In this study, we extracted effective factors of pipe burst from the status data of water asset, operating data of pressure, volume and etc. and 7 years' pipe burst and repair records. The extracted factors were sorted by each attribution and then a statistical analysis was performed to generate a p...
In this study, we extracted effective factors of pipe burst from the status data of water asset, operating data of pressure, volume and etc. and 7 years' pipe burst and repair records. The extracted factors were sorted by each attribution and then a statistical analysis was performed to generate a pipe burst probability function using the logistic regression model. As the result, material, diameter, length, laying year, pressure and road width affected to pipe burst significantly. Especially, in case of small diameter, laying year was most effective factor and in case of steel pipe, external loading was main cause of burst, and in case of cast iron, PE, PC, HP pipes, the deterioration of joint was main cause. The other side, as a result of Hosmer-Lemeshow goodness of fit test the models are turned out significant statistically. Also the classification criteria were determined to minimize the total cost from classification errors, when the predicted probability was more than 18% this pipe could have a chance of burst.
In this study, we extracted effective factors of pipe burst from the status data of water asset, operating data of pressure, volume and etc. and 7 years' pipe burst and repair records. The extracted factors were sorted by each attribution and then a statistical analysis was performed to generate a pipe burst probability function using the logistic regression model. As the result, material, diameter, length, laying year, pressure and road width affected to pipe burst significantly. Especially, in case of small diameter, laying year was most effective factor and in case of steel pipe, external loading was main cause of burst, and in case of cast iron, PE, PC, HP pipes, the deterioration of joint was main cause. The other side, as a result of Hosmer-Lemeshow goodness of fit test the models are turned out significant statistically. Also the classification criteria were determined to minimize the total cost from classification errors, when the predicted probability was more than 18% this pipe could have a chance of burst.
* AI 자동 식별 결과로 적합하지 않은 문장이 있을 수 있으니, 이용에 유의하시기 바랍니다.
제안 방법
(1) Above all, explanatory variables and the category of explanatory variables that was selected through the analysis of the characteristics of large-water-pipeline bursting were used. As for the variable selection, a logistic-regression model was developed using the forward selection method.
Data on pipe bursting and leakage cases were used to improve the large water pipeline management system and to facilitate pipe replacement projects. As a result, the factors that influence pipe bursting the most were identified, and each pipe bursting and leakage case was analyzed by studying the characteristics of the leakage cases.
Data Collection and Scope of Research: In this study, various data were analyzed(including GIS) to investigate and examine the documents or records of the aged water pipes of South Korea and their repair records, as well as the geologic conditions of the areas where water mains are installed.
Determination of the standard for the categorization of the pipe bursting risk : In this study, the observed values were analyzed by assigning T' for bursting and “0” for no bursting. The errors in the predicted variable found through the functional formula can be demonstrated in two ways.
In this study, a new system for leakage maintenance was developed, and the statistical percentage of water pipe bursting was determined by identifying the factor that influences pipe bursting the most. The related accidents were also analyzed by case, using the data obtained from the research that was conducted on the different cases of damage and leakage of water mains that occurred over seven years, along with the information on the present condition of the water utilities throughout South Korea.
In this study, models for leakage management and statistical bursting probability were developed. Data on pipe bursting and leakage cases were used to improve the large water pipeline management system and to facilitate pipe replacement projects.
The time of leakage was categorized into different seasons, months, and days and night. Moreover, the pipe burst probability model was designed by selecting the factors that influence pipe bursting.
It was also found that there are various reasons for pipe bursting. Pipe bursting was classified and analyzed according to the material, diameters, installation environment, water pressure, age, seasonal or monthly changes, causes of bursting, and damaged parts.
Research procedure- Various research data were used in this study: the report on the present condition of the wide pipe networks throughout South Korea, the detailed records of the water mains that were broken and that leaked from 2001 to 2007, and the data obtained from an analytical study on the present condition of the water services on GIS(the environmental factors in the areas where water pipes are installed). Research was conducted using the method shown in Fig.
This amount was determined by dividing the service expense for the regular inspection of the pipes as preventive measures by the average number of pipe bursting cases. The amount of loss incurred by the errors was computed according to the classification standard, by multiplying the amount of loss in one case with the misclassification error rate. The point at which the total amount of loss in all the cases was minimal was chosen as the classification standard.
