Pipeplus recommends the use of the Time History loading for waterhammer analysis. The text below is traight from the manual.
To ensure a meaningful time history analysis, you should pay special attention to the following items:
·Include at least eight modes. PipePlus time history analysis uses the modal superposition technique to summarize the history response. If too few modes are included, the analysis results are not accurate. Once you know the highest frequency extracted from the modal analysis and the mode shape associated with it you will be able to better assess the quality of the dynamic solution. For example if you have a load that acts for 20ms., the frequency that would be excited by a repeated application of this load would be 1/0.020 = 50 Hz. Arguably modes 20-to-40% higher could also undergo some dynamic excitation due to the 20ms load. This would put the highest frequency needed in the solution at (50)(1.4) = 70 Hz. The shape of the mode is important for point loads, (such as slug flow, waterhammer or relief loads). If the shape associated with the higher frequency mode is not associated with the point of loading, and in the direction of the loading, then the mode will not participate significantly and will not be a concern, regardless of its frequency.
·Use a short time step. The recommended time step is 1/12 or 1/4 of the highest frequency you want to consider. For example, if the highest frequency to be considered is 30 Hz, then the time step should be smaller than (1/30)/12 or .00278. Time steps much smaller than a millisecond are rare. If you find that you need such small time steps then some further mass refinement of the model might also be necessary to detect and simulate the higher order modes accurately. You should probably use 1/4 as the time step to begin the calculation so that it proceeds as quickly as possible. As you begin to refine the solution, reduce the time step to be sure that the results do not change. In addition to the frequency, the loading curve must also be simulated properly in the integration. If the load curve has characteristics that vary every 4 ms, a timestep much larger than that will not reflect that variation.
·The analysis time duration must be long enough. Duration equals the time step times the number of steps. The best way to judge if the duration is long enough is by studying the time history plot, available through the "Dynamic:History" command in the Results Processor. If the time duration is long enough, the history curve should either converge or repeat itself in a noticeable pattern. The duration of the simulation should typically be the time of the loading plus the period of the lowest frequency that will participate in the response. When the load stops acting, the system will begin responding in a motion that is the superposition of its individual modes. One full period of unique response will occur in the period of the lowest mode. The maximum response during this time will only be repeated during each subsequent period.
However to undertake the analysis yopu need more information as regards the loading from the waterhammer transient. At least a pressure versus time plot at the various nodes exposed to the maximum waterhammer transient. Software such as AFT's Impulse can be used to model the system to provide this data.
