[학위논문]상수도관망의 누수탐지 및 잔류염소농도 최적화 모델 연구 A Study on the Optimization Models for Leakage Location Detection and Residual Chlorine Concentration in Water Supply Networks원문보기
다가오는 4차 산업혁명 시대는 정보통신기술(ICT, Information and communication technology)/사물인터넷(IoT, Internet of things)에 의해 물리적 공간과 가상의 공간이 연결되고 머신러닝과 인공지능(AI, ...
다가오는 4차 산업혁명 시대는 정보통신기술(ICT, Information and communication technology)/사물인터넷(IoT, Internet of things)에 의해 물리적 공간과 가상의 공간이 연결되고 머신러닝과 인공지능(AI, Artificial intelligence)에 의해 사물을 지능적으로 제어할 수 있는 플랫폼이 사회 전체와 모든 산업에 영향을 미치게 된다. 최근 ICT의 발전으로 상수도관망에서 발생하는 빅데이터는 물 사용량과 미세한 누수 등에서 기인하는 관망해석 상의 불확실성을 해소하고 관망의 설계 및 운영상의 다양한 문제점들을 해결하는데 활용될 수 있을 것으로 판단된다. 우리나라에서도 4차 산업혁명과 상수도 현대화 사업 등으로 관망 내 각종 센서, 스마트미터 등의 도입이 점차 확대되는 것을 고려할 때 수집된 실시간 자료들을 활용하여 관망의 설계 및 운영을 최적화할 수 있는 관망해석, 수질 모델링 등 해석기술 기반의 의사결정 모델의 개발이 필요하다. 본 연구에서는 원격검침 자료와 SCADA 시스템을 통해 수집되는 유량, 수압, 수질 자료들을 활용하여 시시각각 변동하는 물 사용량에 대한 고려와 누수량의 배분 그리고 수압과 수질을 검·보정하는 방법들을 제안하여 보다 정확한 관망해석을 실시하고자 하였다. 수리해석 시 누수에 대한 모의는 선행 연구들에서 제시한 수압-누수량 관계식의 계수들의 평균값들을 적용하여 왔다. 하지만, 매설된 관로에서의 실험을 통해 상수도관망 내에서 동일한 수압 조건이라 하더라도 매설조건에 따라 누수량이 상이하게 나타남을 검증할 수 있었다. 따라서 본 연구에서는 실험을 통해 도출된 수압-누수량 관계식과 관망 내에 설치된 원격검침 및 SCADA 데이터를 활용하여 임의 지점에서 누수가 발생하는 경우 수리해석을 통해 누수의 위치와 누수량을 탐지할 수 있는 GA기반 누수위치 추정 최적화 모델을 개발하였다. 원격검침이 실시되고 있는 250전 규모의 소블록을 대상으로 누수공의 크기(4, 6, 8, 10 mm)와 수압측정 지점수(5 ~ 10지점)에 따른 누수위치 추정 시나리오 분석 결과, 세대수 40,000, 수압 오차율 5 %, 누수량 오차율 0.1 %, 5지점의 수압 측정 조건에서 7.49 m 이내로 누수위치의 추정이 가능하였다. 중블록에 대한 모델 적용 결과, 관로 상에 임의 절점들을 추가하여 수두손실의 계산을 세밀하게 함으로써 누수량 및 누수위치 추정의 정확도를 향상시킬 수 있음을 확인할 수 있었다. 또한, 중블록 내에서 관망의 특성이 상이한 3개 지점에서의 누수모의 및 누수위치 추정 결과, 10개소의 수압 측정 조건에서 블록 유입점 인근의 수지상식 관로의 경우 평균 12.62 m 이내에서 누수위치 추정이 가능하였다. 