Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Mechanical Engineering


Dr. D. S. Weaver


The purpose of this thesis is to discover the mechanism of excitation and methods of alleviation of self-excited vibrations in a swing check valve following rapid pump shut-down. The problem was first encountered when the valve manufacturer incorporated an adjustable spring-damper into the original design to prevent its violent slamming. Tests on the modified design showed that, rather than eliminate the slamming, the valve disc bounced several times on its seat at a well-defined frequency. With increased damping the number of oscillations as well as the amplitude increased while the frequency decreased. For sufficiently high damping a stable limit cycle oscillation is established. This limit cycle oscillation continued until the valve pivot shaft pins failed. These vibrations are clearly hydroelastic in nature, the oscillations being perpetuated through a transfer of energy from the fluid flow.

A two-dimensional geometrically-similar model of the valve was constructed with perspex sides for flow visualization. A central portion along the base of the model was also laminated with perspex to allow the projection of a collimated sheet of light. Aluminium powder tracer preparation was injected into the flow and cine-photography of the flow during vibration carried out. In addition, dynamic measurements of upstream and downstream pressures, valve angular displacement and the load on the damper arm were synchronized with the films. The data collected in this way for a number of restraining spring rates and initial spring deflection angles allowed a detailed stability map of the valve's dynamic behaviour to be plotted. The essential characteristics of the instability observed in the model are the same as those found in the prototype valve tests although the model was not scaled dynamically. This was necessary in order to guarantee the structural integrity of the model over the long period of tests.

The results of the research show that there is a sudden increase in the hydrodynamic closing load as the valve approaches its seat, primarily as a result of the changing discharge characteristics. Although upstream and downstream waterhammer waves are produced as the valve slams onto its seat, the valve responds only to the pressure difference across it. It remains closed until this pressure difference reduces to the point where it either cracks the valve open or allows the damper spring to pull it open. On the opening part of the vibration cycle the hydrodynamic closing load is substantially lower than the load at the same angle during closing. This hysteretic effects shows that there is a net energy input from the fluid during each cycle and the motion is perpetuated.

Tests on the model further show that if the damping spring is stiff enough to eliminate the slamming, either the valve will never close or it will exhibit limit cycle oscillations. Clearly, neither alternative is acceptable. Based on the aforementioned results, it was realised that another possible means of alleviating the problem is to alter the discharge characteristics of the valve at small angles of closure by suitable changes in geometry. In the second part of the thesis, a number of such changes were made in the model and the experiments repeated. It was discovered that by making the rate of change of discharge a more gradual function of the valve closure angle, the dynamic instability in the model could be entirely eliminated.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."