The figure demonstrates how the low order thermal resistance model successfully captures the cooling behaviour of the four geometries, with the order of performance unchanged from that identified via the CFD and experiments. As was the case with the CFD, the greatest discrepancy was identified in geometries G3 and G4. This is perhaps unsurprising given the internal heat transfer coefficient used in the low order model were obtained from the CFD data. It is also worth noting that geometry G3, which demonstrates the greatest deviation, had the largest impingement jet Reynolds numbers – around twice those of the next highest geometry G4, and four and eight times greater than geoemtries G2 and G1 respectively (see [2]). It is therefore anticipated than the complex flow features associated with impingement cooling were less successfully captured by the steady-state RANS turbulence model used in the simulations and the errors introduced here were propogated through via the heat transfer coefficeints specified in the low order thermal resistance model. Additionally, as further elaborated upon in [2], the form of the experimental test piece and consequent CFD setup also resulted in a flange around the edge of the test piece introducing a small additional conductive pathway which is inherently disregarded in the thermal resistance model developed here. Whilst the effect of this, along with the minor effects on the internal flow are not expected to be great, they have likely contributed to the deviation between the low order model, and the CFD and experimental data shown for geometries G3 and G4. Indeed, in geometries G1 and G2 where these additional effects were observed to be reduced and the impingment jet Reynolds number lower, the performance of the thermal resistance model is observed to be impressive with only a small deviation observed at the highest coolant mass flow rates. It is noted that the overall effectiveness based on the impingement wall temperature was also assessed and compared to the experimental and CFD data similarly well.