[American Institute of Aeronautics and Astronautics 29th AIAA Applied Aerodynamics Conference -...

________________________________________

*Research Professor, University of Alabama Birmingham, AIAA Associate Fellow. †Senior Principal Engineer, Dynetics, Inc., Arnold Engineering Development Center. ‡Senior Software Engineer, Aerospace Testing Alliance, Arnold Engineering Development Center. §Engineer/Scientist, Aerospace Testing Alliance, Arnold Engineering Development Center

1 American Institute of Aeronautics and Astronautics

Approved for public release; distribution is unlimited.

Firebolt v1.0 - Coupling of Transient and Steady Engine Performance Models with a High-Fidelity Navier-Stokes Code

R. H. Nichols*, A. G. Denny†, J. A. Calahan‡, S. A. Savelle§, and B. D. Heikkinen§ DoD HPCMP/CREATE Firebolt Team, Arnold AFB, TN 37389

Transient engine operation can have significant impact on the external aerodynamics of a flight vehicle. Engine throttle transients can also produce unsteady flow in the inlet system that may impact the performance and operability of the engine. Currently there is no capability short of flight testing for evaluating the coupled behavior of the engine, inlet, and aircraft. CREATE-AV is developing software products to allow virtual simulation of the coupled engine, inlet, and aircraft system at any point in the life cycle of a flight vehicle. Engine performance models have been successfully coupled with a high-fidelity Navier-Stokes CFD. Both transient and steady-state coupled solutions have been demonstrated. This new capability will serve to improve the fidelity of aircraft numerical simulations and can be used as a design tool or diagnostic technique to improve the quality and reduce the cost of future weapon systems. The new coupled engine model/CFD capability will be available with the release of Kestrel v3.0 scheduled for the spring of 2012.

Nomenclature

A = area a = speed of sound axw = aircraft acceleration along its flight path D = aircraft drag D2 = inlet distortion index FG = gross thrust Fnoz = nozzle exterior drag FR = engine ram drag Fspil = inlet spillage drag g = acceleration due to gravity M = Mach number �̇� = mass-flow p = static pressure Po = total pressure P0Avg = engine face average total pressure P0N = individual engine face probe total pressure PR = total pressure recovery R = gas constant To = total temperature V = velocity

29th AIAA Applied Aerodynamics Conference27 - 30 June 2011, Honolulu, Hawaii

AIAA 2011-3190

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.



W = aircraft gross weight α = aircraft angle-of-attack γ = ratio of specific heats γp = pitch angle ρ = density τ = thrust inncidence angle Subscripts 0 = inlet free-stream conditions 2 = engine face conditions 9 = nozzle exit conditions s = conditions ahead of a standing shock N = rake index x = upstream conditions ∞ = free-stream value

I. Introduction

ransient engine operation can have significant impact on the external aerodynamics of a flight vehicle. Engine throttle transients can also produce unsteady flow in the inlet system that may impact the performance and operability of the engine. Currently there is no capability short of flight testing for

evaluating the coupled behavior of the full-scale engine, inlet, and aircraft. Problems that are discovered during flight test can significantly impact the cost and schedule of a flight system since they occur late in the development process. A virtual capability to allow simulations of the coupled performance of the aircraft and the propulsion system at all stages of the design process is highly desirable. This paper outlines such a virtual system that allows low order steady-state or transient engine models to be coupled with high-fidelity Computational Fluid Dynamics (CFD) aircraft models to simulate the coupled transient and steady-state performance of the aircraft and propulsion system.

The Department of Defense High Performance Computing Modernization Program (DoD HPCMP) submitted a POM08 initiative to improve DoD acquisition program timeline, cost, and performance through the use of Computational Science and Engineering (CSE) tools for ships, aircraft, and antenna design and analysis. The resulting program is called the Computational Research and Engineering Acquisition Tools and Environments (CREATE) Program1 and is managed by Dr. Douglass Post of the DoD HPCMP with oversight by the Deputy Undersecretary of Defense, Science and Technology, Dr. Andre Von Tillborg. The CREATE Program is a $360M 12 year program executed by a tri-service team under the direction of Dr. Post. The air vehicle portion of CREATE is referred to as CREATE-AV and headed by Dr. Robert Meakin of the DoD HPCMP.

Although funding did not begin until October 2007, CREATE-AV planning began in earnest during the summer of 2007. One of the first elements of CREATE-AV to be put in place was a requirements gathering process that would result in software products with a realistic expectancy of impacting the aircraft acquisition process from conceptual design through sustainment. This process is conducted yearly by the CREATE-AV Planning Team with membership from each of the services and as many relevant organizations as possible.

