Applies To Product(s): AutoPIPE, Version(s): 2004, XM, & V8i Environment: N/A Area: Confirm Message Subarea: Original Author: Bentley Technical Support Group Problem: When analyzing a model the followng occurs: --------------------------- Confirm --------------------------- The load case did not converge. Do you wish to continue iterating? --------------------------- Yes No Cancel --------------------------- What should I do? Answer: Seleting YES: The program will re-run the analysis for more iterations in hopes of converging below the non-linear tolerances specified. Selecting No: Stops the analysis procedures and presents the following: --------------------------- Confirm --------------------------- Do you want to see .LOG file for magnitude of unbalance error? --------------------------- Yes No Cancel --------------------------- Selecting Yes: AutoPIPE Viewer is automatically displayed with the *.LOG for review. Selecting No or Cancel: The dialog is closed and user is able to enter the next AutoPIPE command. Selecting Cancel: The dialog is closed and user is able to enter the next AutoPIPE command. Additional Comments, Questions, and Answers: Item #1 - Question: What to do when this dialog appears? Answer: Please see the following AutoPIPE help location. Help Contents Contents Tab Reference Information Analysis Considerations Static Analysis Nonlinear Analysis - Gaps, Friction and Soil Yielding Included Failure to Converge. There are 4 solutions offered. In addition, consider reviewing how the model is moving under the offending load case. Sometimes support gaps need to be decreased / increased. In other cases, additional supports are needed in key locations around the model to prevent the movement to cause a model to not converge. Or a combinations of all 5 solutions is needed to resolve the issue. Item #2 - Question: How do I read the *.LOG file? Answer: When the file is opened by AutoPIPE Viewer or any Word editing application, search for the following word: "Case". See the following example information: Since a nonlinear analysis does not necessarily converge exactly, the force tolerance is used to check for convergence of a load case. If the force error does not exceed this value at any support, the analysis is assumed to have converged. Make adjustments to the model based on information from the LOG file until the model converges Item #3 - Question: Does AutoPIPE as a whole have the same type of converging issues as seen by other programs? AutoPIPE has a more advanced non-linear analysis than other products, developed in the early 80's out of University of California, Berkley. Very rarely does it not converge. Typically when Users enter incorrect soil properties or very closely spaced supports e.g. 1-2' spacing may cause some oscillation of the solution. The program will not hang but the user can chose to stop the analysis and view the results at that iterative state. We have heard complaints from many Caesar Users about Caesar's converging issues. In fact one major German Company using Caesar sent us this quote. "If friction coefficients are defined for supports in CAESAR II, there is a risk, if no convergence is reached the program is hanging. In order to fix this, the user should suppress the friction coefficients on 1 or more supports. This problem is NOT present in AutoPIPE." However it does occur, we encourage our users to contact our technical support in order to help solve such issues. Item #4 - Question: My piping model with gaps and friction does not converge after 5 iteration, can I continue? Answer: Yes, generally the default number of nonlinear iterations = 5 (in version 6.0 or earlier) is adequate for most models but for models with large gaps and friction it is not uncommon for convergence to occur after 20-30 iterations particularly with the thermal, wind and earthquake cases being analyzed in one model. If the analysis did not converge within the specified number of iterations, a confirmation dialog will be displayed requesting whether or not to continue iterating. A "yes" selection will allow the user to specify an additional number of iterations to perform. A "no" selection will display another confirmation dialog requesting whether or not to display the error log file (*.LOG). Press OK to close the &rsquoEquivalent Linear Analysis’ dialog. If at least one load case converged, the analysis results may be examined using the options found in the Result menu regardless if other load cases did not converged successfully. In the case in which the analysis did not converge, an error message will be displayed after selecting any Result menu option (except Filter Criteria) indicating that the analysis did not converge. In addition, a double asterisks is printed before the combination names in the Code Compliance Combinations listing and Other User Combinations listing in the Load summary sub-report indicate that the analysis results for those marked combinations did not converge. Product: AutoPIPE Version: xx.xx.xx.xx Area: GENERAL INFORMATION Problem ID#: 28297 solution 500000055309 See FAQ 11. Item #5 - Question: How to change