This is a typcial relief scenario. You assume all valves fail in their normal fail safe position EXCEPT for one, and that is the one that will fail in the position such as to give you the greatest relieving rate.
Assume a few different issues. One is the instrument air failure per individual user. This should be most common.
Also assume a system wide loss of air. However in this case assume that it is dropping in pressure instead of a change from 90 psig to zero. To accommodate the latter you want some pressure transmitters on the shutdown system (yes triple transmitters would be my suggestion) that are inputs to the Safety Instrumented System. Provide a trip setting that is higher than the pressure that shutdown valve actuators may randomly begin closing. One objective would be an orderly plant shutdown rather than individual items going closed on their own. If you have spring opposed piston shutdown valves the actuators were selected based upon the valve torque, perhaps at a minimum pressure such as 80 psig and perhaps with another 25% safety factor. Different design criteria exist - along with a lack of a shutdown valve actuator design criteria.
You must assume you can lose air and power to all control and shutdown systems.
I've always designed for fail safe, whether open or closed. Then during the PHA that design is verified. I've worked with some pipeline companies that have fail to supply the customer not safe, I disagree.
