Ideally yes, it is required. However, normatively it is depending on the customer binding/required norms; terms and conditions; as well as valve design/function.
Some common standards used for testing are: API 598, API 6D and ISO 5208. Even in those (updated) standards, Closure/Seat testing for some Classes and sizes is optional.
Just try to think out-loud here in correspond with Seat A and B testing logic. This is my interpretation, and might be differ with others in some cases
Gate valve:
> Cavity relief (CR) --> with CR (one side only); without CR (both side)
> Unidirectional (one side only high pressure side); bidirectional (both side); integral double block and bleed (both side)
> Type of valve: Wedge (both side); double expanding (can be both, can be one side only); through conduit (can be both side); Slab (can be both)
> End user perspective: Is this for double barrier (both side tested); Pressure can be from both side (both side tested); one side is pressurized one side is atm/flare (can be one side); a lot number of valves (can be represented by some percentage), etc.
> type of Seat and or back seat ring: both seat floating or spring energized (both side tested); there is Graphite act as primary seal over the seat (optional for High pressure, might be required for low pressure)
> End user function: Isolation valve (required both side); critical for process (required both side)
>etc.
No, Globe valve by design is completely different with gate valve.
Please refer to snip below (courtesy) API 598. For bigger size and bigger pressure the table also similar.
Low and high pressure closure test meaning that it is tested from both side of the seat.
Some background or justification:
- even though it is supposedly uni-directional (flow can be under the plug or above the plug), sometimes end user still install it the other way around. Hence assurance that it is tight from both direction is a must
- it is torque seated and some design is pressure assisted. Meaning that the higher the torque (not to exceed yield point though) then the likelihood of zero leakage is better.
- Some ill-informed end user consider manual globe valve leakage rate is similar with control globe or choke valve. Typically when manual globe valve act as bypass control valve. This statement is not true. Control valve and manual valve leakage fall into different leakage rate standard. Usually control valve have more stringent leakage rate (Class IV, V or VI), whereas manual metal seated valves allows several bubbles per miute
- etc.
Other valve practitioner in this forum may provide better insight.