number of iterations for static analysis? Answer: Convergence is an iterative process that requires multiple passes in order to reduce the errors associated with initial estimates to an acceptable level. After each iteration, the [U] matrix is investigated for any effect on the stiffness matrix [K]. If [K] is affected, a new stiffness matrix is formed and the analysis is repeated until a solution is obtained. Since each loading condition can affect the system differently, a separate [K] matrix may be required for each defined load case. The iterative solution stops when all convergence criterion have been met (bearing, friction, & soil force and displacement). Convergence problems may be alleviated by reducing the friction restraint stiffness. Lower friction stiffness will more-readily distribute friction loads throughout the system and speed or allow nonlinear convergence but this lower stiffness will also affect the accuracy of the results. Lower stiffness values permit more "non-sliding" movement, but given the indeterminate nature of the friction problem in general, this error may not be crucial. To change the number of iterations in AutoPIPE select Load Static analysis sets Modify an analysis set Gaps/Friction/Soil option must be checked ON, press OK button, on the non-linear analysis dialog screen change iteration setting to different value. Item #6 - Question: My piping model with Wave does not converge, what should I do? Answer: Analysis that considers wave, if convergence fails recommend reducing wave height Item #7 - Comment: According to the Log report, there was a problem with the friction factor with T2 load-case. Not sure how to solve this issue with a friction force problem. Answer: Look at the LOG file, see: Load case = Thermal 2 Iteration number = 99 Maximum normal force error = 0.0000E+00 Kgf Maximum gap closure error = 4.2603E+00 mm for support V-STOP at point A20 Maximum friction force error = 1.5379E+02 Kgf for support V-STOP at point B03 Playing with support at node B03 (even setting friction to zero) did not help. Then reviewed support at node point A20. Noticed that the two supports at A17 and A20 are very close. Deleting one of these supports from node A17 or A20 made this model analysis converge. Item #8 - Question: My model with beams AND soil properties does not converge, where is the problem? Answer: Rule out each group by removing it from the model and try to again run the analysis. First create a copy of the original model. With the copied file, try and increase all the soil properties to extreme stiffness values as if it was rigid. Does the system converge in an analysis? No. Then try an remove all beams / frames in the model where all supports are connected to ground. Does the model converge? Yes. Add the original soil properties back into the model. Does the model converge? Yes. Recommend in depth review of each beam to be sure they are all correctly connected and has the sufficient beam properties. Add / Remove beam as required to narrow down the offending beam / frame. Note: if modeling piping resting atop the soil, you may need to consider remodeling using V-stop Supports Item #9 - Comments: Notes about converging issues with soil and under water piping Buried pipe have a lot more nonlinear supports since every soil point has 4 nonlinear springs. In one hand having more soil points would increase the number of nonlinear equations. On the other hand having less soil points especially were needed, like near pipe change in direction, can also cause convergence problems. Consider increasing the soil stiffness (if feasible) and or decreasing your Maximum Spacing may alleviate the non-convergence issues. Another issue is that if the soil bearing capacity P1 is too low, it can cause very large displacements and hence convergence. Under gravity, displacement down can get very large if pipe weight cannot be supported by P1. Upward displacements can also occur due to buoyancy and these can also be very large if uplift soil support P1 is not sufficient. I suggest using nonzero K2 values (0.001 - .0.1) in order to avoid unlimited displacements. Using larger K2 values temporarily (K2=K1/1000 or so) can help you see if P1 values are sufficient by checking displacements. You may need to add concentrated or distributed weight to submerged pipe to prevent it from uplifting (in practice concrete mats are added in some cases). Using V-Stop at a reasonable spacing (that will make it pass sustained stress check), may be a way to get around convergence issues. You may use large friction factors (0.5 to 1.0) for these supports. Also please note that AutoPIPE does not apply hydrodynamic factors to soil points and hence these factors, especially CL (under Xtra data/Hydrodynamic) should be increased to compensate for not applying hydro lift loads to soil points. Hydro factors under wave loading (drag and inertia) are applied to intermediate soil points. Lift factors can only be applied to actual nodes. Using V-STOP instead of soil can avoid this problem with lift factors See Also 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!
↧