Final Report Summary - COMROTAG (Development and Testing of Computational Methods to Simulate Helicopter Rotors with Active Gurney Flap)
Executive Summary:
The project COMROTAG „Development and Testing of Computational Methods to Simulate Helicopter Rotors with Active Gurney Flap” was realised by Institute of Aviation, Warsaw, Poland, based on Clean Sky call for proposals no. JTI-CS-2013-01-GRC-01-014.
The duration of the project, including its extension by Amendment 2 was 38 months, budget was 320 000 EURO with Clean Sky JU financing 75% of the costs. The main requirements of the Call for Proposal were to create dedicated methodology to simulate AGF deployment on both 2D airfoils and 3D helicopter rotors. The developed computational tool had to be validated against a large suite of available test data (that were acquired during several 2D and 3D Wind Tunnel test entries during execution of the project) on standard rotor blade configurations to prove the level of correlation. The testing of the methododlogy included also ‘blind-cases‘ for a possible future flight tests campaign. During the execution of the project all planned goals of the project were achieved, including preparation of the computational tool, modelling of two- and three-dimensional flow test cases, verification of the computational results based on the available results of wind-tunnel tests and conducting simulation of “blind cases”. The results of the project were presented on three conferences, including European Rotorcraft Forum and Congress of the International Council of Aeronautical Sciences.
Project Context and Objectives:
The project COMROTAG addresses requirements of Call of Proposal SP1-JTI-CS-2013-01-GRC-01-014 for a computational method to simulate helicopter rotor equipped with Active Gurney Flap (AGF). Application of AGF on a rotor is meant to enhance lift force on reatreating blade and is one of Active Rotor Technologies investigated by Green Rotorcraft Consortium that should enable a helicopter to operate with reduced power consumption or reduced main rotor tip speed whilst preserving current flight performance capabilities, especially in terms of retreating blade stall. Lower power consumption should lead to reduced fuel usage and exhaust emissions, while reduced main rotor speed should significantly reduce rotor noise.
Active Gurney Flap is a small ‘wall’ perpendicular to the surface of the aerofoil and is located on the lower surface of the blade, in its trailing edge area. Its operation involves continuous change of its height, from zero to its maximum height which is usually in the range of 1-2% of blade chord. Its deflection on retreating blade is meant to increase lift of retreating blade which is otherwise limited by low relative velocity of retreating blade and blade stall.
Operation of AGF on a helicopter rotor blades is a challenge for computational methods of Computational Fluid Dynamics as it involves continuous deformation of blade shape as it undergoes complex rotational and oscillatory motion which is a result of control inputs and also of aerodynamic effects of AGF. For this reason simulation of operation of AGF on a rotor blade involves more modelling than “pure” CFD as it must be coupled with simulation of blade dynamics or otherwise obtain information about blade transient motion.
Requirements for the computational method, set in the CfP included, among others, its capability of conducting “blind” simulations of the effects of AGF. For this reason it was decided that the computational methods developed within the project COMROTAG must have the following capabilities:
- accounting for deformation of the blade as effect of operation of AGF which meant that computational mesh had to be modified in accord with deflection of AGF,
- accounting for blade rotational and oscillatory motion,
- accounting for rotor control.
In order to fulfil these requirements it was decided to begin work from modification of already being in development in the Instytut Lotnictwa simulation module “Virtual Rotor” which is a set of User Defined Functions working with commercial CFD software ANSYS FLUENT enabling modelling of blade dynamics and rotor control. These functions involve special, hierarchical structure of mesh blocks which have to be prepared by the user as well as provided by the solver capabilities of dealing with moving and deflecting meshes.
It was decided that another level of meh subdivision will be introduced to the “Virtual Rotor” code in order to model blade deformation as a result of deflection of AGF which involved designing of a special function transforming computational space in order to enable continuous and reversible deformation of mesh in a zone surrounding the AGF as it is activated. Other important challenges encountered and solved during execution of the project involved specific trimming procedures of blade motion and maintaining proper quality of mesh elements in the whole computational space which had moving boundaries as result of blade oscillations.
The computational methodology developed within project COMROTAG made it possible to fulfil all the requirements set for the numerical tool and to provide information about advantages and disadvantages of application of Active Gurney Flap in different flight conditions, including hover and forward flight with different rotor thrust.
Project Results:
Main Science/Technology results and foregrounds
The main scientific and technological results of the COMROTAG project consist in the development and testing of an original methodology of simulation of deployment of Active Gurney Flap (AGF) on
a helicopter blade, involving significant, reversible and continuous modification of computational mesh in the space surrounding the AGF and at the same time enabling simulation of blade and rotor dynamic motion. The methodology is described in detail in the report together with results of computational simulations used for its verification.
1. Research methodology
Modelling of Active Gurney Flap in the developed methodology requires deformation of computational mesh in a zone surrounding the AGF. In this approach, during gradual deflection of the Gurney flap the computational mesh is gradually stretched on the flap, maintaining the coherence and high quality of the mesh cells. During retraction of the Gurney flap similar deformations are performed in an inverse direction. The dynamic deformations of the computational mesh near the moving AGF are performed using the analytical function which locally deforms 2D or 3D space. Proper definition of the space deforming function is the essence of the developed methodology and it guarantees maintaining proper quality of mesh elements during the deformation, especially in the boundary layer. The proposed method prevents deformed cell from unfavourable degradations and ensures repeatability of cell shape in subsequent cycles of AGF deflection. Additional advantage of such approach is that there is no need to generate a special mesh around AGF, because it will be done automatically during the simulation. At every phase of AGF movement the structure of this automatically deformed mesh remains coherent. The computational method involves hierarchical structure of mesh zones used in discretization of the space surrounding helicopter rotor. Zones surrounding AGF are placed inside cylindrical zones of structural mesh surrounding rotor blades which are placed inside a larger cylindrical zone of unstructured mesh The structural-mesh zones surrounding rotor blades move with the blades as they rotate and oscillate in flapping, pitching and lead-lag motion. Motion of the blades is modelled by the integration of dynamic equations of blade motion. The full set of User-Defined Functions prepared for the computational simulation of rotor dynamics and aerodynamics is called hereafter the Virtual-Rotor-3D software.
Methodology of simulation of motion of rotor blades in balanced flight involves trimming of the rotor for given thrust. The full methodology may be explained in the following example involving “AGF off” and “AGF on” configurations:
1. Initially, the "AGF off" configuration is taken into consideration. For this configuration, few revolutions of rotor in forward flight are simulated, with blade pitch controls and flapping defined based on delivered by AW results of initial simulations by other method. The blade-lag angles are established arbitrary, the same for each blade.
2. The final flow state obtained in the step 1, are being used as initial flow state for next simulations, which are continued independently for two configurations: "AGF-off" and "AGF-on". In the former case, the fixed flapping-and-lead-lag motion of blades are replaced by switched on solution of flapping-and-lead-lag-motion equations implemented in Virtual-Rotor-3D module. In the case of "AGF-on" configuration, additionally the modelling of AGF motion is activated. Configured in this way simulations, for both configurations proceed until a satisfactory convergence, which is assessed based on convergence of global aerodynamic characteristics of the rotor and based on convergence of blade flapping-and-lead-lag motion. Usually, the final flow state obtained in the step 2, for both the "AGF-off" and "AGF-on" configuration, doesn’t meet the requirement that obtained rotor thrust was equal to the thrust defined for a given flight conditions. In such cases, the rotor-trimming procedure is conducted.
Rotor-trimming procedure in the presented approach consists in establishing the blade-pitch controls involving collective and cyclic pitch-control components so as to obtain required thrust and moments generated by the rotor. The applied rotor trimming procedure was based on classic approach. The new, corrected values of blade-pitch controls are evaluated according to the formula involving determination of gradient matrix whose elements are derivatives of thrust vector componets w.r.t. pitch controls. Properly evaluated the gradient matrix, to a large extent determines the success of the rotor-trimming process. In presented research, the matrix was evaluated using the Least Mean Square method, based on results of rotor-flight simulations conducted several times (at least four) for different values of blade-pitch-control vector. A variation of the trimming procedure was used for hover, where trimming involved also zero first harmonic of blade flapping with cyclicly deployed AGF. In this case, if we define the thrust vector T={C_T, B1s, B1c}, where: C_T – rotor thrust coefficient, B1s, B1c – first harmonics of blade flapping, then the trim conditions may be described by the following equation: T=Tr= T={C_Tr, 0, 0} where C_Tr is required value of thrust coefficient. In some cases, the trimming procedure had to be applied several times, iteratively.
In case of simulations of rotor in hover AGF deflected to constant height or retracted the mesh and methodology of simulations can be simplified. Applied simplifications, for a four-blade rotor, may be briefly described as follows:
• Only one quarter of flow field was analysed, utilising conditions of rotational periodicity.
