Testing a theory

Angus Sangster explains the work he carried out on a theoretical fire model for waste and how best to fight waste fires of different types.

During September and October 2017, a series of full-scale fire tests were conducted at the National Fire Service College, Gloucestershire. The fire tests were designed to represent the worst-case fire types likely to be faced by the fire service at a normal waste site. An additional output was the testing of different extinguishing agents, their application and effectiveness.

The materials selected were refuse derived fuel (RDF), e.g. black bag waste collected from domestic street collections, and refined baled plastic. In addition, pre-crushed wood was used to demonstrate the test method and to evaluate the effectiveness of video heat and smoke detection technologies.

Two separate firefighting tests were conducted using RDF and one using plastic bag residue. Each of the tests were conducted within purpose-built bays constructed using blocks supplied by Legioblock®. The first test consisted of setting alight a pile which contained pre-crushed wood. The purpose of this initial test was to establish a reliable method of initiating the deep seated fire. After a number of failed attempts an oxy-propane lance was selected as the pilot for this test program. It was observed that highly volatile fuels consumed all of the available air too quickly. Even attempts to supply air through a pipe could not sustain a sufficient fire to generate a self-sustaining fire.

The second set of tests consisted of a stack of baled plastic which was ignited from the upwind direction and allowed to burn for forty minutes before firefighters extinguished the fire.

Results from conducting these tests have been analysed and the significant findings are discussed herein. This report explains various extinguishing media, the methods of application and relative effectiveness the output from which can provide essential recommendations for firefighting operations.

The objectives of the tests were to verify the theoretical model developed as part of a PhD thesis that describes how waste fires develop in waste and based upon this theoretical model. Using these assumptions, I have developed an approach to firefighting in both piled waste and stacked baled waste. These methods would be applied using a range of water additives.

Theoretical fire models

In 2014, the WISH real fire test programme started with a series of laboratory tests on one tonne samples at the FPA’s facilities in Gloucestershire, and concluded in October 2016 in Essex. During the programme, I have developed the following theoretical fire models. the experiments have been repeated with several materials and at three, six and nine tonne piles to eliminate as far as possible any scaling issues.

The test data and the analysis was reported to the WISH board and is incorporated within the WISH Guide 28 (#1). These firefighting test not only serve as a platform with which to develop firefighting tactics they also serve as validation tests for the theoretical models set out below.

Surface fires

Impermeable fuel bed

This is typically a pile of material that has been crushed down so it forms a single homogenous mass which prevents the movement of convection through it.

Semi permeable fuel bed

A pile of material with sufficient tensile strength in the individual particles to allow the passage of convection currents to pass through the mass of the pile, but the repose of the individual particles of the mass create holes that are smaller than the flame height for the material. This prevents flame passage into the mass of the pile and results in a surface fire that is ash and char controlled.

Permeable fuel bed

A pile of material with sufficient tensile strength to allow the passage of convection currents to pass through the mass of the pile and the repose of the individual particles create holes that are greater than the flame height for the material; thus allowing the flame to pass through the mass of the pile. 

Piles of material display tendency to degrade from permeable fuel beds, over time or due to degradation as a result of burning.

A fire initiated on the surface of a pile of semi permeable or impermeable fuel beds will remain on the surface and tends to become ash and char controlled. Surface fires do not burrow into the material to become deep seated fires (see figure 1).

When the individual particle that makes up the pile have sufficient tensile strength to maintain gaps of sufficient magnitude to allow the flames to pass into the mass of the pile then the whole of the materials will become involved in fire (see figures 2 & 3).

Figure 1
Figure 2
Figure 3

Deep seated fires

This is a fire that initiates within the mass of the pile, will form a hot core in excess of 250C and can develop anaerobically, i.e. without a supply of oxygen. The developing fire will find a passage to the surface of the material and ultimately, breach the surface. This breach will provide an air supply to the hot core gasses resulting in a fuel-controlled fire burning at the peak heat release rate for the material. This fire behavior has been observed in coal heaps and documented (#2), (see figure 4).

Figure 4 - click to expand
Figure 5 - click to expand

Stacked bale fires

When baled material is stacked above 1m in height, a surface fire that affects the vertical space between the individual pillars will generate vortices driven by the convection currents. The fluid dynamic flows produced by the vortices can be sufficient to strip any ash and char away from combustible material and accelerate phase change in low grade plastics to a point where the process is almost instantaneous. This can result in an unusually high heat release rate for any given material within this region (see figures 5 & 6 and 7).

