![]() | Only 14 pages are availabe for public view |
Abstract Two numerical models based on the method of characteristics were adapted to simulate the column separation phenomenon caused by the rapid closure of a valve located at the upstream end of a pipe. The column separation models utilized were the Discrete Vapor Cavity Model (DVCM) and the Discrete Gas Cavity Model (DGCM). Both models arc capable of modeling the effects of the unsteady friction and the viscoelastic behavior of pipe walls. The numerical results predicted by the DVCM and the DGCM were compared with the experimental measurements in a copper (elastic) pipe, the comparisons showed accepted degrees of agreement with either models. Two methods for the modeling of the upstream valve boundary were described. One of the methods assumes that the flow velocity decreases linearly with time as the valve closes. The other method utilizes the energy equation to calculate the flow velocities during valve closure stroke. An experimental setup was constructed to provide the means of obtaining reliable experimental data for transient flow in a viscoelastic pipe to verify the numerical models. The comparison between the numerical and experimental data showed significant discrepancy in case of solenoid globe valve closure at the upstream end of the pipe. Modifications were carried out to the experimental setup in order to find out the reasons for the discrepancy. Of these modifications, the solenoid globe valve was replaced by a ball valve manually operated. The real pattern of closure of the ball valve was numerically simulated; excellent degree of agreement was achieved between the real and simulated valve patterns. The transient flow measurements utilizing the ball valve were compared with numerical predictions; accepted degrees of agreement were achieved over a broad range of Reynolds numbers. The DVCM showed better agreements than the DGCM. Video photographing for the column separation phenomenon during the vapor cavity formation, growth, and collapse was processed for the cases of solenoid and ball valve closure at the upstream end of the pipe. Video films were transformed into frames using computer software. The frames representing the cavity length and the pressure measurements at the upstream valve were compared with the cavity length and pressure traces predicted by the DVCM at each time step of the framing process at the same location. |