Loading...

Quantitative Risk Assessment on Water Conduits by Regression Model and Its Application in Predicting the Cost of Repairing and Replacing

Ganjeh, Arash | 2012

547 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 43094 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Eshraghniaye Jahromi, Abdolhamid
  7. Abstract:
  8. Since early civilizations underground water pipes have played a major role on promoting human being life style. Today, centuries after designing water mains equipment, utilizing a water distribution and sewage collection system is matter of great necessity in modern metropolises. In current research the pilot area is located in the city center .the major motivation for opting such area was extraordinary breakage rate which was unseen on other city areas and nearby cities. Data pertaining to four commonly used pipe material were collected. Preliminary data quality analysis was conducted to detect distinct contradictions. On basis of collected data, breakage rate plots were depicted for each variable. Drawn plots unveiled that cast iron pipes contribute substantial part of submitted breaks. Therefore a multiple linear regression model was developed for as mentioned pipes which constitute thirteen independent variables and "remaining useful life" as dependent variable. Developed model was examined by Fisher test and regression triple assumptions criteria which resulted in rejection of proposed model. In order to modify the model data was adjusted on three major steps namely: outlier's identification, normalizing transformation and pair wise correlation diagnostic. Adjusted data was fitted in multiple linear regression model and verified and accepted by two as mentioned criteria. Due to temporal and budgetary restrictions for new model data collection researcher proposed examining a subset of full model. A tradeoff between model downsizing and model predictability was considered and all possible subset regressions were fitted. Finally a subset with six independent variables was proposed
  9. Keywords:
  10. Remaining Useful Life ; Multiple Linear Regression Analysis ; Water Main ; Best Subset Selection

 Digital Object List

 Bookmark

No TOC