Figure 6
Figure 7.1
Figure 7.2
Figure 7.3

As the stack bindings fail, together with partial thermal expansion of the material at the top of the stack, a section of the stack will fall and bury burning material. Stacked bales tend towards degeneration into piles in this way and as a result, the pile of material will generally develop several buried deep-seated fires.

This behaviour is occasionally mistaken for a surface fire burrowing into a pile. When the bales have a high plastic content by volume the fires tend to vaporise the fuel bed which results in a significant reduction in the calving behaviours. During test with high grade plastic, HDPE plastic bottles, the stack was observed to become unstable which resulted in a burning bale falling off of the top of the stack this is obviously an additional hazard for firefighting.

Note: the RDF tested generally had a plastics content of 30 to 40% this was sufficient to suppress the fire growth. The other materials tested contained 85 to 95 % plastic and in all of these cases the fires were very intense. 

Piled waste tests

The piled test of wood (see figure 8) presumed that the provisions of the WISH Guidance 1 regarding fire resistant separation had been applied. The configuration used presents some key challenges for firefighting crews compared to an open pile of material. These challenges are:

  1. The material is only accessible from one side and therefore all firefighting operations must be undertaken from this elevation.
  2. The material displays a high degree of water resistance.
  3. The volume of material involved which will require the use of heavy plant to aid in the excavation of the pile to expose the core of the fire (see figure 9).

Two bays were filled with 17 tonnes of RDF and ignited by the application of an oxy-propane lance flame directed to the middle of the pile through a 1.5m section of stainless steel pipe with a 100mm diameter. Application of the ignition source in this way helps simulate ignition of a deep-seated fire and a full justification for this method will be included in a thesis that will be submitted in support of a PhD to be completed by 2020. The test method provided a good correlation to data obtained from real fire data.

Both RDF fires were ignited at the same time as baled plastic was ignited. Both RDF fires were allowed to develop for 48 hours and were monitored using thermocouples buried throughout the mass of the piled waste and thermal imaging cameras.

Both fires developed significant anaerobic pyrolising cores with temperatures recorded between 400C and 500C. Waste fires have apparent similarities to coal seam fire behaviour in this respect.

The surface of both of the RDF fires was eventually ignited manually, approximately 52 hours after initial ignition, as neither fire had breached the surface as expected. The surface fires were then allowed to develop for two hours prior to the fire extinguishing test.

Figure 8
Figure 9 - click to expand

Extinguishing methods

Water test

Water jets were applied to the surface fire and it was noted that application of the medium had a good knock down effect. However, water did not have any impact on the temperature reading obtained from the thermocouple 1m below the surface, and the application of copious amounts of water did not appear to have any impact of the water’s ability to penetrate the pile; the water merely ran off.

Wet Class A Compressed Air Foam System (CAFS)

The foam solution was applied to the RDF fire was noted to have displayed a similar ability to knock down the flames; however, with CAFS, it was also noted that the temperature was reduced 1m below the surface of the fire showing a degree of penetration. Again, the application of copious quantities of foam did not show any better performance and just contributed to runoff. Foam however, displayed a degree of resistance to burn-back not display by water.

Wetting agent

The water and wetting agent solution was applied to the RDF fire test. It was noted that the wetting agent did knock the fire down quickly resulting in a reduced temperature 2m below the surface of the pile. But it was also noted that, in common with the other agents, the application of copious quantities did not improve its efficiency but just added to the water runoff.

Re-ignition of piled waste

A water and wetting agent solution was applied, and was noted that the wetting agent did knock the fire down quickly, resulting in a reduced temperature 2m below the surface at the pile.

The thermocouple readings were observed to have dropped to ambient temperature and therefore, it was assumed that the fires had been extinguished and the site was left overnight for approximately 12-hours.

However, the next morning it became apparent that the cores of both fires had returned to their original fire conditions and had reached the pre-extinguished temperature. This behaviour was also observed at the earlier test site at Barling in Essex, which leads the author to believe that the core residue has become chemically predisposed to re-ignition.