또한, 관말 부분의 수지상식 관로의 경우에는 23.24 m 이내에서 누수위치를 추정할 수 있었다. 하지만, 격자식으로 관망이 구성된 지점에서 누수가 발생하는 경우 누수공의 크기가 증가함에 따라 거리 오차가 감소하나 평균 155.56 m의 오차가 발생하였다. 중블록의 총 관로 연장과 수요절점수를 고려할 때 관망에 설치된 1개 수압계가 120 개의 수요절점과 3.2 km의 관로를 모니터링하게 됨을 의미한다. 또한, 6 mm 누수의 경우 새벽 3시 기준으로 급수량 대비 5.73 %에 해당하는 유량으로 1,200 전 규모의 중블록의 경우 10개 수압계 설치만으로도 누수 발생시 누수 위치의 추정이 가능한 것으로 결과가 도출되었다. 다만, 격자식으로 이루어진 관망의 경우에는 누수위치 추정의 정확도 향상을 위해 수압계의 적정 개소수 및 최적 설치 위치 등을 결정을 위한 연구가 필요할 것으로 판단된다. 개발된 누수량 및 누수위치 추정 최적화 모델은 향후 스마트미터와 블록시스템이 구축된 지역에 적용함으로써, 누수음 측정, 상관식 누수탐지기의 설치 이전에 수압계, 유량계 및 스마트미터 데이터를 이용하여 누수의 징후가 나타나는 범위를 한정하는데 활용될 수 있어, 보다 경제적이고 효율적인 누수탐지에 기여할 수 있을 것으로 판단된다. 소비자의 서비스 요구수준과 물 사용패턴을 고려하여 관망 내 잔류염소농도의 시·공간적 편차를 최소화하고, 경제적인 소독이 가능한 수질관리 방안을 도출하기 위해 유전알고리즘 기반의 최적 재염소 위치 선정 및 운영 스케줄링 모델을 개발하였다. J 정수장 급수구역에 대한 모델 적용 결과 수질최악조건에서도 목표 잔류염소농도인 0.1 ~ 0.5 mg/L를 충족할 수 있는 최적 재염소시설 설치지점은 3개로 도출되었으며 총 110,991 원/d의 비용이 소요되는 것으로 나타났다. 수온이 17.1 ℃로 높은 1차 실험일에 대한 재염소시설의 최적 운영 스케줄을 도출한 결과 정수장의 잔류염소농도를 0.5 mgL로 고정한 경우(1,576 원/d)가 가변적으로 운영하는 경우(1,866 원/d)보다 더 경제적으로 운영이 가능한 것으로 나타났다. 수온이 4.1 ℃인 2차 실험의 경우 정수장에서 잔류염소농도를 가변적으로 운영하는 경우(1,126 원/d), 0.5 mg/L로 고정한 경우(1,319 원/d)보다 비용이 낮게 도출되었으며, 0.39 mgL로 고정하는 경우(1,101 원/d)에서 가장 비용이 낮게 도출되었다. 대상지역의 수온과 물 사용량에 따라 차이는 있겠으나, 안정적인 소독시설의 운영과 관망 내 잔류염소농도의 편차를 줄이는 측면에서 정수장의 잔류염소농도를 가급적 낮게 유지하고, 재염소시설의 운영을 최적화하는 방안이 바람직할 것으로 판단된다. EPANET-msx를 이용하여 J 급수구역의 THMs을 모델링한 결과 실측값과의 오차(RMSE)가 2.622 ㎍/L로 모의되어 정확도가 양호한 것으로 나타났으며, 재염소를 실시하는 경우 정수장에서 유출되는 잔류염소농도가 낮아짐에 따라 관망 내에서 생성되는 THMs가 저감될 수 있음을 확인할 수 있었다. 따라서 본 연구에서 개발된 잔류염소농도 최적화 모델은 수질 만족도 측면에서의 잔류염소농도 관리뿐만 아니라 수질 안전성 확보를 위한 THMs 농도 관리에도 활용할 수 있을 것으로 판단된다.
