This paper presents a multidegrees-oI-ireedom non-linear Multibody Dynamic model oI Iront suspension. DOE and Stochastic study are perIormed using Altair HyperStudy. DOE study has helped us in identiIying the critical Variables.
This paper presents a multidegrees-oI-ireedom non-linear Multibody Dynamic model oI Iront suspension. DOE and Stochastic study are perIormed using Altair HyperStudy. DOE study has helped us in identiIying the critical Variables.
This paper presents a multidegrees-oI-ireedom non-linear Multibody Dynamic model oI Iront suspension. DOE and Stochastic study are perIormed using Altair HyperStudy. DOE study has helped us in identiIying the critical Variables.
. ABSTRACT Due to large numbers oI mechanical systems, a high cost oI prototypes, and relentless pressure on cost, saIety and perIormance, the automotive industry has always been at the vanguard oI the use oI Functional Virtual Prototyping technology. A recent popular trend in the automotive industry is called the Six Sigma process-intended to get as close as possible to zero deIects- in part by assuring robust quality and perIormance oI components, subsystems, and systems. To satisIy the demands oI the Six Sigma trend, the automotive industry now uses Functional Virtual Prototyping soItware such as Altair HyperWorks as an important tool in Robust Design. This paper describes new soItware Ior Robust Design, called Altair HyperStudy, and illustrates how the behavior oI a mechanical system model in Altair MotionJiew can now be understood more completely over a range oI system parameters and tolerances. This paper presents a multidegrees-oI-Ireedom non-linear Multibody Dynamic model oI Iront suspension. The model incorporates all sources oI compliance, stiIIness and damping, all with non-linear characteristics. The vehicle model is created in MotionJiew Iormulation. The model is used Ior the purpose oI predicting Kinematics oI the Suspension. Suspension Hard points, Bushing StiIIness and Coil Spring StiIIness were considered as Design variables and Ride Steer, Ride Camber and Roll Center height were considered as Design Responses. DOE and Stochastic study are perIormed using HyperStudy. DOE study has helped us in identiIying the Critical Variables. This has given us the relation oI sensitive variables to the Responses. The InIluencing or critical parameters were taken Irom DOE study and a Stochastic Study is perIormed. A correlation with experimental results Ior K&C Ior Iront suspension subsystems is also presented. Introduction HyperStudy allows you to perIorm Design oI Experiments (DOE), optimization, and stochastic studies in a CAE environment. The objective oI a DOE, or Design oI Experiments, study is to understand how changes to the parameters (design variables) oI a model inIluence its perIormance (response).The objective is to Iind out how the variations in manuIacturing aIIect Ride Steer, Ride Camber and Roll Center oI Height.
Driving Innovation With Enterprise Simulation 2
Figure 1: Front Suspension Model in MotionView The paper consists oI development and the methodology Ior simulation oI a Iront suspension model in MotionJiew, a Multi-body Dynamics tool. The stages involved in this methodology are Front Suspension component modelling and K&C correlation. Problem Definitions. Robust Design reIers to a product design that behaves predictably and reliably, even as manuIacturing tolerances, material properties, or operational parameters change. Within the context oI this paper, Robust Design is combined with Functional Virtual Prototyping to determine the region oI the design space where the perIormance oI the system changes insigniIicantly when design Iactor settings are varied a small amount. There are errors and variations in Design and ManuIacturing. A tolerance band is deIined Ior proper Iunctioning oI the system.
For our study the variation considered are in hard points and bushing stiIIness. Design intent manuIacturing variation allows in hard points are -3 mm and Bushing stiIIness are -10 oI the base StiIIness value. Mahindra SUV Front Suspension Module SLA is modelled in MotionJiew (Iigure 1). Suspension Hard points and Bushing StiIIness are considered as Design variables and the study on Ride Steer, Ride Camber and Roll centre height is carried out. MODELLING SUBSYSTEM MODELLING The hard points (co-ordinates) oI vehicle suspension are taken Irom CAD models. The Cg locations, mass, mass centres oI suspension components, Iront anti roll bar, wheels and steering subsystems are directly taken Irom CAD package. The bushing stiIIness properties Ior suspension are measured. Jounce and rebound stopper properties, anti roll bar stiIIness, torsion bar bush stiIIness oI steering system, are all measured. Static Ride Simulation runs were perIormed on the halI models oI suspension and steering to correlate kinematic and compliance characteristics as shown in table 1 & 2. The correlation is excellent Ior suspension.
