GIS and Spatial Analysis in the Utilization of Residual Biomass for Biofuel Production

Abstract

The main goal of this study is to investigate the possibility of using residual materials (biomass derived from used cooking oils and lignocellulosic biomass from plant waste) on a large scale for producing renewable fuels and, in particular, the best way to collect them. The methodology of Geographic Information Systems (GIS) as well as spatial analysis (SA) techniques were used to investigate the Greek case for this. The data recorded in the geographic database were quantities of waste cooking and household oils as well as quantities of lignocellulosic biomass. The most common global and local indices of spatial autocorrelation were used. Concerning the biomass derived from used cooking oils, it was found that their quantities were important (163.17 million L/year), and these can be used to produce green diesel in the context of the circular economy. Although the dispersion of the used cooking oils was wide, there is no doubt that their concentration in large cities and tourist areas is higher. This finding suggests a collection process that could be carried out mainly in these areas through the development of small autonomous collection units in each neighborhood and central processing plants in small regional units. The investigation of the geographical–spatial distribution of residual lignocellulosic biomass showed the geographical fragmentation and heterogeneity of the distributions. The quantities recorded were significant (4.5 million tons/year) but widely dispersed, such that the cost of collecting and transporting the biomass to central processing plants could be prohibitive. The “geography” of the problem itself suggests solutions of small mobile collection units in every part of the country. The lignocellulosic biomass would be collected and converted in situ into bio-oil by rapid pyrolysis carried out in a tanker vehicle. This would transport the produced bio-oil to the nearest oil refineries for the conversion of bio-oil into biofuels through deoxygenation processes.

1. Introduction

The utilization of minerals for the preparation of catalysts in conjunction with the collection and utilization of residual raw materials for the production of renewable fuels is a new and very promising perspective [1,2,3,4,5,6]. More specifically, from the food crisis of 2007–2008, it has become clear that first-generation biofuels could pose a risk to the world’s food balance. Thus, research has focused on second- and third-generation biofuels that do not compete with food for their production, as they are derived from residual feedstock. There are two types of biomass used for the production of these new biofuels: lignocellulosic biomass from plant waste and biomass derived from used cooking oils, animal fats, distilled fatty acids and coffee oil. This second source of biofuels contains triglycerides and is very good for the production of mainly green diesel. Until recently, the catalysts used for this process were entirely synthetic, but a few years ago, excellent catalysts based on the very cheap natural mineral “palygorskite” were introduced. Palygorskite is found in large quantities in various parts of the world. By using such natural precipitants, the biofuel production process can be made “green” and integrated into the circular economy. Can all of the above be achieved on a large scale? And if so, what is the best methodology for this? The main goal of this study is to investigate the possibility of using cheap minerals and residual materials on a large scale for producing renewable fuels and, in particular, the best way to collect and use them [7,8]. This paper examines the Greek case for this.

The methodology of Geographic Information Systems (GIS) as well as spatial analysis techniques were used [9,10]. GIS information belongs to the general category of intelligent systems and can be understood as a model of a smart digital map with hyperlinks. The geographical, cartographical and spatial data of this digital map are uniquely related to descriptive data. Thus, the relational connection between descriptive and spatial information can be considered the basic method of GIS. Recording data in a GIS database is a prerequisite for visualizing map distributions. In many studies, this visualization is sufficient, as it can provide a comprehensive picture of the issue under consideration. However, for a deeper investigation, it is necessary to develop a methodology in the context of spatial analysis [11,12]. In the modern literature, spatial analysis is frequently connected to the use of GIS. The basic methodology followed in the present study in terms of spatial analysis was that of spatial autocorrelation [13]. The fundamental logic of spatial autocorrelation is based on Tobler’s so-called First Law of Geography [14], which is the methodological basis of all theories and practices of spatial data analysis. According to Tobler’s law, everything is related to everything else, but near things are more related than distant ones [15]. Some indicators are used to calculate spatial autocorrelation. The most common are the global and local indices of spatial autocorrelation; Moran’s I, Geary’s C, and Getis’ G, which were used in the present study. Software such as the R version 4.4.1 programming language and the open-source software GeoDA version 1.22 and QGIS version 3.40 were used to calculate the indicators.

