Convergence Understanding Convergence – Lecture UnderstandingConvergence.mp3 Understanding Convergence Using Convergence to determine whether there is a viable analysis result from Mechanica. Convergence defined Solution does not change appreciably as a function of mesh characteristics. In the case of Mechanica, the mesh characteristics are the p-level of each analysis elemen edge. Evaluating results: Review analysis summary o Max Edge Order o Resultant force = Applied Loads o Error Estimates Convergence graphs o Slope approaches zero Review stress error fringe plots Review general fringe plots o Locate singularities & discontinuities. Mesh Elements for Analysis Stress vs. P Loop Pass Lecture Notes Defining Convergence . you do not control the convergence tolerance. Evaluating MPA and SPA Results Knowing if a solution has converged is a critical step in evaluating the quality of a solution. you will examine the procedures available in Mechanica to estimate solution quality. and so on) against the p-pass. Selecting this option enables you to create a fringe plot for the analysis which shows the estimated local stress error for each element. then continue to use the SPA convergence criteria for efficiency. If these errors are acceptable. Convergence is an important characteristic of the solution because it enables you to determine whether you have a viable solution. remember to turn on “Display Element Edges”) of the quantities of interest. This error can be expressed as a percentage normalized by the maximum model stress. ensure that the resultant force in the model equals the load applied in the model (all the loads applied in the model need to be accounted for). In the case of SPA analyses. you can opt to have Mechanica calculate error estimates for each element in the model by selecting the Local Stress Errors check box in the Calculate section on the Output tab for static analyses. This may indicate that you’re dealing with a potentially unconverged solution. A value of 9 indicates that at least one edge went to plevel 9. View local stress error estimates: For SPA and MPA analyses. View and investigate the fringe plots: You can create fringe plots (with “Average” and “Continuous Tone” turned off. Also. A good example of this is shown on the graph shown above: you can see the slope of the max_stress_vm approaches zero as the p-level increases (the P Loop Pass axis). This is probably the best way to identify the quality of the solution. The convergence can be quantified as a graph that has a zero slope towards the higher p-levels and a clear trend approaching a value. Mechanica will provide a message indicating whether the analysis failed to converge within the specified error percentage. Within recent years. but you do need to examine the stress error estimates reported in the run summary. a percentage normalized by the maximum element stress. Here.Converge can be described as a situation in which the solution found does not change appreciably as a function of mesh characteristics (be it a polynomial order or number of elements). This is useful for locating and investigating discontinuities between elements. To determine whether a solution is converged. Generate convergence graphs (available only for MPA analyses): You can create a graph plotting the quantity of interest (max displacement. max stress. however. you should perform the following reviews: Review the analysis summary content (available for both MPA and SPA): In the case of MPA analyses. Mechanica has refined the Single-Pass Adaptive (SPA) method to the extent that it has become a more rigorous method to estimate Best Practices . Mechanica specifies the Maximum Edge Order for each Pass it goes through. If that’s the case. or in terms of its raw value. A run usually fails to converge for one of two reasons: a singularity is present and the engine is trying to capture a high stress or flux gradient. or a highly distorted element is trying to capture a smooth stress or flux field. Error error_plate. Open the model. Use your best judgment when specifying the convergence percentage in the case of MPA analyses. larger models analyzed using the SPA method require less disk space and computational memory. Understanding Convergence – Demonstration UnderstandingConvergence_demo. Therefore. The default value is 10%. In general. check whether that quantity converged. On the other hand. 3. . Singularities cause edges in a model to require a high p-level for convergence thus resulting in a longer solution time.prt Task 1. 1. when using the SPA method you have less information regarding the quality of the solution. Additionally.mp4 Understanding Convergence – Procedure Procedure: Understanding Convergence Scenario Review a stress error estimate fringe plot and a convergence graph of a Mechanica analysis model. For instance. always monitor and understand stress singularities in the model. and you must rely on experience to detect improperly converged results. You only need to rerun the study if the quantity of interest did not converge or come close to converging. but it can be lowered or increased to different values based on the geometry/loads/constraints available in the model. and improve solution accuracy. Right-click the ERROR analysis in the Analyses and Design Studies dialog box and select Edit. run an analysis. the result of singularities can cause anything from abnormally high results to models that will not converge. 2. Click Mechanica Analyses/Studies from the Main toolbar. you should use SPA convergence when able. If you are primarily interested in a quantity other than the convergence quantity. Click Applications > Mechanica. and review the element stress error estimate results. Click OK to close the Static Analysis Definition dialog box. 8. 5. . The dialog box should appear as shown. Verify that ERROR is still selected in the Analyses and Design Studies dialog box and click Results to start Results mode. 9. 4. Select the Output tab and select the Local Stress Errors check box as shown. 7. click Close in the Diagnostics dialog box. 6. When the analysis completes. Verify that the second field is set to Normalized by Maximum Model Stress.Note that the ERROR analysis has already been defined as a Static Analysis with appropriate Load and Constraint sets and MPA convergence. The analysis should complete in just about 30 seconds. Click Start Run > Yes in the Analyses and Design Studies dialog box. Select Element Stress Error Estimate from the first drop-down field in the Quantity tab. Select Graph from the Display Type drop-down menu. Click OK and Show from the Result Window Definition dialog box to show the result. 14. 12. 11. Click Define Measure . . The dialog box should appear as shown. Click Copy from the main toolbar.10. select max_stress_vm from the Measures dialog box and click OK. 13. Select Measure from the drop-down menu in the Graph Ordinate section of the dialog box. 15. Additionally. . Note that although not shown here. the local stress error estimate graphs show no large errors. Click OK and Show from the Result Window Definition dialog box to show the result. the local stress error estimate fringe plots are also available for analyses using the SPA convergence option. nor do they show any significant errors in areas of interest such as high stress areas. The curve of the convergence graph smoothly approaching a maximum value suggests a well converged solution. we can conclude that this problem is well converged and probably fairly accurate. Given these pieces of information. . If necessary. click File > Exit Results > No to close the results window. 19. 18. This completes the procedure. After reviewing the result window. Click Save from the main toolbar and click OK to save the model. Click File > Erase > Current > Yes to erase the model from memory. 17.16. return to the Standard Pro/ENGINEER mode by clicking Applications > Standard.