|
real world success
Easy Repair and Optimization of Surface Meshes Leads to Fast and Accurate MPA Analysis on STL Models!
By Koen Engelborghs, M.Sc., Ph.D., Development Manager, Materialise
RapidParts (UK), Materialise's rapid tooling site, has a unique approach to tool design and manufacture: they routinely use the STL format throughout the entire design and manufacturing process. One of the secrets behind RapidParts' short lead times is the emphasis they place on analysis. All parts are analyzed with Moldflow Plastics Advisers® (MPA®) software to anticipate manufacturing issues and product quality. With hundreds of projects each year, RapidParts relies on MPA as a valuable tool for ensuring efficient and cost-effective production. In order to maximize the advantages of using MPA, RapidParts looked for a way to guarantee fast and accurate analysis for any STL model.
Rachel Collier, a CAD and tool designer at RapidParts, explains: "Because RapidParts aims at keeping analysis lead-times short, while optimizing the analysis' added value, we studied ways to maximize the quality of analysis results. Our research revealed that the 'garbage in, garbage out' rule also applies here. So we investigated ways to improve the quality of the
input data."
"Garbage In — Garbage Out"
RapidParts examined a random sampling of 100 recent tooling projects. Their researchers were able to establish three categories for distinguishing input data. In category 1 are STL models RapidParts receives that don't require adjustments in order to allow correct MPA analysis. Category 2 encompasses models whose flaws make them unusable for analysis. Most of these defects originate from incorrect or poor export of CAD files. Finally, RapidParts identified category 3, files that look fine at first sight, but either cannot (completely) run in MPA or return sub-optimal results.
"In terms of analyzing these problematic files, only two options are available: review the results with caution or stop the analysis. If the part is quite complex and analysis by MPA is required to ensure good tool design and molding, this restriction can be a big disadvantage. Failing to recognize inaccurate results can mean drawing the wrong conclusions regarding design and manufacturing. Imagine the additional time and costs involved," says Collier.
Optimizing MPA Input Data with Magics Software
Next, RapidParts looked for easy ways to transform problematic files (of categories 2 and 3) into files that return accurate analysis results. Because a defective mesh cannot be repaired in MPA, RapidParts uses its own Magics software before analyzing the file to fill holes and small gaps, remove noise and double surfaces, invert wrongly oriented triangles and repair untrimmed surfaces. In this way, category 2 files are easily made suitable for import in MPA. The study showed that 70 to 80 percent of these watertight models can be analyzed correctly and completely with MPA.
In 20 to 30 percent of the projects, however, the MPA software either indicates "potential mesh problems" or cannot process the parts completely. Nine percent of these files can't be processed at all. These are the STL files classified above as category 3. Closer inspection revealed low quality surface meshes; all files contained triangles with a quality aspect ratio below the one advised by MPA.
"With Magics Remesh module, which was released in September 2003, we could finally deal with such low quality surface meshes. We now systematically optimize every STL file with potential mesh problems, with a normalized quality parameter set at a minimal required value (advised by MPA). This results in a much better STL surface definition. After importing the optimized file into MPA, we can thoroughly perform each analysis without any problems. We now have a 100 percent success rate!"
Figure 1 shows the surface mesh of a part received at RapidParts. Rachel Collier was especially interested in identifying and eliminating sink marks, so the part was imported into MPA. "MPA predicted a short shot, something unexpected for this size part. Also, we could not visualize where the short shot occurred. This prevented further analysis phases, including the sink analysis, because you need a complete shot to do it. So I had to conclude the analysis of the original file had failed," Collier says. MPA advises maintaining an aspect ratio (the ratio of the height perpendicular to the longest side, to the length of that side) higher than 0.17. The quality histogram shows that the original file included a great many triangles with a height/base ratio below 0.17 (the green line).
Figure 1. Surface mesh of the original file. |
Figure 3. Surface mesh of the optimized file. |
Figure 2. Quality histogram of the original file. |
Figure 4. Quality histogram of the optimized file. |
Magics easily optimized the mesh. The histogram (Figure 4) shows that all triangles have moved to the right side of the 0.17 line, well above the recommended quality. Imported in MPA, the optimized model was correctly analyzed and the sink mark analysis could be run.
Quickly Analyze Every STL File with More Accurate and Complete Results
This is just one example from the 100 STL files in RapidParts' study. A comparison was made between the MPA analysis results and calculation times from the original and optimized files. Rachel Collier summarizes the results of the study: "By using different STL optimization tools, we obtain a wider range of results, since all files can be fixed and prepared for the MPA analyses. The results are more accurate and complete because the mesh meets MPA quality requirements. An average time-saving of 16 percent was realized because the STL surface definition is better and contains fewer triangles (an average reduction of the number of triangles of 23 percent), allowing faster calculations."
You can get more information about Magics at www.materialise.com or by attending Materialise's presentation at iMUG 2004, May 17-19, Frankfurt, Germany.
| 
