Available online at www.sciencedirect.com
ScienceDirect Procedia Engineering 154 (2016) 58 – 61
12th International Conference on Hydroinformatics, HIC 2016
Decision support system for daily and long term operations of the system of Milan, Italy Quan Pana*, Mario Erik Castro-Gamaa, Andreja Jonoskia, Ioana Popescua a
UNESCO-IHE, Institute for Water Education, Westvest 7, Delft, 2611AX, Netherlands
Abstract This study introduces the development of a Decision Support System (DSS) for daily and long term operations of the Water Distribution Network of Milan operated by the utility Metropolitana Milanese S.p.A (MM), developed during the European project ICeWater. The DSS has been developed for the two main problems of the utility by applying multi-objective optimization for pump scheduling and sectorization of the system. The DSS was built based on open source software on the server and the client side, making its applicability to other utilities possible. This paper presents the architecture of the DSS components and shows the advantages in the application of such tool in the operation for MM. A test for validation of the DSS for pump scheduling has been applied in a subsector named Abbiategrasso. Some results are presented showing the benefits for the utility by using the DSS in their daily operations. © by Elsevier Ltd. by This is an open 2016Published The Authors. Published Elsevier Ltd.access article under the CC BY-NC-ND license © 2016 (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of HIC 2016. Peer-review under responsibility of the organizing committee of HIC 2016 Keywords: Water Distribution Networks, Decision Support System, Milan (Italy).
1. Introduction Numerous decision support systems have been widely developed in the previous decades. A great number of tools to support the complex decision making processes have been applied for policy makers, stakeholders and scientists. The existing tools range from simple spreadsheets, databases and networks, to geographical information systems, expert systems, virtual reality systems, intelligent agents and machine learning ensembles which have been widely used as decision support systems [1, 2]. Fast delivery of accurate information has become the crucial factor
* Corresponding author. Tel.: +0-000-000-0000 ; fax: +0-000-000-0000 . E-mail address:
[email protected]
1877-7058 © 2016 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of HIC 2016
doi:10.1016/j.proeng.2016.07.419
Quan Pan et al. / Procedia Engineering 154 (2016) 58 – 61
in the blooming era of knowledge and information technologies. Even, since the beginning of the century there is a need for the development of mixed interactive participatory tools [3]. Being water one of the key resources for mankind development, the water management (WM) field, can greatly benefit from such development. In WM, such exploration has been steadily increasing as there is a rise on volumes of information, and technologies available. Efficient management of water resources and water quality are keys to sustain needs of growing human population and address the growing water demand, especially in densely populated urban areas. The largest source of water for customers in large cities is usually a Water Supply System (WSS). WSS’s often suffer from a daily and seasonal demand-supply mismatch, steady increase of water losses due to leakages and pipe aging, and waste due to improper operation. Further, energy costs comprise a significant share of the operational expenses of water utilities. Solutions existing today for energy reduction are inadequate as they do not provide capabilities for fine-granular, real-time monitoring and optimization, and management of the water supply based on customer needs and energy costs. ICeWater project was funded by the European Union’s Seventh Framework Programme for research. The project aims to use a holistic approach to manage the “water energy nexus” [4]. Milan WSS, located in Italy, is a case study in the project. The WSS is operated by Metropolitana Milanese S.p.A. It is a complex system that provides water to around 1.3 million inhabitants and 3 million commuters [5]. The whole network consists of two connected systems, Water Transmission Network (WTN) and Water Distribution Network (WDN). There are 642 pumping wells in WTN and 28 pumping stations with total 101 booster pumps in WDN. The total length of pipes of the WDN is around 2,300 km. A plan view of Milan water supply network is presented in Fig. 1. The article is presented as follow in section 2 the DSS architecture implemented is presented, in section 3 some results of the use of the DSS, and finally some conclusions are presented. 2. DSS architecture and implementation 2.1. System Architecture A common web architecture is a three-way interaction under client server environment. The proposed architecture of the web DSS for ICeWater project separates the application tier into two parts, an optimization server and application servers (problem solvers). In this research, the optimization server included two evolutionary algorithms, NSGA-II [6] and AMGA2 [7]. Open source hydraulic solver EPANET2.0 was applied as problem solver for water supply network, see Fig. 2. 2.2. Implementation All the implementations were performed under PHP environment. Two PHP extensions for the evolutionary algorithms, NSGA-II and AMGA2 were implemented as a default module. The source code of NSGA-II was available in C language and of AMGA2 in C++ language as a class. For this development it was chosen to use C++, as an object-oriented language, for both algorithms in the optimization module. Thus the source code of NSGA-II has been rewritten as a C++ class. One of the challenges for the module integration is the definition of the entries between C++ class and PHP. For this purpose the Zend engine has been used, which has APIs available to wrap C and C++ codes as PHP extensions [8]. The server environment is LAMP (Linux, Apache, MySQL and PHP). Besides MySQL database, one PostgreSQL database was deployed which is located on the application server. On the top of PostgreSQL, PostGIS, which is a spatial database extender was installed. PostGIS database contains the network elements of the Water Supply System (WSS), i.e. junctions, tanks, reservoirs, pipes, pumps, valves, needed for setting up of an EPANET model. GeoServer loads the geo-spatial data from PostGIS database and delivers them to the map-based GUI on the web browser as standardized layers. In this application the Web Mapping Service (WMS) and Web Feature Service (WFS) standards of the Open Geospatial Consortium (OGC) were used for delivering the geospatial data to the client [9]. Due to the simple and extensible architecture, the simulation module suppliers can be involved together to share their knowledge and potential solutions. Each node of the DSS network contributes to the web-based DSS as a
59
60
Quan Pan et al. / Procedia Engineering 154 (2016) 58 – 61
component of the integrated application. In addition, the Internet-deployed resources can be used more efficiently, reducing the need for duplication of the same work.