다가오는 4차 산업혁명 시대는 정보통신기술(ICT, Information and communication technology)/사물인터넷(IoT, Internet of things)에 의해 물리적 공간과 가상의 공간이 연결되고 머신러닝과 인공지능(AI, Artificial intelligence)에 의해 사물을 지능적으로 제어할 수 있는 플랫폼이 사회 전체와 모든 산업에 영향을 미치게 된다. 최근 ICT의 발전으로 상수도관망에서 발생하는 빅데이터는 물 사용량과 미세한 누수 등에서 기인하는 관망해석 상의 불확실성을 해소하고 관망의 설계 및 운영상의 다양한 문제점들을 해결하는데 활용될 수 있을 것으로 판단된다. 우리나라에서도 4차 산업혁명과 상수도 현대화 사업 등으로 관망 내 각종 센서, 스마트미터 등의 도입이 점차 확대되는 것을 고려할 때 수집된 실시간 자료들을 활용하여 관망의 설계 및 운영을 최적화할 수 있는 관망해석, 수질 모델링 등 해석기술 기반의 의사결정 모델의 개발이 필요하다. 본 연구에서는 원격검침 자료와 SCADA 시스템을 통해 수집되는 유량, 수압, 수질 자료들을 활용하여 시시각각 변동하는 물 사용량에 대한 고려와 누수량의 배분 그리고 수압과 수질을 검·보정하는 방법들을 제안하여 보다 정확한 관망해석을 실시하고자 하였다. 수리해석 시 누수에 대한 모의는 선행 연구들에서 제시한 수압-누수량 관계식의 계수들의 평균값들을 적용하여 왔다. 하지만, 매설된 관로에서의 실험을 통해 상수도관망 내에서 동일한 수압 조건이라 하더라도 매설조건에 따라 누수량이 상이하게 나타남을 검증할 수 있었다. 따라서 본 연구에서는 실험을 통해 도출된 수압-누수량 관계식과 관망 내에 설치된 원격검침 및 SCADA 데이터를 활용하여 임의 지점에서 누수가 발생하는 경우 수리해석을 통해 누수의 위치와 누수량을 탐지할 수 있는 GA기반 누수위치 추정 최적화 모델을 개발하였다. 원격검침이 실시되고 있는 250전 규모의 소블록을 대상으로 누수공의 크기(4, 6, 8, 10 mm)와 수압측정 지점수(5 ~ 10지점)에 따른 누수위치 추정 시나리오 분석 결과, 세대수 40,000, 수압 오차율 5 %, 누수량 오차율 0.1 %, 5지점의 수압 측정 조건에서 7.49 m 이내로 누수위치의 추정이 가능하였다. 중블록에 대한 모델 적용 결과, 관로 상에 임의 절점들을 추가하여 수두손실의 계산을 세밀하게 함으로써 누수량 및 누수위치 추정의 정확도를 향상시킬 수 있음을 확인할 수 있었다. 또한, 중블록 내에서 관망의 특성이 상이한 3개 지점에서의 누수모의 및 누수위치 추정 결과, 10개소의 수압 측정 조건에서 블록 유입점 인근의 수지상식 관로의 경우 평균 12.62 m 이내에서 누수위치 추정이 가능하였다. 또한, 관말 부분의 수지상식 관로의 경우에는 23.24 m 이내에서 누수위치를 추정할 수 있었다. 하지만, 격자식으로 관망이 구성된 지점에서 누수가 발생하는 경우 누수공의 크기가 증가함에 따라 거리 오차가 감소하나 평균 155.56 m의 오차가 발생하였다. 중블록의 총 관로 연장과 수요절점수를 고려할 때 관망에 설치된 1개 수압계가 120 개의 수요절점과 3.2 km의 관로를 모니터링하게 됨을 의미한다. 