WP1: Fixed Wing Aircraft Aerodynamics
Flight Mechanics Simulations using CFD (Badcock, Barakos; Industrial Partners: BAE Systems)
Recent effort has focused on predicting the aeroelastic response of complete aircraft [3]. A parallel effort has looked at the validation of the prediction of the vortical flow on highly swept wings [4]. Finally, the rolling response of a delta wing in the presence of vortex breakdown has been simulated, and shown to feature significant hysteresis. Attention is now turning to the exploitation of CFD to predict nonlinear dynamics of rigid and flexible aircraft.
Two projects will require access to supercomputing facilities. First, EPSRC grant EP/D504473/1 (April 2006-March 2009) aims to approach this task from two directions. A fast nonlinear frequency domain method will be developed to calculate dynamic derivatives for conventional flight mechanics aerodynamic models. In parallel the CFD solver will be coupled with a six degree of freedom solver (and active control) to allow the direct calculation of flight mechanic behaviour. The two approaches will be compared where the aerodynamics is likely to be beyond the derivative modeling. In the FP6 project SimSAC (June 2006- May 2009) the validation of CFD predictions of derivatives will be made against measurements for the DLR F-12 configuration (generic civil jet). In parallel, CFD generated data will be used to develop reduced aerodynamic models for conceptual design.
The effort to use CFD for flight mechanics applications is justified on two grounds. First, wind tunnel tests at large scale and transonic conditions are currently used to develop the aerodynamic models. This entails significant expense. Secondly, the derivative aerodynamic models are inadequate for certain flow regimes (involving moving shocks, separation and vortex breakdown), potentially leading to unexpected (uncommanded) responses in flight. Access to supercomputer resources through UKAAC-2 will allow assessment of turbulence treatments at extreme (high incidence) flow conditions (for example allowing a DES calculation as a reference for URANS predictions) and more detailed grid refinement studies to be made.
Development of Hybrid RANS-LES Capabilities for Simulating Aircraft Undercarriage Geometries (Leschziner; Industrial Partners: Airbus, ARA, DLR)
Airframe noise is a significant component of aircraft noise during the final approach phase of the landing. For large aircraft, in particular, landing gear noise is becoming the dominant source of airframe noise. In order to quantify the sources of noise generation, prior to a consideration of noise-abatement strategies, the unsteady flow around the landing gear needs to be resolved. This can only be done by use of LES. However, wall-bounded flows, in which the gross properties depend upon wall friction and the boundary layers, are very difficult to simulate, because the resolution requirements tend to rise with Re2 (as contrasted with Re0.4 away from wall). This has led to the development of a whole range of hybrid RANS-LES techniques. At Imperial College, one particular modelling strand pursued is a zonal scheme, in which the near-wall flow is modelled by parabolized URANs equations in the near-wall layer, coupled to the inner LES domain. In this project, Leschziner will extend the DLR code TAU (originally written for RANS computation) on the basis of these developments in the RANS-LES area. TAU is an unstructured-FV scheme for compressible flow, is applicable to LES and is parallelised. The code will then be applied to a model aircraft undercarriage geometry for which experiments are being undertaken. This code will be adopted (alongside ELSA) by Airbus as its principal work horse for CFD work.
The project is fully funded by Airbus and involves collaboration with ARA and DLR. The project started in April 2006 and will extend over at least 3 years, likely to extend to 4 years.
Aircraft Icing/De-icing (Thomson, Savill; Industrial Partners: Airbus
In-flight aircraft icing will be modelled by Thomson and Savill. This involves the coupling of steady RANS CFD with ice accretion computational models for an aircraft. The initial application will be to a civil airliner 3D wing in isolation; this will subsequently be extended in the context of the whole aircraft configuration (established pre-competitive European test cases to be provided by Airbus in CAD form for ICEM or alternative meshing). A challenging range of icing scenarios involving anti-icing, de-icing and (possibly) simple models of ice shedding will be considered. The combined CFD and icing codes will produce information which may eventually be used to provide input to aircraft stability and control assessments.
