This computational study focuses on the dilution process and cyclic flow variation of a rotary engine system. In light load regions, rotary engines tend to become less stable than reciprocating engines. As visualization experiments have shown[ 1], the flow fields inside rotary engines themselves vary from cycle to cycle. Under certain conditions, even the dominant vortices during intake-compression process entirely change their directions very often. Fuel distribution inside rotary engines and the combustion may hence not be the same, even if operating conditions are kept constant. Furthermore, dilution gas, which thins fresh air-fuel mixture, makes engines’ stability worse by slowing combustion process, thereby leading to higher fuel consumption and other performances[ 2]. The two factors therefore hold the key to improve the performance.
More than a dozen numerical studies of rotary engines have been carried out under various conditions, and their focus has been mostly on the intake or combustion process[ 3, 4]. Simulations of the dilution process and cyclic flow variations require the following additional computational facilities: (1) The computational domains should include the two ports and all three combustion chambers. (2) The simulations have to be robust and accurate enough to track the exhaust gas getting into exhaust and intake ports.
To simulate the phenomena, the authors have developed a two-dimensional compressible Navier-Stokes simulation code whose calculation domain includes all three chambers and both the ports and a large tank connects them, so that the dilution process takes place on its own. This configuration makes it also possible to avoid any artificial boundary value settings at the open side of the ports, which may change nature of the intake-exhaust interactions. The Chakraverthy third-order TVD scheme is employed to meet the second condition. Neither air-fuel mixing nor combustion process is included. Numerically heated air, however, well plays exhaust gas, since its major role is to give high pressure and temperature gradients near the intake and exhaust port.
Keeping the engine speed at 100 revolutions per second, the calculations are carried out with two different throttle valve widths to see effects of change in load. In each case, after three consecutive motoring cycles, three artificial firing cycles are given in a row. Not only do the results show thermodynamic or macroscopic adequacies of the simulation, but also unveil mechanisms of the dilution process and the cyclic flow variation of rotary engines.