Driving Innovation With Enterprise Simulation 3 Table 1: Front Suspension Kinematic Characteristics Comparison For With Test Data Front Suspension Kinematic Characteristics Comparison With Test Data Characteristics " Correlation Ride Steer 95 Roll Steer 88. 5 Ride Camber 92 Roll Camber 90 Roll Center Height 95 Roll Center Height migration 95 Anti Dive 96 Wheelbase Change(Hub) 94 All gradients are initial gradients about 0 wheel travel.
Table 2: Front Suspension Compliance Characteristics Comparison with test Data Front Suspension Compliance Characteristics Comparison With Test Data Characteristics " Correlation Lateral Parallel (In Phase) Toe Change 85.72 Camber Compliance 82.6 Lateral StiIIness TCP 88
Lateral Opposed (Out of Phase) Toe Change 80
Two Wheel Longitudinal (Braking) Toe Change 82 Castor Compliance 98.5
Driving Innovation With Enterprise Simulation 4
Single Wheel Longitudinal (Braking) Toe Change 87
Longitudinal (Force Applied at hub center) Longitudinal stiIIness Hub Centre 87.4
AIter correlation and reIinement oI the subsystems DOE analysis is perIormed. The inIluencing parameters Irom DOE were taken Iurther Ior stochastic study. Double Wishbone Independent suspension DOE and Stochastic Study is perIormed. Table 3 and 4 shows the hard points and Bushing stiIIness considered as design variables. Table 3: Suspension Geometry Hard points considered Ior DOE Study nr tierod ball jt- 9 Spring upr- 8 UCA rear bush- 7 UCA frnt bush- 6 LCA rear bush- 5 LCA frnt bush- 4 Otr tierod ball jt- 3 Upr ball jt- 2 Lwr ball jt- 1 Z (mm) Y (mm) X (mm) Parameters Hard Points (Design variabIes) nr tierod ball jt- 9 Spring upr- 8 UCA rear bush- 7 UCA frnt bush- 6 LCA rear bush- 5 LCA frnt bush- 4 Otr tierod ball jt- 3 Upr ball jt- 2 Lwr ball jt- 1 Z (mm) Y (mm) X (mm) Parameters Hard Points (Design variabIes)
DOE approach can help determine which tolerances are most critical to maintain Ior a robust design.
Results and Discussion: DOE STUDY: SoItware Used: Altair HyperStudy No. oI Design Variables: - 35 a) Hard points 25 Variables I) Variation consider with base value - 3mm b) Bushing StiIIness 10 Variables I) Variation consider with base value - 10 Study DOE a) DOE Class: - Fractional Class b) No. oI Levels: - 2 c) Allocations: L 64 Arrays d) Response: Ride Steer, Ride Camber and Roll center height. Table 5. L 64 ARRAYS 75.13 11.61 2.87 ----- ----- ----- 738.36 200 57.53 11.43 9.01 ----- ----- ----- 738.58 199 67.26 12.06 12.95 ----- ----- ----- 739.56 198 77.77 11.42 2.01 ----- ----- ----- 740.71 197 73.36 12.47 6.23 ----- ----- ----- 740.67 196 I I I I 52.47 10.02 19.74 ----- ----- ----- 740.53 7 66.32 9.94 4.35 ----- ----- ----- 742.30 6 69.93 11.88 5.43 ----- ----- ----- 740.27 5 74.85 13.19 8.60 ----- ----- ----- 737.68 4 79.80 13.81 4.14 ----- ----- ----- 741.21 3 53.83 9.80 10.15 ----- ----- ----- 739.40 2 85.05 13.88 7.28 ----- ----- ----- 741.58 1 RCH Ride