Independent Investigation Report of the Nov 2005 Drum Fire at the Idaho National Laboratory Site

Organizational and Process Weaknesses

The excavation process and procedures and identified organizational and process weaknesses. The more significant weaknesses involved failure to follow procedural requirements, lack of needed detail in the operational procedures, and a deficient Management Self Assessment for start of operations after contract transition. These weaknesses did not contribute to nor were they identified as causal factors for the fire.

1.2          Process Description

Excavation was started in a designated area of Pit 4. REs are constructed and relocated to other SDA pits as necessary. Soil overburden may be removed and stored for pit backfill. Based on visual examination, waste-zone materials are segregated into two waste groups: a targeted waste group and a nontargeted waste group. Targeted waste is repackaged into 55-gal drums and nontargeted waste is returned to the pit. The nontargeted waste consists of items such as combustibles, scrap metals, interstitial soil, and miscellaneous sludges. Targeted waste is contaminated with volatile organic compounds (VOCs), various isotopes of uranium, and TRU radionuclides. It consists of graphite, filter media, uranium roaster oxides, and Series 741 and 743 sludges. The excavator operator places nontargeted waste into soft-sided containers that are staged outside of the pit or relocates nontargeted waste within the pit to allow access to targeted waste. If an operator determines that waste at the excavation is nontargeted, it is not retrieved. As room becomes available in the excavation pit due to the removal of targeted waste, the excavator operators return the staged nontargeted waste containers to the pit using the appropriate excavator end effector. As excavation progresses, the pit is backfilled with nontargeted waste, potentially contaminated soil, previously removed overburden, or clean soil brought into the RE from outside. At the end of the excavation campaign, the pit is covered with enough clean overburden soil to prevent the spread of contamination.

As the waste is excavated, the excavator operator watches for intact drums. If an intact drum of targeted waste is found, the excavator pierces and depressurizes it with spikes (thumbs) located on the bucket to release potential hydrogen that may have accumulated in the drum. The excavator then places the drum or loose waste materials on a waste transfer tray for transport to the DPS. Waste may also be staged outside the pit (aboveground) until it can be processed.

Technicians at the DPSs open the vented drums and internal containers, if necessary, using electric and pneumatic hand tools and select screening samples from the potentially targeted waste. The screening technicians use visual assessment, gross alpha/beta swipes, direct low-energy gamma detectors, and/or organic vapor analysis to identify targeted waste. If the technicians determine that the waste is targeted, then they assign the applicable shipping codes, collect physical samples to meet shipping requirements, and remove any Waste Isolation Pilot Plant (WIPP) prohibited items. WIPP-prohibited items are either returned to the pit or containerized for further processing. Transfer trays with nontargeted waste are returned to the pit are to allow content removal.

The ARP safety support systems include fire protection systems (detection and suppression), the excavator and telehandler safety features, and the breathing air system.

1.3.1      ARP Fire Protection

The SDA is not provided with a fixed water distribution system. The nearest fire hydrant is located 1,500 ft from the ARP site, at the west edge of the RWMC operations area. Water for any extended suppression efforts will be supplied by the fire department through an extended large diameter hose lay. To provide the minimum required water supply, a trailer is provided with 3,000 ft of pre-connected 4-in.-diameter hose used to reach the RE. An additional 400 ft of 4 in. hose and 500 gallons of water are available from the fire department engine.

To supplement the available water supply, a 4,000-gal firewater storage tank is installed near the RE. The water tank is equipped with an isolation valve and a 4-in. standardized connection to allow prompt connection to the fire department water tender or engine pump inlet.

1.3.1.1              Fire Detection and Alarm Systems. The RE is provided with smoke detection and a building fire alarm and occupant notification system to notify occupants and the fire department in case of emergency. The system includes manual-pull stations at each exit and occupant notification devices. Additionally, telephones are provided and some operations personnel are equipped with two-way radios to notify the shift supervisor and to communicate with the fire department in case of a fire or other emergency during operations.