This is not understood at present and will require further research to establish the cause of these repeated re-ignitions. Firefighting crews should be cautious and isolate burnt material for a number of days to establish that the material will not reignite.

Fire resisting block work

The fire resisting block work was formed using blocks provided by Legioblock® although, walls formed of concrete blocks supplied through another supplier may be as effective, as would walls constructed of concrete or other fire-resistant material.

The blocks were arranged into three bays. Bay 3 was sealed using a proprietary own brand intumescent mastic sealer purchased from a builder’s merchant; and Bays 1 and 2 were left unsealed. The construction were subject to the tests outlined in the table above (see also figures 10, 11 & 12).

Figure 10
Figure 11
Figure 12

The pre-crushed wood fire was used to demonstrate the test method and to evaluate the effectiveness of video heat and smoke detection technologies. Both technologies proved very effective with this type of material and detected the thermal lance before the pile had been involved. The resulting deep-seated fire took a further 4 hours to breach the surface (see figure 13).

Figure 13

The configuration of the bay acts like a room and as a result the flame is stretched up the “overboard ” by air being entrained, (see figure 12). This is a well-documented phenomenon in fire dynamics. In practical terms it is unlikely that the overboard could be raised high enough to overcome this problem although it may theoretically be possible. This is discussed in An introduction to Fire Dynamics (#3) by Dougal Drysdale.

It is more likely however, that fire crews would have to consider this effect and position covering jets to the material either side of the bay that is on fire (see figures 14, 15 & 16).

Figure 14
Figure 15
Figure 16

Although the Legioblock walls had been exposed to temperatures in excess of 900C for 20 hours and then rapidly cooled with firefighting jets, the spalling was relatively minor at around 10 to 15mm maximum. This bay was subjected to a further intense plastics fire for 2 hours at around 1100C and again rapidly cooled with water jets. The blockwork did not display any signs of instability or failure.

Baled plastic fire tests

Recycled mixed-plastic bales were piled three bales high and three bales in each other direction forming a stack of 27 bales. Plastic was selected due to the intense heat release rate observed in earlier tests. All the fires were piloted by the application of a naked flame at the base of the gaps between the stacks on the up-wind elevation. All the fires behaved in a similar way.

Fire spread in the virgin material was quick: from point of ignition to full involvement within four minutes. The test beds were built 6m from the previous fire to establish the effectiveness of the separation distances suggested in the Environment Agency Fire Prevention Plan (FPP) guidance document. In figure 17 below, which was the water with wetting agent test, all three test beds were fully involved in under 12 minutes of the pilot flame being applied to the first stack.

In all, baled plastic stacks have been the subject of nine fire tests using various plastic compositions. In these tests, it was noted that the plastic did not enter the liquid phase but rather appeared to enter the vapour phase and combust almost instantaneously. The heat output therefore between the different grades and types of plastic was not as significant as earlier research would have suggested and would appear to be a function of the fluid dynamic regime that the orientation of the fuel bed has on the fire behaviour.

Figure 17

Typically, the surface temperatures of the burning stacks are likely to be in the region of 1,100C to 1,200C or more. The impact of the heat flux on fire crews and surroundings cannot be underestimated and it is suggested appliances and equipment should be a minimum of 40m away from a plastic bale fire (see figure 17).

Extinguishing methods

Water tests

Two 70mm jets were applied. The jets were applied to the base of one of the gaps to use the fluid dynamic flows driving the system to transmit the steam through the fire as postulated following work package 1, 2 and 3 of the WISH fire test program. One jet remains in the original position to prevent the fire burning back while the second jet is worked around the base of the stack to extinguish the fire. This system worked well and supports the theoretical model. Water usage was 19,000 litres (2×70 mm jets at 7 bar delivering 475 l/min for 20 mins).


The attack used in test 1 was emulated using CAFS. A Class A wet solution was used and was markedly more effective. Water usage 1,800 litres of water (7 mins 2x 128 l/min)

Wetting agent

The wetting agent resulted in the quickest knock down of all the media used and produced the least run off. (2x 45mm jets at 7 bar using 0.3% induction of agent (5 litres). Time to extinction: 2 minutes (see figures 18, 19 & 20).

Figure 18
Figure 19
Figure 20

Case study

On the 4th August 2014, a fire was reported at the Shanks MBT plant at Rainham in East London. London Fire Brigade, who attended this fire, made the following observation regarding large piles of RDF (#4).

