微型汽車變速器設(shè)計(jì)【兩軸式四檔手動(dòng)】,兩軸式四檔手動(dòng),微型汽車,變速器,設(shè)計(jì),兩軸式四檔,手動(dòng) Experimental analysis of a composite automotive suspension arm M. PINFOLD and G. CALVERT (University of Warwick/Rover Group Gaydon, UK) Received 11 November 1992; revised 26 March 1993 In applications where weight saving and parts integration can be achieved, the Rover Group has been investigating the design and manufacture of components from composite materials. The methods used in the different steps in the design- to-manufacture cycle in the high volume automotive industry are relatively well known for a steel component, but are not so well established for a composite component. A design methodology for composites has been emerging in which a principal procedure is design analysis. One of the most established methods of analysisis that using the finite element technique, and this is being supplemented with experimental tests on prototypes using photoelastic analysis and stress pat- tern analysis by thermal emission, coupled with conventional strain gauge moni- toring. Little work has been undertaken to correlate the results obtained from these different test methods and to compare the results with measurements made on an actual component. This paper presents some of the work undertaken concerning the analysis and testing of a composite automotive suspension arm. The results obtained from the three different analysis techniques are compared with experi- mental test results, and their accuracy is discussed. Key words: autmotive suspension arm; stress analysis; finite element method; photoelastic analysis; SPA TE; strain gauges; sheet moulding compound Sol and de Wilde state that composite materials have been used increasingly as structural materials. A reason for this., is that composite materials have high strength to weight and high stiffness to weight ratios which can significantly reduce the weight of a structure. Perhaps the most important feature ofcomposite materials is that their mechanical p:operties can be tailored to meet a specific criterion. However, Johnson et al? suggest that composite design, analysis and fabrication technology must undergo major developments and successful demonstrations before significant structural components will be incorporated in production automobiles and trucks. Composite materials have to compete with steel within the engineering environment. Within the automotive industry this requires a certain amount of technology transfer from places such as the Advanced Technology Centre at the University of Warwick, which work with material manufacturers and automotive engineers to enable understanding about these materials as an alter- native to the traditional materials such as steel. If com- posites are to compete with traditional materials in a real sense, then automotive designers need to be fully aware 0010-4361/94/010059-05 of their strengths and limitations so that they can be one of perhaps many options considered at the concept stage of the design. For this to happen automotive engineers need to catch up on the techniques of designing, testing and manufacturing components from composites. This will include understanding how various methods such as finite element (FE) analysis, stress pattern analysis by thermal emission (SPATE) and photoelastic analysis can be applied to composite components in their design and development. Thus far little work appears to have been undertaken to study whether the results obtained from these different analysis methods correlate with one another or with actual experimental results obtained from testing a real component. In order to study the application and corre- lation of the different analysis methods to composite materials, a composite component - an automotive lower suspension arm - was manufactured. This com- posite component was analysed by the three methods described above and also tested under realistic loading conditions, with experimental results being obtained from strain gauges. 1994 Butterworth-Heinemann ktd COMPOSITES . VOLUME 25 . NUMBER 1 . 1994 59 , B a l l J o i nt Housing Fig. 1 The composite suspension arm DESIGN The existing steel lower suspension arm consists of nine pieces welded together whilst the re-designed composite component-which can be seen in Fig. 1-is a single moulded part. The material used to manufacture the suspension arm was a sheet moulding compound (SMC), comprising a polyester resin bonding agent with a 30% content of randomly arranged short glass fibres and cal- cium carbonate fiIler. The weight of the steel suspension arm is 2.53 kg whilst the re-designed SMC suspension arm complete with bushes and ball joint weighs 1.5 kg. The material properties used for the composite suspension arm in these analyses, obtained from tests carried out at Rovers materials laboratory, were Youngs modulus = 10.5 GPa, Poissons ratio = 0.26 and density = 1.8 x 10 -6 kg mm -3. EXPERIMENTAL TECHNIQUES Prior to undertaking experimental analysis of an actual