A 10 row impingement heat transfer configuration with a single sided exit at the end of the impingement gap was modelled using conjugate heat transfer CFD. The predictions were compared with experimental results for an electrically heated, 6.35mm thick, metal wall of nimonic-75, which was impingement cooled. The geometry investigated was a square array of inline impingement 10 × 10 holes with X/D of 4.66 and Z/D of 3.06, where D = 3.27mm. The use of metal walls enabled the local surface averaged heat transfer coefficient h, to be estimated from an imbedded thermocouple that logged the rate of cooling when the heating was removed. Conjugate heat transfer analysis provided local h values, which were surface averaged for comparison with the measured h. The CFD results also provided velocity, turbulence and Nusselt number distributions on the target and impingement jet surfaces. The aerodynamics data enabled the pressure loss of the system to be predicted, which compared well with experimental measurements. The predicted surface distributions of Nusselt number were similar to the surface turbulence kinetic energy distributions, which demonstrated the importance of turbulence in convective heat transfer. Surface averaged heat transfer coefficients were predicted and are in good agreement with the measurements for five coolant mass flow rates. The predicted and measured results for surface averaged h were similar to measurements of other investigators for similar impingement geometries.

Impingement flat wall cooling, with 15.2 mm pitch square hole arrays, was investigated in the presence of an array of interrupted rib obstacles. These ribs took the form of rectangular pin-fins with a 50% blockage to the cross flow. One side exit of the air was used, and there was no initial cross flow. Three hole diameters were investigated, which allowed the impingement wall pressure loss to be varied at constant coolant mass flow rate. Combustor wall cooling was the main application of the work, where a low wall cooling pressure loss is required if the air is subsequently to be fed to a low NOx combustor. The results showed that the increase in surface average impingement heat transfer, relative to that for a smooth wall, was small and greatest for an X∕D of 3.06 at 15%. The main effect of the interrupted ribs was to change the influence of cross flow, which produced a deterioration in the heat transfer with distance compared to a smooth impingement wall. With the interrupted ribs the heat transfer increased with distance. If the heat transfer was compared at the trailing edge of the test section, where the upstream cross flow was at a maximum, then at high coolant flow rates the increase in heat transfer was 21%, 47%, and 25% for X∕D of 4.66, 3.06, and 1.86, respectively.

Conjugate heat transfer CFD studies were undertaken on the influence of the number of impingement holes/unit surface area or hole density n (m−2) for n from 1076 to 26910m−2 at a constant X/D of 4.7, with n varied by varying the hole diameter D from 1.31 to 6.54mm and pitch X varied from 6.1mm to 30.5mm. Square array impingement cooling geometries for the jet holes were used with a 152.4 × 152.4mm experimental wall area. The impingement gap had a single sided exit which generated a cross-flow in the gap. The number of impingement holes N in the cross-flow direction was 5, 10, 15 and 25. A coolant mass flux G of 1.93kg/sm2bar was investigated at a constant impingement gap Z of 10mm (Z/D 1.53–7.65 as n was varied). This high coolant mass flow simulated the coolant flow for regeneratively cooled combustors using all the combustor air flow to cool the combustor wall prior to entering the low NOx flame stabiliser. The predictions were compared with experimental results for the heat transfer coefficient h, that used the lumped capacitance method. The predictions of the surface averaged h and pressure loss ΔP/P were in good agreement with the measured results. The predictions showed that increasing the number of impingement jet holes resulted in lower h, due to the impact of cross-flow for large numbers of holes. At the other extreme, a very small number of holes were predicted to have high thermal gradients. The maximum heat transfer was found experimentally and computationally to be 4306 holes per m2 for an X/D of 4.7, with acceptable thermal gradients.

Impingement heat transfer with turbulence enhancing obstacles in the impingement gap were investigated and compared with the unobstructed flow situation. The objective was to significantly enhance flat plate impingement heat transfer for gas turbine combustor external wall cooling applications. Large flow blockages were used in the form of ribs and slotted ribs. The latter were found to have the best heat transfer enhancement of up to 23% at the trailing edge. The main effect of the blockage was to considerably reduce the influence of crossflow on the heat transfer by preventing the deflection of the impingement jets by the crossflow.

Impingement flat wall cooling, with 15.2 mm pitch square hole arrays, was investigated in the presence of an array of interrupted rib obstacles. These ribs took the form of rectangular pin-fins with a 50% blockage to the crossflow. One side exit of the air was used and there was no initial crossflow. Three hole diameters were investigated, which allowed the impingement wall pressure loss to be varied at constant coolant mass flow rate. Combustor wall cooling was the main application of the work, where a low wall cooling pressure loss is required if the air is subsequently to be fed to a low NOx combustor. The results showed that the increase in surface average impingement heat transfer, relative to that for a smooth wall, was small and greatest for an X/D of 3.06 at 15%. The main effect of the interrupted ribs was to change the influence of crossflow, which produced a deterioration in the heat transfer with distance compared with a smooth impingement wall. With the interrupted ribs the heat transfer increased with distance. If the heat transfer was compared at the trailing edge of the test section, where the upstream crossflow was at a maximum, then at high coolant flow rates the increase in heat transfer was 21%, 47% and 25% for X/D of 4.66, 3.06 and 1.86 respectively.