Following the requirements gathering process, the CREATE-AV team determined there were four key software products needed by the acquisition engineering workforce that fit within the available budget and were that accomplishable in the CREATE Program timeline. The four software products are Helios, a virtual helicopter simulation tool; Kestrel, a virtual fixed-wing aircraft simulation tool; Firebolt, an airframe-propulsion integration simulation tool; and DaVinci, a conceptual design tool. The Helios and

T



Kestrel teams released initial versions of their software products early in CY10. The Firebolt tools will be developed as a module and will be accessible through Kestrel, Helios, and DaVinci.

The Kestrel software product is a modularized multidisciplinary fixed-wing virtual aircraft simulation tool incorporating aerodynamics, structural dynamics, kinematics, and kinetics. The first version of Kestrel2 (v1.0) was released in CY 2010 and was targeted at subsonic, transonic, and supersonic flight conditions. Kestrel v1.0 includes three major capabilities: a single static grid simulation, a single grid rigid body motion simulation, and a deforming single grid aeroelastic simulation. The second version of Kestrel3 (v2.0) was released in CY 2011 and added a 6 Degree-of-Freedom (6DOF) predictive capability for a single mesh and control surface motion.

Firebolt is being developed as a module to deliver key propulsion system analysis capabilities to the acquisition engineering workforces targeted by the CREATE-AV Project. Firebolt is not a stand-alone product. Firebolt capabilities are delivered to CREATE-AV product users through the Kestrel, Helios, and DaVinci products. At maturity, the Firebolt module will provide high-fidelity, full-physics engine simulation (inlet through exhaust and all physics in between). When coupled with the CREATE-AV products, the Firebolt module will provide the ability to include engine propulsion systems in the analysis of fixed and rotary wing aircraft designs early in the acquisition process. Firebolt v1.0 will provide part of this capability using lower fidelity engine performance models. Firebolt v2.0 and beyond will include high-fidelity simulation capabilities for engine propulsion system performance and operability evaluation and for engine/airframe integration. This paper addresses the initial Firebolt release (Firebolt v1.0) and its coupling with Kestrel. Firebolt v1.0 will be released with Kestrel v3.0 in CY 2012 and with Helios v4.0 in CY 2013.

The over-arching design philosophies being applied to CREATE software products are outlined in Ref. 1. The first is scalability on next generation machine architectures with linear scalability targets on the order of 104 processors/cores. Another CREATE design philosophy is a “legacy to native” software development approach with near term legacy software being rewritten or factored into native software over the life of the program. Another important CREATE software design philosophy is modularity. A common architecture in CREATE-AV is a Python based infrastructure and executive and either C or FORTRAN components. This allows a build-up approach to adding capability and multi-disciplinary physics. It also allows a factored approach to the software, aiding in code maintenance and supportability. One of the most important CREATE design philosophies is to follow a professional software development process incorporating configuration management, automatic unit, integration, and system testing, and user support. To have the desired impact on the DoD acquisition processes, the software has to be maintainable through the life of the program. All CREATE-AV products will have strict version control and configuration management through a Subversion (SVN) repository and continuous integration through nightly unit, integration, and system testing of current versions.

II. Kestrel Software Architecture

The Kestrel architecture is a blend of the CREATE design philosophies discussed above. It is a modular approach factoring traditional monolithic solvers into the Kestrel Infrastructure and Executive (KIE) piece, components to perform fluid dynamic, structures, kinematics and kinetics and other analysis, and the Kestrel User Interface (KUI). Figure 1 depicts a notional view of the Kestrel software architecture. The infrastructure and executive is an event-driven Python infrastructure that is component unaware. The components themselves can produce or respond to events and subscribe to or publish data. This allows the infrastructure and executive to be coded once and the eXtensible Markup Language (XML) input file to specify the use case and contributing components. The inputs to KIE are read in from an XML file generated by the KUI. Efficient data handling by KIE is accomplished by passing pointers to “heavy-weight” data or scalars. The resulting overhead was measured at less than 1% compared to a monolithic solver.

In Figure 1 there are two dashed boxes surrounding the components. The left-hand box denotes those components that are shared objects with the KIE and maintained by the Kestrel and Firebolt development



teams. The right-hand box represents executables from outside sources that will exchange data via an executable-to-executable communication path. This feature will be implemented in later versions of Kestrel and is intended to allow industry or commodity software to work with Kestrel without significant rewrites of their software. An example use of this feature is to allow a commodity CFD solver to be used with all of the other components in Kestrel. Another example use would be to incorporate a “blackbox” autopilot from another contractor into the simulation.

