Effective water-based fire suppression systems Nov 2014

Due to the many catastrophic fire accidents that have occurred in the past two decades, the use of water-based fire suppression systems in tunnels has attracted much attention. Full-scale fire suppression tests are expensive to perform, so model-scale tests have become an alternative for studying specific conditions. Tests were carried out in 1:4 model-scale tunnels using key parameters including fuel load covers (ceiling cover and end blocks), activation time, water flow rate, nozzle type, ventilation velocity, sprinkler section length, and tunnel width.

The 1:4 model scale tunnel with K5 nozzles

Model scale tests

The model tunnel was 15m long, 2.8m wide, and 1.4m high. The scale ratio was 1:4, which means that the corresponding full-scale tunnel would be 60m long, 11.2m wide, and 5.6m high. In some tests, the model tunnel width was changed to 1.88m, corresponding to a 7.5m-wide full-scale tunnel. Three types of nozzles were tested, including one scaled T-Rex nozzle and two normal K5 (K-factor 5) and K9 (K-factor 9) nozzles. The sprinkler section length was either 12.5m (50m full scale) or 7.5m (30m full scale). Wood pallets were used as fire sources based on a validated advanced scaling theory. The fuel load was designed to produce a maximum heat release rate of approximately 100MW in full scale, excluding the target.

Analysis of key parameters

The activation time plays an important role in fire suppression efficiency. For a late activation, the fire could become fully developed. In this case, the fire is much more difficult to suppress and the benefit of a fire suppression system becomes limited. In order to reduce the heat release rate and suppress the fire efficiently, the fire suppression system should be activated as early as possible.

The water flow rate significantly influences the performance of a fire suppression system without ceiling coverage. A water flow rate of 5mm/min (10mm/min full scale) is sufficient to extinguish a fire in this case, whereas 2.5mm/min (5mm/min full scale) is not enough and resulted in a maximum heat release rate of 46% of that in a free-burn test (i.e., a test without fire suppression). However, with ceiling coverage the influence of the water flow rate on the performance of a fire suppression system is insignificant under the test conditions.

Free-burn test with a maximum heat release rate of 3.6MW (115MW in full scale)

The fire suppression systems using normal nozzles cannot effectively suppress fires with ceiling coverage, however, the K9 nozzles that discharge larger droplets perform slightly better. An increase of 50% in the water flow rate only decreases the maximum heat release rate by approximately 9% compared to that of a free-burn test. As a comparison, the fire suppression systems using T-Rex nozzles with a water flow rate of 5mm/min (10mm/min full scale) effectively suppressed most of the fires. In the scenarios tested, the T-Rex system with larger droplets performed better than the two systems with normal nozzles.

The ceiling coverage of the fuel package plays an important role in fire suppression. In the tests with ceiling coverage, the fires were difficult to suppress using normal nozzles. Only the T-Rex nozzles with large droplets performed well in the corresponding situations. Note that the activation time was simulated well in the tests by the commonly used heat detection algorithm (it was even conservative in some tests). Therefore, the simulated scenario should be quite realistic. In other words, the combustible covers normally used will probably not burn out before the activation, with the exception of a very thin and highly combustible ceiling cover, such as thin tarpaulin.

The end blocks of the fuel package (in front of and at the end of the fuels) prevented the fuels from being exposed directly to wind. If they were not present, the high ventilation significantly increased the fire growth rate and thus increased the difficulty of fire suppression. For fuels with end blocks, the fire developed much more slowly and could be suppressed more easily. Therefore, trailers should be constructed with end blocks of fire resistant material.

Fig 1. Heat release rate curves: red dashed line corresponds to 50% of the maximum heat release rate of the free-burn test

Tunnel ventilation affects the performance of a fire suppression system by influencing fire development. Further, under low ventilation conditions, heat or smoke fire-detection systems can be triggered much earlier, thereby the fire suppression system can be activated earlier. Therefore, the fire with low ventilation was suppressed more easily due to both the low heat release rate at the activation time and the slow fire growth.

The sprinkler section length decreased from 12.5m to 7.5m (50m to 30m full scale), but did not affect the performance of the fire suppression system with large-droplet nozzles (T-Rex). The key sprinkler section corresponds to the section covering the fire source.

The tunnel cross-section has some influence on the performance of the fire suppression system. For early activation, the tunnel width has no influence. However, with an activation delay of two minutes, a fire in the narrow tunnel was not suppressed efficiently. The main reason could be that close to the fire source, the fuels, together with the end blocks, increased the local gas velocity by obstruction, which stimulates the fire growth and make the fire more difficult to suppress.

Fire spread to a target placed 1.25m from the rear end of the main fire load (5m full scale) was prevented in all the tests with fire suppression. In the free-burn test, the target was ignited at 13.2 minutes (approximately 1.6MW in model scale and 50MW in full scale).

A design fire with a water-based fire suppression system was the key interest of this study. It was concluded that in tests with normal-deluge water-spray systems operated at a water flow rate of approximately 10mm/min (or higher), the maximum heat release rate was less than 50% of the maximum heat release rate of the free-burn test (Fig 1). Note that these results correspond to a full-scale activation delay of less than four minutes after a gas temperature of 141°C was measured beneath the ceiling, or when the activation heat release rate was not over 16MW at full scale. Comparison of these results to full-scale test data will be carried out to further verify the findings.