engineering component, some initial validation work was required to gain confidence in the techniques when applied to sheet moulding compound. Therefore, fiat plates, beams and discs constructed from SMC were ana- lysed under various loading conditions before progress- ing on to the designed component. Most validation tests were carried out using strain- gauged specimens to correlate with the finite element analysis results. Although it is recognized that SMC is not an isotropie material due to some fibre orientation during processing, for the purposes of analysis the mater- ial was assumed to be isotropic. Also, when the actual SMC suspension arm was cut up and examined, signifi- cant fibre distribution was observed in the ribs. It is felt that the correlation between the experimental and analy- sis results validated this assumption in the case of this particular component. Strain gauge tests Before undertaking the experimental test work, the com- posite component was mounted via its rubber mounting bushes onto a relatively infinitely stiff structure. It is very difficult to cover all of the loading conditions when con- ducting experimental tests and thus a worst-case scenario is usually assumed. The worst-case loading condition on suspension components is known as pot-hole brake. This attempts to simulate the vehicle falling into a deep pot-hole at 30 mph with the brakes fully applied at the point of impact. The resultant fore/aft and lateral loads are then calculated based on the weight and velocity of the vehicle. Due to the limitations of the test rig the full pot-hole loads could not be applied to the component, and thus reduced loads with the same resultant direction as the pot-hole loads were applied and the results scaled. The loads applied for the full pot-hole brake case were 24.2 kN in X and 8.2 kN in Y, and for the reduced load case were 5.9 kN in X and 2.02 kN in Y - see Fig. 1. The strain gauges used consisted of six three-axis rosette gauges and 13 single-grid gauges, with 2.5 mm grid lengths, chosen to fit into the radii of the component in an attempt to measure the maximum strain, Gauges were situated near the ball joint housing, where the loads were applied, and around the radii of the body mounting bushes, where the component would be mounted to the car subframe. Additional strain gauges were situated on some of the strengthening ribs and close to the anti-roll bar mounting position. SPA TE analysis Stress pattern analysis by thermal emission (SPATE) can be used to determine the surface stresses of components by studying the small changes in temperature due to cyclic loading conditions. SPATE equipment comprises a detector unit with scanning head, an analogue signal processing unit and a digital electronic data unit. The system works by detecting the minute temperature changes which occur when a structure is cyclically loaded. The infra-red detector scans the structure and correlates the measured output with a reference signal from the loading system. An electronic data processing system correlates the detected stress-induced thermal fluctuations with the loading reference signal. A colour contour map of the sum of the principal stresses (cr + 4) is then plotted, together with a bar chart giving actual values. This correlation of signals effectively eliminates all signal frequencies other than those caused by the loading system, i.e., all ambient temperature fluctua- tions. The SPATE system has a temperature resolution of 0.001C, and a spatial resolution of less than I mm. This type of analysis has been shown by a number of authors TM to also be applicable to non-isotropic mater- ials such as composites, and the small errors (6%) demonstrated from such studies when compared with theoretical or FE results are felt to be due to inaccuracies in the material data used 4. It is apparent from the studies undertaken that the use of thermoelastic stress analysis to evaluate stresses and strains in anisotropic composite materials is more complex than for isotropic materials. However, it has been shown that the technique can provide valuable qualitative information on stress distri- bution, effects of surface defects and crack growth predictions. It has also been demonstrated that, given accurate details of material properties including expan- sion coefficients, quantitative results can be obtained depending upon the degree of anisotropy of the material. Prior to undertaking a full SPATE analysis of the suspen- sion arm it was necessary to determine a calibration factor for the material used. This can be achieved in two ways, either by loading a disc of the material in compres- sion and comparing the SPATE output with the theoreti- 60 COMPOSITES. NUMBER 1 . 