W. Kurek

Multi-objective optimization of water quality, pumps operation, and storage sizing of water distribution systems.

A multi-objective methodology utilizing the Strength Pareto Evolutionary Algorithm (SPEA2) linked to EPANET for trading-off pumping costs, water quality, and tanks sizing of water distribution systems is developed and demonstrated. The model integrates variable speed pumps for modeling the pumps operation, two water quality objectives (one based on chlorine disinfectant concentrations and one on water age), and tanks sizing cost which are assumed to vary with location and diameter. The water distribution system is subject to extended period simulations, variable energy tariffs, Kirchhoffs laws 1 and 2 for continuity of flow and pressure, tanks water level closure constraints, and storage-reliability requirements. EPANET Example 3 is employed for demonstrating the methodology on two multi-objective models, which differ in the imposed water quality objective (i.e., either with disinfectant or water age considerations). Three-fold Pareto optimal fronts are presented. Sensitivity analysis on the storage-reliability constraint, its influence on pumping cost, water quality, and tank sizing are explored. The contribution of this study is in tailoring design (tank sizing), pumps operational costs, water quality of two types, and reliability through residual storage requirements, in a single multi-objective framework. The model was found to be stable in generating multi-objective three-fold Pareto fronts, while producing explainable engineering outcomes. The model can be used as a decision tool for both pumps operation, water quality, required storage for reliability considerations, and tank sizing decision-making.

Grid Implementation of a Parallel Multiobjective Genetic Algorithm for Optimized Allocation of Chlorination Stations in Drinking Water Distribution Systems: Chojnice Case Study

Solving multiobjective optimization problems requires suitable algorithms to find a satisfactory approximation of a globally optimal Pareto front. Furthermore, it is a computationally demanding task. In this paper, the grid implementation of a distributed multiobjective genetic algorithm is presented. The distributed version of the algorithm is based on the island algorithm with forgetting island elitism used instead of a genetic data exchange. The algorithm is applied to the allocation of booster stations in a drinking water distribution system. First, a multiobjective formulation of the allocation problem is further enhanced in order to handle multiple water demand scenarios and to integrate controller design into the allocation problem formulation. Next, the new grid-based algorithm is applied to a case study system. The results are compared with a nondistributed version of the algorithm.

Multiobjective Water Distribution Systems Control of Pumping Cost, Water Quality, and Storage-Reliability Constraints

AbstractThis work describes a multiobjective model for trading-off pumping cost and water quality for water distribution systems operation. Constraints are imposed on flows and pressures, on periodical tanks operation, and on tanks storage. The methodology links the multiobjective SPEA2 algorithm with EPANET, and is applied on two example applications of increasing complexity, under extended period simulation conditions and variable energy tariffs. The proposed approach enables decision makers to take full advantage of the obtained information on a multiobjective scale for trading-off, cost, water quality, and storage-reliability requirements. Verification of the model outcomes through engineering judgment on all runs for both example applications confirmed the model suitability as a decision tool. Limitations of the proposed model reside in using variable speed pumps with assumed constant efficiency as representing an entire pumping station operation, the storage reliability constraint as an u-priori set...

Genetic Hybrid Predictive Controller for Optimized Dissolved-Oxygen Tracking at Lower Control Level

A hierarchical two-level controller for dissolved-oxygen reference trajectory tracking in activated sludge processes has been recently developed and successfully validated on a real wastewater treatment plant. The upper level control unit generates trajectories of the desired airflows to be delivered by the aeration system to the aerobic zones of the biological reactor. A nonlinear model predictive control algorithm is applied to design this controller. The aeration system itself is a complicated hybrid nonlinear dynamical system. The lower level controller (LLC) forces the aeration system to follow these set-point trajectories, minimizing a cost of energy due to pumping of the air and accounting for system operational limitations such as the limits on the allowed frequency of switching of the blowers and on their capacity. The predictive control is also applied to design the LLC based on a piecewise-linearized hybrid dynamics of the aeration system. Casting the mixed-integer nonlinear optimization problem under heterogeneous constraints due to the limits on the blower switching frequency into the approximated mixed-integer form is done at a cost of introducing large number of auxiliary variables into the lower level predictive controller optimization task. This paper derives another nonlinear hybrid predictive control algorithm for the LLC. It is directly based on the nonlinear hybrid dynamics and logical formulation of the switching constraint. A genetic algorithm is derived with dedicated operators allowing for efficient handling of the switching constraint and nonlinear hybrid system dynamics. The efficiency of the control algorithm is validated by simulation based on real data records.

GENETIC SOLVER OF OPTIMIZATION TASK OF MPC FOR OPTIMIZING CONTROL OF INTEGRATED QUANTITY AND QUALITY IN DRINKING WATER DISTRIBUTION SYSTEMS

Abstract The predictive integrated water quantity and quality yields better performance then different approaches. However, it requires solving a nonlinear, non-convex optimization problem. Therefore, a specialized optimization solver has to be used within designated time. In this paper a dedicated genetic algorithm specially designed in order to efficiently solve predictive optimization task is derived and the simulation results of the case study Gdynia drinking water distribution system are presented.

OPTIMISED ALLOCATION OF CHLORINATION STATIONS BY MULTI-OBJECTIVE GENETIC OPTIMISATION FOR QUALITY CONTROL IN DRINKING WATER DISTRIBUTION SYSTEMS

Abstract Providing water meeting strict quality conditions is now a number one priority for the water authorities. Water chlorination only at water treatment plants forced the usage of high concentration of disinfectant which resulted in bad taste of water near them and still did not solve the quality issues at the far points of the network. The answer to this problems is booster chlorination - placing booster station throughout the network allowing a significant reduction of chlorine concentration and at the same time ensuring proper water quality in the network. The booster chlorination is highly dependent from the demand scenarios.

Adaptive multiobjective model predictive control with application to DWDS

Abstract In this paper a new approach to multiobjective control is proposed. The approach is based on Model Predictive Control but is extended to account for several independent objective functions. This is done by substituting a single objective predictive optimisation task with the multiobjective one. A dedicated method of selecting a unique solution from the resulting Pareto front is then required to be applied to the plant. An adaptive selection mechanism is proposed. It adjusts itself to the current Pareto front by means of updating the utopia point and span of the front based on the current front. Moreover, a way of introducing preferences to the selection mechanism is proposed. The derived adaptive multiobjective model predictive controller is then applied to integrated quality and quantity control of drinking water distribution system

GRID IMPLEMENTATION OF PARALLEL MULTIOBJECTIVE GENETIC ALGORITHM FOR OPTIMISED ALLOCATION OF CHLORINATION STATIONS IN DRINKING WATER DISTRIBUTION SYSTEMS: CHOJNICE CASE STUDY

Abstract There is a group of problems that require big amount of computing power to solve. Computer grids allow building effective computing platforms at relatively low cost. It is expected that algorithms like Genetic Algorithm will perform well on the grid. In this paper, grid implementation of multiobjective distributed genetic algorithm is proposed. A distributed version of the algorithm is based on a modified island algorithm where genetic data exchange is replaced by introduced new Forgetting Island Elitism. The algorithm is applied to booster station allocation in Chojnice water distribution system.