Analysis of Two-Dimensional NACA 642-015 Aerofoil Section - Coursework Example

?DSGN313 – work A – CFD Assignment This 7 credit work should be carried out individually. The deadline for this work is 23.59 on 4th November 2011.You must submit your coursework electronically via the SCOLAR system accessed from the DSGN313 Portal site. You should give due consideration to your personal time management to ensure that coursework is submitted in plenty of time prior to the deadline. Coursework can be submitted at any time ahead of the deadline; however, the Faculty cannot take any responsibility for late submission due to uploading problems, etc. Please note that the University enforces a penalty of zero percent for work submitted after the published deadline without valid extenuating circumstances (see University student handbook on the portal for details). Intended outcomes – By the end of this coursework, you should be able to: Carry out a CFD simulation using ANSYS Workbench/CFX, demonstrating ability to import geometry, produce a mesh, set up and solve a simulation and effectively post-process results. Evaluate grid-dependency of a solution and demonstrate the process of finding a grid-independent solution. Demonstrate ability to compare CFD results with published experimental data, and critically evaluate results with reference to relevant literature. Recognise capabilities and limitations of a CFD analysis in a particular application. Present results of a CFD analysis clearly and concisely, with appropriate output from CFD-Post. Aim To use ANSYS CFX to simulate the flow around a two-dimensional NACA 642-015 aerofoil section at a 5o angle of attack and to assess the accuracy of the simulation. Problem specification It is important to understand the lift and drag characteristics of aerofoil sections when designing devices such as aircraft (wings and tails) or yachts (rudders and keels). Traditionally, foil theory has been used to give performance estimates, along with extensive experimental testing. More recently, CFD has become another possible option when investigating foil performance. In reality, foils exhibit three-dimensional performance, because flow around the tip of the foil affects lift and drag. However, it is useful to determine performance of a two-dimensional foil – that is one that is so long (approaching infinite length) that the effects of flow around the tip are negligible. You are going to use CFD to simulate a 2-d foil, and compare your results to those obtained experimentally in a wind tunnel, detailed in a NACA paper from 1945. Instructions You will not be writing a formal report for this project. Instead, you will work through this document (using it as a template), adding content and answering questions as instructed. You will then submit the completed document for assessment. Note that answers/images, etc. do not have to fit into the space provided – insert extra space as necessary, but keep answers concise. Carry out the steps as follows: Carry out a basic CFD simulation (named “Run_1”) of a NACA 642-015 foil noting the following: The foil geometry has been created in SolidWorks for you – the file aerofoil_CW_2011.SLDPRT can be found on the DSGN313 Tulip site under CFD Coursework. The file Aerofoil_Instructions_2011.doc (also on the portal) explains how to modify the geometry parameters using SolidWorks. Note that it is down to you to modify dimensions to set the extents of your domain in all directions, and to set your foil chord and angle of attack – don’t just run with the dimensions given to you. Your simulation should be for a 5o angle of attack, and a 24 inch chord length (for comparison against the NACA experimental data). You should run your simulation at a Reynolds number of 6x106. Note that the length scale used in the Reynolds number is the chord length of the airfoil section in the model. Use ‘Water’ as the fluid and assume that the flow is incompressible, steady, isothermal and turbulent. Use the k-? turbulence model. Run_1 should be a coarse, unrefined mesh purely to get your simulation working (you will add refinements later). 1. Show an image of your geometry below. Annotate the image to show what you have named each surface of the geometry: 2. For EACH surface in your geometry, state below what boundary condition you have applied, briefly describe what that boundary condition does, state how the software applies that condition at the boundary nodes, and explain why you selected it. For any boundary conditions that require you to set a numerical value, clearly show the calculation you used to determine that value. Region Boundary Condition(s) Reasoning Aerofoil_Section No slip The aerofoil is fixed in place and a solid body so as the fluid approaches the velocity tends to go towards zero. The software treats such boundaries as walls. Fluid_Region_Front Inlet boundary, velocity 1 m/s (arbitrary), normal speed, subsonic Fluid inlet for the simulation, flow is subsonic. The software treats such boundaries as inlets. Fluid_Region_Back Outlet boundary, subsonic, static pressure, relative pressure 0 Pa, turbulence zero gradient Back of aerofoil wall, simple atmospheric pressure applies and flow is subsonic. The software treats these boundaries as outlets. Fluid_Region_Left Free slip The regions define the fluid layers