F2008-SC-005

FORMULA SAE RACE CAR ANALYSIS: SIMULATION & TESTING OF THE ENGINE AS A STRUCTURAL MEMBER
Abrams, Ryan* The University of Western Ontario, Canada KEYWORDS - Chassis Torsional Stiffness, Modes of Vibration, Chassis Deflection, Finite Element Analysis, Beam Elements ABSTRACT Through simulation and testing, the objective of this report was to determine the methods and validate results for using an engine as a structural member of a Formula SAE race car chassis. Modal analysis was used with a beam mesh of the chassis to validate an increase of chassis stiffness. Results confirmed that providing structural support at engine mounting points of the chassis increased the stiffness and changed the first elastic mode of vibration from bending to torsion. Additionally a static analysis was used to measure chassis deflection, and established a decrease in deflection when structural support was provided at the engine mounting points of the chassis. This reports conclusion indicates that using the engine as structural member significantly improves the stiffness of the chassis. Furthermore, with improved software Dynamic testing on the frame and engine block should be analyzed to obtain measurable results, establish the allowable forces on the engine and the direction of their application. The chassis under investigation for torsional stiffness is from The University of Western Ontarios Formula SAE race car. Chassis stiffness can be divided into two components torsional and bending - critical for the formula race car. Defining the systems axis, a stiff chassis opposes rotation about the y-axis in bending, and twisting about the x-axis in torsion.

Figure 1: The University of Western Ontario Formula SAE race car chassis with defined axis

CHASSIS STIFFNESS Torsional Stiffness

An ideal chassis is one that has high stiffness; with low weight and cost. If there is considerable twisting, the chassis will vibrate, complicating the system of the vehicle and sacrificing the handling performance. Thinking of the chassis as a large spring connecting the front and rear suspensions: if the chassis torsional stiffness is weak, attempts to control the lateral load transfer distribution will be confusing at best and impossible at worst 1. Therefore, predictable handling is best achieved when the chassis is stiff enough to be approximated as a rigid structure. There are numerous reasons for high chassis stiffness. A chassis that flexes is more susceptible to fatigue and subsequent failure, and suspension compliance may be increased or decreased by bending or twisting of the chassis1. Torsional stiffness range from 3000 lb-ft/ deg (4068 Nm/ deg) for a small race car to 12 000 lb-ft/ deg (16 272 Nm/ deg) and up for a Formula 1 car, quoted in 1995. 1 Bending stiffness Bending stiffness is less significant relative to torsional stiffness of a car chassis. In static bending, the wheel loads and their distribution deviate only slightly. Also, if a chassis is well designed to handle torsional loads, bending should not be an issue1. MEASURING CHASSIS STIFFNESS Natural Frequencies and Modes of Vibration A modal analysis was used on the chassis to provide relative comparison; with and without structural support from an engine. The chassis, as with any structure, has an infinite number of resonant frequencies. A resonant frequency, also known as natural frequency, is a preferred frequency of vibration, and results when the inertial and stiffness forces cancel. The lone factor controlling the amplitude of vibration in resonance is damping 2. For each of the infinite natural frequencies of vibration which exist, a different shape that the chassis will deform to during vibration also exists. The deformed shapes that chassis will vibrate are also know as modes of vibration. Of the infinite modes of vibration that exist on the frame structure, only the lowest frequencies are of interest. The low modes of vibration maximize the kinetic energy and minimize the strain energy while the high modes act in an opposite manner. This means that the soft and stiff parts of the chassis will be apparent in the low and high modes of vibrations respectively. Modal Analysis Before beginning discussions associated with the results of modal analysis, it is important to first touch on the type of mesh used. The chassis is a structure composed of many thin tubes and complex geometries near the corners. Therefore, it is not recommended to use a solid or shell element mesh type when conducting a stress analysis, even though it may be acceptable for frequency analysis. Beam elements are available on COSMOSWorks 2007 Professional, which can accurately calculate stresses using wire frame geometry and significantly reduce meshing time. A beam mesh is the natural choice for the structural analysis of the chassis. Returning to discussions of frequency analysis; a beam mesh of the chassis model was used for analysis of: mode shape and corresponding natural frequency, or modal analysis. Following meshing and defining on the order of eight to ten deformation plots, the study was created and run for interpretation of results. It may be anticipated that torsion would be listed in the first couple modes of vibration; however, it was seen the seventh. No loads or restraints were applied, therefore the first six modes are translation and rotation or rigid body modes. The structure has 6 degrees of freedom, describing 3 translational and 3 rotational rigid body modes, with are assigned values of 0 Hz . Of interest for analysis, are the seventh, eight and ninth modes of vibration or first, second, and third elastic modes of vibration. The lower