III. Firebolt v1.0 Module Overview

Firebolt v1.0 is a module incorporated within Kestrel as depicted in Figure 1. The objective of the

Firebolt v1.0 development effort is to provide means for both static and dynamic engine performance model simulation capability within high-fidelity Computationally Based Engineering (CBE) simulation software products Helios and Kestrel and access to the same by conceptual designers via the DaVinci product. The coupling of engine models and high-fidelity CFD codes will provide the DoD acquisition engineering workforce with the capability to:

a. Perform high-fidelity CFD simulations with the correct engine inlet mass-flow and exhaust

plume. b. Investigate throttle dependent effects on:

- Aerodynamics (inlet spillage and afterbody drag) - Inlet performance - External structural mechanics - Weapons separation - Weapons carriage loads

c. Investigate maneuvering aircraft effects on inlet performance with and without throttle transients. d. Provide a methodology for representing engine components with lower fidelity models. e. Investigate nozzle and afterbody structural and aero heating. f. Quantify wind tunnel support system design and interference correction. g. Provide a general method of incorporating unsteady flow boundary conditions into high-fidelity

CFD flow solvers. h. Provide engine information for design studies.

The DoD currently has a broad range of turbine engines in its inventory. Some engines are 20 years

old, while others are now just entering the inventory. A wide range of engine performance models have been developed to provide both static and transient performance information for each engine type. These models include Original Equipment Manufacturers (OEM) models, Advanced Turbine Engine Simulation Technique (ATEST)4 models, and Numerical Propulsion System Simulation (NPSS)5 based models. These older models are usually written in FORTRAN, while the NPSS models are developed using C++ libraries. Each of these models was designed to provide stand-alone engine simulation capability, and each has its own input requirements for controlling the simulation and methodology for outputting the results of the simulation. There are non-trivial issues with running multiple instances of the same engine model or a mix of engine models in the same monolithic code (such as common block clashes in FORTRAN).

Firebolt v1.0 defines a standard Application Programming Interface (API) for coupling engine models with a high-fidelity CFD solver. The Firebolt v1.0 library EM0D coordinates the exchange of information between the CFD code and the engine model at each CFD solver time step. EM0D also initializes the engine model and provides output from the engine model for analysis and post processing. EM0D runs each engine model as its own executable process, thus providing a safe interface between flow solvers and engine models. This allows multiple CFD solvers to couple with multiple engine models during a simulation. The time loop for the simulation is controlled by the Kestrel executive KIE.



The engine performance models must be modified to allow EM0D to control their operation. OEM and ATEST models are converted into libraries that allow their time loop to be controlled by EM0D. This requires some minor modifications to the main loop of the code. NPSS allows the engine performance model to be output as a library, so all that is required is the formulation of a simple master module to call the appropriate engine component modules during the simulation. All inputs and outputs for the engine performance models are passed through EM0D.

Many engine performance models are proprietary to the OEMs and access to source code is often difficult to obtain. Access to engine performance model executable is often easier to obtain. Firebolt v1.0 allows the output from an executable code to be tabulated and used as input for the coupled simulation. The tabulated engine input method is a one-way coupling since it does not allow inlet loss information to be passed from the CFD solver to the engine model.

EM0D works by performing model initialization then taking steps for each successive set of conditions. The library is initialized using information contained in the engine setup XML file along with an integer indicating the engine number. The initialization routine returns an engine model instance identifier. The instance identifier is passed to all other EM0D routines. Following initialization the engine model start parameters are set. At this point a handshake occurs with the model executable and the model echoes back the start parameters along with any startup results. Communication/synchronization is done using simple First In/First Out (FIFO) files. Following the startup the model enters the caller (CFD solver) time loop and advances in time. The coupling of the engine performance models and the CFD code is done by exchanging information at the start of each time step. The CFD code provides mass weighted averages of total pressure and temperature at the engine inlet to the engine model. The engine model runs one time step and provides a mass-flow at the engine face and mass-flow, total temperature, and total pressure at the entrance to the engine nozzle. The engine face mass-flow is used in an outflow boundary condition in the CFD code. The nozzle entrance information is used in an inflow boundary condition in the CFD code. EM0D is compatible with engine performance models with multiple inlet or exit streams such as high bypass turbofans. The relationship between the CFD solver, engine models, and EM0D is shown schematically in Figure 2.

Many engine models have a restrictive time step range, which can differ from the desired CFD time step. The EM0D library facilitates this by running an engine model at one time step while the CFD code is run at another. EM0D runs the engine model as many or few times as necessary then linearly interpolates the engine model to provide inputs for the CFD solver.

The output of some engine models can produce non-physical spikes during an engine transient simulation. These spikes can cause numerical instabilities in the CFD solver. EM0D includes the capability to clip and filter the parameters passed between the engine model and the CFD solver to allow the simulation to proceed through these non-physical spikes. This clipping and filtering may be applied to any of the shared variables between the CFD solver and the engine model. Clipping and filtering values can be specified by the user. Clipping bounds the maximum and minimum values for the new (updated) parameters shared between the engine model and the CFD code at each CFD code time. Filtering is done by maintaining a simple running average of the shared variables. The user can control the filter width by specifying the number of solver time steps for the running average. Experience has shown that the filter width should be 10 to 20 engine time steps.