adjacent to the aerofoil and is not fixed so incoming fluid velocity will have some effect. The software treats such boundaries as walls of the fluid that can assume some velocity and hence move. Fluid_Region_Right Fluid_Region_Top Fluid_Region_Bottom 3. Show a dimensioned image of your geometry below. The image should clearly show how far away from the foil the extents of your domain lie (upstream, downstream, above and below). It should also show the width of your domain, and the chord length and angle of attack that you have set. 4. Explain how you decided upon each of the dimensions detailed above (giving references if necessary). The initially set dimensions have been left as is to perform the first run. In case that the flow is not fully developed, the dimensions would be changed. The length being used for the fluid region on each side is around twice the chord length while the height is more than three times on each side. The flow will be turbulent as the Reynolds’s number is high so large dimensions would be required to fully appreciate the developed flow. 5. Show an image of your entire mesh below (and any zoomed-in images necessary to provide the reader with a clear picture of the element shape/resolution at any particular areas of interest). 6. Explain what element type(s) your mesh uses, and why you think they are appropriate for your Run_1 mesh. Also explain what mesh method(s) you used, and what global and local controls you applied to control the size of your mesh – be sure to state not just the type of controls you used, but also the values that you set. Explain why these settings were used. The mesh uses triangular elements with variable sizing in order to deal with regions that display smaller dimensions. The smallest dimension is near the back edge of the aerofoil at 4e-3 m. For the first run the controls have been set as below for a rough run. Maximum spacing 1 m Angular resolution 30 Minimum edge length 0.05 m Maximum edge length 0.5 m 7. Paste a copy of the convergence plot (Mass & Momentum) for Run_1 below. 8. For the above plot, explain the following (you may wish to annotate your plot to make your explanations clearer): What does each axis of the plot represent? The vertical axis represents the amount of error (or the residual left in the solution) while the horizontal axis denotes the number of iterations. What does each of the plotted lines mean, what governing equation do they refer to, and (in words, not formulae) what does each governing equation state? Each plotted line refers to the quantities that could produce errors in the final solution based on the finite element formulation of the exact equation. These quantities are derived from the approximate FEA solution of the Navier Stokes equation for mass, momentum and energy conservation. Is one of the momentum lines much lower than the other two? If so, which one and can you explain why? The momentum line for W is much lower than that for U and V because it represents velocity based momentum on the Z axis which is not affected much (thus has reduced error residuals) because of the geometry of the aerofoil. What caused the run to stop at the point that it did? The run converged to an acceptable level of error so it stopped. Does your plot show adequate convergence? What tells you that it is adequate or not? The convergence is adequate because the error terms are low (all under 1e-4). Going any lower would require large computing power and time without much improvement in the existing results. Does the information that you gain from your convergence plot give you confidence in your CFD results and why? As far as the solution is concerned, the convergence plot provides adequate confidence as listed above. However, when the mesh geometry is considered the results are inadequate as they deviate from the actual dimensions of the aerofoil by a sizeable margin. 9. Determine the Drag and Lift forces on the foil from your simulation. Write the down the values you obtained for Drag and Lift. L = -1.21 N D = -0.106 N Explain exactly how you extracted this data. Using plots of forces on the x-axis and the y-axis and seeing the region around the aerofoil the upper and lower colour values were averaged. This gave Fx and Fy which were 6.142e-10 N and -1.212 N respectively. These were used with the following formulas to get lift and drag. With = 5o for the angle of attack. For both drag and lift, calculate the percentage of the force that is due to viscous force and the percentage due to pressure force. Show your calculations and explain how you obtained the input data for these calculations. Input data was obtained from shear and pressure plots for x and y. Areas were neglected as they would remain the same. These values were tabulated as below: Shear Pressure X 2.01 -1.449e-3 Y -1.438e-3 1.33e3 Shear component of drag = 2.01 / (2.01 + 1.449e-3) * 100 = 99% Pressure component of drag = 1% Shear component of lift = 1.438e-3 / (1.438e-3 + 1.33e3) = 1% Pressure component of lift = 99% Briefly explain how (i) pressure distribution and (ii) viscous effects can impart a force on a foil. Pressure distribution produces pressure differences above and below the aerofoil which in turn creates lift while when viscous effects come into play, the flow over the aerofoil is negatively affected. Viscous