대상 데이터
The total number of pipe bursting and leakage cases over the aforementioned seven-year period was 377; from these, damage artificially occurred in 11 cases. The specific characteristics of the pipe bursting were studied based on 326 cases, excluding the 11 cases where damage artificially occurred. The primary causes of pipe bursting were found to be aging(151 cases), defective materials(56 cases), external weight pressure(49 cases), pipe expansion and contraction caused by temperature changes(24 cases), water pressure within the pipes(21 cases), and others(25 cases).
데이터처리
Logistic regression was performed using the variable shown above, and forward selection, a stepwise regression, was chosen as the logistic-regression coefficient. As for the significance of the regression coefficient, using Wald statistics, a coefficient that was significant within the significance level of 5% was chosen.
It is believed that correctly identifying the pipes that have been damaged is crucial, for which reason the prior mistake was designated as type I error. To identify the optimal criteria of pipe bursting probability for pipe replacement and rehabilitation, the amounts of loss incurred by the two aforementioned errors were determined through approximate calculation. That is, for the type I error, as the pipes that have actually been broken were categorized into pipes that have never been broken, the amount of leaking water caused by the neglect of the pipe bursting was regarded as the amount of loss.
이론/모형
bursting were used. As for the variable selection, a logistic-regression model was developed using the forward selection method. As a result of such development, the null hypothesis that “the coefficient of the independent variables included in the models is 0” was rejected.
성능/효과
(2) In addition, the optimal classification standard was chosen to minimize the amount of loss incurred by the errors in the categorization of the pipes with past bursting records, and it was concluded that pipes with 18% bursting probability have a considerable potential to burst. This shows that the pipes with 18% bursting probability should be the first to be repaired or replaced when applying the pipe bursting probability model to the actual situation.
At the significance level of 5%, the null hypothesis of the pipe bursting model developed in this study(“No difference between the predicted dependent variable and the observed values was detected”)was not rejected because the significance value was larger than 0.05 by Chi-square statistic. Therefore, the pipe bursting model is a valid model.
Even though the external pressure from the road traffic has a relatively lower impact on the large water pipes than on the small pipes and tubes, the potential damage from such should not be ignored. It was found in this study that aging and defective materials are the major causes of leakage, accounting for 64% of the cases. It was also found that there are various reasons for pipe bursting.
참고문헌 (10)
Ahn J.C., Kim H.I., Lee G.S., Park S.H., Yu M.E., Cho H. and Koo J.Y., 2004, Predicting water pipe breaks using neural network, Proceeding of JWWA Conference
Kleiner Y. and Rajani B., 2001, comprehensive review of structural deterioration of water mains: statistical models", Urban Water, 3, pp. 131-150
American Water Works Association, 2008, Manual of Water Supply Practices M36. Water Audit and Loss Control Programs, 3rd ed., Denver, Colo.
Julian Thornton, Reinhard Sturm, George Kunkel P.E., 2008, Water Loss Control Manual Second Edition, MacGrow Hill,.
J Thornton, A Lambert, 2005, Progress in practiacal prediction of pressure: leakage, pressure: burst frequency, pressure: consumption realtionship, Leakage 2005 Conference Proceedings
J Thornton, M Shaw, M Aguiar, R Liemberger, 2005, How low cau you go? A practical approach to pressure control in low pressure systems, Leakage 2005 Conference Proceedings
Kim, J. W., Bae, C. H. and Kim, J. H., 2006, Evaluation models of deteriorated water mains for replacement/rehabilitation by field surveying, 8th Annual Water Distribution Systems Analysis Symposium
Kleiner, Y. and Rajani, B., 2002, Forecasting Variations and Trends in Water-Main Breaks, Journal of infrastructure systems, Vol 8, No 4, pp.122-131
P.R Bhave and R.Gupta, 2006, Analysis of Water Distribution Network, Alpha Science International Ltd. Oxford, U.K.
R.Sturm, J.Thornton, 2007, Water Loss Control in North America: More Cost Effective Than Customer Side Conservation - Why Wouldn't You Do It?!, Proceedings of IWA International Conference Water Loss 2007, Bucharest, Romania.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.