Final Report Summary - FLIRET (Flight Reynolds number testing)
The 'Flight Reynolds number testing' (FLIRET) project's objective was to improve the accuracy of performance measurements at flight Reynolds number in cryogenic wind tunnels where the highest measuring accuracy is needed to predict the flight behaviour and performance of new aircraft.
Despite considerable progress in computational aerodynamics, wind tunnels are still the prime tool to measure and to predict aircraft performance for take-off and cruise, design and off-design conditions However, conventional wind tunnels face physical limits in matching Reynolds and Mach number ranges required to realistically simulate cruise conditions. A means to overcome this limit are cryogenic wind tunnels as for instance, the European transonic wind tunnel (ETW).
Another objective of FLIRET was to better integrated CFD simulation capabilities and wind tunnel testing. It was intended to clarify the advantages and the disadvantages of numerical and experimental work to take maximum benefit of synergy effects. The FLIRET approach offered a lot of opportunities for identifying weaknesses of each method and to combine their strengths.
The following work was done:
- Designing and manufacturing of several model mounting devices (stings)
- Applying and harmonising CFD and prediction tools including the numerical meshes
- Analysing the test results of each FLIRET work package
- Analysing the applied model quality, manufacturing and handling strategies
- Deriving recommendations for industrial testing in cryogenic tunnels
Based on numerical simulation, new and improved designs were used of straight stings, fin stings and twin stings. Ten test campaigns were performed in the ETW, the ETW pilot tunnel and the Aircraft research association (ARA) tunnel with four new sting configurations and two existing ones. One test with a new 2D-model and three half model tests were performed. One of them required three different peniches, which ended up in nearly three small, but separate measuring campaigns.
For most of the tested configurations improvements were found. It is estimated that FLIRET managed to raise testing accuracy in cryogenic tunnels, but in particular in the ETW by about 10% with reference to the state of the art at the start of the project. This was demonstrated by utilising FLIRET's new sting configurations. Unsurprisingly, the new stings have each to be used under specific model and wind tunnel conditions. A universal sting which allows excellent measurements under any condition is not feasible. For example, the minimum size straight sting provides reference data in a limited loads window and the blade sting guaranties very stable model behaviour in the wind tunnel.
A full design process including a detailed analysis of model/support interferences was performed with two sting designs being selected for detailed analysis of sting interferences. CFD results showed encouraging results showing our ability to reduce the interference between stings and models. This has shown the large benefit from designing a support to the model itself, instead of trying to adapt an existing support. The analysis has also shown the difficulty to define wind tunnel corrections since the efficiency of a support is strongly dependant on the way the correction is defined.
The aero-lines of the Straight and Fin Stings have been worked out leading to the minimum size straight sting and the optimised fin sting designs. Their performance has been assessed with Navier-Stokes codes and compared to the reference stings. This comparison is showing a clear reduction in the level of interference. The design of the blade sting was successful completed because it keeps free the wing of the model because the blade is attached to the model front fuselage. This was validated in wind tunnel test. The result is the new blade sting which is ready for industrial testing in cryogenic environment.
For improving buffet onset understanding and prediction the first step was to investigate the capability of numerical tools to predict buffet onset including parametric study of grid refinement, turbulence model effect and using different codes. These numerical means were to be used to determine wing deformation, Reynolds, Mach effect on buffet onset and separated flow characteristics. In the next step a specific high Reynolds wind tunnel test campaign for buffet onset investigations had to be prepared and conducted. After the test a deep analysis of the wind tunnel test results in term of physical understanding, parametric effects (Reynolds, Mach, wing deformation,) and steady and unsteady pressure characteristics was performed and different prediction methods of buffet onset with wind tunnel test measurements were compared.
For investigating the model vibration problems encountered in wind tunnels, but here particularly in the ETW the effectiveness of the advanced version of an vibration suppression system was to be investigated. During that test potential sources of vibration had to be identified. An aero elastic model of the model mounted on the ETW sting-balance including unsteady aero loads had to be built for simulating the dynamic response of the model in the test section. Finally the dynamic response of the mounting to the unsteady effort due to ETW turbulence and compare it to the model vibration measured during the test had to be quantified.
The parametric study shows that the start of non-linearity buffet onset is not so sensitive to turbulence model and mesh refinement in the range of mesh and turbulence model tested. The CFD analyses demonstrates the influence of the wing twist deformation that is of primary importance for a reliable buffet onset prediction
The wind tunnel test and the CFD analysis confirmed a significant Reynolds effect on buffet onset prediction especially in the range of 6 to 10 million where there is quite large transition effect. For the highest Reynolds number Re=32 Mio to Re=54Mio a slight effect on the CL of flow separation appearance is observed (DCL about 0.01- 0.02).