The CFD code (NEWT) will be loosely coupled to the icing code (ICECREMO II) in the following manner. An initial high definition CFD run will be performed to determine the the flow field and especially the boundary layer information required to initiate a subsequent first icing computation. The icing code has two stages: firstly a lagrangian droplet calculation is performed to calculate the impinging mass flux. This is then linked to a surface water (runback) and icing model which uses an Eulerian formulation of the extended lubrication equations. There is also integral coupling of heat transfer in the skin of the airframe. The surface mesh used for the CFD computation is also used for the conduction in the substrate. As ice is formed, an additional (typically) prismatic layer encompassing this is formed to handle heat transfer through the ice itself (also allowing a limited representation of re-melting processes). As the surface geometry changes significantly due to ice build up and the development of surface roughness, a remeshing step is then carried out before performing a new cycle of CFD and then icing computations to complete the next iteration. Typically ten of these iterations are required for complex configurations.
Industrial backing has been offered from Airbus UK with funding for an available EngDoc studentship to be confirmed. An additional self-funded PhD student is already in place at Cranfield who will perform local complementary CFD/icing computations for an F-18 configuration.
Adjoint-based aerodynamic optimization for novel aircraft configurations including flow control (Qin; Industrial Partners: Airbus)
Qin's group at Sheffield will carry out adjoint-based optimisation of novel aircraft designs including flow control devices (shock and separation control devices), flight dynamic and structural considerations. The adjoint based optimisation strategy allows for large number of design variables in the design and the multi-disciplinary consideration makes it possible to incorporate the planform parameters such as wing sweep, thickness, and chord length spanwise distribution in the design process in addition to the optimisation of the master section profiles. Optimisation carried out in UKAAC 1 involved over 650 design variables, one of the largest aerodynamic optimisation computations carried out so far. As under UKAAC-1, the Adj-MERLIN code will be used for the project. Tests have indicated a bronze rating (speedup of 74% from 128 processors to 256 processors) for optimisation runs with 2 million mesh points and 650 design variables. A key issue to be investigated is the interaction of flow control with configuration design. The inclusion of flow control devices can substantially simplify the configuration design and improve the vehicle's performance. Access to high performance computing facilities under UKAAC-2 will enable computations to be carried out that will enable further insight into this interaction. Projects at Sheffield funded by dstl, Airbus and RR in the related topics will provide the man-power for the specified research.
WP2: Helicopter Aerodynamics
In contrast to fixed-wing aircraft where CFD solutions for full configurations are already common in the literature, rotary-wing vehicles have so far received less attention and CFD computations of helicopter configurations are still in their infancy [5]. The main reason behind this delay is the difficulty in simulating the complex flow associated with rotorcraft, the complex geometric configuration, the coupling between aerodynamics and aeromechanics and the need for accurate turbulence modelling strategies. In addition to the above, the inherently unsteady flow field around the moving rotor blades and the fuselage requires a substantial amount of grid points and long computations to allow the CFD solutions to converge. The few examples of CFD computations for helicopter configurations available in the current literature involve several simplifications and approximations which make their use limited to specific purposes [6]. Regardless of these difficulties, the importance of helicopters and other rotary-wing vehicles is paramount. Search and rescue operations, ability to hover, independence of runways are just few of the advantages helicopters can offer.
At present, the UK rotorcraft industry is exploiting CFD (based on the combined work of Liverpool and Bristol Universities as members of the rotorcraft aeromechanics DARP [7,8]) for design calculations at design conditions only and for isolated rotor cases [9]. The work under UKAAC-2 will attempt to bring CFD computations to conditions close to the edges of the operational envelope of helicopters and, in addition, to provide solutions for full helicopter configurations. This is a timely proposal due to the availability to the participating universities of (1) a range of codes ready to undertake the proposed challenge and (2) unique sets of experimental data currently being generated from European Framework Projects. The utilisation of the available codes and the experimental data in a systematic way will allow for conclusions to be extracted both on the accuracy of the various CFD techniques and the turbulence modelling strategies for such complex flows. It is also fortunate that such developments are aligned with work undertaken by all partners either as part of national projects (DARP) or European projects (Helifuse and Goahead). There are three main datasets to be used for UKAAC-2 and the programme within this work package is organised according to these.
