Unless you test on the actual system being protected, you will not be able to 100% recreate the relieving scenario - you would not want to create a fire case or a blocked outlet etc. etc., would you?
Even laboratory flow/coefficient testing for PRV certification (National Board) is using just 'simple' parameters for testing.
This is why the usual only option, is the workshop test bench test. Here you use air or nitrogen for PRV's on compressible fluid applications, and water as the test medium for liquid PRV use. For multi phase flow, take the manufacturers recommendation. Steam safety valves should be set on steam where possible. For workshop test bench testing you also use the CDTP - Cold Differential Test Pressure for the effects of static back pressure (non balanced PRV's) and operating temperature (manufacturers recommendation for testing at ambient conditions).
The above is generally in most, if not all recognised standards such as API-527.
Other testing methods such as in-situ testing, are done on line, but this verifies the set point but not the true relieving condition that may occur.
*** Per ISO-4126, the generic term
'Safety Valve' is used regardless of application or design ***
*** 'Pressure-relief Valve' is the equivalent ASME/API term ***