“It should be noted if the waste is a large stack of refuse derived fuel (RDF) and the sprinklers are located high up in the roof space, by the time they are activated then due to the make-up of the RDF a crust forms and any water runs off the crust and does not penetrate to extinguish the fire,”

The fire became deep seated and despite all the fire suppression systems the fire burnt for over 48 hours. Interestingly, this is consistent with the above text. Demonstrating, both anecdotally from real fire experiences and from test data that piled waste materials are surprisingly water resistant. In such cases, the fire was not dealt within the four-hour target. This individual site was heavily fire engineered with a substantial emphasis on fire suppression. However, it is clear to see fire suppression alone is not a guarantee of an early resolution to a fire of this magnitude.


Water or other firefighting media should not be continuously directed onto a pile of waste as it has no effect other than adding to pollution. Jets should be used to knock flames down and then stopped once excessive water runoff is observed. At this stage the surface material has reached a point where saturation is at its greatest and the application of water serves no purpose.

It is quite possible that the flames are being emitted from an exposed core. Such a core must be fully exposed to allow for it to be extinguished or smothered to suppress the smoke and flames with earth or sand. Research conducted during this program would indicate that partially burnt waste materials possess a pre-disposition to reignite readily and therefore the wisdom of land-filling partially burnt waste materials is questionable.

Piled materials should be excavated and the “muddy puddle” principle is currently the most appropriate approach. A ‘muddy puddle’ is where fire crews excavate a gently sloping hole which is filled with water. The waste is excavated from the main pile and submerged in this hole where fire crews can spray the material with water. Where this is not a practical approach to extinguishing a fire, the material should be buried under earth or sand to reduce noxious emissions.

The material is likely to continue to pyrolyse slowly until the fuel is consumed. Controlled burns may be an option, but the mass loss rate is fairly slow, so this could make extinguishment protracted. Flooding a Burning pile of material with water is not likely to be successful and is likely only to protract the incident and cause pollution.

Surface fire in piled material can readily be extinguished but the surface should be stripped back to unburnt material to ensure that the fire was not piloted by a deep seated and concealed fire. Crews should not walk on piles of waste as deep seated cores are not obvious and may not give any indication of their presence on the surface of the pile.

Temperature readings of the surface material are not a guarantee that there is not a hot core below the surface, as waste materials are good insulators. Baled and stacked materials are predominantly surface fires and consequently extinction of the fire is relatively achievable.

Stacked baled material fires are driven by the convection current in the gaps between the stacks. Jets should be directed at the base of the stack and into the gaps, and crews should resist the temptation to direct the jet at the flames.

RDF, Solid recovered fuels (SRF), paper and textile bales burn with less ferocity than plastic and consequently given time the bales fail and form piles. These piles should be fought as pile fires. Waste management sites could consider providing a supply of wetting agent that is compatible with the types of waste they process and with their local Fire and Rescue Service’s pumping equipment.


The full-scale fire tests and the success of the firefighting interventions validate the theoretical model developed by the author. Concrete walls and concrete blocks including, for example Legioblock® are an effective means of providing compartmentation provided the joints are sealed.

Six metre spacing between baled storage is not adequate and is potentially dangerous in the case of plastic. In general terms, wetting agent proved to be the most effective medium for extinguishing waste fires, whilst water proved to be the least effective. Standard firefighting tactics are deemed to be ineffective and alternative firefighting operations should be considered when dealing with an incident of this type.

Proactive waste site management can reduce the severity of an incident through measures such as compartmentation and stack positioning. A number of fire detection technologies were use at these trials; however, I have not covered the test results in this article as the analysis is too complex and is deserving of a specific article in the future.


#1  Waste Industry Safety and Health Forum, (2017) Reducing Fire Risk At Waste Management Sites

#2  Sloss L.L., (2015) Assessing and managing spontaneous combustion of Coal,  IEA Clean Coal Centre

#3  Drysdale D., (1998) Introduction to fire dynamics. 2nd ed. Chichester: John Wiley and Sons, p.137.

#4  Letsrecycle.com (5th August 2014), Fire still burning at Shanks’ East London MBT

Angus Sangster is a Senior Fire Safety Engineering Manager at International Fire Consultants Group