- ROBIN
ROBIN is an idealised fuselage with a simple rotor design and is suitable for assessment of methods and CFD techniques. It has so-far been used for validation methods and Bristol University is to base their work on this test case using the unsteady structured multi-block compressible CFD code, ROTORMBMGP. During UKAAC-1 this code was used to perform detailed grid dependence studies for rotors in hover and forward flight [10] and has received a gold star for parallel performance on the HPCx system. Liverpool University will employ the Helicopter-Multi-Block (HMB) solver which is the main rotor design CFD code at Westland Helicopters. This flow code was originally developed at the University of Glasgow and is currently being assessed for star status on HPCx. Several rotor studies have already been performed with HMB [5,9,11]. The proposed work of Liverpool and Bristol on the ROBIN test case will allow a comparison between Bristol’s multigrid solver and Liverpool’s fully implicit method. This is a demanding flow so even small benefits of one method over the other may lead to substantial savings in CPU time.
- HELIFUSE
The HELIFUSE experiment has already been exploited by Manchester (for isolated fuselage conditions) and it is now to be repeated using a Navier-Stokes solution for the fuselage combined with a blade-element method for the rotor. This hybrid approach should be sufficient to show the effect of a rotor on the helicopter fuselage loading. Liverpool has already asked for the release of the data and is to perform full N-S computations for this case.
The HELIFUSE case will subsequently be used to allow a comparison between blade-element method (Manchester) and full blade approximation (Liverpool). The conclusions from this work will guide the industry as to which method is best for the computation of fuselage characteristics within the shortest possible time. Exploitation of recent data obtained as a part of Fifth FRamework projects (like HELIFUSE) is a key issue, due to the accuracy of recent experiments and the superior techniques used.
- GOAHEAD
The experiments of the GOAHEAD project will be exploited by Liverpool and Cranfield in an attempt to determine the limitations in flow predictions due to turbulence modelling. Liverpool will compute this case with URANS and DES while a new LES model will be employed by Cranfield. The work of Liverpool and Cranfield will result in guidelines regarding the preferred turbulence modelling/simulation technique for full helicopter configurations and a detailed assessment of the benefits LES has to offer for helicopter flows. Any conclusion regarding turbulence modelling will be a major step towards the simulation of full helicopter configurations since recent research [12-13] reveals that hybrid LES/RANS, Detached Eddy Simulation and MILES approaches are effective and promising for high Reynolds number practical aerospace engineering problems. As LES approaches are computationally expensive, only a few key flight conditions will be calculated. Cranfield University, in close collaboration with DLR (Germany), is implementing high-order methods into the FLOWer code and carries out investigations of LES for helicopter configurations. The code uses MPI and has been ported to the Cambridge-Cranfield HPC (CCHPC) facility. The work is carried out in the framework of the EU project GOAHEAD. In the framework of the UKAA Consortium the code will be implemented on HECToR and will be used in large-scale high-order-based LES of helicopter configurations.
WP3: Propulsion Systems
Hall, Hills, Page, and Tucker will be using the Rolls-Royce code HYDRA. This incorporates the acoustic and aerodynamic modelling required and, under UKAAC-1, has obtained a gold star for parallel performance on HPCx. The current proposal will further the collaboration between Rolls-Royce plc and several universities that are applying the method to complex propulsion problems.
- Simulation of the flows within real multistage compressors
The level of efficiency of modern compressors means that traditional blade design methods, based on steady state CFD in an idealised annulus, are no longer able to yield the efficiency improvements needed to develop a competitive compressor. Future design improvements will come from understanding and control of (a) the parasitic effects of real geometry, such as cavities, gaps and leakages; and, (b) unsteady interactions between neighbouring components. In fact, there are many examples, in both industry and academia, where the poor performance of a rig is caused directly by larger than expected parasitic flows resulting from, for example, a poorly performing seal.