• Computational mesh consisted of one fluid zone surrounding the single blade and one quarter-of-cylinder-shape fluid zone corresponding to far field of flow.
• Moving Reference Frame approach was used to model rotational motion of the rotor.
• In simulations, the cone angle ?0 and average lag angle ?0 were set based on results of wind tunnel tests.
The computational method of simulation of AGF was implemented also in two simplified versions: purely two-dimensional and two-and-a-half dimensional, for a constant-chord blade segment. The second version was used in the project in simulations of wind-tunnel investigations of blade segments in order to compare results of computations with experimental results. The methodology of
flow simulations for 2D and 2.5D versions of the method is as follows:
• pitch oscillations of rotor blade are simulated by periodic changes of angle of attack,
• blade flapping simulated by downward-upward oscillations of the airfoil (segment) using Moving Reference Frame method,
• lead-lag motion simulated by forward-backward oscilalions of airfoil (segment) using Moving Reference Frame method
• AGF motion is simulated by User-Defined Functions redistributing mesh point
As far as flow model is concerned, the CFD simulations have been conducted using the following flow model of ANSYS FLUENT solver:
• URANS
• Unsteady
• Compressible, air model: ideal gas
• Viscous, turbulent, model of turbulence: k-omega SST
The choice of turbulence model was made based on experience gained by the Project Team in other research projects, including Clean Sky projects STARLET and ESTERA, where the k-omega SST turbulence model produced results close to experiment at flow conditions involving high angles of attack and flow separation.
2. Results of simulations for constant-chord blade segments.
Simulations for constant-chord blade segments were conducted in Work Package Two. One Project Task task concerned simulations of several cases investigated in the University of Twente (UT) at low Mach numbers, approximately between 0.17 and 0.19 and the other Task concerned simulations at Mach number of 0.3-0.4 with simultaneously modelled oscillatory motion of the blade segment, and harmonically changing angle of attack. These cases were investigated experimentally in the CIRA icing wind tunnel.
2.1. Results of simulations of University of Twente cases
In investigations conducted in the University of Twente the AGF was implemented on NACA 0012 airfoil The dimensions of the wind-tunnel model, flow parameters, location of AGF, modes of its deflection and dimensions of computational space are shown in pdf-version of this report together with original designation by UT (Phase,Case). For each of the computational cases, the analysis of results of CFD simulation included:
• Time-varying lift coefficient CL compared with experimental results
• Time-varying pitching moment coefficient Cm compared with experimental results
• Time-varying drag coefficient CD
• Frequency domain analysis of time-varying pitching moment coefficient Cm compared with experimental results
• Chordwise distributions of pressure coefficient CP compared with experimental results for selected moments of time
• Velocity-Magnitude contours in proximity of airfoil trailing edge and the AGF compared with experimental results of PIV (if accessible) for selected moments of time
• Vorticity-Magnitude contours in proximity of airfoil trailing edge and the AGF compared with experimental results of PIV (if accessible) for selected moments of time
• Q-criterion contours in proximity of airfoil trailing edge and the AGF compared with experimental results of PIV (if accessible) for selected moments of time
• Sequences of Vorticity-Magnitude contours and Q-criterion contours presenting the vortex shedding in selected phases of the AGF deployment-retraction cycle
Representative results of flow simulations including time change of aerodynamic coefficients compared with experiment are shown for case designated as Phase3/Case 41 with ramp-type AGF deflection in the pdf-version of this report. As a summary of results of simulations for cases investigated in UT it can be stated that a good agreement between time-dependent aerodynamic coefficients and pressure distributions was observed. This was also true with respect to momentary contours of velocity magnitude, vorticity or Q-criterion, for which qualitative agreement between results of flow simulations and processed results of PIV experiment was observed. In case of frequency-domain analysis the experimental results for deflected AGF, compared with simulations, lacked some dominant frequencies associated with trailing-edge vortex shedding, althoug these were observable for clean airfoil.
2.2. Results of simulations of 2D CIRA test cases (“blind” tests).
The computational domain of the CIRA cases is shown in in pdf-version of this report. The AGF was located at distance of 5% of blade chord from the trailing edge. The simulations have been conducted for three static configurations (constant angle of attack):
"CLEAN" - clean blade segment,
"Thagf=1.0mm" – blade segment with Fixed Gurney Flap of 4.0 mm height and 1.0 mm thickness,
"Thagf=2.5mm" – blade segment with Fixed Gurney Flap of 4.0 mm height and 2.5 mm thickness
and for two oscillatory configurations:
"Dynamic Baseline 1" - clean blade segment, and
"Dynamic AGF 1" - blade segment with AGF Gurney Flap of 4.0 mm maximum height and 1.0 mm thickness.
Because of not-availability of experimental results at the time of completion of flow simulations only the numerical results are presented in the report.
Flow parameters were as follows:
Mach number: 0.3
Reynolds number: 2795814
Air density: 1.225 kg/m3
Air static pressure: 101325 Pa
Air temperature: 15 deg C
In static cases: angles of Attack: AoA = 0, 2, 4, 6, 8 , 10, 12, 14, 16, 18 deg,
In oscillatory cases of 3Hz frequency, “Dynamic AGF 1”:
mean AoA: 8 deg,
amplitude of AoA: 6 deg,
In oscillatory cases of 7Hz frequency, “Dynamic AGF 9”:
mean AoA: 14 deg,
amplitude of AoA: 6 deg,
For each configuration and for each considered angle of attack, the computational procedure has been conducted as follows:
1. Initially the stationary solution of RANS equations was found.
2. Next, starting from this solution, the unsteady (URANS) simulation has been conducted until the flow state was either quasi-static or oscillatory. For oscillatory cases the ramp-type deflection of AGF is shown in pdf-version of the report.
During the unsteady simulation of the flow, several global and local flow parameters were monitored, including:
• transient global coefficients: CL – lift coefficient, CD – drag coefficient and Cm pitching moment coefficient, obtained by integration of static pressure and wall-shear vector along the narrow, central strip of the blade segment,
• frequency-domain analysis of time-variable pitching moment coefficient (Cm),
• momentary distribution of pressure coefficient (CP) and skin-friction coefficient (CF) in the middle cross-section of the blade segment,
• Q-Criterion contours in the middle cross section of the wind-tunnel chamber.
Based on obtained results of static-case simulations, it may be concluded that:
• Aerodynamic characteristics of "Thagf=1.0mm" and "Thagf=2.5mm" configurations are very similar to each other,
• Configurations with Fixed Gurney Flap in comparison to the CLEAN configuration indicate for the same angle of attack higher values of lift, drag and negative pitching moment, which is
obviously due to aerodynamic properties of classic Gurney flap,
• CLEAN configuration in comparison to the Fixed-Gurney-Flap configurations indicate higher aerodynamic efficiency for lower values of lift coefficient: CL< 0.9
• CLEAN configuration in comparison to the Fixed-Gurney-Flap configurations indicate lower aerodynamic efficiency for higher values of lift coefficient: CL> 0.9
• Highly-unsteady vortex shedding (measured as high-frequency oscillations of pitching moment coefficient) is observed for the CLEAN configuration within the range of angles of attack: 0 deg <= AoA <= 6 deg. Frequencies of these oscillations are included within the range: 3112 Hz ÷ 3256 Hz.
• Highly-unsteady vortex shedding (measured as high-frequency oscillations of pitching moment coefficient) is observed for the Fixed-Gurney-Flap configurations within the range of angles of
attack: 0 deg <= AoA <= 14 deg. Frequencies of these oscillations are included within the range: 1056 Hz ÷ 1409 Hz. The higher the angle of attack, the lower frequencies of vortex shedding are observed.
The results of simulations of oscillatory cases at 3Hz may be concluded as follows:
• The flow simulations have been conducted for flow conditions (Mach number, range and frequency of oscillations of AoA) where strong dynamic stall rather does not occur (which is also confirmed by transient pressure-and-skin-friction-coefficient distributions),
• In the case of "Dynamic AGF 1" configuration, the deployment of Gurney flap at higher angles of attack, compared to the configuration "Dynamic Baseline 1", leads to:
o increase of maximum lift coefficient of about 0.35
o significant increase of drag coefficient,
o significant growth of negative pitching-moment coefficient, up to 500%
o decrease of aerodynamic efficiency (CL/CD) for the same value of lift coefficient (CL), except the highest values of CL, unreachable for the clean configuration,
• In the case of "Dynamic AGF 1" configuration, the strongest, high-frequency oscillations of pitching-moment coefficient occur in the AoA-drop phase, within the range of AoA: 12.5 deg ÷9.5 deg. These higher-amplitude oscillations are the result of the vortex shedding, which is the strongest in the mentioned range of dropping AoA. It is worth to mention, that the described phenomenon does not occur in the case of clean configuration "Dynamic Baseline 1".
