Composite Optimisation of an F1 Front Wing

Transcription

1 Composite Optimisation of an F1 Front Wing Jonathan Heal Senior Stress Engineer, McLaren Racing Ltd McLaren Technology Centre, Chertsey Rd, Woking. GU21 4YH Abstract Formula one is a fast paced sport where a small advantage can mean the difference between winning and losing. There is a desire to get models from Aerodynamics to the track as quick as possible, which makes optimisation tools essential to generate tailored structural solutions in compressed timescales. Since 2003 we have found that composite optimisation tools are essential in this process and without them it would be much more time consuming to tailor the performance of composite structures. This paper looks at the optimisation process involved in developing a composite laminate for the front wing of a Formula One car. For this work OptiStruct 11.0 is used along with and a newly developed HyperMesh Composite Optimisation Setup tool, developed in close association between McLaren and Altair, to reduce the burden of setting up composite optimisation studies. This study uses Free Sizing optimisation to identify potentially efficient ply shapes. The detailed sizing section uses a PCOMPG(P) definition of a laminate, which includes material draping. The optimisation process is performed with both thickness and orientation as design variables. Keywords: Composite Optimisation, Free Sizing, Detailed Sizing, OptiStruct 1.0 Introduction Formula One is a fasted paced industry where performance will be measured at 19 circuits around the world in The first battle is to design the fastest car through the winter period. In order to win races throughout the season it is necessary to increase the performance of the initial car by more than 3 seconds a lap by the end of the season. This means that the battle in the factory is as important as the battles on the track. Parts need to be delivered from the wind tunnel to the track in as short a time scale as possible. Weight is one of the key drivers in the performance of the car so it is important that an optimal lightweight structure can be developed within a very tight project schedules. Since 2003 composite optimisation has played an essential role in delivering solutions to the track that meet all of the key design drivers of the applications that it is used in. This paper uses the example of a 2008 Formula One Front Wing to demonstrate the process involved in the design of an optimal composite structure using OptiStruct Two separate composite optimisation studies have been performed in this paper. The first optimisation approach uses free sizing optimisation to identify areas of efficient composite material usage. The second is a detailed sizing optimisation, which uses a newly codeveloped interface for setting up pre and post processing aspects of this type of composite optimisation. This interface has been influenced by the processes that have been used at McLaren Racing since This paper looks at some of the features of detailed sizing optimisation that need to be considered so that an optimal solution is obtained. Altair Engineering

2 Figure 1: FE Model of the Front Wing Used in this Study 2.0 Common Model Settings 2.1 Load cases Two load cases are used in this study. The first requirement of the wing is to pass the allowable deflection test prescribed by the FiA. This regulation can be found in Article of the FiA Formula 1 technical regulations [1] and states that the front wing must deflect less than 5mm when loaded at the prescribed position. The second case is an aerodynamic loading case of the structure calculated at the maximum achievable speed of the car. This pressure mapping of the structure is obtained from CFD runs on the wing. This case is used to assess the overall strength of the structure. A deflection characteristic under loading can also be targeted using this case in order to obtain an optimal aerodynamic bending characteristic. The FiA deflection test and the Aerodynamic loading cases are applied symmetrically about the centre of the wing so only half the model is run and a symmetry condition is applied at the centre of the wing. 2.2 Optimisation Objective Lightweight design is the objective of all of the structure on a Formula One car. This front wing is no exception so this is the objective used in these studies. 2.3 Optimisation Constraints In order to maintain consistency between the optimisation studies performed in this paper the constraints are identical. In this case the deflection for the FiA was set to be between 4.75mm and 5.0mm for these runs. The rotation of the wing is also constrained for the Aerodynamic loading case in order to produce the desired aerodynamic performance for this wing as the wing deforms. Altair Engineering 2011 Composite Optimisation of an F1 Front Wing 2

3 3.0 Free Size Optimisation 3.1 Overview This section of the optimisation is mainly aimed at identifying areas of efficient material usage. In this case UD material will be preferred over Woven material of a similar fibre type. In reality the final desired structure will be a mix of UD and Woven materials however the designable plies used in this section of the study are limited to UD material. This part of the optimisation is aimed at obtaining a better understanding of the structure. As the geometry and the loading of a front wing are constantly changing it is not always possible to use the previous design ply shapes as a guide to optimal ply shapes. 3.2 Basic Laminate Setup This study was set up with an array of designable UD material at -30º to 30º in 10º increments. Using Laser projection technology it is possible to use a smaller increment than this, although using incremental angles as small as 1º is likely to make interpretation of the results difficult. Previous experience has shown that this array of UD angles is sufficient to capture the overall shape of an optimal ply. Previous work on this type of front wing showed that the first three main elements of the wing were influential on the overall bending performance of the wing. The End Fence and the 4 th element have a very small effect on the bending of the wing so prescribed laminates are used on these two elements. 3.3 Free Sizing Results The results of the Free sizing optimisation (Figure 2) indicate that the second element is the most efficient element, which can be seen in the thickest areas below. This is consistent with previous work on this generation of front wing. This data is used to develop ply shapes for the detailed sizing section of the study. Figure 2: Optimal thickness distribution of material for the Free Sizing section of the optimisation Altair Engineering 2011 Composite Optimisation of an F1 Front Wing 3

