Regenerative cooling of low NOx gas turbine combustors was investigated using impingement heat transfer with all the combustion air used for wall cooling prior to passing to the flame stabiliser. 10 rows of impingement holes were modelled. Three obstacles were compared with smooth wall impingement heat transfer. The CHT/CFD methodology used was that validated against experimental results in previous publications of the authors. The impingement heat transfer enhancement geometries investigated were circular pin-fins, dimples and zig-zag ribs, which were aligned transverse to the direction of the cross-flow on the impingement target surface. The obstacles were equally spaced on the centre-line between each row of impingement jets transverse to the cross-flow. One heat transfer enhancement obstacle was used per impingement jet air hole. The CFD calculations were carried out for an air mass flux G of 1.08, 1.48 and 1.94 kg/sm2bara, which are the high flow rates used for regenerative combustor wall cooling. Comparison of the current CFD predictions and previous CFD work, that have experimental data, were made for the flow pressure loss and the surface and locally X2 average HTC, h. It was concluded that none of the obstacles in the impingement gap a significant increase in the surface averaged heat transfer coefficient (HTC). The impact of the obstacles was to increase the flow maldistribution due to the increased pressure loss. This resulted is less heat transfer from the reduced air mass flow in the first 4 holes and increased heat transfer in the last 4 holes, relative to the smooth wall results. The main effect of the obstacles was to increase the heat transfer to the impingement jet surface. The dimpled surface was predicted to have a very poor performance, with significantly reduced impingement heat transfer. This was due to the impingement jets being deflected away from the target surface by the shape of the dimples and this reduced the surface heat transfer.