This paper reports on the first phase of an experimental evaluation of four different methods for the mathematical modelling of unsteady gas flow in a pipe system containing an area discontinuity. The four methods under investigation are the non-homentropic method of characteristics, the two-step Lax-Wendroff method with flux corrected transport, the Harten-Lax-Leer upstream difference method and the GPB finite system method.
The experimentation is conducted using the QUB SP (single-pulse) pressure wave generator consisting of a cylinder, connected via a sliding valve to a long duct. The pressure waves it creates closely mimic those to be found in i.c. engines. The initial cylinder pressure may be set to simulate either an induction or an exhaust process. Various ducts are attached to the pressure wave generator to simulate both sudden and gradual area changes. Each duct is sufficiently long as to permit pressure wave observation without superposition effects. Pressure and temperature are recorded at various locations in the apparatus and stored using a high speed data acquisition system.
A computer simulation of the test apparatus has been written for each of the four theoretical methods incorporating the Queen's University of Belfast (QUB) non-isentropic treatment of the boundary conditions. The accuracy of each prediction method is then correlated with the experimental results.
The new non-isentropic formulation of the sudden area change boundary condition is demonstrated to be accurate for each of the sudden enlargements and contractions tested. It is shown that the non-homentropic method of characteristics is unsuitable for predicting flow in steep diffusers, while the GPB finite system method is demonstrated to be the most accurate in this situation.