Literature Review

In recent years, a number of scientific studies have focused on the problem of biomass utilization for the production of biofuels. Notable research has focused on the optimization of the supply chain [16] but also on the use of GIS and mathematical programming systems for supply chain planning [17,18]. In parallel, research has focused on case studies looking at different feedstocks, such as residual biomass from crops [19], Waste Cooking Oils (WCOs) [20], animal fats [21], etc. Also, significant research has been carried out on the optimization of biomass and biodiesel supply chain design models [22,23,24,25], bioethanol [26,27], etc. In addition, significant efforts have been made towards estimating the total amounts of residual biomass from various sources (such as agriculture, forestry, food waste etc.) on a large scale [28]. The above studies address the problem from the perspective of process and supply chain sustainability using mathematical optimization models. However, in our research we do not focus on optimization or sustainability, but we focus on one step before. We focus mainly on the following question: what is the optimal scale (local, regional or national) for the collection and utilization of multiple sources of residual biomass? Due to its particular geography and its many islands, the case of Greece is typical of the problem. In this paper, we make use of spatial analysis and spatial autocorrelation indices for the first time in the study of a similar problem. In

Table 1. Research summary in the area of biomass utilization for the production of biofuels.

Ref. Nom	Author	Biomass	Application Area	Method/Analysis	Goal of the Study
[16]	Mondal et al. (2023)	Bio-ethanol	West Bengal, India	Fuzzy Decision Making Trial/Evaluation Laboratory (DEMATEL)	Supply chain optimization
[17]	Kim et al. (2011)	Logging residual, grasses	U.S. southeastern region	Monte Carlo simulation	Optimal design of biomass supply chain networks under uncertainty
[18]	Huang et al. (2010)	Cotton, rice, corn stover, wheat straw	California, USA	Deterministic flexible programming/spatial analysis	Minimizing the cost of the entire supply chain
[19]	Habib (2022) et al. (2022)	Animal fat	Pakistan	AF-BSCND optimization model/RPFP solution	Reducing supply chain costs and risks
[20]	Babazadeh (2017)	WCOs, Jatropha seeds	Iran	Deterministic programming	Biofuel supply chain optimization
[21]	Habib et al. (2020)	Animal fats	Pakistan	Robust possibilistic chance-constrained programming	Minimization of environmental impact and SC cost
[22]	Kanan et al. (2022a)	Animal manure, municipal waste	Punjab, Pakistan	Flexible and possibilistic programming	Decision improvement of the BG-SCND model
[23]	Kanan et al. (2022b)	Domestic wastes	Pakistan	Grey-fuzzy programming	Investment to achieve social well-being goals
[24]	Ghelichi et al. (2018)	Jatropha Curcas	Iran	Stochastic programming	Supply chain optimization
[25]	Mirhashemi et al. (2018)	Moringa oleifera	Iran	Linear programming/CWDEA model	Νew source supply chain design
[26]	Akgul et al. (2012)	First-/second-generation biofuels	UK	ε-Constraint method/multi-period model	multi-objective model: impact of carbon tax, economic and environmental performance; trade-off between economic and environmental objectives, etc.
[27]	Ghaderi et al. (2018)	Switchgrass	Iran	Possibilistic programming	Economic, environmental and social objectives
—	lycourghiotis	WCO and HO, etc.	Greece	G.I.S. and spatial analysis	Optimal spatial solution model: local, regional or national

, a summary of the above studies is presented.

