Abstract
The flow investigated consists of a pressure wave travelling along a pipe of constant cross-section 1" in diameter. The wave is generated by a piston, actuated by an electromagnet and travelling in a cylinder connected to the pipe trough a convergent nozzle, the ratio of diameters being 4. The maximum acceleration attained by the piston is of the order of 30g. and the maximum flow Mach number obtained in the pipe is equal to 0.334. The increase in pressure with) time in the cylinder and at eight stations placed along the pipe is recorded by means of capacitance pressure gauges and C. R. tube oscillograph aquipped with a drum camera. An accurate time pulse is superimposed on the record, together with a series of electrical im- pulses indicating the position of the piston in the cylinder at a given instant. This last series of impulses affords the means of comparison and indication of reproducibility of the pulse generated. The calculations of the change in the flow and state parameters along the pipe are performed on the assumption that the flow is isentropic, and are based on the method of characteristics for non-steady one-dimensional flow. The difference between the calculated and experimental results, which are well cutside the experimental error range are shown to be predominantly due to friction in the pipe. The magnitude of the deviation of the experimental results from the isentropic theory is found to increase both with increasing Mach number and length of pipe. For Station 5, at 96 diameters from the inlet to the pipe the deviation of the actual pressure from the one calculated on the assumption of isentropic flow is found to be approximately equal to 5% of the isentropic at M equal to 0.31, and 1% at M equal to 0.18. for Station 7;240 diameters distant from the inlet, the deviation from then isentropic at the same Mach numbers increases to 10% and 3% respectively. It is shown that the isentropic assumption is sufficient only in the range of lengths below 48 diameters of the pipe where the deviation does not exced 2% at the highest Mach number obtained. Outside this range the effect of friction in the boundary layer has to be taken into account. An attempt to allow for friction effects by calculating the pressure decrease along the isentropic particle path is shown to be insufficient to account for the losses observed. An attempt is made to develop a step-by-step method of integration for the partial differential equations of linear non-steady flow, with friction included in a way similar to that accepted in the linear and steady flow of a compressible fluid.