Prevention is the best approach to landfill fires, but when that fails, you better be prepared with a timely and effective response.

By Paul Henderson and Tony Sperling

Background
Why the Fire Started
Fighting the Fire
Monitoring Program
Conclusions and Lessons Learned
References

Small surface fires at MSW landfills are relatively common as a result of the presence of potentially flammable materials and such ignition sources as hot vehicle exhausts, improperly disposed of cigarette butts, and loads that inadvertently contain smoldering materials (Ettala et al. [1996] report 285 surface fires in Finland’s 633 sanitary landfills each year). These fires can generally be quickly suppressed through the use of water or cover material or the removal of the burning material from the active landfill area. Underground fires are a much greater threat because it is often difficult to determine the extent of the fire, and suppression using conventional means is often ineffective.

On October 18, 2000, City of Vancouver, BC, staff discovered an underground fire in an area of the Vancouver Landfill filled with woodwaste demolition material. Although the area obviously affected by the fire was only about 50 m², the total volume of demolition material in the area in question was approximately 300,000 m³. A fire in a private demolition landfill located near the Vancouver Landfill involving approximately 250,000 m³ of demolition material in the fall of 1999 cost approximately $4 million to extinguish over a period of four months.

Given the potential impacts of a large underground fire, it was critical to ensure that the Vancouver Landfill fire was brought under control as quickly as possible and that any risk of future fire was minimized.

Background

The Vancouver Landfill is located at the southwest corner of Burns Bog in the municipality of Delta, approximately 20 km south of Vancouver. The facility is owned and operated by the City of Vancouver and receives approximately 400,000 metric tonnes of MSW each year.

Demolition materials consisting primarily of woodwaste from the demolition of wood-frame buildings are accepted at the Vancouver Landfill for use in the construction of a "mattress layer" under MSW. The mattress layer provides a foundation layer on top of the compressible peat established in the vicinity of the Vancouver Landfill. The demolition material also acts as a conduit for drainage of leachate from within the landfill to a perimeter leachate-collection ditch.

The demolition layer has been traditionally filled to a thickness of approximately 3 m prior to filling with MSW. The area where the fire occurred was previously an open ditch system between two adjacent landfill cells. As part of a strategy to increase the height of the landfill, the demolition layer in this area was filled in two lifts to create a drainage layer approximately 8 m thick. A thicker drainage layer was installed to account for future differential settlement of the underlying peat soils. Demolition materials were installed in the area over approximately three months commencing in the spring of 2000. As shown in Figure 1, the total area that received two layers of demolition material was approximately 80 x 700 m.

Figure 1. Landfill Area Receiving Demolition Material. Click here for enlarged views

Why the Fire Started

Spontaneous combustion is the outbreak of fire without application of heat from an external source. Spontaneous combustion can occur through the storage of organic materials, such as woodwaste, coal, tire chips, compost, or hay.

In organic materials, spontaneous combustion occurs when heat initially produced through biological degradation is not allowed to dissipate, thus raising the temperature of the material. Biological degradation will generally cause temperatures to increase to approximately 70° C. At temperatures much higher than 70° C, microorganisms die. Temperatures beyond 70° C are associated with the chemical oxidation of organic materials in the presence of oxygen. For wood, exothermic oxidation commences around 200° C and combustion with open flame commences around 300° C (National Fire Protection Association, 1976). The following table is a practical reference summary for evaluating landfill fire temperatures based on our experience with six major landfill-fire projects.

For spontaneous combustion to occur, conditions must be ideal. The following conditions can lead to spontaneous combustion:

• a pile of sufficient size to retain heat (the Ontario Fire Marshal’s office recommends that woodchips be stored in piles less than 4 m high and 8 m wide with an overall volume of less than 1,000 m³ to avoid spontaneous combustion [Government of Ontario, 1998]),
• moisture content around 25% on a wet basis (Legislative Assembly of Ontario, 1998) (drier conditions prevent biological activity, wetter conditions reduce porosity and prevent temperatures from increasing beyond biological levels),
• supply of oxygen (many spontaneous-combustion fires occur near the windward edge of a pile of material during windy conditions),
• sufficient insulating capabilities to retain heat in the pile (Swedish researchers found that uncovered, loosely packed piles of demolition material piled up to 5 m high did not spontaneously combust, whereas compacted piles regularly spontaneously ignited [Hogland et al., 1996]),
• prolonged storage of organic materials (the Ontario Fire Marshal’s office recommends storage of woodchips for less than three months [Government of Ontario, 1998]).

In the case of Vancouver Landfill, the fire occurred behind an approximately 3-m-high bank of demolition material adjacent to an internal site-access road. The road was located on top of the first layer of demolition material, and the bank represented the edge of the second layer of demolition material. The bank had been partially covered with soil, but sufficient void space was present to allow wind to blow into the bank. The bank was located on the windward side of the demolition lift. Material in the area of the fire had been in place approximately six months.

