PROCISION Frequently Asked Questions - Technical



Q: Which variable loading are available in Procision?

Q: Can we apply moments?

Q: What is a Suspended Load?

Q: Can we output a file with points and the corresponding result values?

Q: How do we Print?

Q: Can Procision show surface contours?

Q: Can we look at vector plots for results?

Q: How do I apply a load to a REGION?

Q: Since PROCISION won't allow us to analyze an assembly, is it proper to use 'Merge' under 'Adv Utils' to merge several parts of a structural assembly into a single part and analyze it in part mode?

Q: What is Traction BC Error?

Q: Is there any comparisons between PROCISION and theoretical or other analytical solutions?

Q: I found that different ways of splitting on identical model can cause up to 35% max stress variation and convergence was achieved in all cases. How to explain it?

Q: How to handle the situation when run terminates with messages of "Traction Error not improving" but with "0 non convergence" on all items (subparts, boundary and intercon)? Keep splitting?

Q: Is there a way to apply two or more types of loads on the same face?

Q: Can notes be added on the plots?

Q: Can the background color be changed to white for plot/screen capture?

Q: Should I apply Loads and Constraints to part surfaces or edges / points?

Q: There is a very big difference in running with only DISP % and no LOCAL TRAC ERR %, can you explain?

Q: How does PROCISION handle symmetry problems?

Q: What guidelines should I follow while splitting up a part?



Q: Which variable loading are available in Procision?

A:Procision handles: uniform and non-uniform loading, non-uniform loading is applied by defining control points (ie. 3 locations on a surface with set values of LOAD/UNIT AREA) and the magnitude is interpolated linearly in-between. In addition, it presently supports bearing loads around holes and suspended loads.



Q: Can we apply moments?

A: Torsion (or moments) is applied either as a set of forces (manual technique) or using a suspended load.



Q: What is a Suspended Load?

A: Suspended loads allow you to pick a point in space and either a set of surfaces or a REGION to react the load upon. For example, if moments are applied to the suspended point, PROCISION calculates the force distribution that needs to be applied to the surface.



Q: Can we output a file with points and the corresponding result values?

A: You can use the PICKINFO option to display numerical result values at any DATUM point or part location. The GRAPH function may also be used to display result values along a user specified line or part edge. These items can be output to an HTML file.



Q: How do we Print?

A: Choose command Print from File menu in the appropriate application.



Q: Can Procision show surface contours?

A: Yes, though contours are shown as color bands. Choose Options -> color ramp -> number of colors in the color plot.



Q: Can we look at vector plots for results?

A: Vector plot results are not currently available.



Q: How do I apply a load to a REGION?

A: First create the region; for a REGION-SURFACE, you must form a closed loop of datum curves (even if the region joins up to a surface edge). To select the region, use query select (to be safe), pick inside the region, and use NEXT until the parent surface of the region highlights. The region itself may or may not become visible during this selection process, in fact it may appear when one of the component datum curves becomes highlighted. Ignore any appearance of the region symbols. ACCEPT the selection when the parent surface is highlighted. Once you do that then the region should appear.



Q: Since PROCISION won't allow us to analyze an assembly, is it proper to use 'Merge' under 'Adv Utils' to merge several parts of a structural assembly into a single part and analyze it in part mode?

A: Yes, merge is fine, in fact some people like to use merge even if it is a single part. Note also that different material properties may be assigned to different subparts.



Q: What is Traction BC Error?

A: Traction Boundary Condition Error is an absolute measurement of error in the solution. PROCISION is unique in its ability to provide such an absolute (versus an estimated) measurement.

If you consider a part surface which is not loaded, then the tensile stress normal to the surface, right on the surface, must be zero. Similarly, the shear stresses tangent to the surface, right on the surface, must also be zero. On a loaded surface, the stress must equal the applied load. Therefore, we have a known or target value for these components of stress, all around the part boundary.

Traction stress components can also be calculated from the computed stress solution (using standard stress transformation functions). Thus, we have known values, and calculated values of these stresses. The difference in these values is Error. This value of error can be displayed as a color fringe plot, showing the user the quality of his solution.



Q: Is there any comparisons between PROCISION and theoretical or other analytical solutions?

A: Yes, and a free Verification Manual is available. You may download the manual in postscript format from: verif.ps . Alternatively, contact your local Rand office for a hardcopy.



Q: I found that different ways of splitting on identical model can cause up to 35% max stress variation and convergence was achieved in all cases. How to explain it?