For the RE, occupancy provides an effective detection means when the building is occupied, and is relied on as the primary detection method during operating periods. A fire watch is instituted when the area is unattended and the periodicity of observation is specified in operating procedures. The fire watch may be a roving fire watch or be provided via a camera-based system covering the RE. The camera-based system consists of web cameras monitored by the INL Fire Alarm Center on dedicated monitors installed at CFA-666. Web cameras are installed on opposite walls of the RE, and one in each room of the airlocks. Additionally, thermal-imaging cameras that cover the excavation area are also provided. While the web cameras have a fixed field of view, the thermal-imaging cameras have limited pan, tilt, and zoom to allow them to focus on the immediate excavation area. While the thermal-imaging cameras are installed to provide adequate coverage of the excavation as it progresses, only a single camera is in use at any time when required for fire watch. As part of the approved fire protection exemption, a high-sensitivity smoke detection system has been provided to the RE. The high-sensitivity smoke detection system is employed in conjunction with the fire-watch cameras during nonoperating periods only to avoid false alarms from suspended dust that accompanies the excavation.

1.3.1.2              Fire Suppression Systems. The RE has no fixed, building fire suppression system because of the temporary nature and low combustible loading of the structure. As part of the approved fire protection exemption for the RE, several local-application fire protection features have been provided to aid in incipient-stage fire suppression. These include an automatic dry chemical suppression system on the excavators and telehandler, automatic and manual dry chemical suppression systems for the drum packaging stations, and traditional portable fire extinguishers.

The drum packaging station fire suppression systems are designed and installed to control Class A, B, and C fires that originate within the drum packaging stations without the use of water. Based on the area of each drum packaging station, only a single nozzle discharging within the station is required. A second nozzle and additional dry chemical agent ensure proper operation without regard to placement of the bag hoist, a potential discharge obstruction, the drum packaging station openings, or drum packaging station ventilation. Each suppression-agent release panel is provided with a local alarm and is connected to the building alarm panel. Discharge, whether initiated by automatic, manual pull, or direct manual activation, activates building alarms and reports to the INL Fire Alarm Center.

The excavators and telehandler engine compartments are equipped with onboard, commercially- available, automatic, dry-chemical fire suppression systems. The systems are capable of automatic detection and actuation and remote manual actuation. The dry chemical is discharged through fixed nozzles into the protected areas, suppressing the fire. When the system is activated, an interlock shuts down the vehicle diesel engine.

1.3.2      Excavator and Telehandler Forced-Air Ventilation Systems

The retrieval area excavator and telehandler cabs are pressurized to protect the vehicle operator. By maintaining the cab at a positive pressure, the operator is protected from the potentially high radioactive and particulate airborne contamination levels outside the cab. The cabs are not designed to be airtight; therefore, they rely on a positive pressure to minimize contamination levels in the cab. Nor are the cabs designed to be contamination free. Radiological personnel decontaminate the cabs as practical to prevent high contamination levels, but contamination is expected to be tracked into the cab as the operators enter the cab from contaminated areas.

Each vehicle is equipped with a radial blower to pressurize the cab. The airflow is from outside the cab, through a prefilter, through the fan, and through two parallel, double-stage, nuclear-grade HEPA filters before pressurizing the cab. Differential pressure gauges are installed to measure the pressure drop across the HEPA filters. These gauges are used for indication of HEPA filter performance. The cabs are designed to operate at or above 0.75 in. water column (WC). An alarm system notifies the operator in the event of low differential pressure in the cab. The alarm is set to actuate when cab pressure is less than or equal to 0.45 in. WC and provides the operator with both a visual and audible indication.

