API 617 Analysis and Boundary Condition Assumptions 1

[Mark8507]

First time posting - I have searched the forums and have yet to locate a thorough discussion of the following. Any input is much appreciated.

1st - API 617 is unclear regarding the point at which forces and moments should be resolved for the 2Fc + Mc check. Is the specific location on the axis typically left for the manufacturer to define? Any thoughts on which of the following locations are more appropriate? Flange face of largest nozzle, or intersection of largest nozzle centerline with compressor vertical centerline?

2nd - I am analyzing a large centrifugal piping system (30" piping), and modeled the compressor as an assembly of vertical and horizontal rigid elements. An anchor on the suction and discharge is connected via C-Node to the compressor assembly. When I assume these anchors are rigid (CAESAR II default 1e12), the loads are astronomical and cost prohibitive to reduce. I have discussed the issue of assuming nozzle stiffnesses with two other stress engineers with experience totally 45 years between them. They suggested that assuming an arbitrary stiffness of 1,000,000 to 5,000,000, and an allowable imposed nozzle deflection of 10-12 mils according to engineer A, but only 1-2 mils according to engineer B is a reasonable approach for practical design purposes. With either approach (engineer B being more conservative), the loads are of course arbitrary, but at least dependant upon a predefined allowable nozzle displacement. The problem with this is, I have never seen a manufacturer publish allowable imposed nozzle displacements for a compressor. I'm working on it, but suspect in the current situation that they (manufacturer) will not provide insight on this issue. Opinions on this are much appreciated.


Best Regards,

MD

 

[richay]

1) Our interpretation of API-617 is that the distances are from the largest (suction or discharge) nozzle to the other nozzles.

2) Assuming the nozzle is a rigid anchor (or a rigid displacement) is conservative - you'll get the biggest restraint loads. Lowering the stiffness (to perhaps something more realistic) will lower the loads. However, picking an arbitrary value to get things to work is not a good idea - you have to be able to justify the value you use.

Richard Ay
COADE, Inc.