12 Summary

The following is a brief summary of important facts dealt with in the report, the main issues, the views of the Commission and its recommendations. We would like to point out that in summarising the main points in this manner, certain shades of meaning may be lost. As far as the recommendations are concerned, only the individual recommendation itself has been included, without the text explaining its basis. The Commission has based its recommendations on the situation as it was on 5 April 2000.

12.1 Appointment of the Commission and its work

On 7 April, two days after the accident, the commission of inquiry that had been appointed following the Åsta rail accident was also asked to investigate the rail accident at Lillestrøm station on 5 April 2000. The Commission’s mandate was formally expanded by Royal Decree on 14 April 2000. The following were appointed members of the Commission:

1. Judge Vibecke Groth, Borgarting Court of Appeals, chair

2. Øystein Skogstad, chartered engineer, SINTEF (Foundation for Scientific and Industrial Research at the Norwegian Institute of Technology)

3. Finn Mørch Andersen, chartered engineer, Directorate for Fire and Explosion Prevention

4. Ingemar Pålsson, chartered engineer, Det norske Veritas, Gothenburg, Sweden

5. Marika Kolbenstvedt, sociologist, Institute of Transport Economics.

At the request of the Commission members, the Commission was expanded to include Joakim Böcher, engineer, Det norske Veritas, Denmark, on 26 July 2000. Jacob Ferdinand Bull, associate of the law firm Arntzen, Underland & Co., was secretary to the Commission. Jens-Henrik Lien, research assistant at Oslo University, was also secretary from 15 October onwards.

The Commission’s mandate was to undertake the necessary investigations to establish the facts of the accident at Lillestrøm station and its cause. The mandate made particular mention of the fact that it would seem particularly appropriate to assess the premises on which the transport of dangerous goods is based.

The Commission enjoyed close cooperation with the police of the Romerike police district in determining which investigations should be conducted.

Det norske Veritas (DnV) was appointed expert adviser and requested to undertake a technical examination of the gas tanks on the train, and to see whether they complied with the current requirements relating to technical standard and the regulations for the transport of propane gas. DnV was also commissioned to perform a general examination of the goods train’s braking systems, the level of maintenance and the general condition of train No. 5781 when it left Alnabru station on the day of the accident.

NSB BA (Norwegian State Railway) issued a report on the condition of the brakes on train No. 5781 after the accident. This work was supervised and monitored by the police and the Commission. Sven A. Eriksson, chartered engineer, former manager of the braking systems section at SJ (Swedish State Railway) from 1988–1998, now employed by Green Cargo AB, was engaged by the Commission to assess the report from the NSB BA.

In addition, investigations have been conducted of the leaks in the damaged tank wagons and analyses and calculations have been made of how close the accident at Lillestrøm was to a catastrophe.

The Commission has interviewed four witnesses. The Commission has had at its disposal all the statements made to the police, and any documents or other information that might be of interest from the police investigation conducted in parallel with the Commission’s own investigations. In addition, the Commission has obtained documentation and other material of importance to its investigations from NSB BA, the Norwegian National Rail Administration, the Norwegian Railway Inspectorate, the Directorate for Fire and Explosion Prevention (DBE), VTG and Statoil.

12.2 The Accident

On Wednesday 5 April at 00.38, goods train No. 5781 left Alnabru station for Mosjøen. It was delayed by approximately two hours due to a heavy fall of snow. The locomotive was an E1 16. The train documentation kept in the locomotive stated the weight and length of the train, its braking percentage and that it was carrying compressed flammable gas.

After the train had passed Strømmen station, the driver began to apply the brakes. He registered that there was no braking effect and that the train’s speed was increasing. He then noticed that the distant signal for the entry signal to Lillestrøm station was yellow. This meant that the entry signal was red. The trip recorder shows that when the train passed the distant signal it was travelling at 102 km/h. The train’s maximum permitted speed according to the train documentation was 90 km/h. The ATC (Automatic Train Control) unit was set for 100 km/h.

