分流式下水道最適化建設經濟分析及水量水質預測 模式之研究

Economic Analyses for Optimizing the Construction of Separate Sewer and Water Quantity/Water Quality Prediction Modelling

陳宏銘(Home-Ming Chen)

指導教授 ： 駱尚廉

國立臺灣大學/工學院/環境工程學研究所/博士(2010年)

https://doi.org/10.6342/NTU.2010.01550

摘要

為加速對河川水質改善，分析下水道管網最適化建設，及配合管網之逐步建設而建立污水處理廠水量、水質預測模式，為本研究之兩大重點。最適化建設經濟分析為以即將完成全面用戶接管之下水道系統為案例，採用邊際防治成本(MCC)等於邊際防治效益（MBC）之觀念，輔以河川BOD5為指標，以探討分流式下水道之最適化用戶接管普及率及最適化經濟建設。
分流式下水道建設期間，由於污水處理廠仍為初期營運階段，為提升進入污水處理廠之水量及其污染濃度，以加速對河川水質之改善，除原有之用戶接管工作外，常輔以增建簡易合流式下水道截流設施以加速達成之。但此段期間，污水處理廠之進廠水量、水質，仍常不如預期，且其變異性頗大，不易穩定操作，故本研究根據每年之管網建設資料，如用戶接管普及率、污水處理率及進出廠水量、水質等參數，採用倒傳遞類神經網路（BPNN），建立進出污水處理廠之水量、水質等4種基本預測模式（即A0、A1、A2及A3），並與灰色模式（GM）比較，前述基本模式經結合，如A1與A2結合即為A1+A2多階倒傳遞類神經網路（MBPNN），此多階模式可提供更多之訊息以利污水處理廠之操作及控制。
當達到最適化用戶接管普及率時，河川BOD5指標，已由嚴重或中度污染改善達到接近輕度污染，故辦理分流式下水道建設，當達到最適化用戶接管普及率時，有其建設之成效。同時該分流式下水道，考量加速對河川水質改善，除持續用戶接管工作外，建議餘人口較稀疏未接管區域可因地制宜採用建設成本較低之簡易合流式下水道截流設施，統稱為階段性結合式下水道系統（Hybrid sewer system），並配合不易接管區域之聚落污水處理設施，處理都市廢污水。
污水處理廠放流水質預測模式（A0）與灰色模式之預測效果接近，惟前者需大量資料，後者各分項水質如BOD5、COD及SS等，僅需少量資料即可達到預測效果，由每年建設之管網資料，所建立之進廠水量、水質預測模式（A1），及由進廠水量、水質建立之放流水量、水質預測模式（A2），以及由進出廠水量、水質建立之污泥預測模式（A3），經研究分析均獲得甚佳之預測效果，最後由A1及A2所結合建立之多階模式（Multi-model）可直接由建設使用中之下水道系統之相關建設參數預測放流水量及水質。經綜合分析各個模式之預測效果，上述各模式對水量Q、BOD5、污泥量之預測結果最佳。

關鍵字

邊際防治成本 ； 邊際防治效益 ； 結合式下水道 ； 倒傳遞類神經網路 ； 灰色模式

並列摘要

The objective of this study was to apply the concept that Marginal Cost of Control (MCC) equals to Marginal Benefits of Control (MBC) to develop a method for studying the optimal percentage of household connection to a separate sewer system and the most cost-effective construction of the separate sewer system. Mathematical models were also developed to provide useful information for operating the end-of-the-pipe wastewater treatment plants to meet discharge standards, and for managing the water quality of water bodies that receive the effluent discharges from hybrid sewer systems.
Base on the progress of the construction, back-propagation neural network (BPNN) was applied to predict the wastewater quantity and quality. Four basic models are included in this network: (1) A0 （PIQ）model for predicting influent quality, (2) A1（PIQQ）model for predicting influent quantity and quality, (3) A2（PEQQ）model for predicting effluent quantity and quality, and (4) A3（PQWCWS）model for predicting the quantity and water content of waste sludge. The multi-model (A1+A2), a multi-back-propagation neural network (MBPNN) formed by combining the A1 and A2 models, was used for estimating A2 output parameters by using A1 input parameters directly. Comparing to the A0 model, the predicting results suggest that GM (Grey model) can be used to predict the variation of municipal effluent with insufficient effluent data. The results also indicate that BPNN (back-propagation neural network) and MBPNN are suitable for predicting the wastewater quantity and quality, especially for Q, BOD5, sludge amount, and the water content of sludge in an under-constructed sewer system.
The validity and applicability of the method proposed in this study have been demonstrated by analyzing the optimal household connection percentage to assess the most cost-effective construction of the separate sewer. The results of that the receiving water quality can be improved in a cost-effective manner. The optimal percentage of household connection to the separate sewer will lead to the most cost-effective stage when the stream Biochemical Oxygen Demand (BOD5) meets the water quality standards. For more accurate analyses, the effect of other factors such as human health protection, and animal and plant production should be quantified. The Scenario Analysis Method can be applied for evaluating the total benefits of control (TBC). Once the economic cost of construction is calculated, the relatively more expensive section of the separate sewer will not be constructed. Instead, it will be switched over to a less expensive combined sewer system to make the whole system a hybrid sewer system.
This study also reveals that during the initial construction phase of the separate sewer more household connection will lead to significant BOD5 reduction in the receiving water body. However, at a later stage, additional increase of the household connection will not further improve the river quality as much as it has previously; the receiving body water quality will reach a steady state thereafter. The receiving river water quality as expressed by BOD5 is improved from near “serious pollution” to “moderate pollution”, and it continues to approach “light pollution” when the optimal household connection was reached. This concept of the hybrid sewer system has been implemented for the other cities to alleviate the financial burden of constructing the sewer system.

並列關鍵字

marginal benefits of control (MBC) ； marginal cost of control (MCC) ； hybrid sewer system ； back-propagation neural network (BPNN) ； grey model

參考文獻

75. 邱智慧，2006，徵收水污染防治費對河川水質之影響，國立台灣大學環境工程研究所，碩士論文。

82. 盧瑞山、駱尚廉，2000，類神經網路於土壤復育工程之應用，財團法人中興工程顧問社。

1. Achleitner, S., Möderl, M., and Rauch, W. (2007). City drain - an open source approach for simulation of integrated urban drainage system. Environmental Modelling & Software, 22(8), 1184-1195.

2. APHA, AWWA, WEF, (1995). Standard Methods for the Examination of Water and Wastewater, 19th ed. American Public Health Association/American Water Works Association/Water Environment Federation, Washington DC.

3. Benedetti, L., Dirckx, G., Bixio, D., Thoeye, C., and Vanrolleghem, P.A. (2008). Environmental and economic performance of the integrated urban wastewater system. Journal of Environmental Management, 88(4), 1262-1272.