A: The final message that PROCISION gives is based upon the examination of several parameters, primarily convergence of strain energy within 1% and convergence of traction error within the user set %. Note that convergence of traction error is based on the CHANGE in error not the absolute value of error. Therefore it is possible for the solution to be considered converged if increasing accuracy ceases to cause a reduction in error. The best approach to determine solution quality is to examine the convergence graphs of strain energy and vonmises stress, and especially the traction error contour plots. The rule of thumb for examining the traction error plots is that the displayed fringe plot should be all in the blues (different shades of blue is okay), any greens, yellows, etcetera indicate a non-ideal solution (of stresses) therefore the splitting scheme should be ammended. This should be done regardless of PROCISION's final message.



Q: How to handle the situation when run terminates with messages of "Traction Error not improving" but with "0 non convergence" on all items (subparts, boundary and intercon)? Keep splitting?

A: Result not improving may be due to a singularity in the model. The message is issued when increasing accuracy causes the error to INCREASE. Examine the convergence and error results. Also look for "HOT SPOTS" in the stress results this may suggest where the difficulty is. If you have high traction stress in some areas that can guide you whether more splits are required. (Typically solutions are pretty good when this message is issued with 0 non converged, but you always want to check using the tools included).



Q: Is there a way to apply two or more types of loads on the same face?

A: Yes, it is implemented.



Q: Can notes be added on the plots?

A: Probably the best way to accomplish this is to save the plot as a graphics format and use an image editor to add the text. Please notice that some basic information, such as the displayed result type, is already printed into the file if it is saved from Procision application.



Q: Can the background color be changed to white for plot/screen capture?

A: Yes, choose Options -> background -> white.



Q: Should I apply Loads and Constraints to part surfaces or edges / points?

A: Whenever possible, apply loads to surfaces or surface regions, as you want to avoid singularities (infinite stresses).

A load applied to a point or edge will give an infinite stress in theory (stress=force/area, so if area =0 ...).

PROCISION seems to be forgiving sometimes of line loads but they should be avoided.

Point or line constraints have the same problem, as well constraints of a region ( a surface or part of a surface with tangent neighboring surfaces) the edge of a region also may form a singularity.

One use for edge constraints is if you need to model a pinned constraint... I.E. a simply supported beam... If you make the end surfaces IMMOVABLE, that is welded (i.e. fixed-fixed beam)...

To Model a simply supported beam you could do the following:

1) (Approximate) Apply fixed constraints to the top or bottom edge of each end face.. Since the axial direction is now constrained only at a line, Moments cannot be reacted at the ends... the structure will pivot about the line. (In solids analysis there are no rotational degrees of freedom... I.e. to carry a moment you need to contrain a surface area). This method will give fair results, but to more correctly match beam theory:

2) Create Datum curve lines midway (or neutral axis position) on each end-face (the line should correspond to the desired pin axis). Create REGION-CURVEs from these lines, and constrain the regions as immovable. This will give you the best representation.



Q: There is a very big difference in running with only DISP % and no LOCAL TRAC ERR %, can you explain?

A: If you run only to converge on displacement, then in general only displacement results are reliable. If you want stress results, you must converge on TRAC BC ERR.

Stresses are more difficult to calculate than displacements, basically displacements are converted to strains which are converted to stresses... You can see that Stress has error of Order 2 whereas displacement Order 1. (This is not only in PROCISION. All analysis software does the same thing.)

The main item always to examine to determine solution quality is the Von Mises convergence graph, and the Stress - Trac BC ERR contour plot.



Q: How does PROCISION handle symmetry problems?

A: Due to PROCISION's high speed and low resource usage, employing symmetry is not always the best approach. However, regular symmetry may be used in the same fashion as any solid element (FEM) model. Providing that geometry & LOADS are symmetric, you may cut the part along the mirror plane(s) and apply displacement constraints to the cut surface(s) normal to the surface to model symmetry.

Some analysis tools have Cyclic Symmetry and Axi-symmetric modules; PROCISION has not implemented anything similar to date. Axi-symmetric can be simulated by cutting a small pie-wedge and treating like normal symmetry. For now cyclic symmetric problems should be modeled in their entirety (of course regular  symmetry can be used if applicable).
 


Q: What guidelines should I follow when subdividing a part?

A: With some experience the best partitioning techniques will become more intuitive. The new user should concentrate on verifying the solutions accuracy using the convergence graphs, and especially the traction BC error fringe plot. The rule of thumb for these error plots is that the part should be shaded all in the blues (different shades of blue are okay, but greens, yellows, etcetera, indicate a non-ideal solution, and modification of the splitting scheme is required).

Section "Summary of Splitting Methodology" of the User's Guide should be used as a guideline for all analyses.

Some additional tips include:

1) When possible split parts normal to the part surface.

2) Use the volume rule for your initial strategy, the results from the first run will guide you on how to add more splits (if necessary).