Most engine models do not include a restart capability or a “save” state. These models run quickly, so the full simulation can be run or rerun to reach the desired time or engine state. CFD codes require significantly more computational resources, and always include a restart capability. EM0D includes a capability for restarting engine models. An engine restart file that includes the user specified inputs and the CFD generated inputs to the engine model is written at each engine time step. When a restart is required, the engine model is run using inputs from the restart file until the engine model reaches the time level of the CFD solver. Thus a save step for parameters within the engine model is not required.

The CFD flow solver chosen for integration into Kestrel was the Air Vehicles Unstructured Solver (AVUS), formally known as Cobalt606. AVUS is a cell-centered, finite volume CFD code developed and maintained at the Air Force Research Laboratory at Wright Patterson Air Force Base, OH. AVUS solves



the unsteady, three-dimensional, compressible Reynolds Averaged Navier-Stokes (RANS) equations on hybrid unstructured grids. Its foundation is based on Godunov’s first-order accurate, exact Riemann solver7. Second-order spatial accuracy is obtained through a least squares reconstruction. A Newton sub-iteration method is used in the solution of the system of equations to improve time accuracy of the point-implicit method. Strang et al6 validated the numerical method with a number of diverse applications. Tomaro et al8 converted the code from explicit to implicit, enabling CFL numbers as high as 106. Grismer et al9 parallelized the code, with a demonstrated linear speed-up on thousands of processors. The parallel METIS (PARMETIS) domain decomposition library of Karypis et al10 is also incorporated into AVUS. The main iteration loop of the original AVUS code was removed and segmented into a number of fundamental routine called at the Python level (e.g. spatial operator, temporal operator, applying boundary conditions, etc.) so that the flow solver could be coupled with other components at the subiteration level. The resulting code has been designated kAVUS.

The engine inlet boundary condition used in KAVUS is a straightforward extension of the existing sink mass-flow boundary condition type. For subsonic flow, the fundamental algorithm is based on the relationship between mass flux (mass-flow/area), total temperature, total pressure, and Mach number

𝑚𝐴̇ = �

𝛾𝑅 𝑇0

∙ 𝑃0 ∙𝑀

�1+(𝛾+1)2 𝑀2�

𝛾+12(𝛾+1)

(1)

Given the mass-flow per area, the total pressure and total temperature are extrapolated from upstream of the exit plane. The exit plane Mach number is then computed by iteratively solving Eq. (1). With Mach number and total conditions, the remaining thermodynamic variables can be computed along with a normal velocity from the mass flux. The new values are combined with the values from the previous subiteration via under-relaxation to provide the updated properties in the appropriate ghost cells.

If the flow is supersonic, a slightly different procedure is used. In this case flow through the sink face is related to the upstream flow via an isentropic path just up to the face, with a normal shock standing at the face to render the flow through the face subsonic. The mass flux at the upstream side of the shock can be written in terms of the upstream properties and the Mach number approaching the shock as

�̇�𝐴

= 𝜌𝑥𝑎𝑥𝑀𝑠 �1+(𝛾−1)

2 𝑀𝑥2

1+(𝛾−1)2 𝑀𝑠

2�

𝛾+12(𝛾−1)

(2)

The subscript ‘x’ denotes the upstream conditions, and the subscript “s” denotes the conditions ahead of the standing shock. By assuming that the Mach number Ms differs from the upstream Mach number by a small amount ∆M = Ms - Mx and applying the Binomial expansion, Eq. (2) becomes

�̇�𝐴≈ 𝜌𝑥𝑎𝑥𝑀𝑠 �1 −

(𝛾+1)2 𝑀𝑠∆𝑀

1+(𝛾+1)2 𝑀𝑠

2� (3)

This approach avoids the numerical issues that arise because the mass flux is a double-valued function of Mach number. The iterative solver will return values of Mach number that are supersonic. Solving this equation iteratively yields the Mach number ahead of the shock. The normal shock relations are then used to determine all of the state variables on the sink face. Once these are determined, under-relaxation is used along with the previous subiteration values to provide the updated boundary variables. It should be noted that apart from the under-relaxation, these are standard kAVUS algorithms for handling sink patches.