2. Data

The most critical issue for any geographic survey is the reliable collection and recording of the data in question. For the collection of data, depending on the type and their extent, the following methodologies were used: (a) on-site recordings, (b) sampling recordings and (c) integration of data from open sources. The data recorded in the geographic database were as follows: I. quantities of waste cooking oils (WCOs) and of household oils (HOs), and II. quantities of lignocellulosic biomass. For WCOs and HOs, the data were obtained from the open sources of the Hellenic Static Service (ELSTAT) [29], data from oil collection companies [30] and the estimates of Kiritsaki et al. [31,32]. For the different amounts of lignocellulosic biomass, I used the open government GIS data from the study on biomass potential by municipality in the country [33]. This open data gives quantities in weight and energy for six different sources: point sources (A), arable crops (B), greenhouse crops (C), tree crops (D), vineyards (E) and forests (F)

3. Spatial Analysis: Waste Cooking Oils (WCOs) and Household Oils (HOs)

In this section, we look at the geographical distribution of quantities of WCOs (also known as frying oils) produced by non-domestic households (restaurants, BBQ restaurants, hotels, fast food restaurants) that have to be gathered by collection companies. The quantities of these oils were determined through extensive telephone calls to all companies involved in the collection industry throughout the country. Following this method, we determined these quantities with accuracy in 278 of the 325 geographical units of the country (municipalities). In the remaining 47 units, where there were no data, the individual distribution was estimated on the basis of the population of the respective municipalities that we considered to be representative. Although the estimation involved a significant degree of uncertainty, the small contribution of the 47 units to the total sample was calculated to be only 8.3% of the national total. Thus, based on the variability in the values from the average, the uncertainty of the calculation was significantly reduced. Taking into account the strong correlation observed between per capita income and WCO quantities in each unit, it was estimated that the total uncertainty introduced by these 47 units into the final estimation was 1.8%. At this level, the uncertainty did not substantially affect the results. The geographical distribution of WCOs is shown in Figure 1. Prices are in liters per capita per year.

It can be observed that there were significant differences in average production per capita per year in each municipality. The smallest value was 2.2, and the largest was 7.4 L. However, these quantities were not produced directly by the residents of each municipality but were produced indirectly, since all quantities drawn from the catering companies belonging to a municipality were attributed to the total of its permanent residents. Observing the map, we can see that the touristic areas of Cyclades, Dodecanese, northern Crete and the Ionian Islands presented high values. However, high values were also observed elsewhere, for example, in large urban centers. On the contrary, extremely low values were observed in the areas of Epirus, the southern Peloponnese, Thrace and elsewhere. From these observations, we were led to the query as to whether WCO production was related to the per capita income of each municipality, as given to us by ELSTAT. The correlation obtained is shown in Figure 2.

It can be observed that there was an impressive correlation between the two spatial variables (r = 0.87). This parameter is critical to the methodology we followed in estimating household oils. To identify high and low spatial correlations, we used the three spatial autocorrelation indices (Moran’s I, Geary and G) for distributing WCOs (Figure 3).

It can be observed that the indices showed two areas with high-value correlations in the two major urban areas as well as low concentrations of near values, mainly in the regions of western Greece and southern Epirus. The high values in the two large metropolitan areas were partly explained by the very large quantities of WCOs (per capita) produced there but also due to the close proximity of many municipalities.

The quantities of WCOs now collected constitute a fraction of the total amount that could potentially be utilized if the quantities that could be collected from consumed HOs are also considered. The study of the spatial distribution of domestic oils, i.e., the quantities consumed by households is not an easy task, since there are no data on the consumption per region. For the assessment of HOs, we took into account (a) the data of ELSTAT for the consumption of olive oil [29] per inhabitant, (b) the data of the report on the domestic consumption of other seed oils and (c) the estimates of Kiritsaki et al. [30,31] for total oil consumption per capita. It should be emphasized that although the data provided by ELSTAT are detailed and offer an accurate representation of the quantities of oils sold in each store across the country, it is possible that the quantities of HOs are underestimated. The reason for this is that in Greece, there is a widespread market for olive oil from non-commercial, private producers. These transactions are not recorded in official data. According to the estimate of Kiritsakis et al. [30,31], this unrecorded market may account for more than 20% of HOs.