• Results of frequency-domain analysis of pitching moment coefficient (Cm) made for the "Dynamic AGF 1" configuration indicated two dominant frequencies: 1118.9 Hz and 3016.7 Hz. The former frequency was close to vortex-shedding frequencies measured in the static-case simulations for the Gurney flap fully deployed, while the later frequency was close to vortex-shedding frequencies measured in the static-case simulations for the Gurney flap fully retracted.
Results of frequency-domain analysis of pitching moment coefficient (Cm) made for the "Dynamic Baseline 1" configuration indicated one dominant frequency: 3015.0 Hz. This frequency was close to vortex-shedding frequencies measured in the static-case simulations for the Gurney flap fully retracted.
The results of computational investigations of oscillatory cases at 7Hz oscillations may be concluded as follows:
• The flow simulations have been conducted for flow conditions (Mach number, range and frequency of oscillations of AoA) where strong dynamic stall occurs for both considered configurations,
• Both configurations reach maximum lift in dynamic-stall state of flow,
• In the case of "Dynamic AGF 9" configuration, the deployment of Gurney flap at higher angles of attack, compared to the configuration "Dynamic Baseline 8", leads to:
o increase of maximum lift coefficient of about 0.30
o moderate increase of drag coefficient,
o moderate growth of negative pitching-moment coefficient,
o in phase of growing AoA: just slightly lower aerodynamic efficiency (CL/CD) for the same value of lift coefficient (CL), except the highest values of CL, unreachable for the clean configuration,
o in phase of dropping AoA and for CL<1.4: considerably higher aerodynamic efficiency (CL/CD) for the same value of lift coefficient (CL),
• For both considered configurations, the extremely strong vortex shedding, similar to that observed in the case of "Dynamic AGF 1" did not occur. However weaker, high-frequency oscillations of pitching-moment coefficient were visible for both configurations. This confirms presence of weak vortex shedding in conducted simulations. Especially, it concerns the "Dynamic AGF 9" onfigurations.
3. Results of simulations for three-dimensional rotors equipped with AGF.
Simulations of three-dimensional rotors concerned the following configurations:
• Five-blade rotor with AGF located in a fragment of blade approximately located in the blade centre – “blind tests”,
• Four-blade model rotor investigated in wind tunnel of Politecnico di Milano for which comparison with experimental results was possible in several cases.
Geometry of the rotor blades of investigated rotors is shown in in pdf-version of this report.
3.1 Results of investigations for five-blade rotor
Computational investigations for five-blade rotor were conducted in two Project Tasks and it involved total of 10 flight conditions, defined as “flight priorities”. Each series of investigations involved trimming of rotor for defined thrust according to methodology described in Chapter 1, conducting of computations for transitional phase and monitoring of forces, moments and blade trajectory in converged oscillations. In all flight priorities sinusoidal AGF deflection was applied with maximum for azimuth of 270 deg and zero for azimuth 0 deg. Exemplary results of simulations for five-blade rotor are provided for “Flight Prority 1” of the first series of “blind tests” in the .pdf version of this report.
The results of simulation reveal that the AGF increases nose-down moment, increases flapping moment due to increased lift and increases lagging moment due to increased drag. As a measure of performance benefits of a rotor, resulting from application of AGF the Power Loading (ratio of thrust to power) was adopted. Simulations of flow for flight conditions requiring high thrust, especially when AGF was deployed in conditions of stalled flow on retreating blade, reveal positive increase of Power Loading, from 2% to 7.3%.
Effects of variations of computational time step: The rotor-forward-flight simulations required usually to conduct several dozens of rotor revolutions to complete the hover-wake development as well as to achieve satisfactory accuracy of the rotor trimming. During these simulations certain experiments concerning choice of time step have been conducted. The time step dt=1.4057?10-4 s applied for simulations of forward flight has been selected based on these experiments. Such time step seems to be sufficient to assess the performance characteristics of the rotor. On the other hand the selected time step was acceptable from point of view of finishing the task on time, taking into account accessible software/hardware resources. However, there is a risk that the selected time step is insufficient for modelling the unsteady-trailing-edge-vortex-shedding phenomenon, if it occurs. To check whether a reduction in the time step could result in the appearance of the flow of high-frequency oscillations caused by the vortex-shedding phenomenon series of numerical experiments of rotor-forward-flight simulation, with tenfold smaller time step (dt=1.4057?10-5 s., 14400 time steps per one revolution of the rotor, 0.025deg of blade azimuth per time step) , have been conducted. Due time-and-hardware/software-resource limitations, the simulations were performed only for about one-fifth of revolution of the rotor. Results of these experiments: the aerodynamic moments: M_theta , M_beta , M_ksi with respect to feathering axis, flapping and lead-lag hinges respectively, were analysed as functions of time (or blade azimuth PSI) and compared to the same functions obtained using nominal time step. Results of such time-step-sensitivity analysis indicate that in all analysed cases, the high-frequency oscillations of M_theta were not observed, neither for the nominal time step nor for tenfold smaller. Effect of tenfold reduction of time step on M_theta is shown in the pdf version of this report where it can be seen that this reduction of computational step did not produce any changes in M_theta.
Simulations of hover with Active Gurney Flap (5-blade rotor)
Simulations were conducted with two modes of AGF: 1/revolution and 2/revolution. The subject of investigations was assessment of the effects of AGF on Figure of Merit, which is defined, based on momentum theory as ratio of ideal power necessary for hover to the real power in hower. The reference results were computed for clean rotor (no motion of AGF). For configurations with AGF activated the trimming procedure described in Chapter 1 was applied in order to obtain zero first harmonic of blade flapping.
Simulations were conducted for the following hover conditions:
Altitude: 5486.467 m,
Density: 0.698 kg/m3,
Static Pressure: 50600.900 Pa,
Temperature: 252.488 K,
Rotor Angular Speed 31.040 rad/sec.
Ten Flight Priorities were defined for simulations, differing with thrust and type of AGF deflection(details in pdf-version of this report). Based on completed computational simulations it may be concluded, that:
• In all considered cases, the configurations with activated GF ("1/rev." and "2/rev.") indicate some growth of Figure of Merit, compared to the clean-blade configuration ("no motion").
• For rotor-thrust coefficient CTUS = 0.01010: compared to the clean-blade configuration ("no motion"), the configuration with operating AGF "2/rev." is characterised by relative growth of Figure of Merit, equal: Delta_FoM = 0.67% .
• For rotor-thrust coefficient CTUS = 0.01096: compared to the clean-blade configuration ("no motion") the configurations with operating AGF "1/rev." and "2/rev." are characterised by similar relative growth of Figure of Merit, equal: Delta_FoM = 0.71% and 0.72% respectively.
• For rotor-thrust coefficient CTUS = 0.01225: compared to the clean-blade configuration ("no motion") the configuration with operating AGF "2/rev." is characterised by relative growth of Figure of Merit, equal: Delta_FoM = 1.00% .
• For rotor-thrust coefficient CTUS = 0.01376: compared to the clean-blade configuration ("no motion") the configurations with operating AGF "1/rev." and "2/rev." are characterised by
similar relative growth of Figure of Merit, equal: Delta_FoM = 1.45% and 1.49% respectively.
• Relative growths of FoM for configurations with operating AGF "1/rev." and "2/rev." are similar to each other for the same value of rotor-thrust coefficient.
• In considered range of thrust coefficient: the higher a rotor-thrust coefficient CTUS, the higher relative growth of Figure of Merit is observed for configurations with activated GF in comparison to the clean-blade configuration.
3.2 Results of investigations for four-blade model rotor
Smulations were conducted for a model rotor of 2.2m-diameter in conditions of forward flight and hover. Simulations of hover were conducted only for clean blades and Passive Gurney Flap. For conditions of forward flight two modes of AGF deflection were modelled: sinusoidal and ramp-type. For forward-flight simulations the trimming procedure was used to obtain zero flapping conditions. The following flow parameters were set, similar to conditions of WTT:
Flow velocity: 48 m/s
Rotor rotational speed: 1600 rpm
Air density: 1.225 kg/m_3
Air static pressure: 101325 Pa
Air temperature: 15 deg C
Pitch angle of the rotor shaft: 0 deg
Bank angle of the rotor shaft: 0 deg
Commanded angles of collective pitch: Theta_0 = 4 deg, 6 deg, 8 deg, 10 deg, 12 deg.