1994 cal solution, or by strain gauging directly onto the component in an area of even stress distribution, thereby obtaining a direct comparison with the SPATE output. Both methods were used in this case, but direct calib- ration with strain gauges can overcome a lot of the problems, thus allowing significant information to be obtained from the SPATE output. Photoelastic analysis The majority of photoelastic work investigating the mac- romechanical behaviour of composite materials has been undertaken using photoelastic coating techniques. This is done to avoid the complexities of constructing a photo- elastic model with anisotropic properties and thus con- structing a composite like the original which would lose its transparency and could not be analysed. However, for complex fibre lay-ups this would be the only method of conducting photoelastic analysis, and thus some research has been undertaken investigating the use of the actual composites j7-30. Reasonable results have been obtained from such analyses, but with limitations due to the neces- sity for transparency within the composite. However, the composite component considered in this study was manufactured from SMC and the material was assumed to be isotropic, thus simplifying the creation of a photo- elastic model. A three-dimensional epoxy resin model of the suspension arm was constructed for the photoelastic analysis. The model was then loaded in a representative manner, with scaled-down loads, and subjected to a stress freezing cycle. This involves heating the model up to the mater- ials glass transition temperature, at which point the Youngs modulus changes, and the model deforms under the applied loads. The model is then slowly cooled, avoiding any uneven temperature distribution which could result in unwanted thermal stresses. During the cooling cycle the deformations and stresses are locked into the model. When viewed under polarized light the three-dimensional model is a jumble of interference fringes. In order to determine both magnitude and direc- tion of the principal stresses at any point, a slice is removed and observed under polarized light. By count- ing the fringes the stresses in the model can be calculated and converted into actual stress in the component. This is done by means of proportionality, between the model and component materials, and the loading and dimensio- nal parameters. The lower suspension arm is mounted to the rest of the car via rubber mounting bushes. Investigations were carried out as to the possibility of modelling these mounting bushes. However, experiments with silicon and foam rubbers showed that the required scaled-down stiffness of the bushes during stress freezing at elevated temperatures could not be maintained. The photoelastic analysis thus assumed that the suspension arm was solidly mounted. FINITE ELEMENT ANAL YSIS The composite suspension arm was modelled using approximately 1300 of the STIF45 ANSYS solid ele- ments. The suspension arm is mounted to the subframe via rubber mounting bushes; these were modelled with spring elements to represent the stiffness of the bushes and to create a realistic load distribution throughout the component. Loads were applied to the FE model via beam elements at the ball joint. Three load cases were analysed using the ANSYS FE software. The first load case simulated the full pot-hole brake loads. The second simulated the reduced load used in the tests due to the limitations of the test rig, to enable comparisons with the results from the experimental strain gauge analysis. These two load cases used spring elements to simulate the stiffness of the rubber mounting bushes. The third load case again used the reduced loads but this time omitted the spring elements; i.e., the suspen- sion arm was modelled as being solidly mounted. This third load case was required to correlate with the SPATE and photoelastic analyses. RESUL TS Finite element analysis Analysis of the suspension arm showed that the maxi- mum equivalent stress in the component for the load case considered is very close to the ultimate tensile strength of the proposed material for the pot-hole loading condition, which is the worst loading condition. This means that the component may need to be manufactured from a differ- ent material, or that other materials need to be posit- ioned in areas of high stress to strengthen the component locally. Due to constraints upon the amount of computer disc space available, the number of elements used within the FE model was relatively low and thus the size of the elements within the area of the radii around the body mounting bushes was too large to detect any large stress concentrations. Also, the types of element used around these areas, due to the geometry of the component, were a mixture of brick, wedge and tetrahedral. The latter shape tends to be too stiff to give good results and is not recommended. If more detailed results were required in these areas, then these radii would have to be modelled in greater detail with more and smaller elements in the areas of high stress gradient. Photoelastic analysis The analysis of the photoelastic model of the suspension arm was undertaken assuming that the directions of the maximum principal stresses lay in