Prediction of air–fuel ratio control of a large-bore natural gas engine using computational fluid dynamic modeling of reed valve dynamics

Abstract

Air–fuel ratio control of large-bore, two-stroke, natural gas engines, typically used in the oil and gas field, is critically important to maintain stable operation and emission compliance. Many two-stroke applications rely on reed valves to control air and gas induction, which can involve complicated gas flow behavior; standard gas dynamic relationships are typically insufficient to characterize such behavior. Computational fluid dynamic simulations offer the needed complexity, but even so the computational fluid dynamic models, as shown in this work, must also capture the dynamic behavior of the valves themselves. The current work reports on a computational fluid dynamics–based model representing this type of large-bore, two-stroke, natural gas engine using commercially available computational fluid dynamic software. The engine under study is an AJAX E-565 with rated power of 30 kW (40 HP), a bore of 216 mm (8½″), and a stroke of 254 mm (10″). The large engine geometry makes a relatively large solution domain, hence requiring an intense, time-consuming numerical investigation. This large-bore engine works at a rated speed of 525 RPM with a compression ratio of 6 to 1. Two approaches to modeling the reed valve are investigated: (1) a pressure difference–based user-defined function and (2) a fluid–structure interaction user-defined function. The pressure difference–based user-defined function captures reed valve behavior in a simple, binary fashion (i.e. valves are either open or closed based on the pressure difference between the intake pipe and the engine’s stuffing box). The fluid–structure interaction user-defined function, however, predicts the motion of the reed valve strips based on fluid and body motions; although a more complex solution, the fluid–structure interaction user-defined function accurately predicts the engine’s gas exchange process. In this article, the results of each method are presented and validated to show that the added complexity is necessary to properly predict (and thus eventually improve) the engine’s air–fuel ratio control.