Camber Ride Steer LCA rear bush-kx LCA frnt bush-ktz ---------------- Upr baII jt-Ieft-y Lwr baII jt-Ieft-z Run Design Responses Design variabIes 75.13 11.61 2.87 ----- ----- ----- 738.36 200 57.53 11.43 9.01 ----- ----- ----- 738.58 199 67.26 12.06 12.95 ----- ----- ----- 739.56 198 77.77 11.42 2.01 ----- ----- ----- 740.71 197 73.36 12.47 6.23 ----- ----- ----- 740.67 196 I I I I 52.47 10.02 19.74 ----- ----- ----- 740.53 7 66.32 9.94 4.35 ----- ----- ----- 742.30 6 69.93 11.88 5.43 ----- ----- ----- 740.27 5 74.85 13.19 8.60 ----- ----- ----- 737.68 4 79.80 13.81 4.14 ----- ----- ----- 741.21 3 53.83 9.80 10.15 ----- ----- ----- 739.40 2 85.05 13.88 7.28 ----- ----- ----- 741.58 1 RCH Ride Camber Ride Steer LCA rear bush-kx LCA frnt bush-ktz ---------------- Upr baII jt-Ieft-y Lwr baII jt-Ieft-z Run Design Responses Design variabIes
The table 5 shows the L64 Array used as Iractional Iactorial approach used with level as 2. The Correlation between the Design variables and Design responses were obtained as an output Irom hyper study. The correlation matrix is ASCII Iormat which is readable to any spreadsheet Iormat. The Excel is used as spreadsheet. Figure 2, 3, 4 were plotted Irom the correlation matrix. Figures oI Ride Steer, Ride Camber and Roll Center Height show the individual design variables contribution to the responses.
Driving Innovation With Enterprise Simulation 6
Figure 2. Ride Steer (DOE Study)
Figure 3 Ride Camber (DOE Study)
Driving Innovation With Enterprise Simulation 7
Figure 4 Roll Center Height (DOE Study)
Driving Innovation With Enterprise Simulation 8
Table 6: Percentage Variation in Ride Steer, Ride Camber and Roll Center Height with diIIerent sets oI Hard Point and Bushing StiIIness causing variation more than 3. Delhi OIIice Name e-Mail Cell No.
Table 6 shows us the Critical ten Hard points and two bushing StiIIness Ior response variation more than 3. So the input parameters contributing more than 3 in responses are taken Iurther Ior Stochastic Study.
STOCHASTIC STUDY. SoItware Used: Altair HyperStudy No. oI Design Variables: - 12 a) Hard points 10 Variables I) Variation consider with base value - 3mm b) Bushing StiIIness 2 Variables I) Variation consider with base value - 10 Study DOE a) Sampling Type: - Monte-Carlo analysis techniques b) No. oI Runs: - 200 c) Response: Ride Steer, Ride Camber and Roll center height.
Driving Innovation With Enterprise Simulation 9 Table 6 shows the ten hard points and two bushing stiIIness considered Ior stochastic study.
Figure 5: Ride Steer (Stochastic Studv)
Figure 6. Ride Camber (Stochastic Study)
Driving Innovation With Enterprise Simulation 10
Figure 7. Roll Center Height (Stochastic Study) Figure 5, 6, 7 were plotted Irom the correlation matrix. Figures oI Ride Steer, Ride Camber and Roll Center Height show the individual design variables contribution to the responses in stochastic study.
