Calculation of the Path of Contact under Load

Whenever a machine is designed, there are parts which need to be engineered precisely for strength, accuracy, and reliability. This means that stresses in this part must be calculated. There are two different approaches to solve this problem, the very general finite element method (FEM), and the specific classical methods available for most of the common machine parts. The classical methods are tailored to one specific type of part only, e.g. bolts, gears or bearings. The advantage of these methods is their fast and easy application. In many cases no classical method is available, for instance for housings. In these cases the application of FEM is the only possibility. In other cases the application of FEM would be much to expensive, like for a key and key way on a shaft. This paper will examine a problem for which it is not apparent which of the two methods is superior: The calculation of the stress in the root area and in the flank of a cylindrical gear. To calculate the stress in the tooth of gear during meshing, the contact path under load is calculated. This means a contact problem must be solved: The number of tooth pairs in contact varies by one during the meshing, most often it changes from one pair to two pairs in contact. This effect causes the total stiffness in the engagement to change periodically (see figure 1). The teeth themselves are deflected due to the torque applied. This shifts the point of change from one to two pairs in contact a small amount and leads to a premature contact. Formulae for all values must be derived for a classical method. For the calculation of the stiffness, a model according to Peterson is used. This model covers the deflection of the teeth, the bending of the teeth in the wheel body, the Hertzian pressure, and the deformation due to shearing. The gear is cut into several transverse sections and the stiffness is calculated for these slices. For a spur gear the stiffness is multiplied by the width which leads to the final value. For a helical gear the beginning and end of contact of the slices is dependent upon the position of the slice in across the tooth width. The final course of the total stiffness is calculated by integrating the stiffness functions for the slices over the width, with increasing delay of the begin of contact. Figure 1 shows a graphical representation of the model: a spring is fixed on the path of contact, which means that is located on the common tangent of the two base circles of the gears. This spring has a periodically changing stiffness c(t). In the simulation of the meshing the deflection of the teeth is given by the normal force applied to a single tooth divided by the stiffness. Since the point where the force is applied varies in the height direction, the stiffness will also depend on the meshing position. Further, if the second pair teeth comes into contact the stiffness increases sharply, so the deflection of the first pair of teeth is reduced. To find the correct point of contact, an iteration must be performed. The reward for all this effort is the calculation of the real path of contact under load, the course of the normal force on the flanks, the stress in the root areas of the teeth, the Hertzian pressure and the transmission error. The example to be considered now is a spur gear pair with module 1mm (DP 25.4 1/in) and 17 to 39 teeth. Two variants are calculated. The first without tip relief and the second variant with a tip relief of 10 microns on the pinion and 17 microns on the gear. Figure 3 shows the effect of the tip relief: in the top left diagram the path of contact for the full involute gears is shown. In the middle part the path is a straight line. In the section of contact start and end, however, the line is curved, and the contact follows the tip circle of the pinion or the gear. This is the region of premature (or prolonged) contact. The tip
1/5
4. Juli 2007 W:\Artikel-Papers-Konferenzen\040-FEM-Vergleich\Path-of-Contact-under-Load-Format-KISSsoft-E.doc

Calculation of the Path of Contact under Load

relief is designed to compensate the deflection of the teeth and thus eliminate the premature contact. This, however, only works for a specific torque applied exactly. The right column of diagrams in figure 3 shows the influence of the tip relief. The path of contact is almost a straight again. The only deviation is a nick in the region of the beginning of the tip relief. This marks the rapid change of the pressure angle at this point. The course of the normal force for the gears without tip relief shows a typical picture for spur gears: in the middle of the contact path only one pair of gears takes the full torque. Before and after that, the normal force is shared over two pairs of teeth, thus it is only about 50% of the maximum value in the middle. The gears with tip relief have only one pair of teeth in contact most of the time, so here the normal force is nearly constant but on the same level as the maximum of the gears without tip relief. Nevertheless, the maximum pressure on the flank is 20% less with tip relief since the premature contact leads to a contact shock with very high pressure. This is illustrated in the diagram showing the stresses and the Hertzian pressure. Finally, the amplitude of the transmission error remains the same with this type of tip relief. The tip relief results in a smoother course so that the higher frequencies are reduced. This leads to less acceleration and hence smaller forces induced by the transmission error. To compare this results with FEM calculations the same gears were calculated with ANSYS. When a FEM is applied to the problem, the geometry of the gears must be given. Figure 2 shows the example gears. This geometry is now divided into finite elements, the so-called net. The results are presented in figure 4. Both methods lead to very similar results. For the Hertzian pressure, the FEM results tend to zig-zag more. This is mainly caused by the fact that the stress defined is given for a single point on a grid. Since the real contact point is usually somewhere between two grid-points, the real maximum stress on the flank is usually larger than the plotted result. Since the calculated results are very similar, the main difference between the two methods is the effort to achieve these results: It took two days to get the FE model set up, calculate the stresses and extract them for presentation. With KISSsoft, the same task was accomplished in 15 minutes. Moreover, each variant for the tooth form such as different amount of tip relief, different geometry (such as changed addendum modification), or different tooling takes only a few minutes to analyse. This demonstrates the advantages of the classical approach.

Figure 1: Typical course of the stiffness, and the model with a spring.

2/5

Figure 2: The gears of the example as calculated in ANSYS

3/5

Figure 3: Results from the classical calculation

4/5

Figure 4: Comparison of the results of the FE calculation and the results from KISSsoft.

5/5