This month’s process safety beacon is on trusting instruments. Or to be more accurate, what can happen when you don’t believe an instrument and it turns out that the instrument was correct. It has prompted me to write about an incident I experienced last year. Luckily in our case, we didn’t just ignore the suspect instrument. We took things to their safest state while we investigated further.
In a similar manner to the incident that is detailed in the beacon, ours occurred during abnormal operations. In our case we were starting the plant up to carry out the regen. We needed to partially fill a vessel with water that was getting pumped in from a tanker using a temporary pump and pipework.
The vessel had multiple instruments on the side of it including 3 local level gauges (LG1, LG2, & LG3), a low level switch (LSL) a high level switch (LSH) and low level switch (LSL) which announce alarms in the control system. Finally there is a level transmitter that is used by the control system to control the level in the tank during normal operations (LC). During the operation, the intent was to fill the vessel until the level transmitter read about 15%.
It takes quite a while to fill the vessel and there is always a bit of doubt whether you are still filling or if the flow has stopped. During the fill on the night in question, the low level switch cleared as expected and we continued to fill, waiting for the level to appear on the level transmitter. We were waiting quite a while, but this is normal and sometimes it does take a long time to get any level. Eventually the high level switch activated. Nobody believed it at the time however we did stop filling immediately. We went around the vessel and checked. All the local level gauges were reading no level. So we had the two switches saying water was in the tank. We believed the bottom one because it was lower than any other instrument and we expected to see water there. We generally believe the transmitter because it is the instrument used to control the level under normal circumstances and has a good track record. The local level gauges corroborated with the what the transmitter said, there was no level in the main section of the vessel. Finally the high level switch very rarely gets activated. Usually it is just when getting a routine test. As a result, there is little experience of seeing this switch work and when considering the evidence, it was clear that the high level switch had been activated spuriously.
So we called on the instrument technicians to have a look at the transmitter and work out why it had activated spuriously. Despite being confident that it was spurious, we did not continue to fill the vessel. Just as well that we didn’t.
The control room operator noticed that the level transmitter was sitting inactive. It had been turned off to do some minor maintenance work, and didn’t get reactivated1. As a result it continued to read bad while there was a liquid level over the range the instrument read. Once the point was made active again, the level showed top scale, over the range of the instrument.
The high level switch was correct. The level was genuinely high. All three level gauges were blocked up.
It turns out that four out of six instruments had failed. It was just as well that we took the safest course of action (don’t fill, don’t empty, just sit still) until we had checked everything.
As I have said before, control systems lie. Don’t believe everything the instruments are telling you, but don’t dismiss it too quickly either.
- The maintenance work was changing the range of the instrument to account for the change in density that you get from operating with water in the tank rather than oil that is normally there. The person who made that change, and left it inactive was myself. Not one of my finer regen moments. ↩