Hall, Hills, and Page will be involved in a collaborative project focusing on the compressor aerodynamics. Under the DTI’s Spring 2006 Technology Programme, Rolls-Royce and Airbus are leading partners in the CFMS (Centre for Fluid Mechanics Simulation) initiative being proposed under the Large Projects scheme. Under CFMS, a new 3.5 stage compressor rig will be built at the Whittle Laboratory, including engine representative real geometry features such as stator shrouds and variable guide vanes. Also under CFMS, CFD modelling of this rig and engine geometries will be carried out at Cambridge, Loughborough, and Surrey but will be limited to LES models of individual blade passages and RANS models of whole compressors. Access to supercomputing facilities under UKAAC-2 would enable an LES simulation of the whole compressor to be carried out. The aim would be to simulate the operating range of the compressor through a sequence of points on the operating map. Such a sequence of calculations would provide a unique opportunity to understand the aerodynamic mechanisms associated with the breakdown, or stall, of the flow in the compressor as it reaches its operability limit. LES simulations would remove the uncertainties associated with turbulence models and their ability to simulate strongly stalled flows.
Imregun will focus on the aero-elastic aspects of the compressor. Current 3D steady and unsteady flow simulations of core compressors are restricted to a single speed because of the difficulties of changing the vane scheduling during the run. Recent work at IC using AU3D has demonstrated the feasibility of changing the rotational speed during the run by changing the VIGV angles according to the defined speed schedule. The work will cover both the steady-state and unsteady flow simulations and the entire core compressor will be modelled in both cases. Some of the unsteady flow computations will be conducted using a single-passage multi-bladerow model, usually associated with steady-state flows, with “time-accurate” mixing planes at bladerow boundaries. Such an approach allows the study of compressor surge with a computationally-efficient model. More accurate unsteady flow simulations will be conducted using a hybrid model where bladerows will be represented as whole annulus where detail is needed, and as single passage wherever such a simplification is possible. Here the aim is to develop a methodology to check the aeroelastic behaviour of the final design by covering the compressor map in the same way as would be done in rig and engine tests. The calculations will provide a unique insight to core compressor aeroelasticity by revealing the exact mechanisms for several intractable phenomena such as transition from stall to surge, determination of secondary compressor characteristics, forced response under rotating stall and vane deflection during surge.
The manpower for this project will be funded directly by Rolls-Royce.
- Aeroacoustic simulation of an advanced open-rotor engine
Advanced open rotors have the potential to deliver significant fuel savings relative to turbofan engines and short-haul aircraft driven by open rotor propulsion systems have been identified as a key technology that is needed to reach future emissions targets for aviation. A major challenge to open rotor development is the noise emission and to become acceptable new low-noise solutions are required. One important contribution to both noise generation and noise attenuation is the interaction of the open rotor engine with the installation and with the local airframe surfaces. Noise can be generated by wakes shed from the pylon impinging on the engine rotors but engine noise can also be reduced through shielding by the airframe surfaces.
The only means to resolve all of the open rotor noise sources and their propagation is through a large non-linear unsteady simulation that includes the contra-rotating blade rows, the engine nacelle, the pylon and the nearby airframe surfaces. As well as being a high-resolution unsteady calculation, a high mesh density is required throughout the domain and an initial estimate of the total grid size is 100 million nodes. These simulations would vastly increase the limited understanding of the noise mechanisms in open rotor engines and the results will guide future open rotor engine and installation design towards lower noise.
The manpower for this project is being funded directly by Rolls-Royce.
- Nozzle geometry influence on flow and acoustics of jets
Jet noise is a major contributor to the overall aircraft noise emission, particularly at the take-off condition. Reduction in jet noise is important to allow the continued growth of air-travel without increasing the impact on communities around airports.
Tucker will concentrate on interaction effects between the jet and the pylon. There is significant recent emerging evidence that jets with co-flow (typical of modern bypass aero engines) exhibit metastability with respect to the inflow conditions and that a jet with a pylon (which more realistically represents the full engine geometry) can significantly reduce this. In an attempt to resolve these physical issues, numerical calculations using the modelling approaches described in [14] will be run for both isolated nozzle and nozzle-pylon configurations. These simulations will help understand in detail how a pylon influences the jet flow physics. Comparisons with experimental results in paper AIAA-2005-2845 will be made. Manpower for this project will be available though EPSRC grant GR/T06629/01 on ‘Aerodynamics & Aeroacoustics of Co-flowing Jets’. In addition, Tucker has a Royal Society industrial fellowship whereby 40% of his time is committed to Rolls-Royce projects. Manpower to assist with the complex meshing involved in jet-pylon interaction studies will be available through the recent Swansea EPSRC Platform Grant Award for which Tucker is a co-investigator.
