Downloaded from https://www.everyspec.com APPENDIX C MIL-STD-31000A Usage of 3D CAD model data determines the recommended acceptable threshold values for each of the product data quality criteria. For example, an acceptable value for design might be unacceptable for data exchange or numerically controlled manufacturing. Product data quality geometry criteria are defined in Exhibit A of this guide. Recommended threshold values for representative usage classifications are indicated in Exhibit B. These suggested values should be used as a starting point for refining acceptance criteria for each class of products acquired by the DoD.
For example, a welded steel frame would be expected to have larger threshold values and fewer acceptance criteria than precision parts for an internal combustion engine. It is the responsibility of the Contracting Officer’s technical representative to increase or reduce the threshold values suggested in Exhibit B or apply additional criteria from Exhibit A, if experience with a specific use of 3D model data warrants such deviations. Refinement of quality acceptance criteria should be based upon a scientific method of successive refinements. The following process is recommended for identifying the relevant criteria and their appropriate thresholds for each use case. 1.
Identify the downstream application, the digital model reuse methodology, and the acceptance criteria. Application examples include mesh generation for CAE, translation to another CAD system, and NC toolpath generation. If, for example, a model fails to sew into a valid solid in the target system after it is translated, then the model contains defects that are unacceptable. Likewise if a numerically controlled toolpath can’t be generated with sufficient accuracy, then the source model may be deemed defective. 2. Gather a representative set of models containing features and tolerances common across each category of design data.
Strive for models that are small in size but have varying degrees of complexity and variation. 3. Process the models through the downstream application(s) and separate them into two groups: those which satisfy the acceptance criteria and those which do not. Analyze each test model with all potential data-quality criteria and record their extreme values. For example, identify the maximum Large Edge Gap (G-LO-LG) and the smallest Tiny Edge (G-ED-TI) in each test model. 4. For each potential data quality criterion, chart the extreme values for the two groups of test models. Identify the threshold which best predicts when a test model will be in the unacceptable group.
If for a given criterion, this process does not indicate a threshold defect value, then the criterion may be judged to have no effect on model quality and should be removed from the list of acceptance criteria. Sometimes a test model will contain an outlier, a defect that is not responsible for its unacceptable behavior but which causes the statistical correlation to be skewed. Such outliers can be confirmed by removing them from the test model, without changing the rest of the model, then repeating the acceptance test and the data quality analysis on the redacted model. 70