elastic modes of vibration result in lower natural frequencies. This means the first elastic mode of vibration demonstrates the shape that the chassis is most susceptible to deform because it has a lower natural frequency and hence, lower stiffness. Therefore it is desired to have a chassis with a relatively high natural frequency in the first elastic mode of vibration. It is worth considering not only first elastic mode of vibration, but several others because of cancellation effects from damping that could occur from the suspension dampers and soft engine mount damping materials. A mock engine was created with relatively strong structural beams at the engine mounting points in COSMOSWorks. Diagonals were used to support the main structure of the engine by connecting the mounting points, to help stiffen the engine.

Figure 2:The University of Western Ontario formula race car chassis with structural support at the engine mounting points

Modal analysis results for the chassis with and without structural support from engine are summarized: Chassis without structural engine support Chassis with structural support at engine mounting points

Elastic Mode 1 2 3

Natural Frequency (Hz) 42.95 57.57 70.78

Elastic Mode 1 2 3

Natural Frequency (Hz) 60.27 101 123.7

Elastic Modes of Vibration

Figure 3: Mode 1, in bending

Figure 6: Mode 1, in torsion

Figure 4: Mode 2, in torsion

Figure 7: Mode 2, in torsion

Figure 5: Mode 3 , in torsion Figure 8: Mode 3, in torsion Table 1: Modal analysis summary of chassis without structural support from engine (left column) and with structural support from engine (right column)

Table 1, figures 3-8 show the model (solid yellow) superimposed on the deformed shape (transparent brown). The results of the analysis on the chassis without structural support show the first elastic mode of vibration in Figure 1, which is primarily in bending at the back of the car. Though bending has less of an effect on the dynamics of a vehicle, it should not show in modal analysis results if correctly designed for torsional stiffness, as quoted earlier. Since significant movement can be seen in the back end, additional bracing should be added. However, elastic modes 2 and 3 in figure 4 and 5 are torsion modes. The average difference is 13 hertz between the three modes of vibration. Tests with structural support at engine mounting points show an increased first natural frequency to 60.27 hertz. This is higher than mode 2 without support from engine, and is torsional mode. Further review of the first mode seen in figure 6, illustrate a more even distortion of twisting compared to the excessive bending of the back end seen in figure 3. Elastic modes 2 and 3 in figure 4 and 5 are also in

torsion and the average difference between the three modes increases to 30.87 hertz. Comparing models, an increase in all three elastic modes with the structural support added to the engine mounting points where observed. Further more, in Mode 1 which lacked support in the back end, benefited from the added stiffness at the engine mounting points. This resulted in a change from bending to torsion in modes 1, improving the structural design. In addition, a chassis with already sufficient stiffness may consider using the engine as a structural member for the purpose of redesign for weight savings. Measuring Chassis Torsional stiffness Another approach to compare torsional stiffness is to simulate a conventional method for measuring torsion in COSMOSWorks. A method shown in Figure 9, shows a method to measure torsional stiffness. Knife edges

Dial indicators Jack Screw (both sides)

Figure 9: Chassis testing for torsional stiffness

Knife edges support the outer ends of the A-frames and the suspension is swapped for a rigid member. A screw jack is adjusted to cause a positive or negative torque, resulting in the chassis twisting. Knife edges are also used at the back end to restrict movement. Measurements of displacements are then taken via a dial indicator as the chassis is twisted. Static Analysis Replicating the process in Figure 9 using COSMOSWorks 2007 Professsional can be done with use of static loads and immovable restraints approximated at suspension mounting points. A static load is a load that does not vary with time. Therefore, the force is applied slowly such that inertial effects are considered negligible. Suspension loads were approximated at the suspension points for the front and rear. The magnitude of applied loads was considered since all materials follow a linear model, and yield along with ultimate stress is not modelled. Restraints limit a point on the chassis model to three or zero degrees of freedom. Beam elements have 6 degrees of freedom; immovable restraints only set the translational degrees of freedom to zero, therefore, more closely approximating the method in figure 9. Results for this static study without and with structural support added from the engine can be seen in table 2.