A wing twist effect (dynamic pressure effect) has been also identified. When dynamic pressure increases, model deformation increases as well and then the twist becomes nose down (lower local incidence) on the outer wing. Then, the CL buffet increases. The order of magnitude is around Delta-CL about 0.02 for a delta twist at the tip of the wing of around 0.6 degrees, quite representative of flight deformation. As a result, there is now a better understanding of the Reynolds effect on buffet onset characteristics at high speed.
The unsteady flow pattern in a cavity at the model / sting interface could be a source of the vibration. CFD investigations and wind tunnel test analysis have shown a very complex 3D/unsteady internal flow driven by rear end geometry. The complexity of the phenomenon and the current limited information of the unsteady flow characteristic don't allow concluding in the frame of FLIRET. Further investigations are required.
Another potential excitation of the model is the wind tunnels atmospheric turbulence. Two different modelling have been performed: the first based on the unsteady measurement on the wing and the second on the pressure fluctuations measured in the holes of the wind tunnel test section. Unfortunately the pre-test wasn't optimised for ETW vibrations. The modelling developed by
Airbus using the unsteady sensor on the wing delivered quite low excitation levels which can't explain the model vibration. A simulation using the pressure on the wall delivered relatively high level of vibration when using the aerolastic model. Therefore it is difficult to conclude at this stage and further investigations are required.
The knowledge about surface finish requirements for high Reynolds number testing gained led to a better understanding of the boundary layer characteristics being responsible for the measured lift behaviour. The results show that the present CFD methods are able to consider surface roughness and transition prediction at least on clean wings at maximum lift. The findings will contribute to allow cost savings on model manufacturing and to develop concepts for the prediction of maximum lift in free flight. The excellent data base achieved experimentally and numerically enables proposals and recommendations to derive principles for half model testing at flight Reynolds numbers at high lift.
The following conclusions can be drawn:
- No half model-peniche configuration for a high lift at high Reynolds number testing can match the free flight result because the major inboard peniche influences as flow displacement and aspect ratio effect which dominate the half model flow strongly interfering with the tunnel wall boundary layer.
- The experimental and numerical results show that the complexity and the non-linearity of the peniche model flow doesn't allow a correct wind tunnel correction as applied in today's wind tunnels
- An improvement of correction methods seems to be not feasible.
- The results of this task clearly demonstrate that only the use of validated CFD methods can open the possibility to compare the wind tunnel test with the free flight condition. The use of the 'Numerical wind tunnel' i.e. the numerical simulation of the complete wind tunnel flow and the concerning free flight model can be the way out.
- Improved correction rules for half model testing based on the evaluation of the numerical and experimental results are afflicted with problems and only possible by the use of the 'Numerical wind tunnel'.
Despite considerable progress in computational aerodynamics, wind tunnels are still the prime tool to measure and to predict aircraft performance for take-off and cruise, design and off-design conditions However, conventional wind tunnels face physical limits in matching Reynolds and Mach number ranges required to realistically simulate cruise conditions. A means to overcome this limit are cryogenic wind tunnels as for instance, the European transonic wind tunnel (ETW).
Another objective of FLIRET was to better integrated CFD simulation capabilities and wind tunnel testing. It was intended to clarify the advantages and the disadvantages of numerical and experimental work to take maximum benefit of synergy effects. The FLIRET approach offered a lot of opportunities for identifying weaknesses of each method and to combine their strengths.
The following work was done:
- Designing and manufacturing of several model mounting devices (stings)
- Applying and harmonising CFD and prediction tools including the numerical meshes
- Analysing the test results of each FLIRET work package
- Analysing the applied model quality, manufacturing and handling strategies
- Deriving recommendations for industrial testing in cryogenic tunnels
Based on numerical simulation, new and improved designs were used of straight stings, fin stings and twin stings. Ten test campaigns were performed in the ETW, the ETW pilot tunnel and the Aircraft research association (ARA) tunnel with four new sting configurations and two existing ones. One test with a new 2D-model and three half model tests were performed. One of them required three different peniches, which ended up in nearly three small, but separate measuring campaigns.
For most of the tested configurations improvements were found. It is estimated that FLIRET managed to raise testing accuracy in cryogenic tunnels, but in particular in the ETW by about 10% with reference to the state of the art at the start of the project. This was demonstrated by utilising FLIRET's new sting configurations. Unsurprisingly, the new stings have each to be used under specific model and wind tunnel conditions. A universal sting which allows excellent measurements under any condition is not feasible. For example, the minimum size straight sting provides reference data in a limited loads window and the blade sting guaranties very stable model behaviour in the wind tunnel.
A full design process including a detailed analysis of model/support interferences was performed with two sting designs being selected for detailed analysis of sting interferences. CFD results showed encouraging results showing our ability to reduce the interference between stings and models. This has shown the large benefit from designing a support to the model itself, instead of trying to adapt an existing support. The analysis has also shown the difficulty to define wind tunnel corrections since the efficiency of a support is strongly dependant on the way the correction is defined.