또한, 6 mm 누수의 경우 새벽 3시 기준으로 급수량 대비 5.73 %에 해당하는 유량으로 1,200 전 규모의 중블록의 경우 10개 수압계 설치만으로도 누수 발생시 누수 위치의 추정이 가능한 것으로 결과가 도출되었다. 다만, 격자식으로 이루어진 관망의 경우에는 누수위치 추정의 정확도 향상을 위해 수압계의 적정 개소수 및 최적 설치 위치 등을 결정을 위한 연구가 필요할 것으로 판단된다. 개발된 누수량 및 누수위치 추정 최적화 모델은 향후 스마트미터와 블록시스템이 구축된 지역에 적용함으로써, 누수음 측정, 상관식 누수탐지기의 설치 이전에 수압계, 유량계 및 스마트미터 데이터를 이용하여 누수의 징후가 나타나는 범위를 한정하는데 활용될 수 있어, 보다 경제적이고 효율적인 누수탐지에 기여할 수 있을 것으로 판단된다. 소비자의 서비스 요구수준과 물 사용패턴을 고려하여 관망 내 잔류염소농도의 시·공간적 편차를 최소화하고, 경제적인 소독이 가능한 수질관리 방안을 도출하기 위해 유전알고리즘 기반의 최적 재염소 위치 선정 및 운영 스케줄링 모델을 개발하였다. J 정수장 급수구역에 대한 모델 적용 결과 수질최악조건에서도 목표 잔류염소농도인 0.1 ~ 0.5 mg/L를 충족할 수 있는 최적 재염소시설 설치지점은 3개로 도출되었으며 총 110,991 원/d의 비용이 소요되는 것으로 나타났다. 수온이 17.1 ℃로 높은 1차 실험일에 대한 재염소시설의 최적 운영 스케줄을 도출한 결과 정수장의 잔류염소농도를 0.5 mgL로 고정한 경우(1,576 원/d)가 가변적으로 운영하는 경우(1,866 원/d)보다 더 경제적으로 운영이 가능한 것으로 나타났다. 수온이 4.1 ℃인 2차 실험의 경우 정수장에서 잔류염소농도를 가변적으로 운영하는 경우(1,126 원/d), 0.5 mg/L로 고정한 경우(1,319 원/d)보다 비용이 낮게 도출되었으며, 0.39 mgL로 고정하는 경우(1,101 원/d)에서 가장 비용이 낮게 도출되었다. 대상지역의 수온과 물 사용량에 따라 차이는 있겠으나, 안정적인 소독시설의 운영과 관망 내 잔류염소농도의 편차를 줄이는 측면에서 정수장의 잔류염소농도를 가급적 낮게 유지하고, 재염소시설의 운영을 최적화하는 방안이 바람직할 것으로 판단된다. EPANET-msx를 이용하여 J 급수구역의 THMs을 모델링한 결과 실측값과의 오차(RMSE)가 2.622 ㎍/L로 모의되어 정확도가 양호한 것으로 나타났으며, 재염소를 실시하는 경우 정수장에서 유출되는 잔류염소농도가 낮아짐에 따라 관망 내에서 생성되는 THMs가 저감될 수 있음을 확인할 수 있었다. 따라서 본 연구에서 개발된 잔류염소농도 최적화 모델은 수질 만족도 측면에서의 잔류염소농도 관리뿐만 아니라 수질 안전성 확보를 위한 THMs 농도 관리에도 활용할 수 있을 것으로 판단된다.
In the coming era of the 4th industrial revolution, physical space and virtual space will be connected together by ICT (Information and communication technology) and IoT (Internet of things), while the whole society and all the industries will be affected by the platforms that can intelligently cont...
