So I do know about sensors and control systems, from the supplier point of view. But maybe like many suppliers I’ve only been on site to troubleshoot a sensor that is reported to be giving incorrect data. So someone else made the decision to question the validity of the sensor outputs.
These days, the nearest I get to regular sensor monitoring is at home, typically with smoke alarms, a CO monitor, and a flammable gas alarm. Plus the plant manager, my wife, is always demanding an immediate solution to any alarm system going off, to continue production.
The problem
A flammable gas alarm was positioned above the gas hob: perhaps in retrospect not the best place, as the instructions said humidity and steam should be avoided. But very quickly we realised that the detector was not very tolerant of any wine added to dishes being cooked on the hob. Then, surprisingly, it alarmed whenever we had bread dough baking in the (electric) oven. Since the detector was said to be set to alarm at 25% of the LEL this was surprising.

The instruction came down – ‘Get me a switch in that alarm circuit, so I can switch the piercing noise of the alarm off!’ So, there was my solution, and a task, so that was done. I did not think it through any further.
Time passes
This system worked well for maybe 3 years. The alarm switch changed power from the alarm to a lamp over the worktop, so we knew to switch it back on after a problem event. But such sensors must have a life, and so when the alarm started going off when the kettle boiled and steam drifted up past the detector, I thought the unit was failing. There were then several late evening alarms, for no apparent reason, and we could not smell any gas (it is difficult to detect these days), nor see anything untoward. Like every engineer I guess, I felt the sensor, to find it very hot. This seemed to confirm the problem, that the sensor was failing, so take the thing out – ‘it was not that hot before!’ Something had changed.
Six months later, this Summer, we had a new gas meter – a Smart unit – installed on the domestic gas supply. Ultrasonic measurement of the flow, wifi connection to the indoor display, and mobile network reporting usage figures to the supplier. That would all be great, except the fitter refused to reconnect the gas to the house, because of a slight leak detected indoors. We had to call out a plumber to deal with our in-house problems. Good job it was Summer, as that took over a week.
Finding the gas leak
The leak was located as somewhere in the piping to the gas hob. The plumber tightened up the connections under the hob, and repeated the pressure loss checks. Still a slight leak, but within allowable tolerances. OK, so he checks once more, to be sure, and starts his paperwork. A last twist of the 90 degree bend directly on the hob (supplied by the Chinese supplier) produced an interesting result: the threaded part of this connector sheared off, almost in two half-round pieces. It looks like steel, but low grade steel, and showed a brittle type fracture all around the fitting.
Presumably the crack that had been there before, allowing a slow leak, had led to the fracture on tightening the connection. That was installed 10 years before, and no-one had done any checks of that or the system post installation. OK, I had never had the systems tested for gas leaks.
What had happened?
The conclusion at this point was that the slow leak presumably collected gas in the lower cupboards, and when this escaped it combined with the wine vapours to trip the alarm. Possibly the steam from the kettle just accelerated the rise of the gas past the detector. The detector was presumably a Pellistor, and got hot because it was burning the gas off. The dough rising in the oven? I don’t know much about bread and dough – but the leak was directly above the oven, so maybe the gas and air, warmed from the sides of the oven, helped the gas rise up past the detector. If that fitting had actually failed totally one night, there would have been a major blast, as I had removed the gas detector.
We now have installed a new detector, further from the cooking (3 metres). Plus the old one is re-installed, as a back-up unit: it is working OK still, next to the boiler. The bad news: the new alarm went off last weekend, when simmering a Paella laced with white wine….
Yesterday the plant manager produced a batch of dough and made bread. Both flammable gas alarms went off, first the unit 3 metres away then the old reserve unit, now even further away in the utility room, with the boiler.
Legislation
In any rented accommodation the landlord has to have a gas system safety check once a year. Because we own our own house there is no such requirement, and the boiler service man, who checks the gas boiler and heating system once a year, is not required to, and does not include, a system gas leak test in his inspection.
Product or system failure?
The gas hob was made by Proline, and installed around 8 years ago by a registered installer. It was a Chinese manufactured unit supplied by Comet as a low-cost own-brand hob to many retail outlets in the UK. The 90 degree bend that failed was supplied as a part of the hob, the gas inlet port. It is not steel, it could be an aluminium or zinc alloy. It appears the design was such that this port could be stressed during installation or tightening, as the bend itself would not rotate to suit the angle of the delivery pipework. It seems the break was on the hob side of the fitting. A combination of a poor quality fitting and a poor design.
The flammable gas alarm seems to work OK in detecting natural gas, but is even more sensitive to alcohol vapours, bread and dough making, and using any window cleaning spray that has any hydrocarbons in the fluid. So beware of using them in a brewery, distillery, bakery, bread shop, pub, restaurant and so on!
There was undoubtedly a small gas leak, around the hob, which has now stopped. Possibly this was from the 90 degree bend fitting, which then completely broke apart on tightening the joint. It remains possible that this failure was an accident waiting to happen.
The domestic plant manager is none too pleased at the moment. So do I leave the sensors installed, take both or one of them away, or fit switches to suppress the noise and turn off the alarm(s)?
Postscript
The supplier of the unit is surprised and upset. He considers these sensors do not give false alarms, when exposed to wine fumes from simmering a paella, or from baking bread in the oven. He has asked me to return the newest one to allow him to test it.
This has been done so we will see what results!
Filed under: Alarm Management, Application story, gas detection, Plant monitoring, Process Safety | Leave a comment »