Investigations
This section contains summaries of reports on drinking water
quality incidents and investigations.
Report on drinking water quality incident at:
Ballachulish Water Treatment Works in July 2007
Scottish Water Executive Summary
Ballachuilish WTW is a filtration works consisting of two pressure sand filters and two granular activated carbon (GAC) filters. Disinfection is achieved by dosing sodium hypochlorite into a chlorine contact pipe, prior to a pressurised limestone tank which pH corrects the filtered water. There are two clear water tanks (CWTs) at the works which are configured to operate in series. The free chlorine residual is continuously monitored pre and post the contact main and on the outlet main of the CWT. The process is monitored by telemetry and there is mains power on site. The works serves a population of approximately 610 people, producing an average of 400 m3/day of treated water.
At 04:10 on Wednesday 4th July a low filtered water chlorine residual alarm from Ballachulish WTW was passed to the duty stand-by officer by the Operational Management Centre (OMC) flight desk. The stand-by officer requested confirmation of the filtered, treated (post contact main) and final chlorine residuals and these were reported as 0.02mg/l, 1.92mg/l and 0.89mg/l respectively. Given that the treated and final residuals were good, the works is designed to shut down on low filtered water chlorine and the CWTs were full, the stand-by officer decided to take no action at this time.
The stand-by officer discusses the filtered water chlorine residual alarm from Ballachulish WTW with the process scientist later that morning and the decision is made to send an operator to site to investigate.
At 13:00 an Operator arrived on site at Ballachulish WTW and found the inlet valve on pressurised sand filter 1 failed to close during the backwash sequence at 03:00. As the works was shut down but the valve had failed to close it was possible for raw water to pass via gravity through pressurised sand filter 1 and the rest of the process without the sodium hypochlorite dosing being operational.
The site operator took immediate corrective action by resetting the works and shock dosing the clear water tanks with sodium hypochlorite. At no time did the free chlorine residual exiting the CWT going into Ballachulish zone, drop below 0.60mg/l. A bacteriological incident sample was taken from the final statutory sample tap and was reported as having passed the next day.
Following the incident on the 4th July, the inlet valve to sand filter 1 failed to close on another three occasions which again resulted in unchlorinated water entering the CWT. Different stand-by officers received filtered and treated water low chlorine alarms on the 21st / 22nd and the 30th July, but the alarms were not appropriately actioned as they believed the works to be shut down. Despite delays of a number of hours in taking remedial action at no time did the final free chlorine residual drop below 0.55mg/l following each of these incidents.
An investigation on the 7th August 2007 established that when a close signal was sent to the actuator of inlet valve on pressurised sand filter 1 it started to close but tripped on high torque levels. As a result the plant was shut down and the configuration settings changed, the function of each valve was then tested by Scottish Water, in both manual and auto control, and operated correctly. The plant start up sequence was initiated and all processes came online as per Control Philosophy.
Actions being undertaken as a result of this incident include:
- Formal fact finding interviews to be carried out with customer operations staff.
- Investigate the possibility of configuring the RTU with logic that would send an alarm if the valve were to fail again.
DWQR Assessment on the incident
The DWQR assessment of this disinfection failure is that it was caused by the inlet valve on pressure filter no.1 failing to close during a backwash. The backwash was set to operate when the clear water tank was full and the works was shut down. Since the valve did not close and the works was in shut down mode, raw water was able to pass through the whole process by gravity without any disinfection. A low chlorine alarm was initiated but the operator took no action since he considered the 24 hours storage capacity and the chlorine residual levels adequate.
Another low level chlorine alarm was passed out about an hour later but the operator does not recall receiving it. This was at 06.00 on the Wednesday 4 July 2007. A decision was finally made to investigate at 10.00 and when the operator arrived at the works at 13.00 he found the inlet valve to the filter open and undisinfected water going into the clear water tank. The valve was closed manually and the plant reset to backwash daily. The faulty valve was reported to the mechanical and electrical (M and E) team, disinfection was increased to protect public health and the public health team were contacted. When M and E investigated the following day they found nothing wrong with the valve.
Two weeks later the operator changed the backwash back to the original setting so that the wash occurred when the clear water tank was full. Early the next morning at 07.00 on Saturday 21 July the valve failed to shut again on a backwash. A low chlorine alarm followed and was reasonably quickly acted on. The operator reset the valve and checked that chlorine residuals were adequate. At 21.40 the valve failed to close again, a low chlorine alarm followed but this time the standby operator did not act since he assumed the works was shut down. Two more low chlorine alarms were generated at just after midnight the same day but again the standby operator did not act since again he assumed the works was shut down.
At 08.00 on Monday 23 July the operator arrived on a routine visit, to find the plant running but that records showed that the valve had failed to close 2 days previous on the Saturday. Having discovered that undisinfected water had entered the clear water tanks twice over the weekend, the backwash mode was changed again to be on a timed or high turbidity basis. The next day on Tuesday 24 July the filter was taken out of service while work was undertaken on the control card for the valve. However, on Thursday 26 July high turbidty required the filter to be brought back into service, the control card was confirmed as working and the filter was set back to wash on a full clear water tank.
On Monday 30 July, the valve failed to close again, a low chlorine alarm was generated but the operator believed this was due to a backwash so did not investigate. A further alarm was generated 1.5 hours later and then another about an hour after that but the operator believed the works was shut down so did not iinvestigate. The works was finally visited 7 hours after the intial alarm to find the valve had failed to close again. The valve was reset, the backwash changed yet again to run on time or high turbidity and disinfection increased as a precaution. Samples were now taken from the works and from distribution.
It took a further week to finally establish the problem with a configuration of the closing signal for the valve which was then changed and checked to ensure it was operating effectively.
DWQR notes the actions being taken by Scottish Water (SW) in this case, notably to investigate the communication failures between the duty and standby operators, to investigate the feasibilty of an alarm on the valve status and another alarm to detect flow through the filter even if the works is shut down.
DWQR notes the ongoing work being carried out by SW in respect to its telemetry system generally and that this should improve the robostness and the ability to react to alarms and to fix problems before they pose a risk to public health. Although the systems appear to be getting better, DWQR continues to be concerned about SW's ability to react to alarms and the all important human element of reacting to alarms.