In the coming era of the 4th industrial revolution, physical space and virtual space will be connected together by ICT (Information and communication technology) and IoT (Internet of things), while the whole society and all the industries will be affected by the platforms that can intelligently control various things through machine learning and AI (Artificial intelligence). The recent advancement of ICT has improved the capabilities of data collection and storage, of transmission through communication, and of data processing and computation. This enables fast analysis of a large amount of data to deduce significant information. The potential of utilizing big data for water supply system has been steadily discussed for a long time. However, Korea is at a low level in developing analytical technology-based softwares that utilize collected data to enable optimal decision making through hydraulic analysis, leakage detection, water quality modeling, and the like. The factors of the highest uncertainty in existing water supply systems have been considered to be water consumption changing every moment, as well as flow rate and pressure affecting water supply network management. In addition, such a fine leakage as not to be exposed to the surface of the ground also contributes to uncertainty in water supply network analysis, thus making water supplier difficult to make an optimal decision. Such hydraulic uncertainty in water consumption and leakage in turn affects water quality in pipe network. There are also some areas for which pipe diameter was designed to be excessively large, because water supply plans had been prepared in anticipation of steady growth in urban population and water demand. Another current problem is large difference in residual chlorine concentration in supplied water between customers residing near water treatment plant and those residing at end node of the pipe. This problem is due to the appearance of low flow rate sections and increase of retention time caused by design the block system in water supply networks. Therefore—at least in order to improve water quality and pipe network design/operation by means of securing proper flow rate, and modeling of residual chlorine and disinfection by-products concentrations—it is demanded to develop a method that can reduce the uncertainty, and so increase the accuracy, of hydraulic analysis through the utilization of customer water consumption data. Accordingly, an effort was made to propose a method that can minimize the uncertainty affecting the accuracy of hydraulic analysis—by considering water consumption that changes every moment and by reasonable distribution of leakage and calibration of water pressure and water quality, through the utilization of automatic meter reading data as well as the data about flow rate, pressure and water quality collected through the SCADA system. The proposed method, by utilizing the Emitter function of EPANET 2.0, performs leakage distribution according to water pressure at node; and enables the calculation and distribution of leakage even when water pressure changes because of water consumption change and leakage occurrence. Conventional strategies for leakage and water quality management are deduced based on manager’s experience, or on data about monthly average water consumption and daily water supply. The proposed method is expected to make such strategies obsolete, enabling optimal decision making through near real-time water supply network analysis. Preceding reports on leakage in buried pipe according to water pressure and leak size showed that the coefficient and power of the water pressure-leakage relation could vary greatly according to pipe burial condition and leak size. Therefore it was considered necessary to calculate coefficient through on-site experiments rather than indiscriminantly apply the average value presented by those reports. It was judged that the water pressure-leakage relation obtained through experiments on the actual pipe burial sites of the study area could be utilized for the proper simulation of hydraulic analysis-based leakage distribution or leakage occurrence. Accordingly, in this research, emitter coefficient according to leak nozzle size was deduced through leakage occurrence simulation for the actual pipeline of the target area, based on measured leakage according to water pressure in buried pipe as well as on the water pressure-leakage relation. In addition, a hydraulic analysis-based optimization model was developed that enables the estimation of the amount and location of leakage when leakage occurs in a block where flow rate and water pressure can be monitored, through more precise hydraulic analysis that utilizes automatic reading and SCADA data. To verify the optimization model of leakage location detection developed for the actual water distribution network, scenario analysis was carried out involving 2 blocks where automatic meter reading was available: a small size block with 250 households, and a medium size block with 1,200 households. In the case of the small size block, the result of the scenario analysis according to leak sizes (4, 6, 8, 10mm) and the number of water pressure measurement points (5 ~ 10 points) showed that location estimation was possible with an error of less than 7.49 m under the following condition: 40,000 generation, a 5% water pressure error rate, a 0.1% leakage error rate, and 5 water pressure measurement points. In the case of the medium