a horizontal plane through the model in the direction of the fore/aft load. Whilst this is not strictly true in practice due to local geometry effects in certain areas, the assumption gave sufficiently accurate results. If obvious discrepancies were found in particular areas then it was possible to take slices from different planes. Maximum stresses were seen to occur in the vicinity of the ball joint housing and the body mounts. Due to the ability of photoelastic analysis to pinpoint very small areas of high stress, the maximum stress values given by photoelasticity tended to be higher than the strain gauge results. For example, maximum stress levels in the internal radius of the leading body mount were found to be 43 MPa compared with a SPATE value of 26 MPa. This difference can be explained by examin- ing the slice taken through the photoelastic model which shows that the maximum stress only occurs at a position COMPOSITES. NUMBER 1 . 1994 61 Table 1. Stress results (MPa) for full load con- ditions Position Strain gauges FE Photoelastic Ball joint housing 176 165 176 spanning 3 mm and that the stress values either side of the maximum are around 25 MPa. SPA TE analysis The initial SPATE scan showed large bands of stress running across the mounting areas and some confusion as to whether these areas were in tension or compression. The problem was identified as excessive movement in the suspension arm body mounting positions due to distor- tion of the rubber bushes as experienced in the strain gauge tests. SPATE is equipped with a motion compen- sator device if required, which deflects the scanning mirrors inside the detector in time with the oscillations of the test-piece, thereby eliminating the movement. How- ever, in this particular case, the geometry and direction of movement could not be eliminated over the entire area at the same time, and thus it was necessary to remove the rubber bushes and to replace them with aluminium ones. The SPATE analysis was repeated with the solid bushes and showed areas of high tensile stress (26 MPa) along the leading edge and around the inner radius of the leading body mounting position. Unfortunately, no SPATE analysis could be undertaken at the ball joint end of the component as it was obscured by the large loading adaptor required to fit the hydraulic actuator supplying the cyclic loading. COMPARISON OF RESULTS It should be clarified that the stress values quoted in the tables from the strain gauge results were calculated from the rosette gauges to give a value of maximum principal stress. The photoelastic analysis also gives maximum principal stresses unless the values are taken inboard of a free edge in which case they are differences in principal stresses (o.- o-,). SPATE analysis gives an output in the form of the summation of the principal stresses (or. + a2) whereas the FE output can be in any form required (in this case yon Mises). Due to the geometry of the compo- nent and the way in which the loads were applied, the values of or2 and cr 3 were always small, and thus direct comparisons could be made between the different analy- sis methods without further conversion. Table l compares the results obtained for the maximum pot-hole load conditions. The maximum stress values all occur at the ball joint area and correlate very well. These resultant stresses for the strain gauges and photoelasti- city were calculated from the results obtained for the reduced load. The model stress was multiplied by a load- ing factor as the ratio between the fore/aft and lateral loading remained constant and in the same proportion as the full pot-hole brake load applied to the suspension arlTI. The results of the analyses undertaken with reduced Table 2. Stress results (MPa) for redTJced loads with mounting bushes Position Strain gauges FE Inner radius of body 25 20 mount Ball joint housing 49 40 Table 3. Stress results (MPa) for reduced loads without mounting bushes Position FE SPATE Photoelastic Inner radius of body 22 mount Ball joint housing 30 26 43 (25) 42 (25) loading but with the mounting bushes included can be seen in Table 2. Table 3 presents the results of the analyses undertaken with reduced loading and without the mounting bushes being used. The stress given by the photoelastic analysis is concentrated at a very small point whereas the stress given by FE analysis is averaged over a relatively large area. In the case of the photoelastic results, an average of the nominal stresses on both sides of the concen- tration point is also quoted in brackets to give a fairer comparison. Compared with the strain gauge results, the values given by SPATE are very similar for the maximum stress. In theory SPATE should be more effective than strain gauges when investigating stress concentration effects, as it is measuring values over a smaller area depending upon its distance from the object during scanning. In this case the measurement point of SPATE was set at I mm diameter compared with a 2.5 mm grid le