Partially capped bank on edge of demolition cell

Fighting the Fire

The fire was discovered at approximately 2:00 p.m. on October 18. Although no flames were visible when the fire was discovered, an area of approximately 50 m² had settled approximately 60 cm, and smoke was venting from it. Steam vents were previously noted throughout the 700- x 80-m demolition area. With the discovery of the fire, there was concern that the steam vents indicated fires were also present in other areas of the demolition cell since various authors have reported that landfill fires may exist without showing any external evidence (Ettala et al., 1996; Carpenter, 1996; and Herold, 1999).

Within two hours of the discovery of the fire, Poschner Construction (the city’s onsite contractor) was delivering water to the fire in three off-road dump trucks with delivery capabilities of 20,000 lit. per load. The trucks are normally used at the landfill to deliver cover material. In the event of a fire, the trucks can be quickly outfitted with a tailgate to allow them to haul water. Water is supplied to the trucks via an overhead delivery system feeding from an onsite dredge pond. The trucks can be filled in about three minutes.

Because the demolition material is essentially free-draining and the surface area where the fire was located was essentially flat, large quantities of water could be applied without reducing trafficability in the area or creating a risk of slope failure. By the end of the evening of October 18, approximately 800,000 lit. of water had been delivered to the fire area.

Off-road dump trucks continued to deliver water to the area of the fire and other hot spots within the demolition area on October 19 and 20. By the end of October 20, the water trucks had delivered a total of more than 2 million lit. of water.

Poschner Construction off-road truck, 20,000 lit. per load

While the off-road dump trucks were delivering water to the fire, a pumping system was erected approximately 2,000 m from the onsite dredge pond for conveying larger volumes of water. The system involved pumping water from the dredge pond into leachate collection ditches. The flooded ditches were used to convey the water to the vicinity of the fire. A 500-m-long pipeline was established to convey the water into a large spray-irrigation cannon. The spray-irrigation cannon was capable of delivering approximately 14,000 lit. of water per hour and operated continuously for a week commencing on October 20.

To prevent air from entering the demolition layer, a cap of silty clay from the dredge pond was installed along the 700-m-long bank of the demolition area (see photo 3).

Clay cover used to eliminate oxygen entry into landfill

A case study described by Hogland et al. (1996) shows a very similar scenario to the Vancouver Landfill fire. In a controlled experiment to monitor self-heating, Hogland et al. constructed a 3.5- to 4-m-high pile of demolition material over a 400-m² area. The pile was compacted during construction and covered with woodchips. Hogland et al. monitored a variety of parameters for several months. Temperatures of around 85° C were attained after about two months. Almost no methane was recorded after two months. Oxygen levels varied as a function of the depth of the pile, but they were around 20% in the middle and upper areas of the pile. A fire occurred after about six and a half months of storage. The fire occurred on the windward side of the pile the day after a strong wind fully oxygenated the pile. In a single day, temperatures in the pile increased from 85° C to approximately 240° C. To extinguish the fire, Hogland et al. capped the pile with clay. After about two months, temperatures in the pile dropped from several hundred degrees to 50° C.

Similarly, fire erupted at the surface in a recent fire incident within a 2 million—m³ demolition material cell in Minnesota. After several days of digging, it became apparent that the fire could not be contained by excavation. Instead, the hot area was covered with low-permeability soil, and water was injected into boreholes drilled into the waste. After several weeks of monitoring temperatures and gas compositions, it appears that the fire abated and temperatures are returning to normal levels.

In the Vancouver Landfill fire, the Minnesota landfill case history, and the fire described by Hogland et al., excavation was not chosen as the preferred method to extinguish the fire. It is difficult to determine the extent of an underground fire, and the risk of excavating is that additional oxygen will be delivered to the fire. To minimize oxygen intrusion, any excavations to dig out burning material or install fireguards should be immediately backfilled with soil or other inert material.

Monitoring Program

Figure 2. Carbon Monoxide Concentration Measurements, Vancouver Landfill Fire. Click here for enlarged view

On October 20, the City of Vancouver retained Sperling Hansen Associates Inc. to monitor the progress of the firefight and to provide recommendations on future actions. Through a simple monitoring program focusing on the installation of subsurface probes in the demolition area, Sperling Hansen was able to show conclusively that the fire had been completely extinguished and that the conditions that had created the potential for a fire had been resolved.

The monitoring program involved collecting temperature and gas-composition data throughout the demolition cell. Readings were taken at all observed steam vents as well as over a grid of regularly spaced "barhole punch" probes installed through the intermediate cover into the shallow subsurface.

