Page 505 - Automotive Engineering Powertrain Chassis System and Vehicle Body
P. 505

Modelling and assembly of the full vehicle   C HAPTER 15.1


                                                      Body slip angle estimate
                                 0.05
                                 0.045
                                 0.04
                                 0.035
                                 0.03
                                 0.025
                                 0.02
                                 0.015
                                 0.01
                                 0.005
                                    0
                                    50   40   30  20    10    0   10   20   30   40   50

                                                      Body slip angle estimate
                                 0.05
                                0.045
                                 0.04
                                0.035
                                 0.03
                                0.025
                                 0.02
                                0.015
                                 0.01
                                0.005
                                    0
                                    50  40    30   20   10    0   10   20   30   40   50

           Fig. 15.1-46 Probability density for body slip angle estimates – Greece 2002 (top) and Germany 2002 (bottom) for Petter Solberg, Subaru
           World Rally Team.

           15.1.13.4 Two-loop driver model                    dynamics task, however, the automotive engineer will
                                                              want to carry out simulations before the design has
           For general use, the authors favour a simple and robust  progressed to such an advanced state.
           two-loop driver model comprising a path follower and  In this case study the level of vehicle modelling detail
           spin catcher, with a separate speed control as appro-  required to simulate a ‘full vehicle’ handling manoeuvre
           priate to the task at hand. Fig. 15.1-48 shows such  will be explored. We will consider a 100 km/h double lane
           a model.                                           change manoeuvre, as a start. The test procedure for the
                                                              double lane change manoeuvre is shown schematically in
           15.1.14 Case study 7 – comparison                  Fig. 15.1-49.
                                                                For the simulations performed in the case study the
           of full vehicle handling models                    measured steering wheel inputs from a test vehicle have

                                                              been extractedand applied asatimedependenthandwheel
           As mentioned at the start of this chapter the use of  rotation (Fig. 15.1-50) as described in Section 15.1.12.3.
           modern multibody systems software provides users with  To appreciate the use of computer simulations to
           the capability to develop a model of a full vehicle that  represent this manoeuvre an example of the super-
           incorporates all the major vehicle subsystems. Clearly  imposed animated wireframe graphical outputs for this
           the development of such a model is dependent on the  simulation is given in Fig. 15.1-51.
           stage of vehicle design and the availability of the data  In this study the influence of suspension modelling on
           needed to model all the subsystems. For the vehicle  the accuracy of the simulation outputs is initially


                                                                                                      513
   500   501   502   503   504   505   506   507   508   509   510