effects tend to decrease the effects of the pressure distribution especially in terms of the lift. Do you believe that the predictions you have obtained for lift and drag are realistic? State the two main reasons why they could deviate from those in reality. The predictions for lift and drag are unrealistic because: The geometry of the mesh does not resemble the actual aerofoil; The flow conditions fail to account for many different effects such as the use of an idealised isothermal heat transfer, predictable turbulence model etc. Calculate your lift and drag coefficients (clearly showing your calculations) and compare with experimental values for the NACA 642-015 airfoil section, which are available in http://naca.central.cranfield.ac.uk/reports/1945/naca-report-824.pdf (see pdf page number 190 (or report page number 447), for the lift and drag curves for this airfoil section). Calculate percentage errors (assuming the NACA values to be accurate). Note that you should be comparing with the R=6x106 values on these curves (with the ? markers). 10. Improve your mesh such that you believe it to be appropriate for use with a k-? turbulence model and re-run your simulation. Call this simulation “Run_2”. If, after running, you discover it is still not appropriate for use with the k-? model, make further improvements to your mesh and run again, calling this one “Run_2a”. You are advised not waste time continuously improving your mesh in search of the “perfect” mesh – two cycles of improvement should be adequate provided you think carefully about the mesh settings that you use. List the attributes that you consider an appropriate mesh to possess. An appropriate mesh should have elements spaced according to the smallest dimension as well as per the associated results. The area with a greater interest should be meshed more detailed. 11. With the aid of appropriate screenshots (annotated as necessary), show that you believe your improved mesh to possess the required attributes. Explain what controls and settings you used to produce these attributes and show any calculations/methods you used to arrive at these settings. In addition, show any areas where you don’t believe your mesh to be ideal and also show an image of your entire mesh. Minimum edge length 0.1 m Maximum edge length 0.1 m The mesh is less than ideal near the trailing edge of the aerofoil as well as near the leading edge. 12. For your improved mesh, calculate the new lift and drag coefficients. How do these compare with the NACA data now? 13. Create and run another two meshes (Run_3a and Run_3b) still using the k-? turbulence model. With these meshes, you should be attempting to reach (or prove that you had already reached) a mesh independent solution. Paste some screenshots of these meshes below and annotate them to illustrate what you changed and explain why you felt that these changes would potentially demonstrate mesh independency. The lowest element size was reduced to improve accuracy of meshing at the leading and trailing edges as well as throughout the block. However the mesh is still too concentrated in more simple geometrical areas. 14. Provide mesh dependency graphs below showing results from ALL your runs. State whether you feel that you have ultimately arrived at a mesh independent solution or not, and annotate your graphs to show why. Annotations should also show exactly what you are plotting on your graph axes, and what your best estimates for lift and drag coefficients are based on the results of your simulations (ignoring any lack of correlation with the NACA data). Run_1 Run_2 Run_3a Run_3b The convergence is improving without doubt and the lift and drag coefficients can be expected to improve as well though quantification may not be possible. 15. Explain why refining a mesh can improve the accuracy of a simulation. The more refined a mesh is, the more closely spaced the nodes are and so the more detailed the analysis is. The presence of a greater number of points indicates that the solution is far more detailed. 16. For the run that you consider to be your most accurate one, state whether you believe flow separation to be occurring at any point on the foil. Provide visual evidence below to support your answer. Flow separation can be seen resulting in the plots presented above. 17. For the same run, produce a line graph to show variation of pressure around the foil. 18. For the same run, produce a streamline plot to clearly show the shape of the flow over the foil. 19. Having run your simulations, do you now believe that the extents of your domain were far enough away from the foil to avoid influencing the solution? For each domain boundary, provide graphical evidence below to show whether it may be considered far enough away from the foil. Based on the stream line plot and the pressure plots above the extents of the domain can be seen as reasonable as the domains ends are not affected by the flow. 20. Based on all the evidence from your simulations so far (and from the NACA data), do you believe your solution to be accurate and reliable? If not, bullet point up to 5 concise reasons why your results may be inaccurate. State your reasons in order of significance, with the one you believe to be most significant first. The solution is not completely reliable because: The geometry is not completely reliable after meshing; The fluid properties are ideal; Viscous effects have not been totally accounted for; Errors in approximate solutions exist; There may be errors in the NACA plots that could promote errors during comparison. 