1.4 Pit 4 Waste Descriptions

The RWMC is located at the INL Site, which is a DOE facility located 52 km (32 mi) west of Idaho Falls, Idaho. The INL Site occupies 2,305 km2 (890 mi2) of the northeastern portion of the Eastern Idaho Snake River Plain. The RWMC is located in the southwestern portion of the INL Site. The SDA (Figure 3) is a 97-acre area located in the RWMC and consists of 20 pits, 58 trenches, 21 soil vaults, Pad A, and the Acid Pit where past waste disposal activities occurred. Pit 4 is located in the approximate center of the SDA (see Figure 3). Pit 4 has a surface area of 107,082 ft2. The total volume of Pit 4 is estimated at 1,600,000 ft3. The selected retrieval area (Figure 4) is approximately 1/2 acre, comprises about 20% of the pit, and is located in the eastern portion of Pit 4. This area is designated by grids that are 15 ft u 15 ft. The data consolidated from shipments from 1964 to 1967 depict locations within the pit by using the Pit 4 monuments. The positional error associated with the data is estimated to be +/- 20 ft (EDF-4478).

Figure 3. Subsurface Disposal Area, with Pit 4 in the center.

x        Line-generated waste: This category is expected to contain various waste materials removed from the plutonium-processing gloveboxes including items such as glovebox gloves, combustible waste, graphite, and filters.

x        Combustible debris: Waste comprising paper, plastic, wood, and other combustible materials was designated as combustible debris.

x        Metal debris: Waste that was predominantly metallic (e.g., pipe, conduit, and empty drums) was designated as metal debris.

1.4.2             Description of Waste in I-2

Historical records provide insight as to the type of wastes stored in the immediate vicinity of the material involved in the fire (EDF-4478). The location of drums being retrieved at the time of the event was Grid I-2 (Figure 5). The likely waste types in or near Grid I-2, are 741-Sludge, 742-Sludge, 743-Sludge, Graphite, RFP Combustible, RFP Noncombustible, and INL Noncombustible. Less likely but possible waste types are roaster oxides, line-generated waste, and beryllium waste. Twenty-one drums of roaster oxide had been excavated at the time of the event: one drum from grid A-3, one from B-5, one from F-2, twelve from F-4, two from G-4, one from G-5, one from H-2, one from H-3, and one from H-5. Roaster oxides are known to include incompletely oxidized metal (DOE 1994). 741-Sludge is known to present a hazard for gas formation and explosion potential as this material was involved in a prior Advanced Mixed Waste Treatment Project drum fire (BNFL 2003). Graphite has a known tendency to burn persistently once exposed to high ignition temperatures.

Figure 5. Grid of retrieval area inside the ARP Retrieval Enclosure (each grid is approximately 15 ft u 15 ft).

2.            INVESTIGATION SCOPE, PURPOSE, AND METHODOLOGY

The independent team was chartered by the Vice President in charge of RWMC to determine the cause of the drum fire/explosion, identify organizational and process weaknesses, analyze potential gaps in hazards analysis, develop recommendations regarding the path forward to sampling and investigation of the fire site, and recommend corrective actions based on our findings. The team included DOE, CH2M-WG Idaho, LLC (CWI), and subcontracted personnel with a wide range of expertise. Additionally, nationally recognized experts in several areas were used as advisors to the team. Biographical information on the team is included in Appendix A. The Investigation began onsite interviews on November 29, 2005, and concluded on February 23, 2006.

A detailed investigation plan is included as Appendix B. Standard event investigation processes were used and included the following:

x        Gathering facts through interviews, document reviews, walk-downs of the work areas and detailed investigation of the event scene.

x        Analyzing facts and identifying causes through approved analysis techniques by qualified cause analysis personnel.

x        Development of recommended corrective actions.

This document provides the team’s analysis, conclusions and recommendations to the CWI Vice President in charge of RWMC Operations.

3.            DESCRIPTION OF EVENT

3.1           Operator Description of the Event

The following description of events is based on the information that was provided by the excavator operator in written statements and interviews. Review of the video evidence of the event supports the operator’s descriptions, although the actual fire and the trench internals are not visible in the video.