The exhaust boundary condition method seeks to set mass flux, total temperature, and total pressure at the source face. In common CFD practice this could be seen as over-specifying the boundary condition if



the boundary flow is subsonic (as it usually is). However, the 0D engine models are effectively uncoupled from the downstream flow. There is no real feedback path along which characteristic information can be fed into the engine model. In this approach, the engine is viewed as a black box that imposes certain boundary values at a point in the flow domain regardless of the local flow state. Hence, this proposed coupling methodology seeks to represent the input of energy, mass, and momentum into the domain. To do this, the mass flux equation

𝑚𝐴̇ = �

𝛾𝑅 𝑇0

∙ 𝑃0 ∙𝑀

�1+(𝛾+1)2 𝑀2�

𝛾+12(𝛾+1)

(4)

is solved iteratively for the Mach number using the values of T0, P0, and (𝑚

𝐴̇ ) which are output by the

engine model. Once the Mach number is known, the thermodynamic variables are computed from the isentropic Mach number relations, and the normal velocity at the source face is computed from the mass flux and density. Under limited testing, this condition appears to do a better job of matching the mass-flow and total conditions when the geometry of the engine nozzle grid is a good representation of the actual flight article. However, it should be emphasized that this boundary condition implicitly assumes that the CFD nozzle geometry matches the geometry of the actual engine.

All of the required input files for the coupled simulation can be constructed using KUI. KUI provides the following functionality:

- Job input definition - Boundary condition specification - CFD code input specification - Engine performance model setup - Job submission preferences - Input validation - Mesh manipulation - Post processing

The job input mode takes the user through general inputs to component specific inputs in an orderly, easy to understand manner with feedback on whether the inputs are complete for each component. This mode is typical in other CSE software packages but is especially critical for Kestrel due to its multi-disciplinary simulation capability. The boundary condition mode is also typical in other CSE packages and simply specifies boundary conditions for each surface and far-field patch in the mesh system. The job submission mode is not commonly offered in current CSE software and one will be useful for both novice and advanced users. This mode will allow a user to transfer files to remote machines, launch jobs, and track progress of jobs. The preferences mode allows users to customize the KUI for the user’s typical workflow. The mesh manipulation mode allows modifications to existing mesh files such as mirroring or adaptive mesh refinement. The post processing mode allows users to make two dimensional plots of output tracking data and engine performance model output, and build up output data by manipulating tracking files into new data.

IV. Thrust Force Accounting with Six-Degree-of-Freedom Models It would be desirable to use the output of the engine model and the CFD calculation to provide the forces to determine aircraft orientation using a six-degree-of-freedom model. The forces on an aircraft along its flight path (wind coordinate system, X-axis) are shown in Figure 3. Summing the forces and assuming constant mass, the following equation results11



(𝑊/𝑔) ∙ 𝑎𝑥𝑤 = 𝐹𝐺 ∙ cos(𝛼 + 𝜏) −𝐹𝑅 − 𝐹𝑠𝑝𝑖𝑙 − 𝐹𝑛𝑜𝑧 − 𝐷 −𝑊 ∙ sin�𝛾𝑝� (5) Net thrust (FN) is defined using the first two terms of Eq. 5

𝐹𝑁 = 𝐹𝐺 −𝐹𝑅 (6)

The gross thrust of a turbine engine is defined at the engine exit (Station 9 of Figure 4.) as

𝐹𝐺 = �̇�9𝑉9 +𝐴9(𝑝9 − 𝑝0) (7)

The ram drag (FR) is defined by the incoming flow momentum (Station 0 of Figure 4)

𝐹𝑅 = �̇�0𝑉0 (8)

Net thrust (FN) is defined as gross thrust (FG) minus ram drag (FR):

𝐹𝑁 = �̇�9𝑉9 − �̇�0𝑉0 + 𝐴9(𝑝9 − 𝑝0) (9)

The aircraft drag (D) can be determined by integrating the pressure and viscous forces on the aircraft wetted surfaces. This integration would include the drag of the inlet duct and would account for the inlet spillage drag (Fspil). The remaining terms in Eq. 5 are defined by the aircraft flight condition and orientation. In practice it is more convenient to define inlet conditions at the engine face (Station 2 of Figure 4) rather than Station 0. Thus the Eq. 5 can be written as

(𝑊/𝑔) ∙ 𝑎𝑥𝑤 = [𝐹𝐺 + �̇�9𝑉9 + 𝑝9𝐴9 − �̇�2𝑉2 − 𝑝2𝐴2] cos(𝛼 + 𝜏)−𝑊 ∙ sin�𝛾𝑝� (10)

Here FA is the total integrated pressure and viscous forces on the aircraft outer mold lines and the inlet. Assuming that the mass-flow at Station 0 equals the mass-flow at Station 2, Eq. 10 can be rewritten in terms of net thrust (FN) to yield

(𝑊/𝑔) ∙ 𝑎𝑥𝑤 = [𝐹𝐴+ 𝐹𝑁 + �̇�2(𝑉0 − 𝑉2) + 𝑝0𝐴9 − 𝑝2𝐴2] cos(𝛼+ 𝜏)−𝑊 ∙ sin�𝛾𝑝� (11)

Eq. 11 can be used to introduce the engine forces into a full aircraft CFD simulation utilizing a six-degree-of-freedom model to define the aircraft motion using parameters available from the engine model, free-stream conditions, and surface pressure and viscous force integrations.

