Fire safety of alcoholic beverages in retail stores

A fire test series to study pool fires of ethanol‐water mixtures in the small, intermediate and large scale has been performed. Additionally, demonstration tests of piles of PET plastic 0.5‐L liquor bottles (39 vol%) in retail store arrays were carried out to illustrate their burning behaviour. On the basis of the test results, the burning of pool fires of different alcoholic beverages and the fire load due to alcohol can be evaluated for the purpose of performance‐based fire safety design. The fire tests of retail store arrays showed that these fires grow slowly, and their heat release rate maxima are relatively low taking into account the mass of the fire load. The estimated heat release rate maxima of four‐layer retail store arrays were ca. 1.5 MW and their growth times were over 600 seconds. Design fires generally used for the fire safety design of shopping centres are typically much more severe than the fire test results of retail store arrays of alcoholic beverages in this study.


| INTRODUCTION
In 2015, the Finnish Safety and Chemical Agency (Tukes) published a guideline concerning the storage of flammable liquids in retail stores. 1 The interpretation of this guideline raised questions about the fire safety of liquor stores, especially related to the storage of alcoholic beverages with an alcohol content of 20 vol% or more. Fire authorities considered that further clarifications and studies were needed, in order to ensure the fire safety of liquor stores. The main goal of this study was to find answers to the following questions: 1. Is a spill collection tray under a retail store array of PET plastic liquor bottles beneficial for fire safety? 2. Shall the spill collection tray be shallow or deep?
3. How easily does a retail store array of PET plastic liquor bottles ignite? 4. What is the heat release rate of the retail store array?

How does a PET plastic liquor bottle behave in fire?
Small-scale pool fire experiments using the cone calorimeter apparatus have been previously reported. 2 The goal of the small-scale tests was to produce data on the heat release rate and effective heat of combustion of ethanol-water mixtures for estimating the heat release and fire load which can be caused by alcoholic beverages in liquor stores. It was concluded that a significant fire load is not formed until the ethanol percentage of the alcoholic beverage is ca. 20 vol% or more. The small-scale data can also be used as an input to simulations of ethanol-water mixture pool fires in the performance-based fire safety design. (0.19 m 2 ) and a depth of 100 mm. In these tests, the volume percentages of the ethanol-water mixtures varied in the range of 20-96.1 vol% and the layer thicknesses in the range of 5-25 mm.
The tests were carried out using the Single Burning Item (SBI) test apparatus, normally used for testing construction products in vertical position according to the EN 13823 standard. 3 The test series of ethanol-water mixtures was run deviating from the EN 13823 standard. In practice, only the hood collecting the exhaust gases, the exhaust duct with its instrumentation, and the data acquisition system were utilised in these tests. A stand for a scale was built on the specimen trolley. The pool fire tray for the liquid specimen was positioned on top of the scale. The specimen was ignited with a hand-held gas burner. During the test, heat release rate and mass loss were measured. This data was needed for estimating the heat release rate in the large scale tests on the basis of mass loss data. In addition, gas temperatures at five heights from the bottom of the pool fire tray were measured (results not presented in this paper). The experimental arrangement is illustrated in Figure 1.

| Large-scale pool fire tests
The large-scale fire tests were performed in the large fire test hall of VTT. In total, 10 tests were run. In these tests, the ethanol concentra-

| Intermediate-scale pool fire tests
The main goal of the pool fire tests carried out using the SBI test apparatus was to measure the heat release rate of the ethanol-water mixture pools and compare it with the mass loss measured simultaneously. The test arrangement provided comparative data for the pool fire tests carried out in the large fire test hall where mass loss was measured but heat release rate measurement was not possible.
Another goal was to study the repeatability of ethanol-water mixture pool fires.
As an example, Figure

| Large-scale pool fire tests
To obtain a scaling factor between the heat release rate and the mass loss rate, the results of the 0.44 × 0.44 m 2 (0.19 m 2 ) pool fire tray tests using the SBI apparatus and in the large fire test hall were compared. Figure 6  On the basis of the literature referred above, the heat release rate estimates of the pool fire tests in this study can be assumed to be on the safe side, that is, thin pools spread on the floor can be assumed to burn less intensely as the test results presented would indicate.  F I G U R E 6 Heat release rates (HRR) and mass loss rates (Dm) of intermediate-scale tests carried out using the SBI test apparatus compared to tests carried out in the large fire test hall. Scale of mass loss rate axis has been selected so that mass loss rate of 1 g/s corresponds to heat release rate of 17.5 kW Figures 7 and 8 show the demonstration tests D1 and D3 (both with shallow spill collection tray) during advanced fire. It is noted that the smoke production in the demonstration tests was characteristically small. Test D3 exhibited a larger smoke production due to the burning of the wooden supports of the non-combustible shelf.