The geographical distribution of domestic oils is shown in Figure 4. Values are also expressed here in liters per inhabitant per year.

This distribution displays very large inequalities in consumption. The smallest value was around 4.8 L, while the largest one exceeded 22 L. In order to identify high and low spatial correlations, we used the three indices of spatial autocorrelation (Morans I, Geary and G) for the distribution of HOs. The results are presented in Figure 5 and Figure 6.

In most areas (235), no correlation was found. Moran’s I index was significant (0.68), but even in this case, the correlation was derived from very few values (top right in graph) that mainly concerned large urban areas. Without these areas, the indices showed no significant spatial correlation. Estimating the values for domestic oils, we found a correlation in terms of high values being observed in the same areas that we found for WCOs, i.e., in the two major metropolitan areas (Athens and Thessaloniki). On the contrary, we observed a concentration of low values mainly in eastern Macedonia, in mountainous areas of Pindos and in some border areas. The general picture given by all three indices is similar. The areas of the big urban centers, in which very high quantities of consumption are concentrated, seem on first sight to be the most promising. Nevertheless, there was no substantial direct correlation between WCO and HO values, and Moran’s I index was only 0.32 (Figure 6). The above results indicate that there is no clear picture concerning the geographically optimal way of collecting residual oils. The methodology has a significant limitation when applied to small islands. Most of these consist of a single unit, making it difficult to accurately estimate spatial autocorrelation indicators. This restricts the main conclusions to the mainland and larger islands, which, however, account for 94% of the total quantities.

Let us move from the distributions to the total amounts of residual oils that could be utilized in Greece. If we add up the annual amounts of frying oils throughout Greece recorded in the present study, a total amount of 44.87 million liters is obtained. Moreover, the quantities in the estimates of domestic oil consumption approach 118.3 million liters. Even if a small fraction (20–30%) of HOs can be recycled, the total available quantity WCOs could reach 80 million liters per year. This is a small but not insignificant fraction of domestic diesel needs.

In order to cover a larger fraction concerning fuel demands, we should turn our attention to the much larger quantities of lignocellulosic biomass, i.e., to residual raw materials from agricultural crops, forests, etc., as these can be converted into biofuels through rapid pyrolysis.

4. Spatial Analysis of Lignocellulosic Biomass (Agricultural/Forest Residues)

Based on the data of the Center for Renewable Sources and Energy Saving (KAPE) [32] used in this paper, the available amount of biomass (in mass and energy) from solid residues that can be used as energy was calculated for each municipal district (a total of 5922 municipal districts). The sources of biomass were classified into the following groups: (a) point sources of biomass relevant to the units of production of significant amounts of residual materials, (b) arable crops, (c) greenhouse crops, (d) tree crops, (e) vineyards and (f) forests. The data in

Table 2. The amounts of biomass (in mass and energy) corresponding to various sources. Τhe first and second lines express the quantities of weight and energy in thousands of tons and thousands of megajoules, respectively. Τhe third and fourth lines express the quantities of weight and energy as a percentage.

Sources of Biomass	Point Sources	Arable Crops	Greenhouse Crops	Tree Crops	Vineyards	Forests	Sum
×1000 Τ	459.55	1854.17	94.45	1090.48	231.50	821.12	4551.27
×1000 ΜJ	7247.71	33,065.18	906.69	18,932.61	4386.84	15,994.40	80,533.44
% (Tones)	10.10	40.74	2.08	23.96	5.09	18.04
% (MJ)	9.00	41.06	1.13	23.51	5.45	19.86

show that the total annual of residual biomass in Greece is estimated at 4.5 million tons, while the energy corresponding to this amount is more than 80 million MJ. The largest amount of biomass comes, as expected, from arable crops (40%), followed by tree crops (24%), forests (18%) and point sources (10%), whereas much lower amounts come from vineyards and greenhouses crops.