Simulations of hover were conducted for fixed cone and lag angles, corresponding to wind-tunnel test data. The results of conducted computational simulations led to following conclusions:
• Results of CFD simulations and WTT indicate good qualitative convergence for both the Forward-Flight and Hover conditions. It especially concerns qualitative evaluation of potential benefits and limits of application of Passive Gurney Flap on rotor blades.
• Quantitative good convergence of CFD and WTT results has been found for the Clean-Blades configuration.
• For Passive-Gurney-Flap-2.5mm configuration the dependency Torque Coefficient (C_Q/sigma) vs. Thrust Coefficient (C_T/sigma), where sigma is solidity, has also indicated good quantitative convergence. However for this configuration, the dependences of Aerodynamic Characteristics on Commanded Collective Pitch (Theta_0) indicate specific, nearly constant offset of CFD results in respect to WTT results. This divergence between CFD and WTT results may be influenced by several factors, including:
o possible differences in computational and experimental mass model of rotor blades equipped with fixed Gurney flaps of height 2.5mm
o possible differences in computational and experimental geometry of model of rotor blades equipped with fixed Gurney flaps of height 2.5mm
o other incompatibilities between conditions of CFD simulations and WTT.
Results of CFD simulations of Active Gurney Flap of sinusoidal and ramp kinematics, conducted for conditions of Forward Flight at speed of 48m/s, led to conclude that:
• Effects of AGF of sinusoidal and ramp kinematics on Torque-versus-Thrust dependency are very similar to each other. It should be noted, however, that this convergence of results was observed for assumed, specific ramp kinematics of AGF. It may be expected, that optimised ramp kinematics of the AGF may have to give some benefits in relation to the sinusoidal kinematics. Undertaking more research on this topic seems to be very appropriate.
• For higher values of Thrust (or Collective Pitch) the AGF-2.5mm configuration, both the sinusoidal and ramp, gives similar benefits (in terms: Thrust versus Power/Torque) as PGF-2.5mm configuration.
• For lower values of Thrust (or Collective Pitch) the AGF-2.5mm configuration, both the sinusoidal and ramp, does not indicate the power penalty, observed for the PGF-2.5mm configuration.
In other words, in this area of Thrust, dependencies of torque on thrust for the Clean-Blades and AGF-2.5mm configurations are very similar to each other and they are more favourable than similar dependencies corresponding to PGF-2.5mm configuration.
• Configuration AGF-2.5mm(ramp) needs slightly lower amplitude of Commanded Cyclic Pitch than configuration AGF-2.5mm(sinusoidal) to trim the rotor to "zero-flapping" conditions.
The results for forward-flight simulations of 4-blade rotor may be summarised in graphs (in pdf version of this report) presenting dependence of power reduction on thrust coefficient. It confirms results obtained in simulations of five-blade rotor where increase of Power Loading was obtained especially at high thrust conditions. It presents also conditions when simple configuration, with Passive Gurney flap is beneficial in terms of reduction of required power.
To summarise results based on determined dependence of power reduction on thrust coefficient it may be stated that:
• For higher values of thrust (dynamic stall) the AGF-sin and AGF-ramp configurations indicate similar benefits in power reduction as (Passive Gurney Flap) PGF configuration.
• For lower values of thrust the AGF-sin and AGF-ramp configurations do not indicate significant power penalty, observed for the PGF configuration.
• Configuration AGF-ramp seems to be more favourable and needs slightly lower amplitude of Cyclic Pitch than configuration AGF-sin to trim the rotor to "zero-flapping" conditions.
4. Final conclusions
The computational method of simulation of Active Gurney Flap proposed by Institute of Aviation has proved its applicability for assessing effects of Active Gurney Flap on performance of helicopters. The method was implemented as User-Defined module in the commercial code ANSYS FLUENT in three version: two-dimensional, two-and-a-half dimensional and fully three dimensional. The two- and two-and-a-half dimensional versions of the method (VIRTUAL ROTOR 2D and VIRTUAL ROTOR 2.5D) provide possibility of modelling steady or oscillating flow conditions with oscillating free-stream velocity, angle of attack, and defined by user scheme of deflection of AGF. The fully three-dimensional version (VIRTUAL ROTOR 3D) apart from simulation of AGF kinematics resolves dynamic equations of motion of rotor blades and enables modelling of rotor control and trimming for required flight conditions.
The computational simulations of operation of AGF on rotors conducted in the COMROTAG project for geometric models defined by Topic Manager company and different modes of AGF deflection have revealed conditions when AGF brings performance benefits in terms of increased rotor thrust or power required for flight.
Potential Impact:
Description of project potential socio-economic impact including wider societal implications
Investigations conducted in the COMROTAG project concern simulation of operation of Active Gurney Flap (AGF) on helicopter blade and determination of transient aerodynamic loads as effects of operation of AGF. Application of AGF on a rotor is meant to enhance lift force on retreating blade and is one of Active Rotor Technologies investigated by Green Rotorcraft Consortium that should enable a helicopter to operate with reduced power consumption or reduced main rotor tip speed whilst preserving current flight performance capabilities, especially in terms of retreating blade stall. Lower power consumption should lead to reduced fuel usage and exhaust emissions, while reduced main rotor speed should significantly reduce rotor noise. Development of computational methods to simulate AGF is an important part of research in the field of Active Rotor Technologies and the importance of the completed investigations may be considered as their potential impact in the following fields:
- aeronautical research and technology development,
- economic aspects of the developed technology,
- societal aspects of the research and development of technology.
In the field of aeronautical research and technology development the results of the COMROTAG project have proven the viability of the assumed modelling approach, in particular of the combination of moving and deforming mesh approaches that make it possible to model in parallel the kinematics of AGF, deformation of blade contour and surrounding mesh zone and dynamics of the controlled rotor. These capabilities have been implemented as a module of User-Defined Functions called “Virtual Rotor” operating with the commercial CFD solver ANSYS FLUENT. Three versions of the software were prepared: fully three-dimensional version, two-and-a-half dimensional one and strictly two-dimensional one. In effect the modelling system created this way is capable of simulating flight of controlled rotor with blades equipped with AGF as well as of simulating of wind-tunnel experiments involving rotors and blade segments equipped with AGF. Capability of computational simulations of every step of experimental investigations should lead to determination of conditions when AGF brings benefits to rotor performance, to be verified by experimental investigations and to determination of conditions creating safety problems, e.g. through intensive vortex shedding requiring increased caution in experimental investigations. So, it is expected, that the computational simulation methods developed in the COMROTAG project will accomplish goals set in general for simulation methods: making the research phase shorter and safer and determining problems requiring increased attention during development of new technology.
The economic aspects of the developed modelling method are linked to its capability of determining flight conditions and modes of operation of AGF that should bring performance and economic benefit in operation of helicopters, such as decreased rotational speed or necessary power at the same thrust, or increased thrust at the same rotational speed or the same power required to produce thrust. Such conditions require subsequent experimental verification, but the capability of aiding directing of investigations towards determination of possible economic benefits of the application of AGF in helicopter rotors is an important feature of the method and it enhances the competitiveness of European industry and research sector. In particular the computational methods developed in the project support improving Technology Readiness Level of the investigated technology.
The societal aspects of the work conducted in the project COMROTAG may be considered in terms of the increased competitiveness of aeronautical research conducted in European companies and research facilities. Complementing of experimental investigations by advanced computational methods increases competences of research teams and shortening of time-to market for innovative solutions promotes job growth in companies exploiting results of research. Project COMROTAG contributes to these goals by providing modelling solutions complementing experimental research that can be exploited in the Topic Manager company as well as in Instytut Lotnictwa.
The societal aspects of the project COMROTAG project include also the potential effects of dissemination of the knowledge gained in the project. Dissemination of new knowledge achieved as an effect of realisation of a research project increases the public awareness of the state of the art of aeronautical research and increases confidence in air transport. Also increased is the awareness of the capabilities of European researchers, designers and manufacturers in the field of aeronautical research. The results obtained in the project have been already presented in three scientific conferences, with two of them very important for presenting results of aeronautical research to the scientific community: the Inernational Congress of Aeronautical Sciences congress in 2016 in Daejon, Korea, and the European Rotorcraft Forum in 2016 in Munich, Germany. In both cases the papers presented modelling capabilities of the computational methods developed in the project COMROTAG as well as results of investigations obtained in the project. Two other papers are planned to be prepared in the near future, after obtaining acceptance for results to be published from the Topic Manager company.
The computational methods developed within the project were also presented to a wider audience during two dissemination/communication events: Reporting Sessions of University of Warsaw Interdisciplinary Center of High Performance Computing in 2014 and 2015 in the form of presentation and a poster session. These presentation were addressed to specialists from different fields of computational simulations
In addition to this a web page dedicated to the COMROTAG: http://comrotag.ilot.edu.pl. This page will be maintained after completing the work in the project.
