分區計量管網之水壓模擬方法

Abstract

台灣地區公共給水系統基礎建設已趨於完善，都市化及公共給水普及率高，但管線破管漏水或管網水質污染事件仍然偶有發生。其中用戶用水行為是重要管網管理上的評估項目之一，掌握用戶用水行為，可以比對是否當前用水已提供足夠的壓力，找出可以滿足用戶需求的最低供水壓力，不僅能減少配水管網輸送過多無謂的水也能降低漏水或破管的發生。目前臺灣水事業單位在記錄用戶需求上，通常採取定期抄表紀錄用戶用水量，但由於人員抄表的誤差以及間隔2個月的抄表頻率，致使無法有效掌握用戶用水行為。 本研究透過蒐集小區管網之基本資料、記錄進水點之流量及壓力數據，輔以售水率與用戶調查分類去建立用戶需水量推估方法，推估現階段用戶用水行為，EPANET 2.0模式能預估未來一星期之節點水壓，作為未來管網管理評估上的重要參考指標。透過實測節點壓力與預估節點壓力值之驗證，相關性(R2)高達0.83，顯示本研究之推估方法確實能夠模擬出此區域之預估節點水壓，且經過測試不同壓力區段下之結果，可得最適預測壓力範圍落在1.5 kg/cm2至4.7 kg/cm2之間。

Water distribution system (WDS) is one of the most important complex infrastructures in urban areas. However, aging infrastructure and high growing of water demand may damage the pipe, causing leakage and contamination. The water consumption behavior of consumers is an important index for water utilities. Understanding the behavior of user consumption not only can understand the current status of WDS (i.e. pressure), but also can figure out the minimum operating pressure that satisfies consumers. Minimum operating pressure can be used to decrease water leakage and reduce the occurrence of pipe burst. At present, water utilities in Taiwan usually use personnel meter reading to record consumer’s consumption. However, its acquisition data of WDS is hard to be analyzed because of occasional errors. This study aimed to evaluate the nodal pressure in DMA for WDS data analysis in Zhubei City. In order to figure out water consumption behavior and minimum operating pressure, the pipe network simulation software EPANET 2.0 was used to simulate nodal pressure of district metered areas. The inflow data of distribution network were acquired by filtering the average leak rate. This method could totally assume the consumer’s water demand pattern. After testing with real data, this method showed good predictability (R2 = 0.86) to the nodal pressure in a future week. Moreover, testing by mean absolute percentage error showed the predict pressure error was less than 13%. Our stimulation model of water demand for Zhubei City’s DMA performed that nodal pressure greater than 1.5 kg/cm2 was a critical condition for the observation of good predictability (R2 > 0.7).