Life Cycle Analysis of Food Waste Valorization in Laboratory-Scale

Abstract

Among the different alternatives for the production of biofuels, food waste could be a favorable bioenergy source. Using food waste as a feedstock has the potential to meet the expectations of the second generation of biofuels, in terms of environmental savings and revenue-generation, and which, along with other valuable co-products, can contribute to biorefinery profits. This study aimed to investigate the early stages of life-cycle assessment (LCA) for restaurant food waste processed into bioethanol, biomethane, and oil, split over different scenarios. Based on a life cycle inventory analysis, the environmental impacts were assessed using an IMPACT 2002+ methodology. The characterized impacts were then normalized against the average impacts, and the normalized results were weighted and aggregated to provide single score LCA results. The overall findings showed that electricity consumption and condensates included VFAs, as well as enzymes, yeast, and n-hexane, were the main contributors to the environmental burdens in all impact categories. Considering the sensitivity analysis, the results demonstrated that the enzyme dosage loading in the hydrolysis process and n-hexane utilization in the fat extraction process can change the environmental performance, along with the process efficiency. This study can provide an approach to foresee environmental hotspots in the very early developmental stages of food waste valorization into biofuels, and for highlighting drawbacks connected to the implementation of conversion processes at pilot and industrial scales.

1. Introduction

Bio-waste is one of the key waste streams for pursuing the potential of the circular economy in Europe. According to the European Environment Agency’s (EEA) assessment, bio-waste is the largest single component (34%) of municipal waste in the EU (European Union) and 60% of this amount is food waste [1]. Based on EEA reports, food waste prevention through decreasing the market demand can considerably reduce the environmental impacts caused by production chains; however, recycling is still a necessity.

Food waste recycling, by finding ways to utilize it or by producing value-added bio-products, is among the main means for climate change mitigation and shifting from fossil fuels to renewable energy, while also addressing the challenges of waste management.

The substitution of fossil fuels began with biofuels produced from food crops containing starch or sugar, as the first-generation biofuels. The conversion of food crops into bioethanol raised concerns about food security on a global scale [2]. Therefore, second-generation biofuels based on non-food resources were considered. Over the past few years, more researchers have focused on the so-called 2G biofuels [3], where residues or by-products are re-utilized; compared to the 1G biofuels, in which sugars and starch are used. Among all the different alternatives for the production of biofuels, food waste could be a favorable bioenergy source. Using food waste as a feedstock has the potential to meet the expectation of 2G biofuels, in terms of environmental savings and revenue generation, and which along with other valuable co-products can contribute to biorefinery profits. Converting FW to biofuels leads to noticeably reduced emissions of greenhouse gases, while soil and groundwater contamination by landfilling is a common consequence of waste disposal. However, landfilling and incineration are still the best known conventional waste management methods [4]. There are preferable approaches, such as composting, vermicomposting, and anaerobic digestion, to transition to value-added products instead of waste ending up in incineration or landfills. In recent years, researchers have worked towards the production of various types of bioproducts, such as liquid biofuels, biochemicals, and biomaterials [5], and whereby production of those bio-commodities may have greater economic benefits than conversion to common biofuels such as biogas [6].

In general, valorizing food waste for both biofuels and bio-commodities represents a step toward a circular economy. The circular economy is gaining increasingly more attention worldwide, as a way to develop sustainable consumption and production. Conceptually, the circular economy could be a solution for replacing the fossil-based linear economy, with sustainable processing of low-value feedstocks into a range of marketable and bio-based energy and material resources. Moreover, this approach could help to mitigate the environmental burdens attributed to urbanization and waste management patterns, especially by reducing the amount of food that goes to landfills. According to the Ellen MacArthur Foundation, with the current agriculture practices and consumption trends of food, for every USD 1 spent on food, USD 2 is paid by society in environmental, health, and economic costs. To understand the effect of such sustainable solutions, innovative tools are necessary to identify the trades-offs of improving the circularity in the food waste chain [7]. Biorefining strategies close the loop of fresh or raw resources, water, minerals, and carbon, along with better waste management approaches, which promise additional environmental and economic benefits. Hence, food waste biorefineries can play a key role in developing a waste-based circular bio-economy [5].