21. Run two more simulations (Run_4 and Run_5) changing any parameters in CFX-Pre or CFX-Solver Manager that you feel may improve your results. For each run, state what you have changed and why you think it may make an improvement. Note – you may also have to change your mesh to ensure it is compatible with your revised settings – if so, present annotated images of your revised mesh. You may also wish to briefly outline any problems that you experienced in carrying out these simulations. The mesh has been revised for Run Four by using mechanical and automatic sizing for controls. This produces a mesh that is sized as per requirement of computation. Run Five has been optimised using meshing controls exclusively for CFD problems. This ensures that the resources are allocated as required. In order to improve the model, it would be better to extend the domain from the left and right end to see how the sides of the aerofoil contribute to drag and lift. 22. Show calculations for lift and drag coefficients for runs 4 and 5, and quantify any improvement (or otherwise) in your results. 23. State whether the choices you made for runs 4 and 5 were influenced by the computational resources you have available, and briefly explain why. If you had access to commercial High Performance Computing (HPC) resources, suggest what you might do differently and explain why you think that would improve your results. If greater computing resources were available, the fluid domain could be expanded and the meshing size could be lowered more in order to improve results. 24. Suppose your results for your single foil at 5o angle of attack were reasonably accurate (say within 5% of the NACA data). If you used the same settings and mesh resolution for a simulation of a pair of 642-015 foils in a biplane configuration (i.e. one above the other, still at 5o angle of attack, Re=6x106), would you have confidence that these results would be reasonably accurate too? Explain why. This cannot be determined as such because the influence of one flow on another has not been dealt with in this simulation. However improvements done in this simulation could be implemented the same way to get similar results. 25. Again, suppose your results for your single foil at 5o angle of attack were reasonably accurate. If you used the same settings and mesh resolution for the same foil at 20o angle of attack, Re=6x106, would you have confidence that these results would also be reasonably accurate? Explain why. Yes because the angle of attack has little bearing on the settings and controls used to simulate these results. Assessment Criteria Marks will be awarded for each step listed in the instructions above, according to how well you have carried out that step. Marks will be deducted if answers are not clear and concise – keep answers short and focused. Make sure all images clearly show what you intend them to show – check they have adequate resolution and that any text is clearly legible. All graphs and charts should follow good engineering practice – clear labelling of axes and units, etc. Calculations and numerical answers should follow good engineering practice – nomenclature and units must be clear throughout. Coursework submission Create a folder on your computer named as follows: Firstname_Surname_StudentNo_DSGN313_CWA Place the following files into that folder: Your copy of this word document, containing all requested content and answers. Name it with the same name as the folder name above. The .def file associated with one of your runs – pick the one that you consider to be most accurate out of Runs 2, 2a, 3a, 3b and 3c. You can find this by going to the working folder for that run, then looking in “[run_name]_files/dp0/CFX/CFX”. It will probably be listed as an “ANSYS CFX Pre File”. The .out file associated with the same run. This will be found in the same folder as the above file. If you ran the solver more than once for this simulation, you will have more than one .out file. Pick the most recent one. Zip up the folder containing the three files listed above, and submit that single .zip file using the SCOLAR system on the DSGN313 portal site. Useful reading The CFD Reading List on the module portal page gives an indication of some relevant books, particularly in the General Fluid Mechanics section (for lift, drag and aerofoil theory). There are many other books in the library that contain information on foil theory. The Workbench/CFX help files are generally the first place to look for information on CFD modelling. The ANSYS CFX Student portal is another useful resource The CFD web links on the module portal page can be very useful – especially the CFX User Forum, which usually contains an answer to commonly encountered problems. The NACA paper from 1945 is useful, not only for data comparison, but also on foil behaviour http://naca.central.cranfield.ac.uk/reports/1945/naca-report-824.pdf (see pdf page number 16 (or report page number 273). The lecture notes and tutorials (on the portal) should proved practical assistance in CFD modelling. Read More