks. To satisfy a more stringent fire resistance period would almost certainly require protection, with a resulting increase in embodied carbon as well as construction and potentially maintenance costs.

An alternative approach to prescriptive fire design (considering a resistance period when subject to a standard fire) is so-called performance-based fire design. By following this design approach, it is possible to calculate temperatures and thereafter the resistance of steel members in fire more accurately. The example below shows that with this greater accuracy the fire protection that would be needed according to a prescriptive approach for a fire resistance period longer than 15 minutes might be eliminated or reduced. In other words, using more elaborate engineering could help meet budget and reduce embodied carbon.

The SCI recently undertook a preliminary study aiming to show that performance-based fire design is a promising tool that could be used to justify reduced or eliminated levels of fire protection for open sided car parks. Pursuant to this goal, an unprotected steel column of an open car park was examined considering distinct fire scenarios to quantify how its buckling resistance varies when different fire design approaches are adopted.

Fire design approaches and fire curves

The prescriptive approach for fire design typically used in the UK considers the ISO standard curve included in BS EN 1991-1-2^[5], which is intended to model gas temperature in a fully developed compartment fire. Notably, this curve does not consider the decaying phase of a fire. Use of the ISO standard curve is reasonable for relatively small compartments where fire load is distributed uniformly. For large compartments or cases where the fire load is restricted to a relatively small area, use of the ISO standard curve is generally conservative. In cases such as fires in open car parks, use of a localised fire curve is more suitable to estimate steel temperatures. The publication SCI P423 Design of columns subject to localised fire^[6] presents a method for determining the temperature of a column subject to a localised fire. This method adopts the software OZone ^[7]. In the current study, both the ISO standard curve and a localised fire curve based on a realistic scenario, described below, were used to compare their effect on column buckling resistance.

Fire scenario and temperature analysis

P423[6] includes a worked example considering an open car park with a length of 60m and a width of 45m. The ceiling height is 3.5m. The standard dimensions of the parking bays are 2.5 m × 5m. The fire scenario considers three large cars and a van parked around a column and the fire starts from the car in the South-West direction of the column then spreads to the van in the North-West direction and the car in the South-East direction after 12 minutes. After another 12 minutes, the fire propagates to the car in the North-East direction. Figure 2 shows the heat release rates (HRRs) of the cars and the van.

In addition to the fire scenario that assumed the ISO standard curve, the fire scenario in P423^[6] was used in this study to develop localised fire curves in OZone^[7]. The column section was assumed to be UC305×305×97, the most similar UK section to the European section HEA300 considered in P423^[6]. As the localised fire model in OZone^[7] can estimate the steel temperatures along the column height, the steel temperatures were recorded at 0.5m intervals. The steel temperatures for the ISO standard curve and the localised fire curve are shown in Figure 3:

The steel temperature for the ISO standard curve is the same for the entire member. The steel temperatures at 30 minutes and 60 minutes were estimated as 750°C and 935°C, respectively. For the localised fire scenario, the maximum temperature recorded for each segment is between 300°C and 400°C except for the top segment whose maximum temperature is about 600°C. The main reason for this is a hot zone forms under the ceiling with a depth of approximately 0.5m. Note that the steel temperature decreases after reaching the peak value at approximately 30 minutes while it increases continuously when the ISO standard curve is considered. This suggests that the ISO standard curve not only overestimates the steel temperature but also fails to predict the shape of the steel temperature curve.

Flexural buckling resistance of column at elevated temperatures

Table 3.1 of BS EN 1993-1-2^[8] tabulates the reduction factors for the stress-strain relationship of steel at elevated temperatures. To determine the flexural buckling resistance of the column, the reduction factors for yield stress (f_y) and modulus of elasticity (E) are required. Clause 4.2.3.2 of BS EN 1993-1-2^[8] outlines a method for compression members; however, the method considers a uniform temperature for the member, which might disguise the full benefit of adopting a localised fire curve.

In this study, an isolated column model was considered to determine the flexural buckling resistance of the column in lieu of clause 4.2.3.2 of BS EN 1993-1-2^[8]. To accurately represent the flexural buckling behaviour, the column was divided into several elements with an initial bow imperfection. The imperfection was perpendicular to the minor axis and assumed to be represented by a sine curve with an amplitude of L/300, where L is the length of the column. The properties f_y and E of each column segment were calculated considering the reduction factors given in BS EN 1993-1-2^[8] per the steel temperatures shown in Figure 2. An axial compressive force was applied until the column buckled. The analysis model is shown in Figure 3.