When he was only able to achieve a slight braking effect, the train driver applied the emergency brakes when he passed Sagdalen block signal post. Train No. 5713 stood waiting at Lillestrøm station on track 7. The train driver of train No. 5781 realized that he would not be able to stop in time and began to signal a danger warning using the train’s horn.

The train driver tried to contact the rail traffic controller via the train radio. He did not make contact until he passed the entry signal to Lillestrøm station, which was red, and he then warned the traffic controller of what was about to happen. The train driver ran back into the engine room and threw himself down onto the floor when the train ran into train No. 5713 at 00.57. The speed of the train at the moment of collision was 62 km/h. Neither of the engine drivers or any other person was injured in the collision.

The collision caused some material damage, but the most important event was that the two propane tanks in train No. 5781 were damaged and propane leaked out. After a short time, the propane ignited. The situation was critical and came very close to a BLEVE that would have killed a large number of people and laid Lillestrøm in ruins. About 2000 people were evacuated from the danger zone estimated at 1000 metres from the tanks.

The danger of explosion was averted and the evacuees were able to return on 9 April 2000.

12.3 Cause of the accident

12.3.1 Direct causes

It was apparent immediately following the collision that the direct cause of the collision was brake failure in train No. 5781. On the basis of this fact, the Commission undertook a thorough examination of the possible causes of the brake failure. The brake failure proved to be extensive.

The technical examinations that have been carried out show that the train’s actual braking capacity was lower than the driver could expect on the basis of the train documentation. The train documentation indicated that the train had a braking percentage of  77. Measurements of brake pad pressure and braking distance show that the actual braking capacity of the train was only 57 per cent. This is 2/3 of the braking capacity stated in the train documentation. However, calculations made by the Commission show that the train with this reduction in braking capacity would have been able to stop in good time before it reached the collision location. Consequently, poor brakes alone cannot explain the brake failure.

Various examinations, calculations and reconstructions of the operation of the train on the night of the accident have been carried out to find the reason why the train’s braking distance was substantially greater than its braking capacity would indicate. In the following section, this part of the brake failure is termed the «main brake failure» or the remaining brake failure.

The investigations made have shown that the train must have been virtually without brakes for about 15 seconds, or for a distance of just over 400 metres after passing Strømmen station, where its speed increased from 95 to 102 km/h in a downward slope of 17 o/oo. Furthermore, the speed profile shows that deceleration was weak but relatively constant from Sagdalen to the point of collision. Investigations have shown that this corresponded to a drop in pressure of approximately 1 bar in the main brake pipe, or that the last 3–5 cars were without brakes.

The main brake failure must therefore be the reason for both the driver’s initial inability to brake and subsequently the fact that braking capacity was weak though constant. Brake failure of this kind can only be explained by a blockage in the main brake pipe and that this blockage changed position in the main brake pipe at Sagdalen, or that there was no air pressure in the brake system.

A plug of ice would be capable of blocking the main brake pipe and thus prevent the drop in pressure that is necessary to activate the brakes on the cars behind the location of the plug. However, the speed profile indicated by the train’s trip recorder does not correspond well with the idea that ice was the cause of the main brake failure. A plug of ice would first have had to block the main brake pipe between the locomotive and the first car, then work loose and block the pipe between the three to five last cars. In addition, inspections at the scene of the accident did not produce evidence of sufficient moisture in the main brake pipe for the Commission to regard it as likely that a plug of ice could have caused the main brake failure. Furthermore, when the brakes were tested before departure, air passed freely through the entire length of the main brake pipe. These factors would indicate that it is so unlikely that a plug of ice was the cause of a main brake failure of this extent that the Commission rules out this possibility.