The spatial distribution of residual biomass throughout Greece for each biomass source is shown in Figure 7.

There is a strong inequality in the biomass distribution in Greece. Moreover, the different sources of biomass present different geographical distributions. It may be observed that the biomass sources relevant to forests and trees are located in different areas with respect to those relevant to arable crops. Thus, the lignocellulosic biomass is “scattered” in almost all parts of the country and is not concentrated in one or a few areas. Therefore, any collection and utilization plan should take this parameter seriously into consideration.

Although the spatial distribution of residual biomass has already been made clear from the previous figures, the use of spatial autocorrelation indices may give us more clear answers. To this end, a spatial correlation effort was made between the various biomass sources (Figure 8).

It can be observed that there is no strong correlation between one source and any other. The indices are, in most cases, below 0.1, which shows that a very low, up to zero, correlation prevails. The diagrammatic, rather than the cartographic, representation of Figure 8 may not provide a fully comprehensible picture. For this reason, in Figure 9, we quote a cartographic correlation between two sources, arable (B) and tree crops (D), which add up to more than 60% of the total amount. It is obvious that although there are areas with some correlation, the predominant situation is that of a zero to a very low correlation.

5. Discussion, Challenges and Limitations

1.

Since residual raw materials have to be collected at hundreds of thousands of points throughout the country, developing and optimizing one or more of the utilization processes is not the only problem. Equally important is the development and optimization of a collection process. The findings of this study aim to contribute to this objective. It has become clear that the method of spatial correlation can provide a satisfactory picture concerning the high and low concentrations of a quantity as well as the correlation between multiple quantities.

2.

Focusing on WCOs, it was shown that their quantities represent a very important, partially untapped, source of residual raw materials that can be used to produce green diesel in the context of the circular economy. Although the dispersion of frying oils is wide, there is no doubt that their concentration in large cities and tourist areas is higher. This finding suggests a collection process that should be carried out mainly in these areas. Concerning the utilization of residual oils in these areas, the development of small autonomous collection units in each neighborhood and central processing plants in small regional units seems to be reliable. In the first instance, in these areas it will obviously be necessary to carry out a pilot installation of a collection and processing device. This device would receive residual oil in liquid form from households and enterprises. The oils would be then collected in a small settling tank at central treatment plants, heated, stirred and separated from any impurities. The oil would then be fed to a flow reactor, from which the green diesel would be collected and stored. The necessary hydrogen for the process could be obtained through renewable sources (electricity from wind turbines and photovoltaics and the production of hydrogen by electrolysis of water, which is, of course, a much more expensive solution). Thus, the process would be completely green.

3.

The investigation of the geographical–spatial distribution of residual lignocellulosic biomass showed the geographical fragmentation and heterogeneity of the distributions. The quantities recorded are significant but widely dispersed, such that the cost of collecting and transporting the biomass to central processing plants could be prohibitive. The “geography” of the problem itself indicates solutions of small mobile collection units in every part of the country. In this context, the lignocellulosic biomass would be collected and converted in situ into bio-oil by rapid pyrolysis carried out in a tanker vehicle. This would transport the produced bio-oil to the nearest oil refineries for the conversion of the bio-oil into biofuels through deoxygenation processes.

4.

The methodology used in this study was applied for the first time to analyze this problem. Despite its originality, it involves several limitations and challenges. One of these is that spatial analysis carries significant constraints when geographical units have unique characteristics. For example, the presence of numerous islands in the Greek case leads to distorted estimates of indicators such as Moran’s I and others in these regions. Additionally, while the methodology provides a general overview of the problem, it cannot accurately determine a future collection and transportation chain, as other methodologies can. These limitations present challenges for the methodology, which will be the subject of future research.