Fig. 2. DSS architecture.
3. Results In this section, the results of performance for the solution of a pump scheduling problem in Milan are presented. The computational environment for the test was performed using virtual machines. Oracle VM VirtualBox free software was used to manage the virtual machines under Windows 7 environment. Each virtual machine was assigned one single core cup with 1.0 GB memory and 8.0 GB hard disk. Table 1 illustrates the specifications of the physical and virtual machine. An extensive presentation of the algorithms implemented and the available set ups in the DSS is presented in [10]. Table 1. Specifications of the physical and virtual machine. Specification
Physical machine
Virtual machine
CPU
Intel(R) Core(TM)
Single Core
i5-3210M, 2.5GHz
2.5GHz
Memory (RAM)
4GB
1GB
Hard Disk
500GB
8GB
Network
Broadcom BCM943228HM4L
Bridge Adapter
802.11 a/b/g/n
The parameters of the optimization algorithms are listed in Table 2. For NSGA-II, population size was set as 20 and 50 generations were used. For AMGA2, the same total number of function evaluations was applied, which was 1,000, but for this algorithm an elite population size was set to 20. Regarding general runtime performance, NSGAII costs less time than AMGA2. NSGA-II could save 1,516 s to finish the pump scheduling optimization problem. Table 2. The performance of the DSS on pump scheduling. Optimization algorithm
Milan water supply network
NSGA-II
AMGA2
Pop size
ngen
Neval
Time [s]
esize
Neval
Time[s]
20
50
1,000
8,891
20
1,000
10,407
Quan Pan et al. / Procedia Engineering 154 (2016) 58 – 61
4. Conclusions This article demonstrates a method to build a web-based model-driven DSS for WSS. The system that is presented herein has three main advantages. Firstly, it requires a relatively short period for development. Secondly, it uses freely available and open source libraries. Finally, it is extensible for other application servers. Future development needs to be sought to perform parallelization of the simulation and optimization modules inside a distributed environment, the establishment of secure data communication, and robust application interfaces for online DSS community contribution. Acknowledgements The research leading to the results presented herein has been conducted under the EU FP7 ICeWater project. We would like to thank Metropolitana Milanese S.p.A. for sharing the information of their WSS. References [1] Janssen R.; Goosen H.; Verhoeven M.L.; Verhoeven J.T.A.; Omtzigt A.Q.A. & Maltby, E. Decision support for integrated wetland management, Environmental Modeling & Software, 2005, 30, 215–229. [2] Xie Y., Wang H.W, Efstathiou J., A research framework for Web-based open decision support systems,Knowledge-Based Systems, 2005, 18, 309–319. [3] Pereira A.G. & Corral Quintana, S.From Technocratic to Participatory Decision Support Systems: Responding to the New Governance Initiatives.Journal of Geographic Information and Decision Analysis, 2002, 6, 95-107. [4] Fantozzi, M.; Popescu, I.; Farnham, T.; Archetti, F.; Mogre, P.; Tsouchnika, E.; Chiesa, C.; Tsertou, A.; Castro Gama, M. & Bimpas, M.ICT for Efficient Water Resources Management: The ICeWater Energy Management and Control Approach, Procedia Engineering , 2014, 70, 633 - 640 . [5] Gama, M. C.; Lanfranchi, E. A.; Pan, Q. & Jonoski, A., Water Distribution Network Model Building, Case Study: Milano, Italy, Procedia Engineering , 2015, 119, 573 – 582. [6] Deb, K.; Pratap, A.; Agarwal S & T, M. A fast and elitist multiobjective genetic algorithm: NSGA-II, IEEE Transactions Evolutionary Computation, 2002, 6, 182–197. [7] Tiwari, S.; Fadel, G. & Deb, K., AMGA2: Improving the performance of the archive-based micro-genetic algorithm for multi-objective optimization Engineering Optimization, 2011, 43, 377-401. [8] Achour M. et al., PHP Manual, PHP Documentation Group, Edited by P. Olson. 1997-2015. [9] Pan Q.; Jonoski A.; Castro-Gama M.E. & Popescu, I.Application of a web-based decision support system for water supply networks. Environmental Engineering & Management Journal, 2015, 14, 2087-2094. [10] Castro Gama M.E.; Pan Q.; Salman, M. A. & Jonoski, A.Multivariate optimization to decrease total energy consumption in the water supply system of Abbiategrasso (Milan, Italy).Environmental Engineering & Management Journal, 2015, 14, 2019-2029.
61