Chassis without structural engine support

Chassis with structural support at engine mounting points

Restraint (green)

Node #:

Figure 10: Standard for defining node number, force and restraint

Force applied on front suspension points

Figure 11: Figure 13: Plot of model superimposed on the deformed shape with forces (purple) and restraints (green)

Static study: displacement plot

3 Node

3 Node

3 Node

Graph 1: Plot of resultant displacements vs. node probed

Graph 3:

Force applied on rear suspension points

Figure 12: Figure 14: Plot of model superimposed on the deformed shape with forces (purple) and restraints (green)

Static study: displacement plot

3 Node

3 Node

Graph 2: Graph 4: Plot of resultant displacements vs. node probed Table 2: Static analysis of chassis without structural support from engine (left column) and with structural support from engine (right column)

Comparing Graph 1 to Graph 3 and Graph 2 to Graph 4, deduce that structural support from the engine reduce the displacement when independently loaded at the front and back suspension mounting points of the chassis. Significantly less deflection can be seen when the chassis is loaded at the back end with the addition of the engine support. These measured displacements are proportional to degrees2. By converting the displacements to degrees and using the applied torques, a torsional stiffness graph could be created by plotting degrees per Newton meter along the length of the chassis. Since the chassis with structural support from the engine reduce the displacement relative to the chassis with unsupported engine mounting points, the torsional stiffness increases. FINITE ELEMENT ANALYSIS (FEA) FEA process Finite Element Analysis was implemented in COSMOSWorks for testing and simulation, and is a method of numerical analysis. The program uses a CAD model which describes geometric data and is the starting point for the FEA. FEA process first involves defeaturing, or the removal of insignificant features, which may or may not be required. This makes the CAD model meshable and saves time in the meshing process. For this report, modification of geometry was not required but the next step, idealization, was made by representing beams

with lines. Subsequently, analysis type, loads, restraints and material properties were applied which formed the mathematical model. The model was then suitable for discretization into a finite element model followed by a numerical analysis for FEA results. Last, but certainly not least, the process involved understanding errors and assumption that can made during the FEA process. Beam Elements on the Chassis The analysis of the chassis was simplified by using beam elements to model weldments. The chassis was created by adding structural members to model the steel tube chassis. As mentioned in the beginning discussion of the Modal Analysis section, because the model contains steel tubes with thin walls and complicated geometries in the corners, solid and shell elements are not suitable. Coupled with long meshing times are meaningless stress results in the corners because sharp re-entrant edges causing stress singularities or infinite stresses. Strains are included into the modal analysis computation and displacements are an integral of stress, one of the primary calculations. Impediments can be avoided with the use of beam elements as used on the chassis. However, one significant drawback with using beam elements is the limitations to model straight lines only. The top of the roll hoops had to be approximated by straight lines for this analysis. Beam elements are represented by curve geometry with 2 dimensions dropped, and assumptions of stress distribution are made in two directions perpendicular to the curve 3.These assumptions include bending stresses distributed linearly, while axial and shear stresses are constant. The tube cross sections are properties only used for area and moment of inertia when deriving a solution to the problem. The solid cross sections of the chassis beams are not meshed, but rather, the wireframe is. This makes it ideal for structural analysis. CONCLUSION Through model analysis and static testing, knowledge has been gained with the use of the engine as a structural member of the chassis. By modelling the frame with a beam mesh and providing support at the engine mounting points, accurate results showed that significant stiffness was added to the chassis. With the additional support, the natural frequency for the first mode of vibration was increased and resulted in torsion in place of the bending mode seen without any support. A Static analysis supported results of increased stiffness showing a significant decrease in deflection when the back of the chassis was loaded. Limitations with COSMOSWorks 2007 Professional included lack of a dynamic analysis but it is now included in the 2008 version of the software and will be a useful simulation and testing tool going into the future. As well, more testing will be carried out on the engine block to discover its limitation as a structural member.

REFERENCES (1) Willliam F. Milliken and Douglas L. Milliken, RACE CAR VEHICLE DYNAMICS, Society of Automotive Engineers Inc., pg 673-667, 1995 Paul M. Kurowski, Engineering Analysis with COSMOSWorks Professional 2007, Schroff Development Corporation, pg 99-104, 2007 Paul M. Kurowski, Engineering Analysis with COSMOSWorks Professional 2007, Schroff Development Corporation, pg 218-219, 2007

(2)

(3)