Figure 8: Histogram, PDF, CDF Ior Ride Steer Response (Stochastic Study) The density Iunction can be deIined as the Iunction, which yields the probability that the random variable takes on any one oI its admissible values. In Iigure 8 the histogram base oI each vertical rectangle represents the change in the parameter or interval oI liIe and the height is no. oI occurrence or no. oI Iailures observed during the no. oI runs or liIe. It is noted that the Irequency oI occurrence is not the same Ior all liIe intervals. For Ride Steer there is high concentration oI occurrence between slopes oI 5 deg/m to 7 deg/m. II a much large sample were tested, a smooth curve could be drawn through the peaks oI the rectangles. The probability curve is drawn through the peaks oI the rectangle bars. The curve represents the no. oI occurrences. It is noted that the probability oI occurrences are not constant over the whole range oI values attained by the variable. And even the value oI probability is not a constant; there seems to be a deIinite trend involved. PDF is one oI the ways to represent the data, there is another method oI plotting the data, which is easier to evaluate. This method involves plotting oI the cumulative distribution Iunction (CDF) verses the random response. CDF is area under the Irequency curve. CDF help us to understand the risk involved in a particular decision. For obtaining Ride steer less than 7.5 deg/m the CDF Is .5 or 50 population may Iall under it.
Driving Innovation With Enterprise Simulation 11
Table 7. Statistical Summary (Stochastic Study)
Table 7 shows the Statistical Summary obtained as a HyperStudy output. The data helps in determine the reliability oI the Ride Steer, Ride camber and Roll Center Height. It also provide us the statistical data like Average Deviation, Standard deviation, Variance, Covariance , Skew ness, RMS and Min and Max values obtained during the simulation. Table 8: Reliability more than 95 %
It is clear Irom the above data that alterations to the coordinate will make a signiIicant change to the Ride Steer, Ride Camber and Roll Center Height in this particular suspension. ThereIore, these parameters must be closely controlled throughout design, component manuIacturing, system assembly, and Iield maintenance to minimize in the interest oI retarding tire wear and Vehicle Dynamics aspect. Please note that this suspension is greatly simpliIied and this conclusion is not to be generalized to any other independent Iront suspension. Also note that many additional Iactors may contribute to Ride Steer, Ride Camber and Roll Center Height these would need to be considered in a more complete study.
Table 9 shows that to get Ride Steer more than 6 deg/m the reliability is 61.5 , Ior Ride camber more than 12 deg/m the reliability is 39.5 and Ior Roll Center Height more than 80 mm the Reliability is 13.5.
Table 8 shows Ior Reliability more than 95 Ride Steer is less than 14 deg/m, Ride Camber is less than 15.027 deg/m and Roll Center height is less than 85 mm.
Table 9: Reliability more than acceptance criteria
Driving Innovation With Enterprise Simulation 12
Conclusion A simulation tool Ior analyzing Stochastic and DOE Study has been created, by using Altair MotionJiew and Altair HyperStudy. DOE study helps us in identiIying the Critical Variables. This has given us the relation oI sensitive variables to the Responses. It can also be used in design adjustment into suspension geometry that will help to make a diIIerent track Ior better Ride and Handling. The InIluencing parameters were taken Irom DOE study and a Stochastic Study is perIormed. The Study shows that the manuIacturing variation has a signiIicant inIluence on the responses. Like the Ride Steer base design value is 7.14 deg/m and the variation may alter it Irom 19.73 deg/m to .191 deg/m. To get the Ride steer more than 6 deg/m the probability oI Iailure is 38.5 . This paper demonstrates how a typical automotive design problem, an independent suspension, can be greatly inIluenced by Robust Design thinking. Using design oI experiments techniques and Stochastic Study on a Iunctional virtual prototype to understand how to achieve the most eIIective balance between competing design parameters and perIormance metrics, and meet today`s Six Sigma requirements Ior product quality. Future Scope 1) Look at interaction with all geometry hardpoints & see how they aIIect the response curve as a whole. 2) PerIorm an Optimization study to optimize the best possible suspension geometry and Bushing stiIIness to achieve the target response. 3) Use oI Advanced DOE Class and Stochastic Sampling methods will be checked. ACKNOWLEDGMENTS I would like to express my gratitude to Mr. Adam Aslam, DGM CAE, Product Development, Mahindra and Mahindra Ior the granting permission to publish this work and the timely support and guidance provide by him. REFERENCES 1. Monte Carlo Simulation 'Mahindra and Mahindra Practitioner`s guide. 2. James E. Gentle. Random Number Generation and Monte Carlo Methods. New York:Springer-Verlag, 1998. 3. DOE and Optimization using MotionView HyperStudy document.