//quoted: PSV is designed for protecting overpressure, of course if the PSV is failed, the system won't be protected......//
In this case off course the system very well protected. The set pressure is 3.5 barg only much lower than the system design pressure (20 barg). Hence, in the occurrence the psv failed at indicated set pressure the system very well safe. Then the system is still safe with or without psv.
//quoted: But that doesn't mean your PSV is not needed. In deed it's code mandated for pressure equipment per ASME VIII//
That's quite funny, so the design engineer have to install something because the local or international regulation and code said so without trying to figure it out why is it mandatory?
I believe before start to do advance things like to determine the govern scenario in a psv whatever it is, block outlet, control valve failure, check valve failure. Every design engineer must understand the fundamental function of psv.
I didn't say what you told before is wrong. Yes, if we have psv's set pressure higher than the MAWP of protected system then the psv useless and in the case describe by the OP, the psv useless also (on basis the function of psv).
I don't know exactly what it is wrong here but there are three possibilities. First the OP doesn't bring enough data to describe the system, the second very bad documentation, and the third the engineer who was designed the psv didn't know what he/she did.
And absolutely the problem is not about "too much allowable overpressure".