size block, the result of applying the model demonstrated that it was possible to improve the estimation accuracy of the amount and location of leakage through the reduction of pipe length between nodes by adding arbitrary virtual nodes along the pipe, and thus through the detailed calculation of head loss per unit length. With the subdivided pipe network map of the medium size block having reduced pipe length, model verification was carried out by applying a genetic algorithm which involves additional generation (200,000), an adjusted leakage error rate (0.05%) and an adjusted water pressure error rate (0.05%), and utilizing water pressure data measured at 10 points. The result showed that it is possible to estimate leakage location with an average error of less than 12.62 m if leakage occurs in the tree type pipe network near the block entrance with an leak size of 6, 8, or 10mm. In case leakage occurs in the tree type pipe network at the pipe end, leakage location estimation is still possible with an error of less than 23.24 m. However, if leakage occurs at a location in a looped pipe network, although distance error decreases as leak size increases, the average distance error was shown to be 155.56 m. In this case, therefore, it is judged necessary to install additional water pressure gauges and adjust their locations. Based on model application to actual pipe network, this research could deduce the conclusion that it is desirable for selected actual water pressure measurement location to be at the downstream of leakage simulation location, in consideration of the direction of water flow. However, there is no standard for choosing the proper installment locations and number of water pressure gauges. Therefore follow-up research in this regard would lead to the improvement of the accuracy of leakage location estimation. In addition, it is considered that future research applying the developed model to an area where smart meter and block system have been installed would contribute to improving leakage detection efficiency by narrowing the possible area of detection beforehand, if that area adopts existing leak sound measurement and correlation type leak detector. An optimization model for locating and operating re-chlorination facility was developed that can minimize spatial and temporal deviation of residual chlorine concentration, enabling economical disinfection, based on an optimization technique utilizing a genetic algorithm and a water quality analysis model that consider consumer water consumption pattern. The application of the model deduced 3 optimal boosting points that allow the achievement of the target residual chlorine concentration (0.1 ~ 0.5 mg/L) even under the worst water quality condition of the service area of J water treatment plant; while the calculation of the cost for re-chlorination facility installation and operation and for chlorine disinfection at water treatment plant resulted in 110,991 won/d. An optimal boosting operation scheduling model that considers consumer water consumption pattern was developed. This model successfully demonstrated that more economical operation satisfying target water quality standard can be achieved by optimizing residual chlorine concentration at water treatment plant and re-chlorination facility than by maintaining a fixed concentration during chlorine disinfection and re-chlorination. According to the result of the 1st experiment, in which measured water temperature at water treatment plant was 17.1℃, the average residual chlorine concentration in efflux from J water treatment plant was 1.18 mg/L, and disinfection cost at water treatment plant was 2,554 won/d. According to the result of deducing the residual chlorine concentration schedule for the water treatment plant and re-chlorination facility that were suitable for the 1st experiment, operation satisfying residual chlorine concentration standard for demand node was possible at lower cost by fixing residual chlorine concentration at water treatment plant (1,576 won/d) than by not fixing it (1,866 won/d). According to the result of the 2st experiment, in which measured water temperature at water treatment plant was 4.1℃, the average residual chlorine concentration in efflux from J water treatment plant was 1.05 mg/L, and disinfection cost at water treatment plant was 2,232 won/d. According to the result of deducing the residual chlorine concentration schedule for the water treatment plant and re-chlorination facility that were suitable for the 2st experiment, operation satisfying residual chlorine concentration standard for demand node was possible at lower cost by fixing residual chlorine concentration at water treatment plant (1,319 won/d) than by not fixing it (1,101 won/d). Therefore, if a water treatment plant can measure and control residual chlorine concentration in consideration of hourly water consumption and its water level and disinfection capability, the utilization of the developed program is expected to enable efficient disinfection and re-chlorination facility operation regardless of hourly changes in water consumption and seasonal changes in water temperature. By offering the ability to maintain residual chlorine concentration at water treatment plant and in pipe network low enough to satisfy water quality standard, the developed program is also expected to serve as an efficient tool for coping with decreasing drinking rate due to disinfection by-products and chlorine odor.