4 4.0 Detailed Sizing Optimisation 4.1 Laminate Definition and Setup This section of the optimisation is set up with identical constraints as the Free Sizing optimisation. The objective is still to minimise the weight of the structure in order to produce the optimal structure. This section of the optimisation is aimed at obtaining a laminate that meets the overall stiffness and weight targets, strength targets can be added. Using the information from this stage of the optimisation it is possible to go straight into manufacturing of the laminates obtained in this section without any further work. In order to speed up the process for setting up optimisation studies, particularly when there is a large number of prescribed plies that are driven by other requirements, McLaren and Altair worked together to develop a setup tool that allows PCOMPG(P) models to be used and automatically creates all of the DDVAL, DESVAR, DLINK and DVPREL1 cards for the chosen designable plies. This was an essential step so that we could use OptiStruct for all of our composite optimisation. In this example draping of each of the ply patch shapes has been included thus increasing the overall accuracy of the solution and removing any potential need for recreating the model at a later stage in order to determine the effect of draping on the final solution. This solution uses a.layup file definition of the starting model. The process of setting up the optimisation has been streamlined so that any laminated structure defined with a.layup file can easily be turned into an optimisation study. This process starts with the definition of the limits of the material thickness and orientation. To speed up the setup process this is performed at the ply material level at the beginning of the optimisation setup although this can be changed for individual designable plies (Figure 3). This data is automatically passed through the process and is assigned to all of the designable plies of a specific material, which speeds up the setup process. It was important to have a setup tool that was quick enough to set up a complicated model but was flexible enough as to not be too restrictive. Figure 3: Material Setup page of the composite optimisation setup tool The designable plies for thickness and orientation are chosen separately, which allows different plies to be linked together in the thickness or the orientation section. In the orientation section plies can be linked together symmetrically. Using this tool it is possible to read in the.layup file so that the names of the plies are visible to allow easier setup of the model. An example of the thickness setup is shown below (Figure 4), which includes linking Altair Engineering 2011 Composite Optimisation of an F1 Front Wing 4

5 of designable plies. This can reduce the size of the model and create a laminate that can be transferred to production with minimal effort. Figure 4: Thickness variables definition page of the composite optimisation setup tool 4.2 Continuous and Discrete Optimisation Phases In order to obtain the optimal solution two aspects of the optimisation need to be considered. At the end of the optimisation a discrete solution is desired as it is not possible to manufacture with fractions of ply thicknesses. In order to generate the optimal discrete solution two different options are available in OptiStruct Using the DDVOPT card in OptiStruct 11.0 it is possible to perform a completely discrete optimisation, where only discrete ply thickness are used during the optimisation, or it is possible to start with a continuous solution and then change to a discrete solution once the continuous phase has converged. Plotting the solution history for identical models with different DDVOPT values (Figure 5) it is possible to see that the objective history and the final solution are very different between the two approaches. In most cases it has been found that a more optimal solution has been obtained using the continuous solution, which changes to a discrete solution, than just using a discrete solution. Figure 5: Plot of the normalised mass, which is based on the minimum mass of the best discrete solution, at each stage of a fully discrete optimisation and a continuous->discrete solution. Altair Engineering 2011 Composite Optimisation of an F1 Front Wing 5

6 4.3 Global Iterations The second aspect that needs to be considered is whether a solution is a global or local optimum. With gradient based optimisation it is possible to obtain a local minimum especially when thickness and orientation are included in a single optimisation study. In order to ensure that the final solution is likely to be the globally optimal solution a new feature in OptiStruct 11.0 is used. A new card has been added which allows multiple optimisations to be performed in a single run. This is the DGLOBAL card and the format of the card can be seen below. With this card a specific number of global iterations can be specified. In this study 10 global iterations were performed in order to obtain the global minimum mass. At each of the global iterations the starting point of the DESVAR cards is modified and the solution is run until an optimum is found. The global optimal solution is then identified once the solution has finished. When the final iteration of each of the global iterations is plotted (Figure 7) it can be seen that there is a significant variation in the mass of the structure as well as the optimisation constraints. One unique global optimum solution is identified in this run. The more global iterations that are performed the more likely that the global optimum is obtained. Figure 6: DGLOBAL card format that determines the number of global iterations performed and how the starting points are identified for each of the iterations Figure 7: The local optimal solutions for each of the global iterations Altair Engineering 2011 Composite Optimisation of an F1 Front Wing 6

7 4.4 Global Optimal Solution Results When the thickness of the optimal solution is plotted (Figure 8) it can be seen that the thickness of the laminate in the second element is significantly greater than the thickness of first and third elements. The optimal thickness is similar to the Free sizing results. Figure 8: Thickness plot of the Global Optimum solution. The second element of the structure dominates the stiffness of this assembly with the Aerodynamic and FiA test constraints Figure 9: Deflection of the wing assembly for the FiA Deflection test on the Font wing Figure 10: Deflection of the wing for the Aerodynamic loading case Altair Engineering 2011 Composite Optimisation of an F1 Front Wing 7

8 Figure 11: Failure Index plot for the Aerodynamic loading case. The final solution is within the The deflection for the FiA load case has been maximised for this case and the Aerodynamic deflection characteristics have been met for this solution (Figures 9 & 10). Once the required stiffness of the wing has been achieved the strength of the structure is checked (Figure 11) and the solution was found to be acceptable without any reinforcement. With the HyperMesh Composite Optimisation setup tool it is possible to update the.layup file. This solution meets all of the structural requirements given for this project. Using the Composite Optimisation tool the.layup file was updated and the layup information for each of the parts can then be passed to manufacturing 5.0 Conclusions Composite optimisation has been an essential part of the composite design process at McLaren since This paper has shown that it is possible to determine the globally optimal solution for the design of a Formula One Front wing using some of the new features in OptiSruct. Using the HyperMesh Composite Optimisation Setup tool it is easy to setup a detailed sizing optimisation, based on a PCOMPG(P) definition of the laminate, and obtain a globally optimal solution. This makes it possible to turn most composite models into an optimisation study and generate optimal structures in very compressed timescales. 6.0 References [1] FIA Formula One Technical Regulations (2008) Altair Engineering 2011 Composite Optimisation of an F1 Front Wing 8