The "complete unit test" is an ambiguous statement and it is a pitty that nor the API617 and API 616 clarifies clearly what the SCOPE is of such testing. It gets even more ambigouous when you are dealing with variable speed drive motors (do you have an idea of how much auxilliaries you may end up with in this case, for example control systems).

If the end user / buyer does not clearly state how the test is to performed (no load, part load or full load and scope, and numbers of units), nothing objects the contractor from doing no load test and claim to be in line with standards as far as the requirement of "complete unit test" applies. You may ask yourself, why would you mobilize a complete string arrangement of job units/ auxilliaries, with heavy impact on logistics and costs, to do a no load testing, which in facts is a costly mechanical running test. In a lump sum environment, this issue can get very nasty.

A mechanical running test with complete unit elements is already big overhead. On top of this, going for full load has absolutely huge consequences. If you are talking of a gas turbine, a full load test you do it on site otherwise (unless this is relatively small turbine) few facilities around the world can accomodate a full load string test at the factory, but it would be at very high cost (2~3 millions USD is not uncommon).

Like most things, the answer is "it depends". You need to consider what your perceived risks are and how far you are prepared to go to mitigate them. Examples to consider are:

• individual tests you are undertaking on the compressors (performance test in addition to mechanical running test?)
• the complexity of the compressor arrangement (novel vs tried and tested),
• process duty (standard gas vs dense gas)
• end location (local vs remote, onshore vs offshore, 1st World country vs country with importation/customs difficulties)