Barhole-punch monitoring points were created by driving a solid-steel bar approximately 1.5 m below grade with an excavator, then pushing a hollow-steel, 2.5-cm-diameter sampling pipe into the cavity. An oversize bolt was used to prevent the pipe annulus from clogging during insertion. On achieving full depth, the sampling pipe was pulled back approximately 15 cm, dislodging the bolt and allowing subsurface samples of gas composition to be collected.

Maximum temperatures noted at the surface using a hand-held infrared sensor were 54.1° C. On this project, as well as on several other landfill fire projects investigated by Sperling Hansen, temperature readings taken at the surface did not prove to be particularly useful in indicating the extent of the fire zone. Subsurface temperature readings from thermistor strings installed in boreholes within the waste typically provide a much more effective indication of fire conditions in the subsurface. On this project, thermistors were not installed because of the relatively shallow fill and also on the basis that the gas composition data from the barhole-punch monitoring program indicated that the extent of the fire was limited. Gas samples were analyzed using portable sensors for oxygen, methane, carbon monoxide, and hydrogen sulfide.

At the start of the monitoring program, subsurface oxygen levels within the burn area were typically in the range of 15-21% oxygen. As firefighting and capping efforts progressed, oxygen levels dropped consistently. By the time the monitoring program was completed on November 3, oxygen levels in most wells had dropped below 1%.

Similarly, subsurface methane concentrations were below 1% methane in many of the sampling ports at the onset of monitoring. Low methane concentrations indicated that decomposition was occurring in an aerobic regime. As mentioned previously, aerobic decomposition is associated with greater generation of heat and can lead to spontaneous combustion. By November 3, methane concentrations had climbed above 40% in most of the wells.

On this project, carbon monoxide (CO) proved to be the most effective indicator of landfill fire. Initially, CO concentrations up to 315 ppm were noted in the vicinity of the active burn zone. These concentrations gradually declined as the fire was brought under control. Because elevated CO concentrations were not noted in steam vents and barhole-punch sites outside the active burn zone and because there was a direct relationship between CO concentrations and fire activity, we believe that CO provides an excellent indication of subsurface fire activity. Contours showing the gradual reductions in CO concentrations are shown in Figure 2.

Based on observations on this project as well as on experience at landfill fires at Campbell Mountain Landfill, the Delta Shake and Shingle Landfill, and the recent landfill fire in Minnesota, Sperling Hansen has developed the following empirical scale that we now use routinely to assess fire conditions in construction demolition landfills.

Conclusions and Lessons Learned

Based on the experience developed from the Vancouver Landfill fire, a number of conclusions can be reached with respect to prevention of fires in demolition materials and effective fire control methods:

Fire Prevention. Intermediate cover material must be installed on all exposed slopes with particular attention given to windward slopes; fill plans incorporating vertical firebreaks need to be developed for demolition areas to limit the potential extent of any fires; and because of their composition and porosity, woodwaste demolition materials pose an extreme fire risk and must be handled with equal or more caution than MSW.

Firefighting and Monitoring. Sufficient equipment resources and personnel must be available to immediately respond to any landfill fire; the approach of cooling down an underground fire with large volumes of water proved effective in this situation and is expected to work equally well on other demolition-material fires, provided the hot zone is shallow (10 m); cutting off the oxygen supply to the burn zone is an effective way to fight fire in demolition materials; an effective monitoring program is essential in determining the extent of an underground fire and to monitor progress in fighting the fire; subsurface carbon dioxide levels in conjunction with methane and oxygen levels provide critical data in determining the extent of a fire and evaluating the potential for a fire to develop; and to reduce the risk of spontaneous combustion, operators should strive to establish anaerobic conditions within demolition landfills whereby oxygen is totally excluded, with methane and carbon dioxide being the predominant gases.

References

Carpenter, M. "Subsurface Refuse Fires, Evolution and Control: Case Study at Palailai Landfill, Oahu, Hawaii." SWANA 1999 Northwest Regional Symposium, pp. 385-398. 1996.

Ettala, M. et al. "Landfill Fires in Finland." Waste Management and Research, pp. 377-384. 1996.

Herold, G. "Subsurface Landfill Fire Suppression and Monitoring." SWANA 1999 Northwest Regional Symposium. 1999.

Hogland W., T. Bramryd, and I. Persson. "Physical, Biological and Chemical Effects of Unsorted Fractions of Industrial Solid Waste in Waste Fuel Storage." Waste Management and Research, pp. 197-210. 1996.

National Fire Protection Association. Fire Protection Handbook, Fourteenth Edition. Boston, MA. 1976.

Paul Henderson, P.E., is manager for the City of Vancouver Landfill. Tony Sperling, P.E., is president of Sperling Hansen Associates in North Vancouver, BC.