TY  - THES
T1  - A Statistical analysis on Asahi Glass Philippines, Incorporated
A1  - De Torres, Donna
LA  - English
PP  - Diliman, Quezon City
PB  - College of Engineering
YR  - 2002
UL  - https://tuklas.up.edu.ph/Record/UP-99796217610073983
AB  - The study was conducted on ASAHI GLASS PHILIPPINES INCORPORATED in Bo. Pinagbuhatan Pasig City. ASAHI is the sole flat glass manufacturer in the Philippines. The study focused on the 2.1 mm thick glass with laminating quality due to a relatively high rejection rate. The primary objective is to focus on at least three problem areas on this product, to apply statistical techniques to analyze each and to recommend solutions based on the conclusions. By making use of the data available to the company and by the use of pareto analysis, the problem areas were identified. The three problems considered are the defect due to stones, defect due to scratches, defect due to digs and the problem on thickness variations. Each of them was provided with ishikawa diagrams to identify their root causes. For the defect due to stones, a model for randomized complete block design was performed in order to know whether lehr speed has an effect on number of stones or not. This makes use of the region as blocking variable. From the computations, lehr speed has an effect on glass outputs and in order to know which specific level of lehr speed produces a significantly different output, Duncan's Multiple Range Test was done. It was found that the level of 729 produces a significantly different output. Lehr Speed setting of 688 and 728 produces a lesser number of stones in the glass. Thus, these were recommended as best lehr speed setting. For the defect due to scratches, randomized complete block design was also used with brance as the factor and shift as the blocking variable. From the results obtained, it was found that the shift has an effect on the number of scratches. To know whether the branch and the shift are dependent on each other, a test for independence of factors was performed and the result was that they are independent on each other. To know which particular shift produces glass with significantly different number of scratches, DMRT was also performed. The 6-2 shift produces a lesser number of scratches on output. Thus, it is recommended that the best practices of this shift serve as the model for the other two shifts. The problem on defect due to digs also made use of the randomized complete block design with factor branch and blocking variable shift. The results showed that neither branch nor shift affects the number of digs in the glass. It is recommended that the company study the transportation condition because it might be the cause of digs. For the problem of thickness variations, three factors were identified such as the temperature, pressure, and pull speed but due to data availability, a three factor factorial design was not performed. Instead, temperature and pressure were studied using the two factor analysis. It was found that both of them do not affect thickness output. For the pull speed, single factor (one-way classification) was done and the results showed that it doesn't affect thickness output. Thus, it is recommended that these levels be maintained.
NO  - Feasibility study
CN  - LG 993 2002 E66 D453
KW  - Statistics.
ER  -