Recently, many studies have focused on waste biorefineries, to explore the abovementioned advantages of biorefinery systems, and the results showed they can increase production efficiency, along with reducing costs and environmental contamination [8]. According to the definition of the International Energy Agency (IEA), a biorefinery is a method of sustainable biomass processing for a wide range of bio-products of food, feed, chemicals, biomaterials, and bioenergy products [9]. Despite all the advantages attributed to biorefineries, such a synergetic production system is not zero-polluting, because of a wide spectrum of variables during the production chain, such as feedstock diversity, local conditions, and the design and implementation of the conversion process [3]. Hence, to assess the environmental performance of biorefineries and make a comparison to that of their fossil counterparts, a comprehensive environmental assessment is required.

In view of this challenge, in recent years, life cycle assessment (LCA) has been widely performed, to measure and assess the environmental impact of biorefineries [10]. Some of these studies conducted LCA for multiple substrates, including FW, while others considered only FW as a feedstock. Moreover, the majority of LCA studies focused on biogas or energy production, with only a few reporting on the synthesis of high-value-added biochemicals [11]. For instance, Karka et al. (2017) performed a large-scale LCA of producing various biochemicals and biofuels (23 products) from a broad set of biomass waste (wood chips, municipal solid waste, rapeseed oil, wheat straw, and waste cooking oil) [12]. Opatokun et al. (2017) performed a life-cycle assessment of energy production from food waste through different methods [13]. In the same context, Xu et al. (2015) performed an LCA for scenarios of biogas generated from food waste [14]. In recent years, the number of LCA studies presented for valorizing food waste into bioproducts rather than biogas has been increasing. Although some of the recent LCA studies were carried out based on experimental data, the application of LCA at laboratory/pilot scale can lead to a framework for assessing the environmental impacts of food waste valorization processes at an industrial scale. Lam et al. (2018) performed a life-cycle assessment on food waste valorization into hydroxymethylfurfural (HMF) in experimental conditions. Their study showed how LCA guides the selection of the best valorization processes, in order to obtain environmental savings [11]. Elginoz et al. (2020) used laboratory data to prepare life-cycle inventories for comparing the environmental performance of producing a volatile fatty acid-rich (VFA) supernatant from food waste and landfilling, as a conventional waste management approach [4]. As previously mentioned, LCA evaluates the future environmental aspects of a system, which could be helpful to reveal hotspots at industrial scale [15]. Furthermore, the results of LCA at laboratory or pilot scale can guide decision-makers and industries in developing sustainable technologies and processes [16].

In this paper, a laboratory-scale LCA for valorizing restaurant food waste into bioproducts was carried out. The study aimed to investigate an early-stage LCA for food waste conversion into bioethanol, biomethane, and oil, split over different scenarios. The novel aspects of this study consist of laboratory-scale scenarios, to guide the decision of selecting the most environmentally friendly scenario for valorizing food waste in a future scale-up. This study could represent an approach to foresee environmental hotspots in the very early developmental stages and highlight drawbacks connected to the implementation of conversion processes at pilot and industrial scale.

2. Materials and Methods

2.1. Descriptions of Food Waste Conversion at Laboratory Scale

In this section, laboratory-scale procedures for the conversion processes are presented. Laboratory experiments were conducted by the Unit of Environmental Science and Technology (UEST), at the National Technical University of Athens (NTUA), Greece, to investigate the potential improvement and optimization of the restaurant food waste conversion process into bioethanol. The feedstock considered in the present study was collected from typical restaurants situated in the central touristic zone of Athens. This mostly included wastes of raw food from food making, food scraps, rotten raw foods, and leftovers.

Table 1. Characteristics of the RFW considered in this study.

