Description: Discussion of AutoPIPE Advanced Non-Linear Analysis Engine
Product: AutoPIPE
Version number: V8i (v9.2)
Submitted by: MU
Revison : 0.0

Non-Linear Friction

Introduction

Friction is often ignored or not easily understand across many industries. If in doubt companies will apply a nominal friction factor = 0.3, but some industries like Nuclear typically ignore friction. It is well documented static friction factors between steel on steel can vary from 0.3 to 0.8 but as industry consultants like Dr Tony Paulin state below the true stress range from Cold to Hot to Cold causes the friction forces to reverse and thereby doubling the frictional stresses in the pipelin. Since pipe stress programs analyze only a Cold to Hot or Hot to Cold range, some companies to capture this increased stress due to friction forces double the steel on steel friction factor to 0.6.

A simplified explanation is below

Imagine friction "building up" on the piping after thermal load. It hasn't broken friction or closed a support gap yet, but it's building up to that point of doing so. Wind or seismic just gives it that little push to break friction and or close the gaps, but it does require that extra 'push' to make it happen. Therefore, the boundary conditions change, but only AFTER the wind load hits, so boundary conditions are different between the thermal and wind load analysis. This is only captured using a Non-Linear incremental approach and not Total Loads approach.

Should we Consider Friction forces for Seismic Loads?

Note: It is not uncommon in fact and stated by many building design codes like UBC and IBC that friction should NOT be considered for occasional type loads like Seismic. This is only possible using an incremental load case analysis engine like AutoPIPE when other non-linear conditions like support gaps need to be considered. AutoPIPE has an option to ignore friction for seismic ('E') and/or gravity load cases.

Friction is calculated as friction factor x bearing force. The vertical or lateral support bearing force (hence friction) changes for each different load case applied by the piping on the support. Of course the lateral bearing force is also dependent on the support gap which is changing by load case.

The Total Load approach like with Caesar does not understand the boundary conditions and bearing force (hence friction) are changing by load case i.e. path dependent friction. It assumes constant boundary conditions from one Total load vector to another e.g. GR+T1+P1+E1 compared to GR+T1+P1 which is incorrect.

Reference: Dr Tony Paulin Webinar February 2012 "New Pipe Stress Capabilities in PCL Gold"

Discussion of the importance of path dependent friction.

Extract Paulin Research at 5/3/2012
http://www.paulin.com/library/prg2009featureslist.pdf

Reference: The common forces: conservative or nonconservative? (2006)

By David Keeports

Department of Chemistry and Physics, Mills College, Oakland, CA 94613, USA

Reference: TREATMENT OF SUPPORT FRICTION IN PIPE STRESS ANALYSIS (1989)

By

Liang-Chuan Peng

Pang Engineering

The author has stated in a couple of highlighted sections it is preferable to perform incremental analysis such as in AutoPIPE, “to ensure no extreme load is overlooked” the complication of the support stop and the friction has to be adjusted.

Most pipe stress programs like Caesar use an old total load approach as described in ref 5.

Reference: IMPROVED ITERATION STRATEGY FOR NONLINEAR STRUCTURES
(1981)

By GRAHAM POWELL AND JEFFREY SIMONS
Department of Civil Engineering Division of Structural Engineering and Structural Mechanics, University of California

AutoPIPE is based on advanced displacement increment method
developed and published from the University of California (Berkeley)

AutoPIPE Load Sequencing

Nonlinear Solution Load Case Sequence
For a linear analysis, the results for each load case are obtained all at once. However, for a nonlinear analysis the results are obtained sequentially. There are two reasons for this. First, the analysis of a nonlinear system requires iteration (successive trials), and different load cases will usually require different numbers of iterations. Second, (and more important) the result for any AutoPIPE load case will generally depend on the initial state for that case. For example, the result for a thermal expansion load will generally depend on the state of the system after gravity load is applied (e.g. which gaps are open and which are closed). The solution for the gravity case must thus be obtained first, and used as the initial state for the thermal expansion case.

For a nonlinear analysis, the user has the option of selecting a default load sequence or of specifying a user defined sequence. It is important to note that in an AutoPIPE analysis, each load case is an increment of load, not a total load. To illustrate the difference, consider two alternative procedures for obtaining thermal expansion effects.

If analyses are performed for total loads, the steps are:

1. analyze for gravity;

2. analyze for gravity plus thermal; then

3. subtract Step 1 from Step 2 to get thermal.

If analyses are performed for load increments, the steps are:

1. analyze for gravity; then

2. analyze for thermal, specifying gravity as the initial state.

AutoPIPE uses the second of these procedures. Thus, to obtain the results for gravity plus thermal, a load combination must be defined in using the commands in the Result menu.

Total vs. Incremental Loading
A load case (e.g. gravity, thermal, wind, etc.) represents an increment of load, not a total load (except for gravity). Hence, pipe forces, displacements, support forces, etc. calculated for a load case represent the increments produced by that case regardless of the type of static analysis performed (linear, or nonlinear). In particular, the results for a thermal case define the changes in the forces and displacements due to thermal expansion, not the total effects due to combined gravity and thermal. In order to obtain total load effects, combinations must be defined which include the load cases that have been used to hold the specific loads of interest. "Superposition" of load cases is a commonly accepted principle for a static linear analysis. However, it is not so straight forward for a nonlinear analysis.

Linear: Perform linear analyses. Initial states are not needed. All gaps are assumed to be closed and friction is ignored for occasional load cases. Linear soil properties are used, and yielding of soil is not taken into account.

See Also

Non-linear Load Sequencing Explained

Bentley AutoPIPE

External Links

Bentley Technical Support KnowledgeBase

Bentley LEARN Server

Comments or Corrections?

Bentley's Technical Support Group requests that you please submit any comments you have on this Wiki article to
the "Comments" area below. THANK YOU!