In the coming era of the 4th industrial revolution, physical space and virtual space will be connected together by ICT (Information and communication technology) and IoT (Internet of things), while the whole society and all the industries will be affected by the platforms that can intelligently control various things through machine learning and AI (Artificial intelligence). The recent advancement of ICT has improved the capabilities of data collection and storage, of transmission through communication, and of data processing and computation. This enables fast analysis of a large amount of data to deduce significant information. The potential of utilizing big data for water supply system has been steadily discussed for a long time. However, Korea is at a low level in developing analytical technology-based softwares that utilize collected data to enable optimal decision making through hydraulic analysis, leakage detection, water quality modeling, and the like. The factors of the highest uncertainty in existing water supply systems have been considered to be water consumption changing every moment, as well as flow rate and pressure affecting water supply network management. In addition, such a fine leakage as not to be exposed to the surface of the ground also contributes to uncertainty in water supply network analysis, thus making water supplier difficult to make an optimal decision. Such hydraulic uncertainty in water consumption and leakage in turn affects water quality in pipe network. There are also some areas for which pipe diameter was designed to be excessively large, because water supply plans had been prepared in anticipation of steady growth in urban population and water demand. Another current problem is large difference in residual chlorine concentration in supplied water between customers residing near water treatment plant and those residing at end node of the pipe. This problem is due to the appearance of low flow rate sections and increase of retention time caused by design the block system in water supply networks. Therefore—at least in order to improve water quality and pipe network design/operation by means of securing proper flow rate, and modeling of residual chlorine and disinfection by-products concentrations—it is demanded to develop a method that can reduce the uncertainty, and so increase the accuracy, of hydraulic analysis through the utilization of customer water consumption data. Accordingly, an effort was made to propose a method that can minimize the uncertainty affecting the accuracy of hydraulic analysis—by considering water consumption that changes every moment and by reasonable distribution of leakage and calibration of water pressure and water quality, through the utilization of automatic meter reading data as well as the data about flow rate, pressure and water quality collected through the SCADA system. The proposed method, by utilizing the Emitter function of EPANET 2.0, performs leakage distribution according to water pressure at node; and enables the calculation and distribution of leakage even when water pressure changes because of water consumption change and leakage occurrence. Conventional strategies for leakage and water quality management are deduced based on manager’s experience, or on data about monthly average water consumption and daily water supply. The proposed method is expected to make such strategies obsolete, enabling optimal decision making through near real-time water supply network analysis. Preceding reports on leakage in buried pipe according to water pressure and leak size showed that the coefficient and power of the water pressure-leakage relation could vary greatly according to pipe burial condition and leak size. Therefore it was considered necessary to calculate coefficient through on-site experiments rather than indiscriminantly apply the average value presented by those reports. It was judged that the water pressure-leakage relation obtained through experiments on the actual pipe burial sites of the study area could be utilized for the proper simulation of hydraulic analysis-based leakage distribution or leakage occurrence. Accordingly, in this research, emitter coefficient according to leak nozzle size was deduced through leakage occurrence simulation for the actual pipeline of the target area, based on measured leakage according to water pressure in buried pipe as well as on the water pressure-leakage relation. In addition, a hydraulic analysis-based optimization model was developed that enables the estimation of the amount and location of leakage when leakage occurs in a block where flow rate and water pressure can be monitored, through more precise hydraulic analysis that utilizes automatic reading and SCADA data. To verify the optimization model of leakage location detection developed for the actual water distribution network, scenario analysis was carried out involving 2 blocks where automatic meter reading was available: a small size block with 250 households, and a medium size block with 1,200 households. In the case of the small size block, the result of the scenario analysis according to leak sizes (4, 6, 8, 10mm) and the number of water pressure measurement points (5 ~ 10 points) showed that location estimation was possible with an error of less than 7.49 m under the following condition: 40,000 generation, a 5% water pressure error rate, a 0.1% leakage error rate, and 5 water pressure measurement points. In the case of the medium size block, the