A hydraulic air compressor (HAC) is an isothermal gas compressor that uses hydropower to compress air, originally developed by Charles Taylor in the 1890s to supply industry with compressed air. In the modern revival of this technology, the hydropower will be provided by pumps rather than natural sources. As such, energy efficiency is an important driver of component design;​ all of the hydropower is consumed either to overcome irreversibility or to compress air. The compressor relies on the increasing pressure of water flowing downward in a downcomer to compress air in the form of bubbles being dragged along with the flow. The air entrainment process at the top of the downcomer is facilitated by a mixing head. At the bottom of the downcomer, the bubbles are separated from the flow in a separator vessel. The objective of this thesis is to develop the design methodology for the air entrainment and air-water separation components on either end of the downcomer process.

Several mixing heads were tested on a small (4.5 m height) prototype HAC. The test without a mixing head successfully entrained air, confirming that air entrainment is a system effect. Two heads with dissimilar geometry were associated with the lowest irreversibility, leading to the conclusion that the best design at that scale is a mixing head incorporating some form of vortex breaker. Air entrainment is driven by a system energy balance and not exclusively by a local Venturi geometry.

The fraction of the air successfully captured in the plenum of the separator is called the separator effectiveness. Mechanistic models have been created to characterize both the irreversibility and separator effectiveness of two types of gravity separator (horizontal and vertical orientation) for the design of separators for future commercial-scale compressors. The separator effectiveness models require as input the flow field information from computational fluid dynamics analysis and the bubble size distribution at inlet. The bubble size distribution was measured on the small prototype and used to select a bubble size prediction model for testing on a much larger scale (29 m height) demonstrator HAC. The displacement model for horizontal separators matched the actual performance at the prototype scale well, particularly at high flow rate. The vertical velocity model produced a good match for the separator on the demonstrator HAC, but not for the same bubble size model identified on the small prototype.