| Demonstration tests of retail store arrays
To estimate the heat release rate of retail store arrays, the scaling factor of 17.5 kW/(g/s) for mixtures of 40 vol% of ethanol was used, since the alcoholic beverage in the bottles of the arrays had ethanol concentration very close to that (39 vol%). Figure 9 shows the estimated heat release rates of the demonstration tests, compared with t 2 fires of ultra-fast, fast, medium and slow fire growth rates (growth time t g of 75, 150, 300 and 600 seconds, respectively). The t 2 fires shown grow according to square-law until the heat release rate reaches 1600 kW, after which the heat release rate is kept constant. Figure 10 shows the heat release rate of the demonstration test D1 with the t 2 fires and some design fires found in the literature. (For clarity, only D1 is presented since it is more severe than D2 and of the same order as D3.) The design fires shown in Figure 10 are the following: • "Shoe storage": A design fire often used for the fire safety design of shopping centres in Finland. 7 The heat release rate curve describes a shoe storage fire going to flashover, based on fire tests performed in Australia. 8 • "Clothing rack": A fire test by NRC, Canada. 9 Total mass ca. 36 kg, with 86, 2 and 12 mass percent of textiles, plastics, and wood and paper, respectively.
• "Clothing store": A fire test by NRC, Canada, with two clothing racks in a closed room. 10 Total mass ca. 71 kg, with 86, 2 and 12 mass percent of textiles, plastics, and wood and paper, respectively.
• "Potato chips (NIST)": A fire test by NIST, US, with a fire load of 27.1 kg of potato chips in small bags (33.8 g/bag). 11 • "Potato chips (SP)": A fire test by SP, Sweden, with a fire load of 275 kg of potato chip and cheese nibble bags. A 5.4 m-long threeshelf high shelving unit tightly packed. 11 On the basis of the demonstration tests, retail store arrays containing alcoholic beverages cannot be considered as a major fire safety F I G U R E 7 Demonstration test D1 ca. 16 minutes after ignition. The bottle array has collapsed, and practically all bottles are contributing to fire F I G U R E 8 Demonstration test D3 ca. 12 minutes after ignition. The bottle array under the shelf has collapsed. The bottles on the shelf do not contribute to fire. The wooden boards supporting the shelf are burning F I G U R E 9 Heat release rates in demonstration tests D1-D3 compared with ultra-fast, fast, medium and slow t 2 fires (t g of 75, 150, 300 and 600 seconds, respectively). The peak of D3 at ca. 1600 seconds is due to the collapse of the shelf with 40 liquor bottles above the basic array F I G U R E 1 0 Heat release rate in demonstration test D1 compared with t 2 fires and different design fires found in the literature risk. This type of fire load burns moderately and its heat release is rather small compared to many other fire loads in retail stores and groceries (see Figure 10).

| CONCLUSIONS AND IMPACT
The results of the fire tests performed can be used in the assessment of the burning of different alcoholic beverages and the fire load due to alcohol for the purpose of performance-based fire safety design.
The fire tests of retail store arrays of alcoholic beverages showed that these fires grow slowly, and their heat release rate maxima are rela- The design fires used in the fire safety design of shopping centres in Finland are typically much more severe than the fires of alcoholic beverage packages can be according to the tests performed. If a liquor store is a part of a shopping centre or in connection of a large supermarket, no additional performance-based fire safety design is needed for it.
As a result of this study, the Finnish Safety and Chemical Agency updated its guideline on the storage of flammable liquids in retail stores. 1,12 The main difference of the 2015 and 2016 guidelines concerning the instructions for the maximum amount of chemicals in shopping spaces is the following (shown underlined): • 2015: Alcoholic beverages and cosmetic products, with the exception of aerosols, are not taken into account in the maximum amounts. No protective measures are required of alcoholic beverages containing less than 20% of ethanol. According to FM Global Property Loss Prevention Data Sheets 8-1, Commodity Classification, "Beer and wine (volume of alcohol 20% or less), considered nonignitable liquid, stored in plastic containers that hold 5 gal (19 L) or less" is classified as a Class 1 commodity. 13 In other words, the fire hazard is similar to noncombustible materials packaged in single-layer corrugated cardboard on wood pallets. This supports the instructions of the Finnish Safety and Chemical Agency dated in 2015. In this study, it was seen that a spill collection tray can be beneficial for fire safety, by limiting the intensity of fire and sideward fire spread. However, the instructions dated 2016 allow the storage of alcoholic beverages in retail stores without protective measures. The demonstration tests showed that typical retail store arrays of ca. 40 vol% alcoholic beverages in plastic bottles cannot be considered as a major fire safety risk, compared to many other fire loads in retail stores.
It is noted that "protective measures" in the quotations above refer to spill collection trays and fireproof cabinets for flammable liquids. Fire safety systems, such as fire alarm and fire extinguishing systems, shall be installed in buildings and premises according to fire safety regulations taking into account their use and fire class.