A Object-Oriented Implementation of a Chemical Waste Consolidation Expert System
Return to front page 1. Background and Rationale 2. Interface Design
3. System Design
4. Evaluation and Results 5. Future Work 6. Conclusions Bibliography and References Appendix A. Example Drum Report Appendix B. Chemical Compatibility Testing Data |
4.2. Software and Logic Testing
It was expected that the core design may need several design cycles to be optimized [Pressman 1997]. The core software has had three major and several minor revisions. Hindsight is a powerful tool in software design, and further optimization will be fully possible from this developed platform. The input files for all the libraries are comma delineated data, and all major spreadsheet programs are capable of opening them. This fact has allowed extensive modification testing. It was explained in section 2.1 that the method of binary compatibility testing is a reliable method of waste consolidation. The chemical compatibility routine was checked by running all possible binary combinations of chemicals from a 22-member representative list. The chemicals selected were representative of each of the major chemical types. The results of the comparisons are listed in Appendix B. These inclusive comparisons allow evaluation of the compatibility logic with entire categories of chemical types. The output from this routine has been checked for errors by several individuals with chemistry or chemistry-related degrees. The twenty-two are described in Table 4.1 along with their chemical type descriptors. Table 4.1 Selected Chemicals For Chemical Compatibility Testing
The optical disk filed with the Department of Computing Sciences of the project documents contains a file with a complete binary comparison of all possible combinations of the entire current chemical library of 130 compounds. These 8385 comparisons were surveyed for error and checked for accuracy.
The tests showed that the rules will flag any correlation between chemicals provided the descriptors will cross reference. Corrections had to be made for the organic base in the list, triethylamine, to be slightly reactive to acetic anhydride. The cross-referencing failed to correlate that the base was reactive to an anhydrous organic acid because the bases incompatibility was for acid and not for either organic acid or acid anhydride. Other examples of this type of minor error have led to an understanding that should be conveyed to the user through demonstration and complete explanation of the software’s algorithm. |