Characteristics	Value	Unit
Initial moisture	50.9 ± 0.01	% w/w w.b.
pH	5.65 ± 0.13	-
Conductivity	2.72 ± 0.10	mS/cm w.b.
BOD5	0.77 ± 0.17	g/gRFW w.b.
COD	0.85 ± 0.09	g/gRFW w.b.
VS/TS	1.05 ± 0.05	% w/w w.b.
Carbon
-TC	50.92 ± 2.12	% w/w d.b.
-IC	ND	% w/w d.b.
-TOC	50.92 ± 1.96	% w/w d.b.
Nitrogen
-TN	1.67 ± 0.02	% w/w d.b.
-TKN	1.67 ± 0.02	% w/w d.b.
Phosphorous
-TP	0.23 ± 0.03	% w/w d.b.
Metallic elements
-Na	6.645 ± 0.125	mg/g d.b.
-Mg	0.767 ± 0.101	mg/g d.b.
-K	4.479 ± 0.053	mg/g d.b.
-Ca	4.597 ± 0.061	mg/g d.b.
-Cr	ND	mg/g d.b.
-Mn	0.012 ± 0.002	mg/g d.b.
-Fe	ND	mg/g d.b.
-Ni	ND	mg/g d.b.
-Cu	ND	mg/g d.b.
-Zn	0.024 ± 0.003	mg/g d.b.
-Cd	ND	mg/g d.b.
-Pb	ND	mg/g d.b.
Nutritional value
-Fats	10.55 ± 0.35	% w/w d.b.
-Proteins	9.43 ± 0.55	% w/w d.b.
-Carbohydrates
Free glucose	1.24 ± 0.14	% w/w d.b.
Total reducing sugars	7.07 ± 0.20	% w/w d.b.
Starch	57.82 ± 0.77	% w/w d.b.
Cellulose	5.79 ± 0.19	% w/w d.b.
Hemicellulose	6.19 ± 0.48	% w/w d.b.

w.b.: wet base (RFW). d.b.: dried base (Dried RFW). ND: not detected. All analyses related to the characterization were performed in quintuplicate and the average values were evaluated.

shows the main characteristics of the raw food waste (RFW) considered in this study [17,18]. All analyses related to the characterization were performed in quintuplicate, and the average values were evaluated; in order to achieve reliable results, those experiments related to fat extraction, enzymatic hydrolysis, fermentation, and anaerobic digestion were replicated 3, 3, 5, and 5 times, respectively [17].

Within the scope of this study, a simplified process flow chart, Figure 1, was constructed, including milling, dehydration, solid–liquid fat extraction, enzymatic hydrolysis, fermentation, distillation, and anaerobic digestion.

The collected RFW was dried and milled using a rotary drum dryer at 105 °C for 17 h in a pilot plant. The main output, as a dried and milled FW, went to the fat extraction process, and the evaporated moisture, including volatile fatty acids (VFA), was condensed and collected, to use in the anaerobic digestion process. The considerable amount of fat in food waste makes it sufficiently valuable to extract it as a potential value-added side product, to use as a ready-to-sell raw material for production of various bio-based products, such as bio-diesel, bio-plastics, industrial soaps, etc.; in addition, it is worth mentioning that the presence of such quantities of fat has some negative effects on the enzymatic hydrolysis and fermentation processes [19]; moreover, the emulsified fat droplets formed during the experiments and processes produced noticeable errors in the measurement of glucose and ethanol. Hence, in the fat extraction step, the Soxhlet method was applied to extract the fat content of dehydrated FW using a solvent. The crude fat was obtained from the liquid output of Soxhlet using a rotary evaporator (BÜCHI Rotavapor-RKRvr 65/45), in which the applied solvent was distilled and recovered under 0.28 mbar vacuum at 55 °C. The defatted food waste (DFW) as the solid output of the extraction process was heated in a drying oven at 105 °C for 10 min, to remove the residual moisture and solvent content. It was notable that 96% of the used solvent was also recovered in the process.

Then enzymatic hydrolysis of defatted FW was conducted at 65 °C for 90 min in an Incubator Shaker IKA-KS 3000i control with an agitation of speed 225 rpm and using amylase NS22109 (noncommercial enzymatic formulation amylase, which was kindly donated by Novozymes, Denmark). After enzymatic hydrolysis, the released glucose and initial glucose were biologically converted to ethanol through the fermentation process by adding Saccharomyces cerevisiae at 32.5 °C for 24 h in the same incubator shaker as above, with an agitation speed of 175 rpm. According to the composition of DFW and the efficiency rate of fermentation, there was no need to add nutrients during the fermentation process. The broth of the fermentation process was distilled under 0.3 mbar vacuum at 60 °C, using the above-mentioned BÜCHI Rotavapor—RKRvr 65/45. Bioethanol was collected as the main output, and the more concentrated part, the so-called stillage, was delivered to the biomethane potential tests (BMP). At this stage, BMP was performed, to investigate the anaerobic digestibility of the stillage from distillation and the effluent output from the dehydration process, for producing biomethane. The inoculum was retrieved from a pilot-scale anaerobic reactor treating wheat straw (VS 5%), and a nutrient basic medium was prepared based on the protocol by Angelidaki et al. (2009) [20]. The samples were placed in a horizontal shaking water bath (J.P. Selecta Unitronic-Orbital 6032011) at 150 rpm and 35 °C for anaerobic digestion. Finally, the co-digestion of the presented mixture showed a positive effect on both the biogas production rate and volume. The biomethane potential was 9.20 L CH₄/L, and 90% of this value was reached after 11 days. Drosg et al. [21] reported higher values of BMP from corn and wheat stillage: 611 and 579 L CH₄/kg VS, respectively. Tang et al. [22] used treated stillage of kitchen waste with 6 g VS/L-d and reported a 32.6 L CH₄/kg waste loading rate. Koike et al. [23] processed the stillage resulting from food waste through dry methane fermentation, and 57.3 L CH₄/kg waste was observed. In addition, Tan et al. [24] used stillage from kitchen and paper waste and achieved 392 L CH₄/kg VS. However, the value observed in this study for the mixture of the streams falls within the range reported in the literature for food waste stillage.

