Site suitability analysis of water tank in Athivilai village of India using an analytical hierarchy process and a geographical information system

The water distribution system (WDS) serves as an infrastructure to supply water to consumers. In India, there is a need for the construction of new water distribution networks (WDNs) in various regions. The design must be such that water is available to consumers throughout their design period with adequate pressure. The adequate supply of water received by the consumers depends on the pressure maintained in the system. This can be fulfilled by the storage tank because it helps to maintain the desired pressure in the mains constantly, even in remote areas (Birdie & Birdie 2010; Garg & Garg 2010; Briere 2014). In the absence of water tanks, the pressure falls as the demand increases. This is because the water tank provides the head that drives the WDN (Mercy & Tiku 2018). Also, the storage tank enables demand management, assures water supply in case of network failure and reserves for emergencies such as firefighting, and allows the modulation of pump flow rate.

The WDN costs about 40–70% of the total cost of the water supply project (CPHEEO – Central Public Health and Environmental Engineering Organisation 1999). Thus, an optimized design is important while designing a new network. The majority of the studies focused on the optimization of the WDN that considered diameter as the main optimization problem because reducing the size of pipes reduces the overall cost of the water supply project (Vasan & Simonovic 2010; Kang & Lansey 2012). Providing a water tank also results in the overall reduction of the size of pumps, pipes, and treatment units (Birdie & Birdie 2010). However, an inadequate design of the WDN or improper location of the water tank increases the cost of pipe and pump operation and reduces the network performance indices such as resilience and reliability (Vamvakeridou-Lyroudia et al. 2007). However, when these tanks are properly designed and located, they are a cost-effective means of improving the overall performance of the network. Gottipati & Nanduri (2014) developed an index called the uniformity coefficient to measure the unity in the distribution of the intermittent WDN and their results indicated that the layout and the location of the tank play a significant role in improving equity in distribution within the network. Thus, the location of the storage tanks plays an important role in the effective design and overall performance of the network.

The research works considering water tank location in optimization are as follows: Ameyaw et al. (2013) developed a multi-objective optimization method GANetXL to improve the equitable distribution of water in intermittent systems. They concluded that equity can be improved by the optimal location and capacity of elevated water tanks. Hooda & Damani (2017) developed an Integer Linear Program (ILP) model in which they considered tank configuration as a variable for optimization. The implementation was done by using Java 7 and GLPK 4.55 Linear Program Solver which uses Google Map's API (Application Programming Interface) for GIS functioning. Abarca & Da Silva (2020) proposed a methodology to locate the water tanks in rural Andean areas by developing a protocol based on the interaction between water supply hydraulics and GIS. The algorithm is based on parameters such as pressure limit, gravity supply, accessibility, stability, and proximity of the largest population density. Basile et al. (2008) developed a two-stage algorithm to optimize the water storage location and capacity. They linked the model with the hydraulic solver EPANET. They applied the algorithm to the existing network and spotted two possible locations for introducing tanks and an analysis was done to find the best among them. From the above literature, it can be noted that the inclusion of water tank location plays a significant role in the optimization of the WDN.

The current development in the Geographic Information System (GIS) serves as a powerful tool for the collection, storage, and management of spatial data in a simplified matter (Goitsemang et al. 2020). GIS plays an important role in viewing the results through spatial and visual interpretations (Saranya & Saravanan 2020). Application of AHP integrated with GIS has developed a lot since the beginning of the 21st century (Marinoni 2004; Mardani et al. 2015). Several studies have been done by integrating GIS and AHP techniques to find the vulnerability characteristics and the spatial distribution of water resources in Guiyang city (Li et al. 2022), to evaluate the dam site for Bortala, in Northwest China (Dai 2016), to delineate the groundwater potential zones (Saranya & Saravanan 2020; Mahato et al. 2022; Sarkar et al. 2022), evaluate suitable site selection for solar farms (Uyan 2013), develop a dam site suitability model for lower Tapi basin (Raaj et al. 2022), and to generate a Seismic Vulnerability Index for Water Distribution Networks (SVI-WDNs) (Marleni et al. 2022).

Although there have been many studies to design the WDN coupled with GIS and AHP, there is no research on site selection for the location of water tanks. Moreover, in previous research studies, authors have developed a protocol or an algorithm linked with GIS to find the tank location. This study aims to find the weights of the factors that influence the tank location using AHP. Then, weighted overlay analysis is used to find the location in the ArcGIS environment. The results are tested by using EPANET by locating the tanks in suitable and least suitable locations.

Various factors such as demand, population, land use land cover, slope, and distance to the road influence the location of water tanks. Multi-criteria decision-making (MCDM) is a method to evaluate the appropriate weights which influence each other. The Analytical Hierarchy Process (AHP) is one of the MCDM methods, which is simple and most commonly used, that helps in making decisions in complex situations and in the field of sustainable engineering (Stojcic et al. 2019). The AHP consists of three levels which include the identification of decision goals, criteria, or factors, the evaluation of pairwise comparisons between each element at every level of the hierarchy, and synthesis using the solution algorithm from the results of pairwise comparisons over all the levels (Wind & Saaty 1980; Saaty 1988, 1990). The AHP method is a process that uses an expert's opinion to determine the weights and ranks of factors by constructing a pairwise comparison matrix based on Satty's scale of importance (Table 1).

