During a hydraulic transient event, the hydraulic-grade line (HGL), or head, at some locations may drop low enough to reach the pipe’s elevation, resulting in sub-atmospheric pressures or even full-vacuum pressures. Some of the water may flash from liquid to vapor while vacuum pressures persist, resulting in a temporary water-column separation. When system pressures increase again, the vapor condenses to liquid as the water columns accelerate toward each other (with nothing to slow them down unless air entered the system at a vacuum breaker valve) until they collapse the vapor pocket; this is the most violent and damaging water hammer phenomenon possible.

Bentley HAMMER makes a number of assumptions with respect to the formation of air or vapor pockets and the resulting water column separation:

• HAMMER models volumes as occupying the entire cross section of the pipe. This may not be realistic for small volumes, since they could overlie the liquid and not create column separation, as in the case of air bubbles, but this does not result in significant errors.
• HAMMER models air or vapor volumes as concentrated at specific points along a pipe. Volume at a node is the sum of the end points (a special case of a point) for all pipes connected to it. However, HAMMER can simulate an extended air volume if it enters the system at a local high point (via a combination air valve or CAV) and if it remains within the pipes connected to it.
• HAMMER ignores the reduction in pressure-wave speed that can result from the presence of finely dispersed air or vapor bubbles in the fluid. Air injection using diffusers or spargers can be difficult to achieve consistently in practice and the effect of air bubbles (at low pressures) on wave speed is still the subject of laboratory investigations.

In each case, the assumptions are made so that HAMMER’s results provide conservative predictions of extreme transient pressures.