1. Introduction

In Mexico, the growing concern to choose efficient energetic sources with low environmental impact is fostering the development and application of renewable energies; in particular, the installation of wind farms to generate electricity. This type of installations complies with policies of energetic diversification and the reduction of CO₂ emissions proposed by international agencies and supported by the Mexican government, as they reduce the dependence on traditional energy sources and limit their negative effect on the environment (^{SENER, 2013}).

The techniques used to select sites for a wind farm consider two essential factors: the strength of the wind (from 5 m/s on), and the favorable conditions of the site, like access roads and being close to an electrical substation (^{Serrano et al., 2011}). The sites frequently chosen are determined through a cartographic review of the natural resources of the region. Afterwards, a wind speed measuring station is installed which verifies the intensity and direction (collecting data during 12 months). Finally, software (like WAsP) is used, which evaluates the wind resource and calculates the energetic yield of the wind turbines, as well as WindPRO, which analyzes the design and planning of the wind farm, or WindFarm, which optimizes the increase of energy or the reduction of energy costs (^{Miranda, 2008}; ^{Artillo, 2017}).

However, more complete studies have been developed that use computational algorithms and mathematical models for selecting the criteria to choose sites and design wind farms (^{Grady et al., 2005}; ^{Herrera et al., 2011}, ^{Guzmán, 2017}; ^{Konstantinos et al., 2019}). Other studies have focused mainly on technical-economic factors to strengthen the decision making about this type of projects, as well as their profitability. In this sense, some analyses add complexity by including environmental and social characteristics as viability criteria in the design and construction of the wind farm. The latter has led to the search and use of better computation techniques that will facilitate the exploitation of this type of renewable energy, considering the economic profitability of the project and its contribution to the social development of communities (^{Berumen and Llamazares, 2007}).

In recent years, methodologies considering multidisciplinary analyses have been incorporated to eolic projects, particularly in issues regarding site selections and performance optimization (^{Falces, 2015}). Different studies have been carried out to determine the possibility of eolic energy by means of computational mathematic models and multicriteria analyses. ^{Fernández-Jiménez et al. (2009)} propose the use of genetic algorithms combined with a geographical information system (GIS) for the selection of wind turbines inside the wind farm. ^{Herrera et al. (2011)1} describe an optimization model of the technical configurations of a wind farm, using algorithms to maximize the conditions of the proposed project.

This study includes and analyzes technical, economic, environmental, and social criteria that are involved in the planning of a wind farm. Since these studies are performed in stages, generally they are scarcely analyzed, and sometimes many criteria are not included (^{SEMARNAT, 2002}). The environmental and social aspects of eolic projects should have the same or greater relevance than the techno-economic conditions, because the project impacts the former in a more significant way. Therefore, this study is aimed at giving a more specific weight to environmental and social conditions in the selection of recommended sites for wind farms. It proposes to assess the potential of the northern region of Mexico to implement wind farms, elaborating a repeatable methodology that will allow assessing the selected criteria and the territorial conditions, using the instrumental methods provided by multicriteria evaluation and GIS.

When selecting sites for wind farms, the most important geographical, ecological, and social criteria for the region must be included in addition to particular criteria like wind power and access roads, as usually done in the development of this type of project (^{Zárate and Fraga, 2016}). In this study we used computational models and algorithms in order to consider issues of multiple criteria (variables) related to decision making that might function as alternative tools for renewable energy projects, in particular for wind farms. The choice of optimal sites for installing wind farms will lead to adequate and beneficial solutions for project developers, the environment, and the local community.

This study initially proposed a regional area of interest, considering the states along the northern border (Baja California Norte, Sonora, Chihuahua, Coahuila, Durango, Nuevo León, and Tamaulipas). Traditionally, the development of eolic installations has been concentrated in central and southern regions of the country, mainly in the state of Oaxaca. Over the last years, due to its characterization as an exploitable region, northern Mexico has been taken into account to explore its possibilities for the development of eolic energy, with the expectation of generating large economic and environmental benefits (^{SENER, 2016}).

2. Methodology

The process to select optimal sites for the installation of eolic parks was developed in four stages (Fig. 1): (1) selection of the study zone, (2) statement of variables with their specific weights, (3) analytic hierarchy process to obtain the variables with the greatest relevance, and (4) use of GIS to visualize results.

Fig. 1 Flow chart of the proposed methodology. 

The methodology requires to determine the study zone based on the digital cartographic information. In addition, a larger period than normal was considered for the measurement of the wind factor, By means of the meteorological model Weather Research and Forecasting (WRF) daily data on direction and amount were analyzed for a period of five years, from 2008 to 2012 (^{CCA, 2016}).

To analyze the variables proposed in this study, we used multiple-criteria decision analysis (MCDA), which helps to make decisions on a posed issue for choosing, classifying, and organizing the proposed elements. Specifically, we applied the analytic hierarchy process (AHP), a powerful and flexible tool for MCDA which is used for issues that require assessing both quantitative and qualitative aspects through a common scale, performing pairwise comparisons between criteria and alternatives. Results were coupled in a GIS, aimed at obtaining a feasibility map of the best sites for the localization of wind farms.

3. Results

Table I shows that the states of Baja California Norte, Sonora, and Tamaulipas are excluded as they do not comply with criterion 3. In this way, the analyzed area considered information on the sites in Chihuahua, Coahuila, Durango, and Nuevo León. Figure 2 shows the area that complies with the set of established criteria (area of active cells).

Fig. 2 Location and limits of the study area. 