List of Websites:
http://comrotag.ilot.edu.pl
The project COMROTAG „Development and Testing of Computational Methods to Simulate Helicopter Rotors with Active Gurney Flap” was realised by Institute of Aviation, Warsaw, Poland, based on Clean Sky call for proposals no. JTI-CS-2013-01-GRC-01-014.
The duration of the project, including its extension by Amendment 2 was 38 months, budget was 320 000 EURO with Clean Sky JU financing 75% of the costs. The main requirements of the Call for Proposal were to create dedicated methodology to simulate AGF deployment on both 2D airfoils and 3D helicopter rotors. The developed computational tool had to be validated against a large suite of available test data (that were acquired during several 2D and 3D Wind Tunnel test entries during execution of the project) on standard rotor blade configurations to prove the level of correlation. The testing of the methododlogy included also ‘blind-cases‘ for a possible future flight tests campaign. During the execution of the project all planned goals of the project were achieved, including preparation of the computational tool, modelling of two- and three-dimensional flow test cases, verification of the computational results based on the available results of wind-tunnel tests and conducting simulation of “blind cases”. The results of the project were presented on three conferences, including European Rotorcraft Forum and Congress of the International Council of Aeronautical Sciences.
Project Context and Objectives:
The project COMROTAG addresses requirements of Call of Proposal SP1-JTI-CS-2013-01-GRC-01-014 for a computational method to simulate helicopter rotor equipped with Active Gurney Flap (AGF). Application of AGF on a rotor is meant to enhance lift force on reatreating blade and is one of Active Rotor Technologies investigated by Green Rotorcraft Consortium that should enable a helicopter to operate with reduced power consumption or reduced main rotor tip speed whilst preserving current flight performance capabilities, especially in terms of retreating blade stall. Lower power consumption should lead to reduced fuel usage and exhaust emissions, while reduced main rotor speed should significantly reduce rotor noise.
Active Gurney Flap is a small ‘wall’ perpendicular to the surface of the aerofoil and is located on the lower surface of the blade, in its trailing edge area. Its operation involves continuous change of its height, from zero to its maximum height which is usually in the range of 1-2% of blade chord. Its deflection on retreating blade is meant to increase lift of retreating blade which is otherwise limited by low relative velocity of retreating blade and blade stall.
Operation of AGF on a helicopter rotor blades is a challenge for computational methods of Computational Fluid Dynamics as it involves continuous deformation of blade shape as it undergoes complex rotational and oscillatory motion which is a result of control inputs and also of aerodynamic effects of AGF. For this reason simulation of operation of AGF on a rotor blade involves more modelling than “pure” CFD as it must be coupled with simulation of blade dynamics or otherwise obtain information about blade transient motion.
Requirements for the computational method, set in the CfP included, among others, its capability of conducting “blind” simulations of the effects of AGF. For this reason it was decided that the computational methods developed within the project COMROTAG must have the following capabilities:
- accounting for deformation of the blade as effect of operation of AGF which meant that computational mesh had to be modified in accord with deflection of AGF,
- accounting for blade rotational and oscillatory motion,
- accounting for rotor control.
In order to fulfil these requirements it was decided to begin work from modification of already being in development in the Instytut Lotnictwa simulation module “Virtual Rotor” which is a set of User Defined Functions working with commercial CFD software ANSYS FLUENT enabling modelling of blade dynamics and rotor control. These functions involve special, hierarchical structure of mesh blocks which have to be prepared by the user as well as provided by the solver capabilities of dealing with moving and deflecting meshes.
It was decided that another level of meh subdivision will be introduced to the “Virtual Rotor” code in order to model blade deformation as a result of deflection of AGF which involved designing of a special function transforming computational space in order to enable continuous and reversible deformation of mesh in a zone surrounding the AGF as it is activated. Other important challenges encountered and solved during execution of the project involved specific trimming procedures of blade motion and maintaining proper quality of mesh elements in the whole computational space which had moving boundaries as result of blade oscillations.
The computational methodology developed within project COMROTAG made it possible to fulfil all the requirements set for the numerical tool and to provide information about advantages and disadvantages of application of Active Gurney Flap in different flight conditions, including hover and forward flight with different rotor thrust.
Project Results:
Main Science/Technology results and foregrounds
The main scientific and technological results of the COMROTAG project consist in the development and testing of an original methodology of simulation of deployment of Active Gurney Flap (AGF) on
a helicopter blade, involving significant, reversible and continuous modification of computational mesh in the space surrounding the AGF and at the same time enabling simulation of blade and rotor dynamic motion. The methodology is described in detail in the report together with results of computational simulations used for its verification.
1. Research methodology
Modelling of Active Gurney Flap in the developed methodology requires deformation of computational mesh in a zone surrounding the AGF. In this approach, during gradual deflection of the Gurney flap the computational mesh is gradually stretched on the flap, maintaining the coherence and high quality of the mesh cells. During retraction of the Gurney flap similar deformations are performed in an inverse direction. The dynamic deformations of the computational mesh near the moving AGF are performed using the analytical function which locally deforms 2D or 3D space. Proper definition of the space deforming function is the essence of the developed methodology and it guarantees maintaining proper quality of mesh elements during the deformation, especially in the boundary layer. The proposed method prevents deformed cell from unfavourable degradations and ensures repeatability of cell shape in subsequent cycles of AGF deflection. Additional advantage of such approach is that there is no need to generate a special mesh around AGF, because it will be done automatically during the simulation. At every phase of AGF movement the structure of this automatically deformed mesh remains coherent. The computational method involves hierarchical structure of mesh zones used in discretization of the space surrounding helicopter rotor. Zones surrounding AGF are placed inside cylindrical zones of structural mesh surrounding rotor blades which are placed inside a larger cylindrical zone of unstructured mesh The structural-mesh zones surrounding rotor blades move with the blades as they rotate and oscillate in flapping, pitching and lead-lag motion. Motion of the blades is modelled by the integration of dynamic equations of blade motion. The full set of User-Defined Functions prepared for the computational simulation of rotor dynamics and aerodynamics is called hereafter the Virtual-Rotor-3D software.
Methodology of simulation of motion of rotor blades in balanced flight involves trimming of the rotor for given thrust. The full methodology may be explained in the following example involving “AGF off” and “AGF on” configurations:
1. Initially, the "AGF off" configuration is taken into consideration. For this configuration, few revolutions of rotor in forward flight are simulated, with blade pitch controls and flapping defined based on delivered by AW results of initial simulations by other method. The blade-lag angles are established arbitrary, the same for each blade.
2. The final flow state obtained in the step 1, are being used as initial flow state for next simulations, which are continued independently for two configurations: "AGF-off" and "AGF-on". In the former case, the fixed flapping-and-lead-lag motion of blades are replaced by switched on solution of flapping-and-lead-lag-motion equations implemented in Virtual-Rotor-3D module. In the case of "AGF-on" configuration, additionally the modelling of AGF motion is activated. Configured in this way simulations, for both configurations proceed until a satisfactory convergence, which is assessed based on convergence of global aerodynamic characteristics of the rotor and based on convergence of blade flapping-and-lead-lag motion. Usually, the final flow state obtained in the step 2, for both the "AGF-off" and "AGF-on" configuration, doesn’t meet the requirement that obtained rotor thrust was equal to the thrust defined for a given flight conditions. In such cases, the rotor-trimming procedure is conducted.
Rotor-trimming procedure in the presented approach consists in establishing the blade-pitch controls involving collective and cyclic pitch-control components so as to obtain required thrust and moments generated by the rotor. The applied rotor trimming procedure was based on classic approach. The new, corrected values of blade-pitch controls are evaluated according to the formula involving determination of gradient matrix whose elements are derivatives of thrust vector componets w.r.t. pitch controls. Properly evaluated the gradient matrix, to a large extent determines the success of the rotor-trimming process. In presented research, the matrix was evaluated using the Least Mean Square method, based on results of rotor-flight simulations conducted several times (at least four) for different values of blade-pitch-control vector. A variation of the trimming procedure was used for hover, where trimming involved also zero first harmonic of blade flapping with cyclicly deployed AGF. In this case, if we define the thrust vector T={C_T, B1s, B1c}, where: C_T – rotor thrust coefficient, B1s, B1c – first harmonics of blade flapping, then the trim conditions may be described by the following equation: T=Tr= T={C_Tr, 0, 0} where C_Tr is required value of thrust coefficient. In some cases, the trimming procedure had to be applied several times, iteratively.
In case of simulations of rotor in hover AGF deflected to constant height or retracted the mesh and methodology of simulations can be simplified. Applied simplifications, for a four-blade rotor, may be briefly described as follows:
• Only one quarter of flow field was analysed, utilising conditions of rotational periodicity.