The aero-lines of the Straight and Fin Stings have been worked out leading to the minimum size straight sting and the optimised fin sting designs. Their performance has been assessed with Navier-Stokes codes and compared to the reference stings. This comparison is showing a clear reduction in the level of interference. The design of the blade sting was successful completed because it keeps free the wing of the model because the blade is attached to the model front fuselage. This was validated in wind tunnel test. The result is the new blade sting which is ready for industrial testing in cryogenic environment.
For improving buffet onset understanding and prediction the first step was to investigate the capability of numerical tools to predict buffet onset including parametric study of grid refinement, turbulence model effect and using different codes. These numerical means were to be used to determine wing deformation, Reynolds, Mach effect on buffet onset and separated flow characteristics. In the next step a specific high Reynolds wind tunnel test campaign for buffet onset investigations had to be prepared and conducted. After the test a deep analysis of the wind tunnel test results in term of physical understanding, parametric effects (Reynolds, Mach, wing deformation,) and steady and unsteady pressure characteristics was performed and different prediction methods of buffet onset with wind tunnel test measurements were compared.
For investigating the model vibration problems encountered in wind tunnels, but here particularly in the ETW the effectiveness of the advanced version of an vibration suppression system was to be investigated. During that test potential sources of vibration had to be identified. An aero elastic model of the model mounted on the ETW sting-balance including unsteady aero loads had to be built for simulating the dynamic response of the model in the test section. Finally the dynamic response of the mounting to the unsteady effort due to ETW turbulence and compare it to the model vibration measured during the test had to be quantified.
The parametric study shows that the start of non-linearity buffet onset is not so sensitive to turbulence model and mesh refinement in the range of mesh and turbulence model tested. The CFD analyses demonstrates the influence of the wing twist deformation that is of primary importance for a reliable buffet onset prediction
The wind tunnel test and the CFD analysis confirmed a significant Reynolds effect on buffet onset prediction especially in the range of 6 to 10 million where there is quite large transition effect. For the highest Reynolds number Re=32 Mio to Re=54Mio a slight effect on the CL of flow separation appearance is observed (DCL about 0.01- 0.02).
A wing twist effect (dynamic pressure effect) has been also identified. When dynamic pressure increases, model deformation increases as well and then the twist becomes nose down (lower local incidence) on the outer wing. Then, the CL buffet increases. The order of magnitude is around Delta-CL about 0.02 for a delta twist at the tip of the wing of around 0.6 degrees, quite representative of flight deformation. As a result, there is now a better understanding of the Reynolds effect on buffet onset characteristics at high speed.
The unsteady flow pattern in a cavity at the model / sting interface could be a source of the vibration. CFD investigations and wind tunnel test analysis have shown a very complex 3D/unsteady internal flow driven by rear end geometry. The complexity of the phenomenon and the current limited information of the unsteady flow characteristic don't allow concluding in the frame of FLIRET. Further investigations are required.
Another potential excitation of the model is the wind tunnels atmospheric turbulence. Two different modelling have been performed: the first based on the unsteady measurement on the wing and the second on the pressure fluctuations measured in the holes of the wind tunnel test section. Unfortunately the pre-test wasn't optimised for ETW vibrations. The modelling developed by
Airbus using the unsteady sensor on the wing delivered quite low excitation levels which can't explain the model vibration. A simulation using the pressure on the wall delivered relatively high level of vibration when using the aerolastic model. Therefore it is difficult to conclude at this stage and further investigations are required.
The knowledge about surface finish requirements for high Reynolds number testing gained led to a better understanding of the boundary layer characteristics being responsible for the measured lift behaviour. The results show that the present CFD methods are able to consider surface roughness and transition prediction at least on clean wings at maximum lift. The findings will contribute to allow cost savings on model manufacturing and to develop concepts for the prediction of maximum lift in free flight. The excellent data base achieved experimentally and numerically enables proposals and recommendations to derive principles for half model testing at flight Reynolds numbers at high lift.
The following conclusions can be drawn:
- No half model-peniche configuration for a high lift at high Reynolds number testing can match the free flight result because the major inboard peniche influences as flow displacement and aspect ratio effect which dominate the half model flow strongly interfering with the tunnel wall boundary layer.
- The experimental and numerical results show that the complexity and the non-linearity of the peniche model flow doesn't allow a correct wind tunnel correction as applied in today's wind tunnels
- An improvement of correction methods seems to be not feasible.
- The results of this task clearly demonstrate that only the use of validated CFD methods can open the possibility to compare the wind tunnel test with the free flight condition. The use of the 'Numerical wind tunnel' i.e. the numerical simulation of the complete wind tunnel flow and the concerning free flight model can be the way out.
- Improved correction rules for half model testing based on the evaluation of the numerical and experimental results are afflicted with problems and only possible by the use of the 'Numerical wind tunnel'.
