Optimization of pump scheduling with dynamic probabilistic methods

Institutes: Systems Engineering Laboratory, University of Oulu
Department of Hydrodynamic Systems, Budapest University of Technology and Economics
Personnel: Enso Ikonen, Ph.D.
István Selek, Ph.D.
Csaba Hős, Ph.D.
József Bene
Funding: Center for International Mobility (CIMO) (2010-2011)
Academy of Finland (2011-2014)
Research plan of the CIMO project:Available here
Short description:

Pump scheduling problems appear as a general problem in various fields of industry. A typical example is that of a water distribution network, but many other problems can be characterized by optimization of transportation between storages. The objective of research is to develop new methods that will enable to optimize pump scheduling type of problems characterized by system dynamics, uncertainties, nonlinearities, discontinuities and large problem sizes. Our main hypothesis is that advantageous algorithms will be found from the class of probabilistic model-based techniques which take into account the system dynamics. In addition to methods, an effort will be put on proper problem formulation for the real life applications. The applicability of the examined methods will be examined and demonstrated via real-life case problems.


Topology of a test network

The project involves joint work between the Systems Engineering Laboratory (SYTE, Oulu, Finland) and Department of Hydrodynamic Systems (HDR, Budapest, Hungary). SYTE has expertize in application of advanced methods for process monitoring, control and optimization. Typically, methods are based on models, built using physical knowledge and/or plant data, requiring understanding of both transport phenomena (process engineering) as well as sampled systems and signal processing (automation). The HDR is highly skilled in modelling, simulation and optimization of hydrodynamic systems; stationary and unsteady analysis of water distribution and sewage systems has been a traditional topic at the HDR. HDR has strong cooperation with waterworks companies, the system at Sopron will be used as a case example.

Algorithms in the area of random and population based dynamic modelling and search techniques have been already examined at the SYTE in our past works. State-of-the-art methods for the pump scheduling problem have already been examined at HDR. This project examines algorithms in a general setting for pump sceduling problems using a) finite Markov chains, and b) particle filtering based techniques and neutrality, c) with applications for real utilities. Given this initial boost, the research is directed via two real life case examples. Careful analysis of the applications will guide us as to where the significant bottlenecks are. The focus of research will be on the design of applicable and flexible techniques for real life problems, as it is believed that the bottleneck of scientific and industrial breakthrough of random and/or population based techniques lies much in the lack of convenient problem formulation setups. The project will promote the role of systems engineering in providing frameworks for better understanding of problems of process and mechanical engineering, in providing scientifically justified results on modern process control, and in providing solutions to problems with high economical and societal significance.


Submitted publications:
  • J. Bene, I. Selek (s. 2011) Water Network Operational Optimization: Utilizing Symmetries in Combinatorial Problems by Dynamic Programming. PERIODICA POLYTECHNICA CIVIL ENGINEERING.

Accepted / published publications:
  • Ikonen, E. and J. Bene (2011) Scheduling and disturbance control of a water distribution network. 18th World Congress of the International Federation of Automatic Control (IFAC 2011), 28 August - 2 September 2011, Milano, Italy.
  • Ikonen, E. and J. Bene (2010) Optimal control of pumps in water distribution network under demand uncertainties. Conference on Simulation and Modelling (SIMS 2010), 14-15 October 2010, Oulu, Finland.
  • J. Bene, Cs.Hős (2010) A Novel Constraint Handling Technique for Genetic Algorithm-based Pump Schedule Optimization. Proceedings of the Seventh Conference on Mechanical Engineering, 25-26 May 2010, Budapest, Hungary.
  • J. G. Bene, I. Selek, Cs. Hős (2010) Neutral Search Technique for Short-Term Pump Schedule Optimization. JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT-ASCE 136:(1) pp. 133-137. (2010)
  • I. Selek (2009) Novel evolutionary methods in engineering optimization - towards robustness and efficiency. Ph.D. Thesis. Acta Universitatis Ouluensis C326.
  • Bene József, Hős Csaba (2006) Computation of cost-optimal pump scheduling for regional waterwork using genetic algorithm. Proceedings of the Fifth Conference on Mechanical Engineering. 25-26 May 2006, Budapest, Hungary.