As you have found there is little in the way of definitive statements regarding friction under seismic action. One of the few is in ALA:
Seismic Design and Retrofit of Piping Systems – American Lifelines Alliance – 2002
Section 8.2 The seismic design should not take credit for the friction force between pipe and support, which tends to reduce seismic motion of the pipe.
Consideration of the behaviour of a piping system in a seismic event must lead you including analysis using the zero friction case. Friction is commonly modelled as 0.3g. However the rolling wave nature of a seismic event causes the vertical acceleration to vary, and therefore the friction component to vary. And it is practically impossible to calculate. So we know that friction forces acting on the piping system lie somewhere between full friction and zero friction. It is recognised that zero friction seismic is generally "toughest" on the pipeline, therefore to not analyse zero friction seismic is to ignore a real pipe design case.
I have found that analysing for zero friction drives you to good pipe supporting solutions. Such solutions generally mean the piping is well restrained, and is therefore also able to resist other types of shock loadings such as water hammer, two-phase flow, slugging, pressure relief loads etc.
- Analysing with full friction tends to reduce the loads on fixed points such as anchors and line stops, and increase the load on other supports.
- Analysing with zero friction tends to put large loads on the fixed points and reduces the loads on other supports. It doesn’t increase the total loads, but it does tend to focus them more.
We design a lot of cross-country pipelines, and one of the consequences of using zero-friction seismic is that you are driven towards putting side guides or clips on almost all supports. This means that lateral seismic forces are shared amongst almost all supports, rather than being taken on say every 3rd or 4th support – and this is a good thing.
We actually run both friction and zero-friction analysis in one model and output both cases for pipe support design (using a different analysis software that has load sets – which makes the setup relatively trivial).
_________________________
Ross Sinclair