 It is around 5.02pm, on the 10th of June 1999, and a usually quiet and calm forest erupts into a fireball. A teenager fishing by the river and two children would die as a result of the explosion, which investigators would later find to be caused in part by unnoticed damage caused to the pipeline five years before. Although a small death toll, the event would cause over $50 million worth of damage, affecting the local water supply and releasing over 277,000 US gallons of gasoline. The explosion showed the inherent dangers of the fossil fuel supply chain when the transportation of fuel from refinery to shipping terminal was required. Not much can be done about this as the constant and convenient supply of petroleum is vital to any country's economy. As such I'm going to rate the Olympic pipeline explosion here seven on my disaster scale, but here six on my legacy scale. This is due to the event not being greatly remembered today. Our story starts in the 1950s. Washington State had a capacity of 125,000 barrels of refined oil products being supplied from fourth refineries. The state has a problem and that is capacity. You see the USA has an unquenchable first for gasoline and as quickly as it is produced it is consumed. Much of the product is transported via truck or barge, which is fairly slow and as such a faster and more efficient mode of delivery is needed. Starting off in the early 1960s the Olympic pipeline company was formed. The organization set about to design and build and network of pipelines to connect the main refineries to terminals in Washington State and Oregon. By 1964 nearly 300 miles of pipeline were installed. Soon after the total system numbers over 400 miles and consists of multiple lines. In its most northern extremities two lines each 16-inch in diameter run from refineries at Cherry Point and Anacortes where they interconnect then join on to a pumping station near Allen. Heading south from Allen station are two lines a 16 inch diameter and a 20 inch diameter respectively that run parallel until reaching a pumping station in Renton. From there a single 14 inch diameter pipeline extends south to Portland with various smaller lines branching off to multiple terminals. The city of Bellingham in 1966 required a new water treatment plant. The location of the Deakin U site landed right over some of the buried section of the Olympic line. But a pipe work to still be accessible for maintenance and to not risk being disturbed. The line and approximately 724 foot long section was re-routed. The move section was constructed of externally coated API grade X52 steel pipe with a wall thickness of 0.31 inches. Before being pressed into service the pipeline was tested to 1820 PSIG on the 20th of June 1966. The maximum operating pressure was worked out to be 80% of the tested pressure resulting in a value of 1456 PSIG. The pipeline safety regulations limited the allowable pipeline pressure to 110% of the maximum operating pressure during surges which is roughly around 1580 PSIG. Now with such a large and intricate network of pipes how is the pressure kept below these maximum values? This was controlled by a series of valves worked by a computer system. To control and manage the whole system a computer administration software was employed and this was called SCADA supervisory control and data acquisition. The system consists of sensors actuators remote terminal units a communications link and the main SCADA computer. The SCADA shows a network on the computer screen to operators who can intervene to reduce pressures across the system by remotely operating valves. If situations develop that exceed preset values an alarm can be given to the operators. Olympic SCADA system used two identical computers which were called OLY1 and OLY2. One of the computers was used as the primary system with the other available as a backup. The system became a victim of its own success and as such in 1998 a new products terminal at Bayview was constructed. The new terminal was two miles upstream from the existing ALIN station. It housed five product tanks and a storage tank capacity of around 500,000 barrels. A 10,000 barrel transmixed tank was also installed as a breakout tank at the facility. To protect the facility several valves were employed. This was to reduce and even stop the flow of fuel into Bayview in the event of a high-pressure situation. The inlet control valve CV1904 was used to throttle product flow into the Bayview terminal and this was set at 600 PSI-G. A relief valve named RV1919 was installed just downstream of the control valve and was intended to work at a pressure of around 650 PSI-G. It was designed to open and transfer excess product to the transmixed tank if the pressure after CV1904 exceeded a set pressure in the relief valve. Upstream of CV1904 was a receiver manifold arrangement consisting of three motor operated and remotely controlled block valves and these also controlled product flow. But were set to close completely if a pressure of over 700 PSI-G was experienced in the facility thus blocking the whole pipeline upstream from it. These were only intended for extreme circumstances. The system also had another over pressurization prevention system and that was the pumps at Ferndale having a set limit of 1400 PSI-G in which if reached was shut down reducing the flow of fuel. When the relief valve was installed and mixed up in specifications meant that RV1919 was set to operate at only 100 PSI-G. When Bayview was opened and the line was repressurized operators noticed RV1919 opening at 100 PSI-G even though this was out of spec. The engineering supervisor instructed staff to find out why aware that the operation of the valve was from a pilot valve and spring that could be adjusted. Thinking it was at its maximum setting and not reading the instruction manual and new spring was ordered. The manual would have told them that the supplier made two types of valve a high and low pressure model both of which used the same spring meaning replacement would yield no change of operation. Regardless of this the spring was replaced and the valve was adjusted allowing the line to be pressurized. This resulted in the inlet block valve closing around 35 times between 1998 and 1999. Operators didn't think this was much of a problem but you see the inlet block valve is a safety system intended to only automatically step in during points of extreme operating conditions of high pressure. This is what the relief valve is there to do but Olympic Pipeline did not investigate these abnormal valve operations properly and this brings us on to June 1999. It is the afternoon of the 10th of June 1999. Two controllers operate the entire 400 mile pipeline system from the Olympic control room in Renton. In addition to this an Olympic Pipeline controller who had been temporarily assigned as a computer system administrator was also working in the control room. Each controller is in charge of a separate section of the network. Throughout the day the SCADA system was operating as normal using computer OLY02 operating as the primary machine with 01 running as a backup. At around 3 10 p.m. the administrator began to note areas being logged on the SCADA system after undertaking database development work. At around the same time one of the controllers began preparing to stop product delivery to the Tosco