Hydrodynamic torque converter operating under dynamic load
Abstract
Unsteady interaction phenomena created by the influence of the blade spacing have been reported in earlier experiments and CFD. However cyclic load changes in start-up and slow-down of the hydrodynamic torque converter operation have been beyond access to the current flow field calculation methods due to the extensive computer run time and memory requirements in the application of time dependent Navier-Stokes solvers to acceleration and deceleration. Therefore computations based on the use of 1D mean-line flow simulation supported by optimised flow correction coefficients tuned in rig-test experiments enable to obtain solutions for engineering-type technical problems.
Keywords
Download Options
Conclusion
In the steady-state CFD analysis by M.Wollnik et al. [9] performance in computational effort required 1.4 Million elements. in the unsteady case. In the present 3D CFD simulation only 676 711 elements had been used for the simulation of a single blade channel in figure 8 and 9. Computer time of the order of one week central processor time had been required for the run indicated in figure 9 from =7 down to =6 during 0.24 seconds. Once the optimisation procedure has set the calibration coefficients correcting flow angle and pressure loss of the 1D method in the curve fit to preceding rig tests, the 1D performance prediction is applicable to the prediction of any torque converter of identical blade shapes and fluid mass at moderate computing times with better accuracy for the situation of acceleration than of deceleration (figures 7 and 8).
At present the truely unsteady analysis on the basis of 3D Navier-Stokes solvers for the torque converter case does not seem economically justified for engineering applications because of excessive computer time requirement.