ARE YOU PERFORMING YOUR PSSR'S CORRECTLY?

A closure of an UV (shutoff valve) was provided in the design of the plant, to avoid two phase flow when carryover occurred.  During an abnormality, the shutdown system activated the UV but the UV did not close. The RCA revealed that the UV did not close due to excessive friction caused by the presence of construction debris left behind that prevented the UV from closing

Separate your shutdown system from control systems

Today,for cutting costs,many plant owners are trying to incorporate shutdown actions using the control system itself. For example, if there is a control valve that has to close when a predetermined shutdown point is reached, the instrument air to this valve is cut off using a solenoid valve and the valve is designed to "fail close".You must always keep your shutdown system independent from the control system. This is important from an emergency safe shutdown point of view. There is an interesting case study presented in this link where the shutdown system was designed to be independent from the control system. The control system valve did not close during an emergency due to a failure of a solenoid valve but the separate shutdown system acted safely.
I have investigated many process incidents where the shutdown system was connected through the control system and it failed to operate. In fact in one of the Ammonia plants in Europe a friend told me that every critical shutdown valve is provided with redundant solenoids for greater reliability. Have a relook at all your critical shutdown systems. Its better to be safe than to be sorry!

1 out of 1 or 2 out of 3?

Today many organisations are going in for two out of three redundant logic systems for trips. I often wonder how I managed to operate an ammonia plant 25 years ago, fitted with standalone pneumatic instrumentation and no DCS! With modern day electronics, isn't it expected that the reliability of an electronic transmitter will be better? Do not go in for two out of three transmitters just because it is more "reliable". Have you obtained data on mean time between failures of electronic transmitters? When you do a LOPA analysis do not go overboard. The risk criteria used in LOPA should reflect your organizations past incidents also. I have seen many overkills of LOPA studies done by consultants who just recommend two out of three systems at the drop of a hat!