V. Applications A. F-16/F110-100

The first example application is a throttle transient for an F-16 aircraft with a transient F110-100 low bypass ATEST engine performance model. This example demonstrates the capability to evaluate unsteady effects on the external aerodynamics of the aircraft and within the engine inlet duct during an engine throttle transient. The fully viscous aircraft grid includes 17 million mixed elements (tetrahedral, prisms, and pyramids). The aircraft inlet was gridded up to the engine face. The engine exhaust duct and nozzle was also gridded. Changes to the exhaust nozzle geometry during the engine transient were not simulated. Two flight conditions were simulated. The first flight condition was a free-stream Mach number of 0.6, angle-of-attack of 6 deg., and an altitude of 20,000 ft. The second flight condition was a free-stream Mach number of 1.4, angle-of-attack of 6 deg., and an altitude of 42,500 ft. The engine Power Lever Angle (PLA) was reduced from 85 deg. to 30 deg. then increased from 30 deg. to 85 deg.



during the 20 sec. simulation as shown in Figure 5. The throttle range for this simulation does not include afterburner operation. The unsteady CFD calculations were performed using kAVUS with the Spalart-Allmaras Detached Eddy Simulation (SA-DES) turbulence model.

The inlet mass-flow for the subsonic flight condition is shown in Figure 6 and the inlet total pressure ratio is shown in Figure 7. The mass-flow decreases and increases with the throttle movement as expected. The total pressure recovery at the engine face show oscillations due to planar waves as the duct responds to the change in mass-flow. These oscillations rapidly damp out and the engine and inlet duct reach steady-state in about five seconds for both the throttle decel and accel.

Two snapshots of the solution at 1 sec. and 5 sec. are shown in Figures 8 and 9. Both figures show isosurfaces of vorticity colored by static pressure. The inlet spillage is seen to stream down the side of the aircraft and avoids the ventral fins at the higher power setting in Figure 8. A separate corner vortex originates in the diverter region and fills the wing-fuselage juncture at the higher power setting. At the lower power setting the spillage envelops the lower fuselage and almost completely covers the ventral fins. Large unsteady flow structures are seen as the inlet spillage moves downstream. This unsteady flow region would also interact with pods located on the lower fuselage. The corner vortex that is present at the higher power setting is no longer present. The vortex has merged with the spillage flow. The engine power level is also seen to alter the size of the LEX vortex on the upper surface, slightly enlarging the vertical structure at the lower power setting. Snapshots of the aircraft centerline pressure distribution are shown at the same time steps in Figures 10 and 11. The change in pressure level within the inlet duct is readily apparent. Also the inlet flow rate is seen to alter the pressure distribution on both the upper and lower surfaces of the aircraft.

The inlet mass-flow for the supersonic flight condition is shown in Figure 12 and the inlet total pressure ratio is shown in Figure 13. The mass-flow decreases and increases with the throttle movement, but also shows a large unsteady component. The total pressure recovery at the engine face show large oscillations due to planar waves as the duct responds to the change in mass-flow. These oscillations are driven by the movement of the normal shock in front of the inlet, and the oscillations do not rapidly damp out in the 10 sec. between throttle movements for this simulations.

Two snapshots of the solution at 1 seconds and 5 seconds are shown in Figure 14 and 15. Both figures show isosurfaces of vorticity colored by static pressure. The inlet spillage is seen to stream down the side of the aircraft and avoids the ventral fins for both power settings. The upper surface of the aircraft shows no effect of the inlet spillage for this flight condition. Snapshots of the aircraft centerline pressure distribution are shown at the same time steps in Figures 16 and 17. The change in pressure level within the inlet duct is readily apparent. The normal shock in front of the inlet duct is seen to move downstream as the inlet mass-flow is reduced.

Comparisons of wind tunnel and CFD inlet performance and distortion are shown in Figures 18 and 19. The CFD results at the engine face were interpolated onto the locations of the 40 probe total pressure rakes used to acquire data in the wind tunnel test. The five probe rakes were equally spaced around the duct. The probes were positioned to produce an equal area weighting for each probe. The average total pressure at the engine face is given by

𝑃0𝐴𝑣𝑔 = 140∑ 𝑃0𝑁40𝑁=1 (12)

The inlet pressure recovery is given by

𝑃𝑅 = 𝑃0𝐴𝑣𝑔𝑃0∞

(13)

The compressor face simple distortion is given by



𝐷2 = 𝑃0𝑚𝑎𝑥−𝑃0𝑚𝑖𝑛

𝑃0𝐴𝑣𝑔 (14)

where P0max is the largest rake total pressure value and P0min is the least rake total pressure value. The CFD results are in good agreement with the wind tunnel results for the two cases simulated here.