• Computational mesh consisted of one fluid zone surrounding the single blade and one quarter-of-cylinder-shape fluid zone corresponding to far field of flow.
• Moving Reference Frame approach was used to model rotational motion of the rotor.
• In simulations, the cone angle ?0 and average lag angle ?0 were set based on results of wind tunnel tests.
The computational method of simulation of AGF was implemented also in two simplified versions: purely two-dimensional and two-and-a-half dimensional, for a constant-chord blade segment. The second version was used in the project in simulations of wind-tunnel investigations of blade segments in order to compare results of computations with experimental results. The methodology of
flow simulations for 2D and 2.5D versions of the method is as follows:
• pitch oscillations of rotor blade are simulated by periodic changes of angle of attack,
• blade flapping simulated by downward-upward oscillations of the airfoil (segment) using Moving Reference Frame method,
• lead-lag motion simulated by forward-backward oscilalions of airfoil (segment) using Moving Reference Frame method
• AGF motion is simulated by User-Defined Functions redistributing mesh point
As far as flow model is concerned, the CFD simulations have been conducted using the following flow model of ANSYS FLUENT solver:
• URANS
• Unsteady
• Compressible, air model: ideal gas
• Viscous, turbulent, model of turbulence: k-omega SST
The choice of turbulence model was made based on experience gained by the Project Team in other research projects, including Clean Sky projects STARLET and ESTERA, where the k-omega SST turbulence model produced results close to experiment at flow conditions involving high angles of attack and flow separation.
2. Results of simulations for constant-chord blade segments.
Simulations for constant-chord blade segments were conducted in Work Package Two. One Project Task task concerned simulations of several cases investigated in the University of Twente (UT) at low Mach numbers, approximately between 0.17 and 0.19 and the other Task concerned simulations at Mach number of 0.3-0.4 with simultaneously modelled oscillatory motion of the blade segment, and harmonically changing angle of attack. These cases were investigated experimentally in the CIRA icing wind tunnel.
2.1. Results of simulations of University of Twente cases
In investigations conducted in the University of Twente the AGF was implemented on NACA 0012 airfoil The dimensions of the wind-tunnel model, flow parameters, location of AGF, modes of its deflection and dimensions of computational space are shown in pdf-version of this report together with original designation by UT (Phase,Case). For each of the computational cases, the analysis of results of CFD simulation included:
• Time-varying lift coefficient CL compared with experimental results
• Time-varying pitching moment coefficient Cm compared with experimental results
• Time-varying drag coefficient CD
• Frequency domain analysis of time-varying pitching moment coefficient Cm compared with experimental results
• Chordwise distributions of pressure coefficient CP compared with experimental results for selected moments of time
• Velocity-Magnitude contours in proximity of airfoil trailing edge and the AGF compared with experimental results of PIV (if accessible) for selected moments of time
• Vorticity-Magnitude contours in proximity of airfoil trailing edge and the AGF compared with experimental results of PIV (if accessible) for selected moments of time
• Q-criterion contours in proximity of airfoil trailing edge and the AGF compared with experimental results of PIV (if accessible) for selected moments of time
• Sequences of Vorticity-Magnitude contours and Q-criterion contours presenting the vortex shedding in selected phases of the AGF deployment-retraction cycle
Representative results of flow simulations including time change of aerodynamic coefficients compared with experiment are shown for case designated as Phase3/Case 41 with ramp-type AGF deflection in the pdf-version of this report. As a summary of results of simulations for cases investigated in UT it can be stated that a good agreement between time-dependent aerodynamic coefficients and pressure distributions was observed. This was also true with respect to momentary contours of velocity magnitude, vorticity or Q-criterion, for which qualitative agreement between results of flow simulations and processed results of PIV experiment was observed. In case of frequency-domain analysis the experimental results for deflected AGF, compared with simulations, lacked some dominant frequencies associated with trailing-edge vortex shedding, althoug these were observable for clean airfoil.
2.2. Results of simulations of 2D CIRA test cases (“blind” tests).
The computational domain of the CIRA cases is shown in in pdf-version of this report. The AGF was located at distance of 5% of blade chord from the trailing edge. The simulations have been conducted for three static configurations (constant angle of attack):
"CLEAN" - clean blade segment,
"Thagf=1.0mm" – blade segment with Fixed Gurney Flap of 4.0 mm height and 1.0 mm thickness,
"Thagf=2.5mm" – blade segment with Fixed Gurney Flap of 4.0 mm height and 2.5 mm thickness
and for two oscillatory configurations:
"Dynamic Baseline 1" - clean blade segment, and
"Dynamic AGF 1" - blade segment with AGF Gurney Flap of 4.0 mm maximum height and 1.0 mm thickness.
Because of not-availability of experimental results at the time of completion of flow simulations only the numerical results are presented in the report.
Flow parameters were as follows:
Mach number: 0.3
Reynolds number: 2795814
Air density: 1.225 kg/m3
Air static pressure: 101325 Pa
Air temperature: 15 deg C
In static cases: angles of Attack: AoA = 0, 2, 4, 6, 8 , 10, 12, 14, 16, 18 deg,
In oscillatory cases of 3Hz frequency, “Dynamic AGF 1”:
mean AoA: 8 deg,
amplitude of AoA: 6 deg,
In oscillatory cases of 7Hz frequency, “Dynamic AGF 9”:
mean AoA: 14 deg,
amplitude of AoA: 6 deg,
For each configuration and for each considered angle of attack, the computational procedure has been conducted as follows:
1. Initially the stationary solution of RANS equations was found.
2. Next, starting from this solution, the unsteady (URANS) simulation has been conducted until the flow state was either quasi-static or oscillatory. For oscillatory cases the ramp-type deflection of AGF is shown in pdf-version of the report.
During the unsteady simulation of the flow, several global and local flow parameters were monitored, including:
• transient global coefficients: CL – lift coefficient, CD – drag coefficient and Cm pitching moment coefficient, obtained by integration of static pressure and wall-shear vector along the narrow, central strip of the blade segment,
• frequency-domain analysis of time-variable pitching moment coefficient (Cm),
• momentary distribution of pressure coefficient (CP) and skin-friction coefficient (CF) in the middle cross-section of the blade segment,
• Q-Criterion contours in the middle cross section of the wind-tunnel chamber.
Based on obtained results of static-case simulations, it may be concluded that:
• Aerodynamic characteristics of "Thagf=1.0mm" and "Thagf=2.5mm" configurations are very similar to each other,
• Configurations with Fixed Gurney Flap in comparison to the CLEAN configuration indicate for the same angle of attack higher values of lift, drag and negative pitching moment, which is
obviously due to aerodynamic properties of classic Gurney flap,
• CLEAN configuration in comparison to the Fixed-Gurney-Flap configurations indicate higher aerodynamic efficiency for lower values of lift coefficient: CL< 0.9
• CLEAN configuration in comparison to the Fixed-Gurney-Flap configurations indicate lower aerodynamic efficiency for higher values of lift coefficient: CL> 0.9
• Highly-unsteady vortex shedding (measured as high-frequency oscillations of pitching moment coefficient) is observed for the CLEAN configuration within the range of angles of attack: 0 deg <= AoA <= 6 deg. Frequencies of these oscillations are included within the range: 3112 Hz ÷ 3256 Hz.
• Highly-unsteady vortex shedding (measured as high-frequency oscillations of pitching moment coefficient) is observed for the Fixed-Gurney-Flap configurations within the range of angles of
attack: 0 deg <= AoA <= 14 deg. Frequencies of these oscillations are included within the range: 1056 Hz ÷ 1409 Hz. The higher the angle of attack, the lower frequencies of vortex shedding are observed.
The results of simulations of oscillatory cases at 3Hz may be concluded as follows:
• The flow simulations have been conducted for flow conditions (Mach number, range and frequency of oscillations of AoA) where strong dynamic stall rather does not occur (which is also confirmed by transient pressure-and-skin-friction-coefficient distributions),
• In the case of "Dynamic AGF 1" configuration, the deployment of Gurney flap at higher angles of attack, compared to the configuration "Dynamic Baseline 1", leads to:
o increase of maximum lift coefficient of about 0.35
o significant increase of drag coefficient,
o significant growth of negative pitching-moment coefficient, up to 500%
o decrease of aerodynamic efficiency (CL/CD) for the same value of lift coefficient (CL), except the highest values of CL, unreachable for the clean configuration,
• In the case of "Dynamic AGF 1" configuration, the strongest, high-frequency oscillations of pitching-moment coefficient occur in the AoA-drop phase, within the range of AoA: 12.5 deg ÷9.5 deg. These higher-amplitude oscillations are the result of the vortex shedding, which is the strongest in the mentioned range of dropping AoA. It is worth to mention, that the described phenomenon does not occur in the case of clean configuration "Dynamic Baseline 1".