The proposed area encompasses an extension of approximately 34 900 km² with maximal distances of 310 × 160 km. The whole area is divided in grids of 100 km², which is a mid-detail and regional measure. It includes important municipalities like Torreón, Matamoros, Viesca, Parras, San Pedro, General Cepeda, Carmen, Ramos Arizpe, Saltillo, García, Mina, Hidalgo, Abasolo, and General Escobedo.

3.2 Feasibility map for wind farms

Firstly, with the vector information in Table V and the distribution of active cells in Figure 2, we calculated the initial optimal maps of variables in Table V. Thereafter, the specific weight value of each cell in these maps was assigned, in order to transform them into raster-type maps.

The eight raster-type maps grouped in pairs and types of variables (technical, economic, environmental, and social) were subjected to a map algebra process, obtaining four maps for each of the proposed types.

The wind farms feasibility map is the result of another map algebra process, which used the previous four maps. Algebra map allows the generation of a feasibility map in a colored chart (Fig. 3), which served to identify the most adequate areas to develop a wind farm. The percentage of feasibility varies according to the colors in the chart of optimal sites.

Fig. 3 Map of optimal sites with feasibility values associated with a table of values to install a wind farm. 

Figure 3 shows the north and northwest regions with the best feasibility for the development of wind farms. Light-green cells indicate a 70 to 80% feasibility, and the yellow cells correspond to a 60 to 69% feasibility. This map favors environmental features, with a good amount of wind and low impact on the fauna. Regarding the social aspects, it does not present organized social groups and corresponds to a secure zone. These two subsections received weights slightly higher than the technical and economic variables.

Regarding the worst sites, they are located in the central and northeast regions, where feasibility is below 50%. According to our data, these regions have social perturbations like robbery and criminality and the required amount of wind is minimal, so they would pose technical risks for the project.

Finally, there are six dark-green cells in the northwest region which comply with the best feasibility conditions (80-90%). At the municipal level (^{INEGI, 2015}), they would correspond to the southwest of Coahuila, specifically to the municipalities of San Pedro (106 142 inhabitants), Matamoros (108 950 inhabitants), Torreón (679 288 inhabitants), and Viesca (21 549 inhabitants). Thus, these are considered the most adequate zones for the development of a wind farm.

4. Discussion

The north of Mexico has a considerable number regions adequate for the development of wind farms, as can be observed from the followed methodology using only digital cartographic information (Table I). This technique is useful to roughly locate sites for exploitation. In addition, if meteorological information from the WRF is implemented, allowing for a lengthier temporal measurement of the wind factor (5 years), the uncertainty for estimating possible sites is reduced. This contrasts with yearly measurements to obtain information on the wind factor, which in some cases are only intermittent and scarcely reliable (^{ERM, 2014}).

The southern part of the state of Coahuila presents the best options for wind farms, as it has industrial infrastructure and a territory with few environmental conflicts (in relation to the limits of natural protected zones). These conditions strengthen the possibility to support this type of projects.

The use of hierarchical algorithm was proposed to ease and optimize the decision-making process when gathering results that contemplate the largest number of criteria and conditions, in order to obtain the best alternatives for this complex project. The posed criteria were divided in four groups covering the most important aspects in the technical, economic, environmental, and social areas (Table II). Based on the literature and prior contributions to this type of projects, representative values were assigned to the variables (Table IV).

In most eolic projects, technical and economic criteria are considered as the most relevant. However, in this study, environmental and social variables were given a greater importance, fundamentally with the idea of causing minimal disturbance to the flora and fauna of the region and providing benefits to the local populations by generating employments, activating the economy, and diminishing the consumption of traditional energy.

The use of a hierarchical process in this study allowed obtaining results that incorporate a large number of criteria for the decision makers, by organizing efficiently and graphically the information regarding the decision-making process. The construction of maps of variables eases the handling and interpretation of data (^{Xu et al., 2012}).

The proposed methodology allows resolving problems that include multidisciplinary aspects, and offers the involved groups a solution that can be easily understood and accepted (^{Janke, 2010}, ^{Kang et al., 2011}; ^{Díaz et al., 2017}), in contrast with some studies on the feasibility of sites for wind farms where only techno-economic aspects (knowing where to place the wind turbines, calculating economic losses, increasing the efficiency of the wind farm, etc.) are taken into account (^{EWEA, 2009}).

5. Conclusions

The process of choosing a location for installing and developing wind farms is a complex task due to compulsory regulations and requirements; hence, this type of study is a good option for Mexico.

The use of multicriteria computational hierarchical algorithms eases the task of decision makers to select the location of a wind farm. The advantage of this methodology is its great usefulness when decision elements are scarcely known and difficult to quantify and assess, as it permits calculating the contribution of each alternative with respect to the others. This methodology can be applied in other regions of the country, as long as the variables of interest are well defined and evaluated.

The AHP, using the PriEsT software, is a useful tool for selecting the most relevant variables for the project. AHP allows estimating priorities based on pairwise comparison judgments. It offers a wide range of optimal solutions based on multi-objective optimization.

The use of GIS to couple the analysis of the best variables yields visual results of the feasible sites.

The resulting optimal sites map reveals numerous sites with values from 10 to 40% and 80 to 90% for the least and most feasible locations, respectively.

The western region of Coahuila presents the best areas with adequate features according to the set of chosen variables. In particular, the municipality of San Pedro, has a surface of 300 km² for the installation of a wind farm with a feasibility of 70 to 80%.