Absorb it. Return it. Prove it.

Two halves, opposite jobs. The buffer takes the energy out. It does so by forcing fluid through orifices whose effective area varies along the stroke — wide early to catch the mass gently, closing progressively to hold the deceleration, easing at the end so nothing arrives hard. That varying area is the design, and reading it back is what the rig exists for. The recuperator does the opposite: charged to a working pressure, it stores energy during the stroke and pushes the recoiling mass back on run-out. Both must be characterised, and so must their interaction — an assembly can absorb correctly and still return badly, and only a test that measures the whole cycle will show it.

The rig must survive what it measures — and, less obviously, must not lie while doing it. The energy arrives in a fraction of a second, so the bed, fixtures and reaction structure take everything the assembly was built to absorb; stiffness, restraint and the rig’s own fatigue life are first-order problems rather than detailing. But stiffness matters twice, because the rig’s compliance is a measurement error: if the reaction structure deflects under load, the stroke recorded is not the stroke the assembly saw, and the curve is quietly wrong in exactly the region where it matters most.

Temperature moves the curve. Fluid viscosity falls as it warms and the gas charge pressure rises, so the same assembly gives a different characteristic cold and hot. A rig that tests only at ambient has measured one point of an envelope. Conditioned testing across the service range is what makes the data mean something — and because endurance running heats the assembly itself, holding fluid temperature during a cycling run is part of the test rather than a convenience.