On November 21, 2005, the excavator operator retrieved, vented, and placed three drums from the pit in a group on the surface level of the RE (referred to as the deck). He saw an additional group of three drums and decided to retrieve these drums prior to assisting the telehandler operator with dirt moving operations. The operator stated that he normally punctures the drums with the excavator “thumb” while he is picking them up and lifts them directly up to the deck. The operator in this case decided to vent all three drums and set them down in the trench prior to moving them to the deck. The operator retrieved and vented one drum. The operator placed this drum in the bottom of the trench and it fell onto its side. The operator retrieved and vented the second drum and set it next to the first. While the operator was in the process of rotating the excavator to retrieve the third drum, he heard a loud noise and almost immediately heard something hit the excavator cab window. When he looked, he saw a large ball of flame around the second drum. The flames appeared to potentially also engulf the first drum. The excavator operator initially attempted to use the dust suppression spray to extinguish the fire. The fire appeared to still be growing. The operator started placing dirt over the fire. The dirt initially knocked down the fire. The fire then began to extend up through the dirt. The operator keyed his microphone and reported the fire to his supervisor and the camera operator. The operator placed additional dirt on the fire and used the dust suppression spray until fire and smoke were no longer visible. The facility was placed in the warm standby mode. Facility management was first informed by operator of the loud bang and material hitting the cab of the excavator at the critique. Facility management then determined that an explosion may have occurred and the appropriate reporting occurred.

3.2           Additional Information from the Evidence Available

Based on review of the video evidence of the event, the team determined that it occurred at approximately 09:56 on November 21, 2005, while the camera operator had moved the camera to look at other operations in the ARP enclosure. There is a period of approximately 20 seconds where the excavator is not in view. When the camera is initially positioned to see the excavator again, there is no visual evidence (such as a plume or overhead material movement) that any event had occurred. After the fire is reported, the camera operator zoomed in on the excavation and white smoke was visible (Figure 6).

The excavator operator and the telehandler operator were the only personnel in RE when the event occurred. A number of personnel were in the airlocks attached to the RE. With the exception of the excavator operator, no one interviewed heard or saw anything when the event occurred.

Figure 6. Smoke coming from fire after initial extinguishing effort.

Interviews of the excavator operator, reviews of the logs, and inspections of the videotapes from the event provided limited information about the fuel involved in the fire. The excavator operator reported a “loud pop” that suggested an explosion/deflagration or rapid pressurization event resulting in a rupture of the drum. The drum was approximately 25 to 30 ft from the excavator cab. The event had enough energy to propel something at least that far. The delay between the operator puncturing the drum and the sound was reported as being roughly 5 seconds (ICP 2005a). Nearly coincident with the noise, the operator noticed a flash in his periphery. When the operator turned to face the drum, he reported observing a large fireball engulfing the second drum and potentially involving the first drum. The flames appeared largely orange or yellow, with some blue initially at the base. The smoke emanating from the fire was white or light gray. The smoke could have been steam generated from the dust suppressant (water). This color smoke is not indicative of common combustibles. The operator did not note any burning debris around the main fireball. Initial reports were made that at least half of the drum was destroyed. The operator indicated during interviews with the team that he did not know how much of the drum was missing. The operator estimated the amount of drum destroyed after repeated prodding by management to provide an estimate.

The operator took emergency actions to extinguish the fire in compliance with the emergency alarm response procedure (EAR-246). The first action involved using the excavator dust suppression system to extinguish the fire. The water appeared ineffective and could have worsened the fire. From the reports and interviews, one can not discern whether the dust suppression system was ineffective due to the fuel type involved (e.g., pyrophoric metal) or whether the limited flow rate proved insufficient to control the fire. The operator then used dirt, alternating with the dust suppression system, to extinguish the fire. The thermal imaging camera (TIC) was not being used at the time of the event. No temperature data of the event is available.

The following day, November 22, the TIC indicated a surface temperature of ~54°F. The temperature decreased approximately linearly to ambient (~45°F) by November 28. The persistent nature of the hot spot suggests a continued reaction (oxidation) of a fuel type that requires limited oxygen or which could continue to react with oxidants present within the waste.