result of applying the model demonstrated that it was possible to improve the estimation accuracy of the amount and location of leakage through the reduction of pipe length between nodes by adding arbitrary virtual nodes along the pipe, and thus through the detailed calculation of head loss per unit length. With the subdivided pipe network map of the medium size block having reduced pipe length, model verification was carried out by applying a genetic algorithm which involves additional generation (200,000), an adjusted leakage error rate (0.05%) and an adjusted water pressure error rate (0.05%), and utilizing water pressure data measured at 10 points. The result showed that it is possible to estimate leakage location with an average error of less than 12.62 m if leakage occurs in the tree type pipe network near the block entrance with an leak size of 6, 8, or 10mm. In case leakage occurs in the tree type pipe network at the pipe end, leakage location estimation is still possible with an error of less than 23.24 m. However, if leakage occurs at a location in a looped pipe network, although distance error decreases as leak size increases, the average distance error was shown to be 155.56 m. In this case, therefore, it is judged necessary to install additional water pressure gauges and adjust their locations. Based on model application to actual pipe network, this research could deduce the conclusion that it is desirable for selected actual water pressure measurement location to be at the downstream of leakage simulation location, in consideration of the direction of water flow. However, there is no standard for choosing the proper installment locations and number of water pressure gauges. Therefore follow-up research in this regard would lead to the improvement of the accuracy of leakage location estimation. In addition, it is considered that future research applying the developed model to an area where smart meter and block system have been installed would contribute to improving leakage detection efficiency by narrowing the possible area of detection beforehand, if that area adopts existing leak sound measurement and correlation type leak detector. An optimization model for locating and operating re-chlorination facility was developed that can minimize spatial and temporal deviation of residual chlorine concentration, enabling economical disinfection, based on an optimization technique utilizing a genetic algorithm and a water quality analysis model that consider consumer water consumption pattern. The application of the model deduced 3 optimal boosting points that allow the achievement of the target residual chlorine concentration (0.1 ~ 0.5 mg/L) even under the worst water quality condition of the service area of J water treatment plant; while the calculation of the cost for re-chlorination facility installation and operation and for chlorine disinfection at water treatment plant resulted in 110,991 won/d. An optimal boosting operation scheduling model that considers consumer water consumption pattern was developed. This model successfully demonstrated that more economical operation satisfying target water quality standard can be achieved by optimizing residual chlorine concentration at water treatment plant and re-chlorination facility than by maintaining a fixed concentration during chlorine disinfection and re-chlorination. According to the result of the 1st experiment, in which measured water temperature at water treatment plant was 17.1℃, the average residual chlorine concentration in efflux from J water treatment plant was 1.18 mg/L, and disinfection cost at water treatment plant was 2,554 won/d. According to the result of deducing the residual chlorine concentration schedule for the water treatment plant and re-chlorination facility that were suitable for the 1st experiment, operation satisfying residual chlorine concentration standard for demand node was possible at lower cost by fixing residual chlorine concentration at water treatment plant (1,576 won/d) than by not fixing it (1,866 won/d). According to the result of the 2st experiment, in which measured water temperature at water treatment plant was 4.1℃, the average residual chlorine concentration in efflux from J water treatment plant was 1.05 mg/L, and disinfection cost at water treatment plant was 2,232 won/d. According to the result of deducing the residual chlorine concentration schedule for the water treatment plant and re-chlorination facility that were suitable for the 2st experiment, operation satisfying residual chlorine concentration standard for demand node was possible at lower cost by fixing residual chlorine concentration at water treatment plant (1,319 won/d) than by not fixing it (1,101 won/d). Therefore, if a water treatment plant can measure and control residual chlorine concentration in consideration of hourly water consumption and its water level and disinfection capability, the utilization of the developed program is expected to enable efficient disinfection and re-chlorination facility operation regardless of hourly changes in water consumption and seasonal changes in water temperature. By offering the ability to maintain residual chlorine concentration at water treatment plant and in pipe network low enough to satisfy water quality standard, the developed program is also expected to serve as an efficient tool for coping with decreasing drinking rate due to disinfection by-products and chlorine odor.
AI-Helper
※ AI-Helper는 부적절한 답변을 할 수 있습니다.