For the flexural buckling analysis, three cases were considered:

• Case (1): Material properties determined for the maximum steel temperature per the ISO standard curve
• Case (2): Material properties determined for the maximum steel temperature along the column height per the localised fire scenario (i.e., the steel temperature of the onerous segment is considered.)
• Case (3): Material properties determined for the maximum steel temperature for each column segment separately per the localised fire scenario (i.e., seven reduction factors calculated for each material property, namely, f_y and E.)

The flexural buckling resistances of the UC section in S355 are given in Table 1. As can be seen from these results, use of a realistic localised fire instead of the ISO standard curve has a substantial impact on flexural buckling resistance. When the more conservative of the performance-based designs (Case (2)) is adopted, the flexural buckling resistance triples for 30 minutes and it is 8 times as much for 60 minutes fire resistance compared to Case (1). Similarly, considering the temperature variation along the column length further positively affects the buckling resistance (Case (3)), leading to an approximately 25% further increase in flexural buckling resistance. Another observation is that because the maximum steel temperature for the localised fire scenario occurs at approximately 30 minutes, the flexural buckling resistance remains almost the same for 30 minutes and 60 minutes fire resistances. On the other hand, when the prescriptive approach is adopted, the buckling resistance decreases significantly with the increase in temperature as the ISO standard curve does not acknowledge the decaying phase of fire. It can be concluded use of performance-based design in lieu of prescriptive design will have a more significant advantage for longer fire resistance periods.

Finally, the results shown in Table 1 were used to determine the maximum utilisation ratios for the ULS combination at ambient temperature beyond which fire protection would be needed to prevent the fire case design governing the member size. The flexural buckling resistance of a UC305×305×97 for a buckling length of 3.5m is given as 3440 kN in the Bluebook^[9]. Assuming a reduction factor of 0.7 for the design load level for the fire situation, the factored design load for the fire situation, beyond which protection would be needed, is 0.7 × 3440 kN = 2408 kN for 100% utilisation at ambient. When the flexural buckling resistances given in Table 1 are divided by 2408 kN, the allowable utilisation ratios for ULS ambient design can be calculated. The results are given in Table 2.

The values in Table 2 show that if a prescriptive approach was used, namely one considering exposure to the standard ISO fire curve, any column with a utilisation ratio at ambient temperature greater than 22% would need protecting to achieve 30 minutes, and any one with utilisation greater than 8% would need protecting for 60 minutes. In other words, they would all need protection.

However, using a localised fire scenario only columns that were more than 79% utilised at ambient would need protection. A small increase in section size could eliminate this need, or indeed a more accurate analysis might increase the utilisation limit.

Further research

It is recognised that because the study reported above is based on an SCI publication that was completed several years ago, the crucial aspect of modern vehicles has not yet been addressed. Its aim was simply to show the potential of this approach. Work is on-going at SCI to consider modern vehicles and extend the study to cover beams and perimeter columns. We will then have fact based quantified evidence, rather than a simple view that modern vehicles must be a significant problem for fire in car parks.

Conclusion

A preliminary study has been undertaken to quantify the advantages of performance-based fire design. Based on the results, it is believed performance-based fire design is a viable option to help prove unprotected columns have adequate buckling resistance during a fire in an open car park if a fire resistance period longer than 15 minutes is required. Consequently, fire protection that would come at significant cost and with embodied carbon implications could be avoided. Further research is ongoing to justify these initial findings.

---

References

1. Approved Document B (Fire safety, Volume 2 – Buildings other than Dwellings), 2019 edition incorporating 20202 amendments – for use in England. Ministry of Housing, Communities & Local Government.
2. Car park design guide. Institution of Structural Engineers. 2003.
3. CROSS safety report. Fire risks in multi-storey car parks. 2020.
   https://www.cross-safety.org/uk/safety-information/cross-safety-report/fire-risks-multi-storey-car-parks-940
4. Car parks in fire. http://www.steelconstruction.info
5. BS EN 1991-1-2:2002. Eurocode 1. Actions on structures – General actions. Actions on structures exposed to fire.
6. SCI P423: Design of columns subject to localised fires. Steel Construction Institute. 2018.
7. OZone. https://sections.arcelormittal.com/repo/Sections/4_19_Setup_OZONE.zip
8. BS EN 1993-1-2:2005. Eurocode 3. Design of steel structures – General rules. Structural fire design.
9. Bluebook. https://www.steelforlifebluebook.co.uk/