Consequently, the only reasonable explanation for the brake failure is, in the view of the Commission, that the train driver must have put the driver’s brake valve in the mid-position and forgotten to move it into the running position before he left Alnabru. With the brake valve in mid-position while a train is travelling, the brake system will not be replenished with air, and with the leakages that have subsequently been discovered, the air pressure will gradually be lost in the course of about 30 minutes. From Strømmen station to Sagdalen block signal post the speed of the train increased by 7 km/h, from 95 km/h to 102 km/h. This is consistent with a situation whereby the air in the brake system has escaped and the train is virtually without brakes. The speed was subsequently slowly reduced up to the point of collision when the speed was 62 km/h. This is a deceleration of 0.14–0.16 m/s² or a braking effect corresponding to a drop in pressure of 0.9–1.05 bar in the main brake pipe. In the opinion of the Commission, this braking power was achieved by the release manoeuvre used by train driver Jensen when he was not able to achieve any braking effect. Release manoeuvres produce a rapid charging of the pressure in the main brake pipe, but not of the whole system. This is consistent with the low, though constant braking effect up to the site of the collision.

These factors would indicate that the main brake system had failed to such an extent because the driver’s brake valve was set at mid-position. This had an impact on the weak braking effect of the train because it prevented replenishment, resulting in a brake system that was virtually empty of air when the brakes were applied.

If the driver’s brake valve had not been in the mid-position, the train would have stopped well before the collision location whether travelling at the actual speed the train had when it passed Sagdalen block signal post or at the maximum permitted speed for this section of the line. Thus, what finally triggered the collision was the fact that the driver’s brake valve was set in the mid-position.

However, the poor brakes on the train may have been a necessary condition for the collision to take place. Theoretical calculations carried out by the Commission show that the collision could have been prevented even with the driver’s brake valve in mid-position if the train’s brakes had the braking capacity stated in the train documentation. However, the assessment obtained by the Commission from S. A. Eriksson, based on theoretical calculations and practical experience, indicates that the collision would have taken place, although at a lower speed. The differences of opinion are due to the uncertainty that has arisen because it is now impossible to say with absolute certainty how the brake control valves on each wagon functioned when the air pressure in the main brake pipe was so low. This will also in this particular case have considerable effect on the calculation of the braking distance. The Commission cannot therefore say with certainty whether the fact that the driver’s brake valve was in the mid-position was the sole cause of the collision or whether inefficient brakes were a necessary condition for the collision to take place.

12.3.2 Underlying causes

12.3.2.1 Deficiencies in NSB Gods brake maintenance procedures

As mentioned earlier, train No. 5781 had less effective brakes than the driver of the locomotive could have expected from the train documentation. Based on calculations carried out by the Commission, the train would normally have stopped before reaching the entry signal to Lillestrøm station, but it would have used the entire safety margin provided by the distance between the distant signal and the entry signal. Travelling at a higher speed, the train would have continued past the entry signal.

In its brake report prepared for the Commission, DnV pointed out that NSB Gods (NSB freight) does not conduct brakepad pressure measurements in connection with annual inspections or otherwise tests train braking power. Thus, no check is made of a train’s theoretical braking percentage to ascertain whether it corresponds with its actual braking capacity. This means that NSB Gods is not fully informed of the actual braking capacity of the trains. DnV’s deceleration measurements of five goods trains and its review of accidents and near-accidents caused by brake failure show that safety-critical deviations between braking capacity and braking percentage can occur without being detected and without any attempt to detect them having been made.

12.3.2.2 No test-braking

The driver did not conduct test-braking after exiting Alnabru as stipulated in the regulations. The requirement of test-braking before the downhill gradient into Lillestrøm was not met either. If test-braking had been conducted before the train began its descent, this would have revealed that the driver’s brake valve was in mid-position. The driver would then have moved the driver’s brake valve to the running position and discovered that replenishment of air to the brake system was blocked. A study of the geographical profile of the line and the train’s trip recorder indicate that test-braking could have been conducted between Lørenskog and Strømmen. If test-braking had been conducted, there would have been enough time to charge the brake system with air before the downhill gradient into Lillestrøm. The train would then have had sufficient braking capacity to avoid collision.

