Due to global climate change, mega-droughts have occurred frequently. Since long-term droughts make it difficult to secure the water resources, water supply needs to be restricted in a reasonable manner. In the event of limited water supply, the waterworks need to develop a restricted water supply strategy. This study showed that analyzing daily water supply could be used to respond to the first stage of a drought. According to an analysis of Korea's major water authorities, there was about 7~21% of room for daily minimum water supply in case of a drought. Restricting the water supply by lowering pressure is a good strategy for local water authorities with high water leakage rate since leakage is inversely dependent with pressure. For this method, it is necessary to quantify water deficiency and pressure at each node using a simulation. Since DDA-based software is not possible to predict changes in demand at nodes with pressure reduction, WaterGEMS, a PDA software, was used to quantitatively predict water shortages and pressures at each node. Locations where water is deficient need to install booster pumps or to be dispatched with water tank truck and bottled water. Without these support, lowering pressure could not be an option for water works. This paper suggests a method for waterworks to plan a drought by lowering pressure to restrict water supply using daily water supply analysis and PDA based simulation.

The variation of the urban water demand and the amount of water in the distribution reservoir was studied with time for a day. The city of Kwangju in Korea was selected as a study area. The population of Kwangju in the end of 1993 was more than one million and two hundred thousand peoples. The average of daily water use in 1993 was about three hundred and fifty thousand tons a day. One day was devided into 12 divisions with a 2 hour increment. The water use demand related to the amount of water in the distribution reservoir was observed for the given time interval of a day. The water use index was defind in percentage that indicates the ratio of the amount of water use for a time interval to the amount of water use for a day. The water use index with consideration of the water stored in the reservoir was found to be useful to manage and to operate the water supply systems in real time.

본 논문에서는 상수도시설을 효율적으로 운영하는 데 필요한 1일 급수량 수요를 예측하는 방식에 대하여 인공지능(Artificial Inteligence)이라 불리는 퍼지 뉴론(fuzzy neuron)을 이용하여 연구하였다. 퍼지뉴론이란 퍼지정보(fuzzy information)를 입력으로 받아들이고 처리하는 퍼지 신경망을 일컫는 말이다. 본 연구에서는 소속함수와 퍼지규칙을 신경망으로 학습하는 기능인 적응식 학습방법을 통하여 1일 급수량을 예측하였으며 연구

수돗물, 송배수펌프의 운전 등 상수도시설을 합리적인 운용을 위해서는 일, 또는 시간 단위의 급수량 사용량의 추정이 필수적이다고 할 수 있다. 급수량의 추정방식은 회귀모형식 및 시계열 분석방법이 있는데, 본 연구에서는 시계열 분석방법인 ARIMA모형을 이용하여 일일 급수량을 추정하였으며 연구대상 지역으로는 광주광역시를 선정하였다. 일일 급수량을 추정하는데 있어서 시계열장을 15, 30. 60, 90일로 나누어 각각의 시계열장에 대해 시행착오법으로 각 모형