• Results of frequency-domain analysis of pitching moment coefficient (Cm) made for the "Dynamic AGF 1" configuration indicated two dominant frequencies: 1118.9 Hz and 3016.7 Hz. The former frequency was close to vortex-shedding frequencies measured in the static-case simulations for the Gurney flap fully deployed, while the later frequency was close to vortex-shedding frequencies measured in the static-case simulations for the Gurney flap fully retracted.
Results of frequency-domain analysis of pitching moment coefficient (Cm) made for the "Dynamic Baseline 1" configuration indicated one dominant frequency: 3015.0 Hz. This frequency was close to vortex-shedding frequencies measured in the static-case simulations for the Gurney flap fully retracted.
The results of computational investigations of oscillatory cases at 7Hz oscillations may be concluded as follows:
• The flow simulations have been conducted for flow conditions (Mach number, range and frequency of oscillations of AoA) where strong dynamic stall occurs for both considered configurations,
• Both configurations reach maximum lift in dynamic-stall state of flow,
• In the case of "Dynamic AGF 9" configuration, the deployment of Gurney flap at higher angles of attack, compared to the configuration "Dynamic Baseline 8", leads to:
o increase of maximum lift coefficient of about 0.30
o moderate increase of drag coefficient,
o moderate growth of negative pitching-moment coefficient,
o in phase of growing AoA: just slightly lower aerodynamic efficiency (CL/CD) for the same value of lift coefficient (CL), except the highest values of CL, unreachable for the clean configuration,
o in phase of dropping AoA and for CL<1.4: considerably higher aerodynamic efficiency (CL/CD) for the same value of lift coefficient (CL),
• For both considered configurations, the extremely strong vortex shedding, similar to that observed in the case of "Dynamic AGF 1" did not occur. However weaker, high-frequency oscillations of pitching-moment coefficient were visible for both configurations. This confirms presence of weak vortex shedding in conducted simulations. Especially, it concerns the "Dynamic AGF 9" onfigurations.
3. Results of simulations for three-dimensional rotors equipped with AGF.
Simulations of three-dimensional rotors concerned the following configurations:
• Five-blade rotor with AGF located in a fragment of blade approximately located in the blade centre – “blind tests”,
• Four-blade model rotor investigated in wind tunnel of Politecnico di Milano for which comparison with experimental results was possible in several cases.
Geometry of the rotor blades of investigated rotors is shown in in pdf-version of this report.
3.1 Results of investigations for five-blade rotor
Computational investigations for five-blade rotor were conducted in two Project Tasks and it involved total of 10 flight conditions, defined as “flight priorities”. Each series of investigations involved trimming of rotor for defined thrust according to methodology described in Chapter 1, conducting of computations for transitional phase and monitoring of forces, moments and blade trajectory in converged oscillations. In all flight priorities sinusoidal AGF deflection was applied with maximum for azimuth of 270 deg and zero for azimuth 0 deg. Exemplary results of simulations for five-blade rotor are provided for “Flight Prority 1” of the first series of “blind tests” in the .pdf version of this report.
The results of simulation reveal that the AGF increases nose-down moment, increases flapping moment due to increased lift and increases lagging moment due to increased drag. As a measure of performance benefits of a rotor, resulting from application of AGF the Power Loading (ratio of thrust to power) was adopted. Simulations of flow for flight conditions requiring high thrust, especially when AGF was deployed in conditions of stalled flow on retreating blade, reveal positive increase of Power Loading, from 2% to 7.3%.
Effects of variations of computational time step: The rotor-forward-flight simulations required usually to conduct several dozens of rotor revolutions to complete the hover-wake development as well as to achieve satisfactory accuracy of the rotor trimming. During these simulations certain experiments concerning choice of time step have been conducted. The time step dt=1.4057?10-4 s applied for simulations of forward flight has been selected based on these experiments. Such time step seems to be sufficient to assess the performance characteristics of the rotor. On the other hand the selected time step was acceptable from point of view of finishing the task on time, taking into account accessible software/hardware resources. However, there is a risk that the selected time step is insufficient for modelling the unsteady-trailing-edge-vortex-shedding phenomenon, if it occurs. To check whether a reduction in the time step could result in the appearance of the flow of high-frequency oscillations caused by the vortex-shedding phenomenon series of numerical experiments of rotor-forward-flight simulation, with tenfold smaller time step (dt=1.4057?10-5 s., 14400 time steps per one revolution of the rotor, 0.025deg of blade azimuth per time step) , have been conducted. Due time-and-hardware/software-resource limitations, the simulations were performed only for about one-fifth of revolution of the rotor. Results of these experiments: the aerodynamic moments: M_theta , M_beta , M_ksi with respect to feathering axis, flapping and lead-lag hinges respectively, were analysed as functions of time (or blade azimuth PSI) and compared to the same functions obtained using nominal time step. Results of such time-step-sensitivity analysis indicate that in all analysed cases, the high-frequency oscillations of M_theta were not observed, neither for the nominal time step nor for tenfold smaller. Effect of tenfold reduction of time step on M_theta is shown in the pdf version of this report where it can be seen that this reduction of computational step did not produce any changes in M_theta.
Simulations of hover with Active Gurney Flap (5-blade rotor)
Simulations were conducted with two modes of AGF: 1/revolution and 2/revolution. The subject of investigations was assessment of the effects of AGF on Figure of Merit, which is defined, based on momentum theory as ratio of ideal power necessary for hover to the real power in hower. The reference results were computed for clean rotor (no motion of AGF). For configurations with AGF activated the trimming procedure described in Chapter 1 was applied in order to obtain zero first harmonic of blade flapping.
Simulations were conducted for the following hover conditions:
Altitude: 5486.467 m,
Density: 0.698 kg/m3,
Static Pressure: 50600.900 Pa,
Temperature: 252.488 K,
Rotor Angular Speed 31.040 rad/sec.
Ten Flight Priorities were defined for simulations, differing with thrust and type of AGF deflection(details in pdf-version of this report). Based on completed computational simulations it may be concluded, that:
• In all considered cases, the configurations with activated GF ("1/rev." and "2/rev.") indicate some growth of Figure of Merit, compared to the clean-blade configuration ("no motion").
• For rotor-thrust coefficient CTUS = 0.01010: compared to the clean-blade configuration ("no motion"), the configuration with operating AGF "2/rev." is characterised by relative growth of Figure of Merit, equal: Delta_FoM = 0.67% .
• For rotor-thrust coefficient CTUS = 0.01096: compared to the clean-blade configuration ("no motion") the configurations with operating AGF "1/rev." and "2/rev." are characterised by similar relative growth of Figure of Merit, equal: Delta_FoM = 0.71% and 0.72% respectively.
• For rotor-thrust coefficient CTUS = 0.01225: compared to the clean-blade configuration ("no motion") the configuration with operating AGF "2/rev." is characterised by relative growth of Figure of Merit, equal: Delta_FoM = 1.00% .
• For rotor-thrust coefficient CTUS = 0.01376: compared to the clean-blade configuration ("no motion") the configurations with operating AGF "1/rev." and "2/rev." are characterised by
similar relative growth of Figure of Merit, equal: Delta_FoM = 1.45% and 1.49% respectively.
• Relative growths of FoM for configurations with operating AGF "1/rev." and "2/rev." are similar to each other for the same value of rotor-thrust coefficient.
• In considered range of thrust coefficient: the higher a rotor-thrust coefficient CTUS, the higher relative growth of Figure of Merit is observed for configurations with activated GF in comparison to the clean-blade configuration.
3.2 Results of investigations for four-blade model rotor
Smulations were conducted for a model rotor of 2.2m-diameter in conditions of forward flight and hover. Simulations of hover were conducted only for clean blades and Passive Gurney Flap. For conditions of forward flight two modes of AGF deflection were modelled: sinusoidal and ramp-type. For forward-flight simulations the trimming procedure was used to obtain zero flapping conditions. The following flow parameters were set, similar to conditions of WTT:
Flow velocity: 48 m/s
Rotor rotational speed: 1600 rpm
Air density: 1.225 kg/m_3
Air static pressure: 101325 Pa
Air temperature: 15 deg C
Pitch angle of the rotor shaft: 0 deg
Bank angle of the rotor shaft: 0 deg
Commanded angles of collective pitch: Theta_0 = 4 deg, 6 deg, 8 deg, 10 deg, 12 deg.
Simulations of hover were conducted for fixed cone and lag angles, corresponding to wind-tunnel test data. The results of conducted computational simulations led to following conclusions:
• Results of CFD simulations and WTT indicate good qualitative convergence for both the Forward-Flight and Hover conditions. It especially concerns qualitative evaluation of potential benefits and limits of application of Passive Gurney Flap on rotor blades.