4.            SCENE INVESTIGATION

The Project developed a recovery plan to obtain the necessary samples to make a positive identification of the fuel source/material of concern. An evaluation of safety significance was written and presented to DOE for approval to ensure a safety basis existed for the sample activities. The sampling activity was conducted in a two phased approach.

4.1           Phase 1 Initial Sampling

The initial sampling operation was planned to include excavation until the drum that was the source of the event was located, placing a portion of the soil/material around the drum top in a tray for sampling and then covering the drum with soil. Additionally, soil from the pile of dirt used to extinguish the fire was placed in a tray and sampled to provide a baseline for comparison. The sampling process was performed in the DPS. If the samples proved conclusive, no further sampling would be performed. This evolution was performed on January 26, 2006. The operator who was in the excavator when the event occurred was the excavator operator for this sampling operation.

Almost immediately upon starting digging operations, what appeared to be the top of a drum and a drum ring were located (Figure 7). It appeared that the drum was only approximately 1 ft below the surface of the soil. Sample material was loaded into two trays and transferred into the DPS. A marker (overpack drum lid with a handle attached for ease of handling by the excavator) was placed on top of the “drum” and soil was placed over the marker.

Figure 7. Photo of initial excavation showing what was suspected as the corroded drum top and ring.

Sample analysis was completed and proved inconclusive. There were no significant differences between the baseline samples and the sample of the material around the “drum.” Sample results are included in Section 5.2.

4.2           Phase 2 Sampling of Drum Contents

Based on inconclusive results, a second sampling evolution was planned that involved opening and sampling the actual contents of the drum that was the source of this event. The plan included excavation and inspection of the drum, opening and sampling the drum contents, and location and inspection of the other drum that had been placed in the area prior to the event. On February 14, 2006, the second phase of sampling was conducted.

After location and removal of the marker, the operator began digging in front of the ‘drum’ location to expose the body of the drum. After digging down several feet, the drum ring slid down and the dirt pile collapsed. Apparently, the drum ring and other debris in the area had formed what looked like a drum top through the camera.

Excavation continued and the drum to be sampled was located a few feet away from and approximately 1-1/2 ft below the drum ring (Figure 8). The drum lid and drum ring were still in place on this drum. There was an opening at the top of the drum and the one drum side was compressed (Figure 9). During movement of the drum, gray material appeared to fall out of the drum opening. Following inspection of the drum externals, the drum lid was removed and an inner 30-gal drum was observed. The lid and ring of this drum were still in place. The top of the drum had what appeared to be a hole possibly from a puncture in it and appeared charred and pitted (Figure 10).

Figure 8. Photo of excavation of drum continued, showing top lid and ring.

Figure 9. Photo of event drum exposed

Figure 10. Photo of excavation of drum continued, showing top of the drum charred and pitted.

The lid to the 30-gal drum was removed and material from this drum was placed in a tray for sampling. This material was black granular/powder in appearance (Figure 11) and created a significant dust cloud (Figure 12) when dumped into the tray. Some sparking was clearly visible in the bucket of the excavator when the excavator picked up the tray containing the 30-gal drum contents. A quickly dissipating set of sparks were seen coming from the area where the excavator jaw is joined to the bucket while the jaw was being closed (Figures 13, 14, and 15). After placing the trays at surface level, a small amount of the 30-gal drum contents was placed in a tray and raked through with the excavator. No sparking was noted.

The drum the operator first vented and placed in the area was also located. It was immediately adjacent to the event drum (Figure 16). There was no visual evidence that the drum had been involved in a fire. The lid of this drum was found partially open, consistent with being squeezed by the excavator during venting. Plastic liner material was visible in the opening. This material displayed no evidence of melting or discoloration from a fire (Figure 17).

The trays were moved into a DPS and the contents sampled. A hand-held gamma spectrometry instrument (ICS-4000 Radionuclide Identifier) was used to perform field analysis of the samples. The material that was outside the inner drum did not have any identifiable isotopes above background. The material from inside the inner drum was identified as U-238. Subsequent laboratory analysis confirmed the material to be depleted uranium consistent with what was disposed from Rocky Flats.