All factors influencing the water tank location are compared with each other in pairs and the comparison matrix is created, as given in Table 2.

From Table 4, it can be noted that the CR is 0.0639 which is less than 0.1; thus, the comparison matrix is said to be consistent.

After finding location suitability mapping of the ungauged study area, validation is carried out in EPANET. An analysis is done by locating tanks in (1) suitable locations and (2) least suitable locations. EPANET is a computer software used to simulate the hydraulic and water quality behavior of pipe networks. The rate of flow and velocity of water in pipes, the pressure at junctions, the height of water in storage tanks, and the concentration of chemical species can be found throughout the network. EPANET models a WDN as a collection of pipes connected to junctions (Rossman 2000). With the help of EPANET, the effective design of a WDN can be done within a short period, even for the complex type of networks (Ramana et al. 2015). EPANET analysis is less time-consuming when compared to Excel programming. Moreover, graphs of demand, nodal pressure, and diameter of links can be obtained without any tedious work (Rai & Lingayat 2019).

For EPANET analysis, the following steps are involved: The junctions are fixed along the road network and the latitudes and longitudes are found at the junctions using GPS surveying. The input data required for the junctions are base demand and elevation. Likewise, the inputs required for pipes are length and roughness coefficient. The demand at the junctions is obtained by multiplying the population at junctions with the rate of supply. The rate of supply is taken as 80 lpcd (liter per capita per day) as per CPHEEO (1999). The elevation and the pipe length can be obtained from Google Earth Pro (Sathyanathan et al. 2016). Furthermore, the area is divided into three zones since the elevation ranges from 27 to 87 m. This is because when the elevation difference varies between 15 and 25 m (CPHEEO 1999), this region has to be divided into three zones. Also, if an individual tank is provided for the whole zone, then the whole system is shut down for repair when the pipe at any point of the network fails, which makes the consumers suffer. Each zone is assumed to have individual tanks. The input data for the junctions, pipes, and tanks are assigned. The diameters are randomly fixed by trial and error method in such a way that the pressures are within the limits, i.e. the diameters are fixed such that the water is available at the endpoints of the junctions with the allowable pressure even at the time of maximum demand (Garg & Garg 2010). In this study, initially, all diameters are assumed to be 100 mm for EPANET analysis and adjusted to get the optimum diameter with respect to the required pressure at the junctions (Ramana & Sudheer Chekka 2018).

The weights obtained through AHP are assigned over the factors and the location suitability map was created. It is divided into five different classes, 1–5, corresponding to unsuitable, least suitable, moderately suitable, suitable, and highly suitable, respectively.

Water demand is the product of population and per capita demand. Water demand is directly related to the population. Thus, as the population of the area increases, the water demand also increases. The demand for the highly populated area is more than that of the less populated area.

For the performance study, a single-period simulation is carried out by considering the peak demand at the junctions in EPANET. Since the unsuitable location falls only in the small area of the study area, analysis is done in EPANET for suitable and least suitable locations, as shown in Figure 8.

The diameters of pipes have been the subject of numerous studies that have used heuristic algorithms or machine learning to solve their optimization problems. This study showed that the location of the water tank has a big impact on the successful design of a WDN. Water tanks are typically placed in very elevated places based on experience. As a result, in this study, AHP and GIS are used to determine a water tank's suitable site. This study consists of four major steps: (i) to identify the factors that affect the water tank's position, (ii) to determine the weights of the components using the AHP method, (iii) to develop a location suitability map using the ArcGIS overlay method, and (iv) to verify whether the location has an impact on the effective design (i.e., the decrease of pipe size). The results depend on the rank-based rankings produced by the AHP approach. As a result, the AHP method is significant in the assessment of factor weights. More factors could be taken into account in order to increase the study's accuracy. The results from this study proved that the location of water tanks plays an effective role in the effective design of the WDN, by reducing the pipe diameters significantly.

The use of GIS and AHP proves to be effective tools for identifying suitable tank locations. In this study, the thematic map layers of all the factors were created using ArcGIS, and the AHP technique was used to determine the normalized weights for the factors. The location suitability map of the water tanks was generated using a weighted overlay tool. They are classified into five classes: highly suitable, suitable, moderately suitable, least suitable, and unsuitable areas of water tank location. The results show that 15.69% of the area is suitable, 52.36% of the area is moderately suitable, about 31.31% of the area is least suitable, and 0.63% of the area is unsuitable for locating water tanks. None of the areas corresponds to a highly suitable location. The results are verified by placing tanks in suitable and least suitable locations, and an analysis was done using EPANET software. It can be concluded that locating water tanks in suitable locations reduces the size of pipes and maintains the pressure effectively. When the tanks are located in least suitable locations, the pipe size increases and a negative pressure is created at some junctions. Thus, the integration of AHP and GIS can help the designers to find suitable tank locations, and also, effective designs of the WDN.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.