B. C-17/F117-100

The second example case shows the capability to perform exhaust plume simulations using the new tools for a C-17 aircraft coupled with an OEM supplied steady-state F117-100 high bypass turbofan engine performance model. The calculations were performed to provide insight into the flow field in the vicinity of the side door for deployment of paratroopers. The fully viscous half aircraft grid includes 22 million mixed elements (tetrahedral, prisms, and pyramids). The aircraft grid included a deployed flap and an open side door with a deployed spoiler configured for paratrooper deployment. The cargo bay of the aircraft was also included in the grid. The aircraft inlets were gridded up to the engine face. The engine exhaust ducts and fan exhaust ducts were also gridded. The flight condition for the simulation was a free-stream Mach number of 0.23, angle-of-attack of 0 deg., and standard day altitude of 6,000 ft. The engine PLA was set to 85 deg. This is a representative condition for deploying paratroopers from the aircraft.

Unsteady CFD calculations were performed using kAVUS with the SA-DES turbulence model. Instantaneous isosurfaces of total temperature colored by pressure coefficient are shown in Figures. 20 and 21. The engine exhaust plume remains nearly cylindrical and well behaved until it interacts with the deployed flap. Once the flap is encountered the flow becomes extremely dynamic with multiple large scale flow structures. This flow outside of the door is seen to include the large flow structures. This new analysis capability could be used to exam combinations of aircraft flight condition, orientation, flap position, and engine throttle position for each engine to minimize the flow dynamics in the vicinity of the door.

VI. Conclusions

CREATE-AV is developing software products to allow virtual simulation of the coupled engine, inlet,

and aircraft system at any point in the life cycle of a flight vehicle. These new products will range from integrating engine models with CFD solvers to fully coupled high-fidelity CFD/CSM simulations of the aircraft and engine. These new capabilities will serve to improve the fidelity of aircraft numerical simulations and can be used as design tools or diagnostic techniques to improve the quality and reduce the cost of future aircraft weapon system developments. The new coupled engine model/CFD capability will be available with the release of Kestrel v3.0 scheduled for the spring of 2012.

Engine performance models have been successfully coupled with the kAVUS high-fidelity Navier-Stokes CFD solver using the Firebolt v1.0 EM0D library with the Kestrel executive KIE. The EM0D library was developed to simplify the coupling and provide the CFD solver with a single interface with one or more engine performance models. EM0D coordinates the exchange of information between the CFD code and the engine model at each CFD solver time step. EM0D also initializes the engine model and provides output from the engine model for analysis and post processing. The time loop for the simulation is controlled by the Kestrel executive KIE.

Both transient and steady-state engine performance model couplings were demonstrated. The subsonic F-16 /F110-100 case demonstrated a flight condition where the flow inside the engine inlet duct was well behaved, but the inlet spillage produced large unsteady flow structures that could adversely affect aircraft components on the underside of the fuselage. The supersonic F-16/F110-100 case demonstrated a flight condition where the external flow beneath the aircraft was well behaved, but the flow inside the engine inlet duct demonstrated large planer waves that could impact the performance and operability of the



engine. The C-17/F117-100 case demonstrated how this new capability can be used to investigate the interaction of the propulsion system exhaust streams with downstream aircraft components.

Acknowledgments

Material presented in this paper is a product of the CREATE-AV Element of the Computational Research and Engineering for Acquisition Tools and Environments (CREATE) Program sponsored by the U.S. Department of Defense HPC Modernization Program Office. Computational resources were also provided by HPCMP.

References

1. D. E. Post, S. Arevalo, C. Atwood, P. Bell, T. D. Blacker, S. Dey, D. Fisher, D. A. Fisher, P. Genalis, J. Gorski, A. Harris, K. Hill, M. Hurwitz, R. P. Kendall, R. L. Meakin, S. Morton, E. T. Moyer, R. Strawn, D. van Veldhuizen, L. G. Votta, S. Wynn, and G. Zelinski, Journal of Physics: Conference Series vol. 125 (2008) 012090.

2. Morton, S. A., McDaniel, D. R., Sears, D. R., Tillman, B., and Tuckey, T. R., “Kestrel – A fixed-wing Virtual Aircraft Product of the CREATE Program,” AIAA-2009-338.