Page will work on jet noise reduction via serrations and micro-jets. Introduction of a serrated trailing edge to a nozzle reduces noise, but also incurs a thrust loss. Alternatively microjets produce a similar physical effect by creating streamwise vortical structures, but only need to be switched on when needed. Large Eddy Simulation has had some success in predicting jet noise for simple single stream jets but applications to realistic coaxial jet flows with short cowl nozzles are rare. Serrations and microjets are computationally challenging due to the grid resolution requirements. Loughborough University is part of the EU Framework 6 programme Computation of Coaxial Jet Noise (CoJeN) which has carried out extensive flow and noise measurements on a realistic short cowl coaxial nozzle. A serrated nozzle was included in the test programme and Loughborough are planning to carry out preliminary LES on local facilities with up to 20 million nodes. However, it is recognized that to achieve reasonable fidelity of the flow around the serrations, a significantly finer mesh will be needed, typically up to 100 million nodes. This would only be viable on a large scale national facility. Microjets are a newer technology with much more limited experimental data for realistic geometries. It is intended to replace each serration with a microjet and initially carry out RANS predictions to determine an appropriate strength and direction of a microjet that would reproduce the same influence on the flow field as the serrations. LES would then be used to produce a high fidelity solution so as to determine the similarities and differences between the microjet and serrated nozzle jet flow. This could lead to improved understanding of this type of flow-field and how to exploit it for noise reduction.
WP4: Hypersonic Aerodynamics
Hypersonic aerodynamics is a new theme in UKAAC-2. Building UK capabilities in simulating hypersonic aerodynamic flows has been identified by private (MBDA) and public (Dstl) industries as critical to their international competitiveness, not only in developing new technologies but also in recruiting skilled and trained R&D staff. However, the focus in this proposal will be on modelling industrial flows with wider civil application, e.g. inlets to scramjets, re-entry of spaceplanes, rather than the more traditional military applications. A relatively small resource is requested for this theme and it is expected that enabling access to national supercomputing facilities will significantly increase the research in this area.
Current expertise at Strathclyde University is on next-generation hydrodynamic codes for rarefied flows, and this is being implemented within the OpenFOAM numerical flow solver. Reese will extend the capabilities of this aerodynamic model by incorporating the 5-species Park thermochemical model for hypersonic air flows, as well as introducing modern high-Knudsen-number hydrodynamic schemes that are appropriate for low-density or high-gradient gas flows. It is planned to port the OpenFOAM code to parallel implementation on HECToR. This will enable the investigation of a range of large-scale simulations of applications including hypersonic flow around a blunt body, base flow, and cavity flow. Strathclyde University currently host a PDRA under EPSRC grant GR/T05028/01 jointly-supported by Dstl. This project will end in November 2007 and a proposal is currently being written to support future work arising from this project.
Drikakis will develop a generalised equation-of-state (GEOS), large eddy simulation (LES) approach to hypersonic flows and couple the GEOS-LES code with the high-order hydrodynamic models being developed by Reese. The Fluid Mechanics and Computational Science (FMCS) group at Cranfield has developed an LES code that has been ported to the HPCx and Cambridge-Cranfield HPC (CCHPC) facilities. The GEOS-LES version of the code currently runs on the CCHPC. The code has been used for simulating compressible cavity flows and is currently used for a broad range of shock tube flows involving 3-D shock wave phenomena, flow transition and turbulent mixing as well as multi-component flows for different gases and gas interfaces. Under UKAAC-2, the code will run on HPCx and will be implemented on HECToR to perform large-scale simulations of hypersonic aerodynamics. Cranfield University will be hosting one funded PhD student in this area from October 2006. The GEOS-LES code development in this project is supported by various EPSRC and industrial projects.
Together, Drikakis, Reese and Emerson will develop a further EPSRC application on coupling of GEOS-LES with high-order extended hydrodynamic models for hypersonic multi-component flows.
|