Figure 11. Photo showing black granular/powder removed from 30-gal drum.

Figure 12. Photo showing dumping of black granules into tray, causing dust cloud.

Figure 13. Photo showing sparks at jaw to bucket joint.

Figure 14. Photo showing sparks on both sides of the excavator bucket.

Figure 15. Photo showing sparks dissipating.

Figure 16. Photo showing drum next to event drum.

Figure 17. Photo showing drum next to event drum with no indication of fire.

5.            DETERMINATION OF THE CAUSE OF THE FIRE

5.1    Sampling Logic

To facilitate identification of the fuel (or waste type) involved, the investigation team recommended the following analyses:

x        Gamma spectroscopy of the retrieved solid and drum contents. Gamma measurements will indicate the relative uranium content and isotopic distribution of the waste. Comparison of this data versus historical knowledge (Haar 2005) for the waste types will help discern the fuel involved.

x        Elemental (Metals) Analysis. Digestion and elemental analysis will identify the primary metals included in the waste. Personnel will look for elevated concentration of uranium, zirconium, cadmium, beryllium, and other potentially pyrophoric metals that may have contributed to the event. The analysis will also indicate if mercury metal is present in elevated quantities.

x        Volatile Organics Analysis (VOA). The visual evidence from the videotape of the fire suggests unlikely involvement of substantial amounts of common combustibles or liquid solvents. However, VOA will provide more definitive evidence. VOA may also indicate if any organo-mercury compounds are present that may be shock-sensitive or potentially explosive.

x        Anion Analysis. Some of the line-generated wastes or other materials from Rocky Flats may include melted salts from pyrochemical operations. Analysis of the anions in the collected soil – and comparison with similar measurements on a control sample of soil from the area – might exclude this waste type from consideration.

x        X-Ray Diffraction (XRD). XRD analysis the samples will provide molecular identification of the material involved. Determination of the degree of oxidation for any uranium present will provide insight as to whether incompletely oxidized roaster oxides were involved in the fire. If present in sufficient mass, this analysis will also determine the presence on nitrate salts and inorganic oxidants.

x        Total Inorganic Carbon/Total Organic Carbon (TIC/TOC). This analysis will determine the amount of carbon present including insight as to the amount of graphite.

5.1.1      Potential Scenario Analysis

Table 1 lists potential fuel types and initiators. The table includes common characteristics of the fires for the various fuels and a ranking by the investigation team of the likelihood that the incident involved the given fuel (or waste) type. The investigation team reviewed a number of references regarding fire characteristics in development of the table (DOE 1994; Zerwekh 1979; Molecke, 1979). Given the description of the event and the potential involvement of multiple waste drums in the event, the potential existed that the initiating event and the persistent fire could involve two distinct waste types.

Table 1. (continued).

Table 1. (continued).

5.2           Sample Results

The first phase of sampling collected replicate samples of the background soil as well as replicate samples from each of the two excavator bucket lifts of material. The samples identifications follow.

x      ARPB0101GR – Lift 1-1 x      ARPB0201GR – Lift 1-2 x      ARPB0301GR – Lift 2-1 x ARPB0401GR – Lift 2-2

x        ARPB0501GR – background sample 1

x        ARPB0601GR – background sample 2.

These samples received analyses as discussed above (5.1 Sampling Logic). Analyses—see

Table 2—showed the presence of trace amounts (53–173 mg/kg) of depleted uranium (0.21 ± 0.04 wt% U-235) in the solid near the drums that only slightly exceeded the concentration (57–85 mg/kg) observed in the background soil samples. Concentrations of other common metals (e.g., iron, calcium, aluminum, potassium, sodium, calcium, and manganese) proved indistinguishable between the samples. No evidence existed of elevated concentrations of any other potential pyrophoric metals characteristic of selected waste types.

Table 2. Uranium analyses of samples.

U-235       U (total)           U-235

mg/kg       mg/kg       Enrichment

Background Soil

Outer Drum