Date of Award
Master of Science (MS)
SUMMERS, JOSHUA D
This thesis presents a case study examining the suitability of design exemplar technology as a CAD query tool for an industrial scenario. The search and retrieval of geometrically similar mold inserts to save tooling cost for a tire manufacturing company is taken as the case study. During the implementation of the design exemplar as a CAD search and retrieval tool, several limitations of it are identified, such as the difficulty and tediousness in authoring exemplars for real world problems. To overcome these limitations, a new mathematical-based exemplar approach (mathematical model) is developed for tire mold insert retrieval. This approach calculates a set of maxima and minima based on the specifications of the target mold insert and identifies similar molds that fall within these specifications. Though the mathematical model requires less effort to author the exemplar queries than the initial boundary envelope approach, it has an unreasonably high time complexity when implemented using design exemplar. A partially developed mathematical-based exemplar takes 30 seconds to run through a sample database of 10 mold inserts. Assuming that a fully built exemplar would take more time when run on a database of 7,000 mold inserts as it would have more number of constraints added to it, the approach was not implemented through design exemplar tool. However, as this exemplar approach gives a satisfactory theoretical solution to the problem, the exemplar is hard-coded in an independent C++ program to suit the requirements at hand. The mathematical-model which is implemented in an independent C++ program successfully searches through the complete database of 7,000 tire mold inserts, retrieving similar mold inserts, in a span of one second, a huge efficiency gain when compared with the traditional exemplar system. Apart from the time reduction for the search and retrieval, the exemplar inspired program has an additional advantage in that no geometrical entities must be handled to build exemplars. However, the mathematical model may result in several false positives that must be manually eliminated. In an experiment, 15 mold inserts were randomly selected from the database and were searched for similar mold inserts using the program developed. In the experiment, it was observed that there were no false negatives and the number of false positive ranged from 0 - 10 among the 11 retrieved mold inserts. Considering the size of the database which is roughly 7000 mold inserts, the mathematical still provides advantages in search and retrieval. It is still good because 7,000 have been reduced on an average to 11. During the experiment, it as been found that 80 percent of the mold inserts which were tested have similar mold inserts. The design exemplar tool could have helped company to save the tooling cost of $184,000 per annum by providing them with a search and retrieval tool. However, it was not implemented because of some of the limitations mentioned above. Still the strategies of design exemplar are valid for querying CAD models. This is proved by the fact that the mathematical model which is an essentially an exemplar can successfully finds and retrieves similar mold inserts.
Srirangam, Narasimha murty, "A CASE STUDY ON THE USE OF DESIGN EXEMPLAR AS A SEARCH AND RETRIEVAL TOOL" (2007). All Theses. 230.