The reason for the message in the BF manual is that inspecting phased resultscan be difficult. There is no question that all loads applied with the correct direction and phase angle is best, except that applying all phased, directional transient loads in a modestly complex piping system due to a repeated acoustic wave passing whether using a fluid simulator to compute the loads (BOS Fluids),or estimating the loads by hand has resulted in a considerable number of errors and unproductive runs. Our recommendation is to evaluate individual, unbalanced dynamic loads in the longest runs individually and without phase angles first.
As more competence and understanding is developed both with the CAESAR harmonic output and with the load development tool then more complex runs can be developed. BOS Fluids produces the unbalanced load for point pairs within the piping system and not for single point loads at each change in direction and so the need for phasing the loads is not as great. Our hope was that this would make developing and applying the transient loadings simpler. The recommendation is essentially to start with simple dynamic models, and to add complexity as needed.

Tony,

Correct me if I'm wrong. I've generated 2D unbalanced force output plot
at midway node (node 100) of the elbow-elbow pair run in BF. peak-to-peak load is 450N, so applied load is 225N at node 100. Should I apply this force at elbow? In CAESAR II user's guide it's said that user should apply loads (Harmonic Force = 0.5 (Pressure variation) (Area)) at elbows with phase angle (phase = [(frequency)(length) / (speed of sound)]360º). As I understood, as far as I apply shaking force at midway between 2 elbows, it's not that important to enter phase angles. CAESAR II approach for harmonic loads' phase angles is using single points at elbows, like time history of pressure wave traveling from source to downstream elbows.

Thanks,
Farhad

________________________________

Farhad,

The questions you ask are general, and so please accept that the responses must be general too. For the significant loads and a first pass at evaluating the system your approach sounds correct to me. This presumes the mode of interest for the load of interest involves an axial movement of the pipe between the elbow pairs where there is no relative movement of the elbows in the axial direction. For typical runs of pipe around compressors, this may be an adequate assumption. For long runs of pipe, this won't be the case. Inspecting mode shapes and frequencies associated with the period of the excitation will give you some idea if these assumptions are valid. For your first runs, I wouldn't worry about the phase angles, keeping in mind that the interactions of loads in time may be important. As you get started, proper inclusion of support or intersection stiffnesses, and all weights is probably the most important items to focus on. Once you've seen that you've got multiple runs of pipe that will participate together to load the system, you can begin to look at phase relationships. If you want to apply point loads rather than unbalanced loads, these can be extracted from the BOS Fluids results for
each point and applied at elbow pairs, (or change in direction pairs). The same can be done with the next elbow pair and the phase relationship provided to CAESAR. The phase angle may not be known, or may not be available from a third party, and in these cases running a zero phase for all unbalanced loads can be done with a selection of load sense that is conservative.
Care should be exercised when inserting supports to reduce displacements and/or loads. New supports will change the frequency of response and may introduce involvement of other modes depending on the geometry.
There is an assumption inherant in the fact that you are applying an unbalanced load at the middle of a piping run. It is that the relative axial deflection of the run of interest occurs at a higher frequency (lower period) than the time scale associated with the loading of interest and that the rigid body axial behavior of that run is sufficient to develop bending moments, torsion, forces, etc, at other places in the system that are susceptible to the unbalanced load studied. This assumption is invalid, for example for impact driven piles, since in the pile driving problem the relative axial deflection of the pile is an
important parameter in the dynamic solution.

Tony