• Quantitative good convergence of CFD and WTT results has been found for the Clean-Blades configuration.
• For Passive-Gurney-Flap-2.5mm configuration the dependency Torque Coefficient (C_Q/sigma) vs. Thrust Coefficient (C_T/sigma), where sigma is solidity, has also indicated good quantitative convergence. However for this configuration, the dependences of Aerodynamic Characteristics on Commanded Collective Pitch (Theta_0) indicate specific, nearly constant offset of CFD results in respect to WTT results. This divergence between CFD and WTT results may be influenced by several factors, including:
o possible differences in computational and experimental mass model of rotor blades equipped with fixed Gurney flaps of height 2.5mm
o possible differences in computational and experimental geometry of model of rotor blades equipped with fixed Gurney flaps of height 2.5mm
o other incompatibilities between conditions of CFD simulations and WTT.
Results of CFD simulations of Active Gurney Flap of sinusoidal and ramp kinematics, conducted for conditions of Forward Flight at speed of 48m/s, led to conclude that:
• Effects of AGF of sinusoidal and ramp kinematics on Torque-versus-Thrust dependency are very similar to each other. It should be noted, however, that this convergence of results was observed for assumed, specific ramp kinematics of AGF. It may be expected, that optimised ramp kinematics of the AGF may have to give some benefits in relation to the sinusoidal kinematics. Undertaking more research on this topic seems to be very appropriate.
• For higher values of Thrust (or Collective Pitch) the AGF-2.5mm configuration, both the sinusoidal and ramp, gives similar benefits (in terms: Thrust versus Power/Torque) as PGF-2.5mm configuration.
• For lower values of Thrust (or Collective Pitch) the AGF-2.5mm configuration, both the sinusoidal and ramp, does not indicate the power penalty, observed for the PGF-2.5mm configuration.
In other words, in this area of Thrust, dependencies of torque on thrust for the Clean-Blades and AGF-2.5mm configurations are very similar to each other and they are more favourable than similar dependencies corresponding to PGF-2.5mm configuration.
• Configuration AGF-2.5mm(ramp) needs slightly lower amplitude of Commanded Cyclic Pitch than configuration AGF-2.5mm(sinusoidal) to trim the rotor to "zero-flapping" conditions.
The results for forward-flight simulations of 4-blade rotor may be summarised in graphs (in pdf version of this report) presenting dependence of power reduction on thrust coefficient. It confirms results obtained in simulations of five-blade rotor where increase of Power Loading was obtained especially at high thrust conditions. It presents also conditions when simple configuration, with Passive Gurney flap is beneficial in terms of reduction of required power.
To summarise results based on determined dependence of power reduction on thrust coefficient it may be stated that:
• For higher values of thrust (dynamic stall) the AGF-sin and AGF-ramp configurations indicate similar benefits in power reduction as (Passive Gurney Flap) PGF configuration.
• For lower values of thrust the AGF-sin and AGF-ramp configurations do not indicate significant power penalty, observed for the PGF configuration.
• Configuration AGF-ramp seems to be more favourable and needs slightly lower amplitude of Cyclic Pitch than configuration AGF-sin to trim the rotor to "zero-flapping" conditions.
4. Final conclusions
The computational method of simulation of Active Gurney Flap proposed by Institute of Aviation has proved its applicability for assessing effects of Active Gurney Flap on performance of helicopters. The method was implemented as User-Defined module in the commercial code ANSYS FLUENT in three version: two-dimensional, two-and-a-half dimensional and fully three dimensional. The two- and two-and-a-half dimensional versions of the method (VIRTUAL ROTOR 2D and VIRTUAL ROTOR 2.5D) provide possibility of modelling steady or oscillating flow conditions with oscillating free-stream velocity, angle of attack, and defined by user scheme of deflection of AGF. The fully three-dimensional version (VIRTUAL ROTOR 3D) apart from simulation of AGF kinematics resolves dynamic equations of motion of rotor blades and enables modelling of rotor control and trimming for required flight conditions.
The computational simulations of operation of AGF on rotors conducted in the COMROTAG project for geometric models defined by Topic Manager company and different modes of AGF deflection have revealed conditions when AGF brings performance benefits in terms of increased rotor thrust or power required for flight.
Potential Impact:
Description of project potential socio-economic impact including wider societal implications
Investigations conducted in the COMROTAG project concern simulation of operation of Active Gurney Flap (AGF) on helicopter blade and determination of transient aerodynamic loads as effects of operation of AGF. Application of AGF on a rotor is meant to enhance lift force on retreating blade and is one of Active Rotor Technologies investigated by Green Rotorcraft Consortium that should enable a helicopter to operate with reduced power consumption or reduced main rotor tip speed whilst preserving current flight performance capabilities, especially in terms of retreating blade stall. Lower power consumption should lead to reduced fuel usage and exhaust emissions, while reduced main rotor speed should significantly reduce rotor noise. Development of computational methods to simulate AGF is an important part of research in the field of Active Rotor Technologies and the importance of the completed investigations may be considered as their potential impact in the following fields:
- aeronautical research and technology development,
- economic aspects of the developed technology,
- societal aspects of the research and development of technology.
In the field of aeronautical research and technology development the results of the COMROTAG project have proven the viability of the assumed modelling approach, in particular of the combination of moving and deforming mesh approaches that make it possible to model in parallel the kinematics of AGF, deformation of blade contour and surrounding mesh zone and dynamics of the controlled rotor. These capabilities have been implemented as a module of User-Defined Functions called “Virtual Rotor” operating with the commercial CFD solver ANSYS FLUENT. Three versions of the software were prepared: fully three-dimensional version, two-and-a-half dimensional one and strictly two-dimensional one. In effect the modelling system created this way is capable of simulating flight of controlled rotor with blades equipped with AGF as well as of simulating of wind-tunnel experiments involving rotors and blade segments equipped with AGF. Capability of computational simulations of every step of experimental investigations should lead to determination of conditions when AGF brings benefits to rotor performance, to be verified by experimental investigations and to determination of conditions creating safety problems, e.g. through intensive vortex shedding requiring increased caution in experimental investigations. So, it is expected, that the computational simulation methods developed in the COMROTAG project will accomplish goals set in general for simulation methods: making the research phase shorter and safer and determining problems requiring increased attention during development of new technology.
The economic aspects of the developed modelling method are linked to its capability of determining flight conditions and modes of operation of AGF that should bring performance and economic benefit in operation of helicopters, such as decreased rotational speed or necessary power at the same thrust, or increased thrust at the same rotational speed or the same power required to produce thrust. Such conditions require subsequent experimental verification, but the capability of aiding directing of investigations towards determination of possible economic benefits of the application of AGF in helicopter rotors is an important feature of the method and it enhances the competitiveness of European industry and research sector. In particular the computational methods developed in the project support improving Technology Readiness Level of the investigated technology.
The societal aspects of the work conducted in the project COMROTAG may be considered in terms of the increased competitiveness of aeronautical research conducted in European companies and research facilities. Complementing of experimental investigations by advanced computational methods increases competences of research teams and shortening of time-to market for innovative solutions promotes job growth in companies exploiting results of research. Project COMROTAG contributes to these goals by providing modelling solutions complementing experimental research that can be exploited in the Topic Manager company as well as in Instytut Lotnictwa.
The societal aspects of the project COMROTAG project include also the potential effects of dissemination of the knowledge gained in the project. Dissemination of new knowledge achieved as an effect of realisation of a research project increases the public awareness of the state of the art of aeronautical research and increases confidence in air transport. Also increased is the awareness of the capabilities of European researchers, designers and manufacturers in the field of aeronautical research. The results obtained in the project have been already presented in three scientific conferences, with two of them very important for presenting results of aeronautical research to the scientific community: the Inernational Congress of Aeronautical Sciences congress in 2016 in Daejon, Korea, and the European Rotorcraft Forum in 2016 in Munich, Germany. In both cases the papers presented modelling capabilities of the computational methods developed in the project COMROTAG as well as results of investigations obtained in the project. Two other papers are planned to be prepared in the near future, after obtaining acceptance for results to be published from the Topic Manager company.
The computational methods developed within the project were also presented to a wider audience during two dissemination/communication events: Reporting Sessions of University of Warsaw Interdisciplinary Center of High Performance Computing in 2014 and 2015 in the form of presentation and a poster session. These presentation were addressed to specialists from different fields of computational simulations
In addition to this a web page dedicated to the COMROTAG: http://comrotag.ilot.edu.pl. This page will be maintained after completing the work in the project.
List of Websites:
http://comrotag.ilot.edu.pl