12.3.2.3 Warning of lack of air pressure in brake system

Locomotive E1 16 is not equipped with any device to warn the driver of a safety-critical drop in pressure in the brake system. If a light and/or sound signal or traction blocking device had been linked to a critical fall in pressure in the brake system, the driver would have discovered before he started braking at Strømmen station that the train’s brake system was virtually empty of air. The collision would then have been prevented because the driver would have discovered that the driver’s brake valve was in mid-position.

12.4 Causes of gas leaks

12.4.1 Direct causes of gas leaks

The gas leaks in the two gas tanks, which created the dramatic situation at Lillestrøm, were a direct result of the collision itself. The tanks were wagons 1 and 2 behind the locomotive in train No. 5781. Technical investigations carried out by DnV for the Commission and the observations made after the accident showed that the tanks in themselves withstood the strain they were subjected to in connection with the collision.

The manholes that were located opposite each other during transport had covers that were attached with protruding bolts. The gas leaks arose as a result of the bolts becoming interlocked during the collision and two bolts on each of the covers were damaged. One bolt was torn off, while another was deformed. Thus the design of the tanks with their protruding bolts was the direct cause of the gas leaks that arose as a result of the collision.

12.4.2 Underlying causes of the gas leaks

12.4.2.1 Design of manhole covers etc.

The two gas tanks involved in the accident were in themselves very strong, and were equipped with a stop valve on the underside of the tank to prevent leakages if the pipelines were damaged in connection with a collision or derailment. If the tanks had also been designed with manhole covers with a smooth surface, it is highly likely in the view of the Commission that the collision would not have resulted in a gas leak. There were also protruding elements at the opposite end of the tanks. The Commission is of the opinion that the leakage would not necessarily have been prevented if the tank had been turned in the opposite direction. The Commission calls for greater awareness with regard to the design of tanks that carry dangerous goods. For example, there are no rules for the positioning and design of manhole covers. With smooth ends the preconditions for avoiding gas leaks in connection with a collision or derailment would have been far better than with protruding parts on the tanks.

12.4.2.2 Location of tank wagons in train, protection wagon, etc.

The two tank wagons were positioned just behind the locomotive. If the gas tank wagons had been placed further back in the train, a greater proportion of the collision energy would have been absorbed by cars not carrying dangerous goods. It is then likely that the collision would not have subjected the two gas tanks to a level of stress that would have resulted in gas leaks.

If a protection wagon had been placed between the locomotive and the gas tank wagons on train No. 5781, the gas tank wagons would have absorbed a much smaller proportion of the collision energy. However, the Commission cannot say with certainty whether the gas leak would therefore have been prevented. A protection wagon between the two gas tank wagons would probably have prevented direct contact between the manhole covers and thus prevented the gas leak, provided the protection wagon had a smooth surface without any protruding parts. The Commission finds it likely that the use of a protection wagon between the locomotive and the first gas tank wagon and a protection wagon between the two gas tank wagons would have prevented the gas leak. In its review of the legislation, the Commission has discovered that there is little in the way of regulations concerning the use of protection wagons in the transport of dangerous goods.

12.4.2.3 Locomotive’s direct brakes not used

In addition to its main brakes, train No. 5781 had direct brakes that only work on the locomotive. Direct brakes work independently of the train’s main brakes, but were not used by the driver. The use of direct brakes would have doubled the locomotive’s braking capacity at speeds above 55 km/h compared to using the main brakes only. Calculations show that using the direct brakes throughout the downhill gradient would have increased the available braking power for train No. 5781 by about 20 per cent. The addition of direct braking would not have prevented the collision, but the speed at the moment of impact would have been reduced to between 40 and 45 km/h. Whether the gas leak would thereby have been prevented is impossible to say. It must be stressed that driver Jensen had not been trained in the use of direct brakes in the event of brake failure.

