Decision Making Algorithms for Hydro-Power Plant Location

The global energy scenario highlights the growth in demand for fossil fuels due to the continuous development in technology and the socioeconomic scenario of the world. This increase in demand of fossil fuels and utilization of the same to sustain development has enforced stress on the resources. Also, uncontrolled use of fossil fuels has increased the total concentration of greenhouse gases, which in turn has become a major cause of global warming. As a result the climate of many places has displayed abnormality. The pollution content of various places is also aggravated due to the rampant use of fossil fuels. That is why an alternative to fossil fuel is now being searched. Some of the sources of energy like solar, wind, and hydro-energy are infinitely available but the technology to convert it into utilizable form is expensive. Among all these renewable energy sources, hydro-energy is found to be relatively inexpensive and available at a greater phase of time than other similar kinds of energy resources. However,most of the factors which are considered in the feasibility studies for HPP is location dependent. The requirement of population displacement from the project watershed depends on location of the project. The amount of utilizable hydro-kinetic energy also is a function of both space and time. Not only the location dependency of factors but influence of all the factors on generation capacity is not uniform. For example the influence of amount of flow available in the project location is more important than the area of forest which are required to be removed from the project area. That is why both the importance of the factors and its location dependency must be considered in any feasibility studies for HPP. If the site selection is performed logically and scientifically, considering the importance of all the socioeconomic, geophysical, and logistical factors, then only such project may optimally satisfy the present demand for energy. In this regard decision-making algorithms like neural network, fuzzy logic, bat algorithms, and analytical hierarchy process along with hybrid models like neurogenetic and neuro-fuzzy which are popular for their intuistic decision making abilities were applied to identify the ideal sites for installation of hydropower.

The hydropower plants can be broadly subdivided into different classes based on quantity of water available, available head, and nature of load. Based on quantity of water available hydro power plant can be classified into Reservoir plants, Run-off river plants with pondage, and Run-off River plants without pondage. In case of available head, hydropower plants can be further subdivided into Low head, Medium head, and High head. With respect to nature of Load any hydropower plant can be grouped into Base and Peak load plants.

In case of multiple alternatives or solutions to a given problem a decision maker often gets confused to select the ideal option which will yield the maximum benefit. That is why there are various algorithms which assist in logical and scientific decision making so that the costs can be reduced and in the same time benefits can be multiplied without compromising the basic requirement. The procedure of scientific and logical decision making involves fixing a goal, defining and selecting the criteria and alternatives, and applying decision making tools to make a decision after comparing all the alternatives with respect to each of the alternatives. Multicriteria Decision Making is nowadays gaining popularity in many field of research as solution of most of the problems came from taking an optimal decision. A suitable but logical and practical procedure to identify alternatives can reduce many discrepancy and inefficiency of a probable solution.

Nature-based algorithms are those algorithms which mimics nature to solve a real-life problem. This kind of meta-heuristic algorithms are popular to search for an optimal answer within a given set of nonlinear complex problems by replicating the way by which nature is solving its problems. For example, bat search for food with the help of the emitted sonar signal which accurately identifies the location of ideal sources of food. This same concept can be replicated in case of real-life problems to estimate the ideal solution of nonlinear problems. In this chapter, the concepts of neural network, fuzzy logic, bat algorithm, and Analytical Hierarchy Process which are applied in the present study for taking a scientific decision in regard to find a suitable location for hydropower plants.

The problem of site selection for installation of hydropower plant can be solved by logical and scientific decision making. In the present problem, authors have used 20 different but related criteria to analyze the suitability of a location for installation of hydropower plants with the help of six popular nature-based algorithms (Neural Network, BAT Algorithms, Fuzzy Logic, Analytical Hierarchy Process, Neuro-Fuzzy, and Neuro-genetic.) Three alternatives from three different places (River Chenab in India, River Danube in Germany, and River Yukon in Alaska) which have various potential of hydropower generation (Very high, medium, none, respectively) was selected to demonstrate the efficiency of the algorithms in taking accord about this problem. The KAPPA Coefficient was utilized to compare the decision from the six different algorithms.

In the present investigation the ideal location for installation of HPP is determined with the help of six different nature based algorithms. The alternative consist three different locations having three different potential of hydro power generations. The algorithms were compared with the real life situations so that an algorithm among them can be selected as the better method for identifying HPP sites accurately but logically satisfying and preventing all the demands and hostilities from the neighborhood people under the requisite geomorphological restrictions. According to the results of comparison with the help of KAPPA method, it was found that BAT followed by Fuzzy algorithms displayed maximum accuracy. The clarity in the decision making of AHP was not satisfactory and the estimated decision of Neuro-genetic model was found to be unreliable. The reason for poor performance of AHP can be attributed to the linear nature of the method whereas for neuro-genetic models the requirement of time and computational infrastructure becomes a constraint. The role of weightage was found to be influential in the accuracy of the output which was evident in the estimated decision from neuro-fuzzy as well as fuzzy logic.

In the present investigation six different nature-based meta-heuristics were used to estimate the decision for selection of locations for installation of hydropower plants. Twenty different factors that can influence the potentiality of hydropower plants were used as criteria and three alternatives have been proposed. The aim of the study was to adjudge the performance efficiency of nature-based algorithms in decision making. That is why all the three alternatives have a known hydroelectricity potential and the decisions were matched with the potential to find the level of accuracy of the algorithm. The BAT was found to be a better performer than the other six models; the reason being that BAT determines weightage at which the difference between coherent and non-coherent variables is the maximum. Algorithms that are linear or where the weightage of importance is not considered have performed poorly. The Fuzzy Logic and BAT, both of which consider weightage for making a decision, has been found to perform satisfactorily. The results from BAT and Fuzzy and according to KAPPA Coefficient of Agreement results make these two algorithms better than the other considered algorithms (neural network, neuro-genetic, neuro-fuzzy and AHP) for prediction of the selection feasibility of a location for installation of HPP. After going through the results it can be concluded that neural network or hybrid algorithms utilized with the same algorithm may have performed poorly due to the requirement of sufficient training data set and computational infrastructures to learn the problem from this given training dataset. Although training dataset is available satisfactorily but limitations in the computational infrastructure may have limited the performance of the said algorithms. In case of hybrid algorithms the same requirement gets doubled. The same study can be repeated with more alternatives and different other algorithms for a more convincing conclusion regarding the potentiality of nature based algorithm in selecting suitable sites for hydropower plants or other related multi criteria decision making problems.