facility and start delivery of gasoline to the Harbor Island terminal in Seattle. At around 20 past 3 the North controller contacted personnel at Cherry Point refinery and asked them to begin transferring product to the adjacent Cherry Point pumping station. The pressure in the pipeline began to rise. This is normally remedied by starting a second pump at the unattended Woodenville station. This required the controller to issue the command via the SCADA system. The system failed to send the signal and almost instantly the pressure in the line rose to 1440 PSIG. The controller responsible for the other sections of the line also noted that their commands into the system had become unresponsive. The administrator attempted to change over to OLY01 computer. At around the same time the controller phoned the pumping station Allen to get them to shut down one of the pumps. As the pressure increased the block valve at Bayview began to close. The pressure sensors in the area recorded a 1494 PSIG reading at 3 28 p.m. Uncommanded shutdown of the pumps at Cherry Point and Ferndale stations also happened at the same time. With the shutdown of the Cherry Point and Ferndale pumping stations the 16 inch pipeline was essentially dead from Cherry Point to Renton. With a single pump at Woodenville the only remaining operational unit left. Strangely just a minute later at 3 29 p.m. the system recorded the pressure dropping back down to 230 PSIG. Little did they know but the section of pipeline next to Dakin new water plant by the city of Bellingham had ruptured. The very same section rerouted back in 1966. Grantically the administrator tried to get the SCADA system back up and running eventually returning to operation on OLY02 at 3 45 p.m. At around 4 p.m. the controller came into the computer room and asked if the pipeline could be restarted. At 4 11 p.m. the inlet block valve reopened at Bayview. At 4 15 p.m. the controller contacted the refinery and asked that product transfer be restarted. At 16 past 4 the controller started the pump at the Cherry Point station followed by at 4 17 p.m. by starting of a pump at the Ferndale station. An Olympic employee traveling on his way home called the control centre to report smelling gasoline at a bridge over Whatcom Creek. By now fuel was pouring out of the ruptured pipeline and the smell of gasoline was becoming unbearable for anyone nearby. As such multiple calls to 9-1-1 were made by the public the Whatcom Creek turned a strange colour. The local fire department escalated the response to a hazmat level incident. At about 4 35 p.m. the controller contacted Arco again and asked that the transfer of fuel be stopped. Emergency responders started cordoning off and evacuating the areas that smelled of petrol. At 502 p.m. fire department staff started to see fire igniting in the Whatcom Creek and Benningham police officers also witnessed the ignition. The fire turned into an explosion releasing a plume of smoke 30,000 feet into the air. The temperature was thought to exceed 1090 degrees Celsius. Three were killed Liam Wood 18 who was fly fishing in the creek drowned after succumbing to the fumes and two children Wade King and Stephen Tezorius both 10 were playing near the creek both survived the blast but sustained severe burns dying the next day in hospital. Eight more people were also injured around an hour and a half later the fire was quelled with most the smoke dissipating by 7 p.m. But what was the cause of the event? As high pressure shouldn't have resulted in rupture especially in a well-established system. The question of why would send investigators down a very worrying path in which the way the system was operated and how work undertaken near the pipeline can have a devastating effect. Investigators found that five years before the disaster the Deakin new water treatment plant had undergone modernization works. Part of the work required excavations near the Olympic pipeline. On the 9th of May 1994 during installation of a 72 inch diameter steel water line that was to cross the line of the 16 inch pipe was exposed. Olympics pipeline was then covered with material on the 12th of May 1994 with the water pipe placed some 36 inches above it. This work was undertaken under the watch of Olympic pipeline officials. More work was undertaken over the pipeline to install a tee between existing water pipeline and a new discharge line from the improved treatment site. An Olympic inspector was present but only for part of the day. More duct and pipe work was installed during the water plant upgrade with most but not all the days being witnessed by an Olympic employee. On the 11th of August 1994 several witnesses noted the pipeline being struck 17 feet north of the centre line of the tee connection. Although Olympic was not informed of this personnel coated the area of the pipeline that had been struck with amastic coating before backfilling over it. When looking at a diagram of where the rupture happened we could see that the works for the water treatment plant and the strike location match up. The pipeline was inspected between 1991 and 1999 and anomalies were discovered in the area of the waterworks but no remedial action was undertaken. Investigators found that the cause of the rupture was due to multiple failures stemming from the initial damage to the pipeline but the disaster could have been averted with proper investigation into the damage. But a failure goes much deeper than that as the way Bayview was brought online with the malfunctioning relief valve led to extended times of high pressures within the network which in turn caused the damaged section of the pipeline to rupture. The warning signs of the block valve repeatedly closing were not rectified by Olympic. This problem was further compounded with the actions of pipeline staff undertaking database development work on the SCADA system whilst it was being used to control the pipeline leading to a complete system failure. With the system unresponsive and the block valve closing precious surges found the path of least resistance. The damaged section of the pipeline which resulted in catastrophic structural failure and ignition. Apart from the human cost property damage was estimated at $58 million most of it caused by the ignition and fireball. Many nearby buildings were damaged with broken windows with one house being completely destroyed. The water treatment plant near the explosion site survived but most of the machinery within was severely damaged. The line was eventually however repaired and brought back into service. Olympic reached a $112 million settlement with the victims and in 2000 the pipeline was taken over by BP and the company still operates the line today. The area around the rupture and fireball still shows signs of damage but like all things nature has done a lot to reclaim and cover the evidence. This is the plain difficult production. All videos on this channel are created commons actuation share like license. Plain difficult videos are produced by me John in a currently wet and windy corner of southeast London UK. Check out my Twitter for all sorts of photos and odds and sods as well as hints on future videos. I've got patreon and YouTube membership as well so check them out if you fancy supporting their channel financially and all we're left to say is thank you for watching.