This paper presents an analysis of 2D PIV model test experiments performed on a series of forward-facing bent-duct type of oscillating water column (OWC) models with varied underwater geometry. Experiments of this complexity involving particle imaging velocimetry (PIV) to compare performance difference between differing geometry is the first of its type. The intent of the experiments was to investigate conversion losses and device performance by modifying the underwater geometry based on previous work by the authors. Four models were tested in total; the base model and three variations that had additional segments to afford different chamber length and lower/upper lip angles (10, 20 and 30 degrees). Dynamic similarity was assumed to be maintained between models by using a constant lower lip depth and a constant ratio of chamber length to projected underwater cross-sectional area. Performance comparisons were undertaken using phase-averaged wave probe, pressure transducer and PIV data. Additional qualitative analysis of velocity fields was performed using temporal averaging. In most test cases the new geometry with a 10 degree lip angle had marginally superior performance in terms of capture width and total power production. However, due to a number of issues, including the difference in damping factor between models; the difference in chamber length; and the void behind the upper lip on some models, it was not possible to definitively conclude that the 10 degree model was the best performer, but rather, the range of all new geometries presented have the potential for significant performance gains over the base geometry with additional modifications. Recommendations for future design of geometry are provided.