12.5 The gas fire

The term BLEVE (boiling liquid expanding vapour explosion) is used about the situation that arises when a tank containing flammable gas or liquid is subjected to extreme heat in connection with a fire and ruptures because of overloading due to internal pressure or other factors. The contents will then immediately evaporate. This will result in an effective mixture of propane and air, resulting in explosive combustion in the form of a fireball some height above ground. A fireball like this will have a surface temperature of more than 1000 °C and will radiate extreme heat. In addition, the shock wave from the blast will cause fragments both from the tank and from the area near the tank to be propelled outward at high velocities, causing damage.

Thermal radiation in the most exposed zone will result in fatal burns for any persons who are outdoors, and wooden buildings and combustible materials inside windows etc. will catch fire. This zone can stretch for hundreds of metres in all directions from the tank.

On the basis of the investigations and analyses that have been carried out, the Commission is in no doubt that a BLEVE would have developed with catastrophic consequences on the night between 4 and 5 April if cooling of the tanks had not been undertaken. In the opinion of the Commission, a catastrophe would have occurred between 3 and 4 a.m. if action had not been taken to start cooling the tanks. At the time the catastrophe would have occurred, evacuation had not been started. It must be assumed that under these circumstances more than hundred people would probably have been killed instantly, and several hundred would have been seriously injured. Many people would perhaps have received life-threatening injuries. It must be assumed that any persons who were outdoors within a radius of 500 metres from the tanks would most probably have been killed by thermal radiation. Furthermore, fires in a large number of buildings at the same time at this time of night would probably have meant that many people would have been unable to get out in time.

The fires that would have developed would have raged from some time before anyone was able to do anything about them. Many fire service personnel would have already lost their lives and most of the firefighting equipment would have been destroyed. Much of Lillestrøm town centre would have been reduced to ruins as a result.

In the opinion of the Commission, a catastrophe the likes of which we have not seen in Norway in peace time was probably less than an hour away when the fire service started cooling the tanks on the night between 4 and 5 April.

12.6 Rescue operation

The police, ambulances and the fire service were quick to arrive at the scene of the accident. It was quickly established what had actually happened and that there were no casualties in connection with the collision.

The organization of the rescue operation ran for the most part smoothly throughout. In the course of the operation a number of important decisions were taken and measures implemented that in combination prevented a catastrophe. Fire service personnel were aware at an early stage that the tanks would have to be cooled with water. Awareness of the importance of cooling the tanks and how this should be accomplished increased as experts were drawn in as advisers.

The incident commander did not possess the necessary expertise to assess the hazards and the response in a major accident involving propane tanks where gas was leaking out and on fire. An early initiative was therefore taken to call in specialists.

In the opinion of the Commission, the efforts of the rescue service prevented the Lillestrøm accident from developing into a catastrophe that would have had enormous consequences. Good emergency preparedness, access to suitable equipment and a good supply of water prevented an accident on a very large scale.

To drain the tanks by means of the leaks that had arisen in connection with the collision could have taken two weeks or more. This would have had a major impact on the Lillestrøm community. The incident commander was notified of an emergency response group composed of operators in the Swedish LP Gas industry called «Gasakuten». This group has five permanent members who have the expertise and the specialized equipment to be able to empty large gas tanks. Gasakuten arrived in Lillestrøm on the afternoon of 6 April and began burning off (flaring) the gas the next day. Both tanks were completely empty by Monday 10 April. On Sunday 9 April the danger of explosion was considered to be over and the evacuees were able to return to their homes.

The accident happened at a location where it was easy to bring in heavy equipment and there was a good water supply. If this had not been the case, it must be taken as likely, in the opinion of the Commission, that the accident would have developed into a BLEVE. In many parts of the country, the conditions for managing this kind of accident are not as favourable as they are at Lillestrøm.

12.7 Recommendations of the Commission

The Commission has the following recommendations: