Page 333 - Cam Design Handbook
P. 333

THB11  9/19/03  7:33 PM  Page 321

                                    CAM SYSTEM MODELING                    321

            For simple systems like this suspension model and many cam-follower systems, the vibra-
            tional modes can often be analyzed intuitively and this information used to simplify models.



            11.2.2 Model Sufficiency and Model Reduction
            Natural frequencies provide some insight into the applicability or sufficiency of the model.
            For mechanical systems, if the system is forced at the natural frequency, resonance (the
            maximum  amplification  of  the  forcing  input)  will  occur.  Intuitively,  since  the  system
            naturally  tends  to  vibrate  at  certain  frequencies  and  in  certain  modes,  motion  at  these
            frequencies is, in a sense, easier to initiate and maintain. Resonance is an important feature
            to capture in a mathematical model since it is not uncommon for the ratio of system motion
            to input magnitude to increase by an order of magnitude or more at resonance if the system
            is lightly damped. In a cam follower system, therefore, if the cam rotates at speeds up to
            3000rpm (or 50Hz), the dynamic model of the follower system must, as a bare minimum,
            include all natural frequencies up to 50Hz to capture the basic motion of the follower with
            reasonable fidelity. Similarly, if experiments show a resonance at 1000Hz but the highest
            natural frequency predicted by the model is 100Hz, the model requires additional refine-
            ment. It is “too simple” to predict the observed behavior.
               Continuing with the example of the suspension in Fig. 11.2, this two-mass model is
            sufficient for simulating the vehicle response to road excitations in the range of around 0
            to 15Hz. If a test driver complains about road vibrations being amplified around 50 or 60
            Hz, the model is too simple, since the mass responsible for this vibration has been lumped
            into either the body or the tire in the two-mass model. Conversely, if the goal of the mod-
            eling process is to capture the fundamental response of the vehicle to vibrations in the
            range of 1Hz, information about the resonance at 11Hz is extraneous and the model is
            more complex than needed. Of course the more complex model could still be used. Instead
            of retaining the additional complexity in the model, however, it is often beneficial to look
            for a simpler model that retains the important characteristics of the original model. This
            process is known as model reduction. In this case, the objective of the reduction is to create
            a model with a single natural frequency that predicts the resonance associated with the
            low-frequency mode (the sprung mass bouncing on the tire). The task then is to choose
            an equivalent mass and stiffness for a simplified model with a single natural frequency
            that resembles Fig. 11.3a. It cannot be emphasized enough that since any model will be
            an approximation of the real system, there is no unique or correct way to simplify. Sim-
            plification is again part of the art of engineering modeling.
               The idea of natural frequencies can provide some guidance, however. Since natural fre-
            quencies are associated with masses and stiffnesses (more formally, with mass and stiff-
            ness matrices) and the goal is to remove the higher natural frequencies, the model can be
            simplified by ignoring masses or replacing certain spring elements with rigid links (thus
            assuming  infinite  stiffness).  These  assumptions  remove  degrees  of  freedom  from  the
            system and reduce the number of natural frequencies and the complexity of the resulting
            equations of motion. It is also possible, in the manner suggested by Chen (1982), to lump
            masses and stiffnesses together without explicitly assuming any of these to be zero or infi-
            nite. Three possible ways to apply these concepts to the suspension system are:

               Approach 1: Assume Infinite Tire Stiffness. Assume that the stiffer spring (the tire)
            is rigid and use the mass of the car and the stiffness of the suspension spring to determine
            natural frequency. In this model, the mass of the tire and the degree of freedom associ-
            ated with the tire are neglected since the rigid spring prevents the tire from moving rela-
            tive to the road. This simplification is illustrated in Fig. 11.3b. The natural frequency in
            this case is given by
   328   329   330   331   332   333   334   335   336   337   338