3. Morton, S. A., McDaniel, D. R., Sears, D. R., Tillman, B., and Tuckey, T. R., “Kestrel v2.0 – 6DoD and Control Surface Additions to a CREATE Simulation Tool,” AIAA-2010-511, Jan. 2010.

4. Chappell, M. A. and Blevins, A. G., "Advanced Turbine Engine Simulation Technique Development and Applications to Testing,” AIAA-86-1731, Jun. 1986.

5. Evans, A. L., Follen, G., Naiman, C., and Lopez, I., “Numerical Propulsion System Simulation’s National Cycle Program,” AIAA-98-3113, 1998.

6. Strang, W. Z., Tomaro, R. F., Grismer, M. J., “The Defining Methods of Cobalt60: A Parallel, Implicit, Unstructured Euler/Navier Stokes Flow Solver,” 37th AIAA Aerospace Sciences Meeting and Exhibit, AIAA- 99-0786, Jan. 1999.

7. Gottlieb, J. J. and Groth, C. P. T., “Assessment of Riemann Solvers for Unsteady One-Dimensional Inviscid Flows of Perfect Gases.” Journal of Computational Physics, 32:101-136, 1988.

8. Tomaro, R. F., Strang, W. Z., and Sankar, L. N., “An implicit algorithm for solving time dependent flows on unstructured grids,” AIAA Paper 1997-0333, 1997.

9. Grismer, M. J., Strang, W. Z., Tomaro, R. F., and Witzemman, F. C., “Cobalt: A Parallel, Implicit, Unstructured Euler/Navier-Stokes Solver,” Adv. Eng. Software, Vol. 29, No. 3-6, pp. 365-373, 1998.

10. Karypis, G., Schloegel, K., and Kumar, V., “Parmetis: Parallel Graph Partitioning and Sparse Matrix Ordering Library, Version 3.1,” Technical Report, Dept. Computer Science, University of Minnesota, 2003.

11. Ray, Ronald J., “Evaluating the Dynamic Response of In-Flight Thrust Calculation During Throttle Transients,” NASA TM 4591, 1994.



Figure 1. Kestrel Architectural Design.

Figure 2. Schematic of coupling of CFD code and 0D Engine Model using EM0D.

CFD Code

Boundary Conditions

EM0D Library

Engine Model 1 Engine Model 2 Engine Model 3

FIFO Communication Synchronization



Figure 3. Aircraft force balance.

Figure 4. Turbojet engine station definitions.



Figure 5. PLA for the F-16/F110-100 throttle transient simulation.

Figure 6. Inlet mass-flow for the subsonic F-16/F110-100 throttle transient simulation.



Figure 7. Inlet total pressure ratio for the subsonic F-16/F110-100 throttle transient simulation.

Figure 8. Vorticity isosurface colored by static pressure with PLA=85 deg. for the subsonic F-16/F110-100 throttle transient simulation.

LEX Vortex

Inlet Spillage

Ventral

Corner Vortex



Figure 9. Vorticity isosurface colored by static pressure with PLA=30 deg. for the subsonic F-16/F110-100 throttle transient simulation.

Figure 10. Static pressure contours with PLA=85 deg. for the subsonic F-16/F110-100 throttle transient simulation.

LEX Vortex

Inlet Spillage Ventral



Figure 11. Static pressure contours with PLA=30 deg. for the subsonic F-16/F110-100 throttle transient simulation.

Figure 12. Inlet mass-flow for the supersonic F-16/F110-100 throttle transient simulation.



Figure 13. Inlet total pressure ratio for the supersonic F-16/F110-100 throttle transient simulation.

Figure 14. Vorticity isosurface colored by static pressure with PLA=85 deg. for the supersonic F-16/F110-100 throttle transient simulation.

LEX Vortex




Figure 15. Vorticity isosurface colored by static pressure with PLA=30 deg. for the supersonic F-16/F110-100 throttle transient simulation.

Figure 16. Static pressure contours with PLA=85 deg. for the supersonic F-16/F110-100 throttle transient simulation.

LEX Vortex


Normal Shock



Figure 17. Static pressure contours with PLA=30 deg. for the supersonic F-16/F110-100 throttle transient simulation.

Figure 18. Inlet total pressure recovery for the F-16/F110-100 throttle transient simulation.

Normal Shock



Figure 19. Inlet distortion index D2 for the F-16/F110-100 throttle transient simulation.

Figure 20. Total temperature isosurface colored by pressure coefficient for the C-17/F117-100 simulation.

Door and Spoiler



Figure 21. Total temperature isosurface colored by pressure coefficient for the C-17/F117-100 simulation.

Door and Spoiler

Date post:	12-Dec-2016
Category:	Documents
Upload:	bonnie
View:	218 times
Download:	0 times

[American Institute of Aeronautics and Astronautics 29th AIAA Applied Aerodynamics Conference -...

Documents