2.2. Descriptions of LCA Scenarios

LCA is a methodology for the comprehensive assessment of the environmental impact associated with a product or process throughout its life cycle (from the extraction of raw materials, to product disposal at the end of its life) and it is sometimes referred to as cradle-to-grave analysis [25]. Herein, the LCA methodology was followed according to the ISO 14040 and 14044 standards, in four steps [26,27]: (1) goal and scope definition: defining the objectives, system boundary, and the functional unit; (2) data collection and inventory analysis; (3) impact assessment: selecting the impact categories for the energy and resource, as well as the emission generated; (4) result interpretation and presentation: reporting the results in as much detail as possible.

2.2.1. Goals and Scope Definition

This step consists in setting goals, to ensure that the precise aims, methods, outcomes, and intended applications are optimally aligned. The goal of this study was to assess and compare the LCA results of defatted food waste and non-defatted FW valorization scenarios at laboratory scale. The scope of the LCA covered the processes of production within a “gate–gate” system boundary, where a chain of material/energy flows occur, to produce the final product of interest. As Figure 2 depicts, the two considered scenarios differed in terms of the fat extraction process. Scenario A presents bioethanol production from defatted FW, including fat extraction through a solvent “n-hexane 95%”, and scenario B, without an oil production step, was modeled using SimaPro version 8.5.2.0 software. The functional unit (FU) was defined as the conversion of 1 kg of food waste substrate. Sensitivity analyses were included to evaluate the significance of the variability of the LCIA results, by varying the critical parameters of the hydrolysis process in the laboratory. The infrastructure of the equipment were excluded from the system boundary, and the study focused on the operation of the system. Collection and transportation of RFW were also excluded from the system boundary.

2.2.2. Inventory Analysis

Life-cycle inventory (LCI) includes relevant input/output information or data of all processes involved within the defined system boundary. Hence, LCI is a significant procedure, where the reliabilities of an LCA study outputs are fully dependent on the quality and adequacy of the LCI content [28]. In this study, the details of the conversion process of FW into bioethanol were obtained from experimental procedures performed in the laboratory. Input/output data were based on the optimized quantity of the materials, chemicals, and energy utilized during the experiments. In addition, LCA modeling of the scenarios involved the selection of data from Ecoinvent v3 [29]. For processes not found in the Ecoinvent databases, input/output data were obtained from the literature. The electricity consumption for the processes was calculated based on the power of the devices and the time of use. The electrical power was supplied by the Greek national electricity grid, which is included in the Ecoinvent database for the year 2014. The Greek electricity grid mix dataset in Ecoinvent in that year comprised 27% natural gas, 25% lignite, 13% oil, 12% hydro, and 10% wind, while 10% of electricity was imported from neighboring countries [30]. We also adopted data on yeast production from the paper published by Dunn et al. (2012) [31]. In the analysis, we also considered the “zero burden assumption”, in which upstream environmental burdens were not included in the analysis. Furthermore, food waste collection and transportation were excluded from the system boundary.

2.2.3. Impact Assessment

A life-cycle impact assessment (LCIA) phase was carried out, to calculate the environmental impact of products/services using an impact methodology. This phase included different methods, to provide indicators for the environmental impacts of processes considered within the system boundary [32,33].

Based on the LCI analysis of the inputs and outputs of the FW conversion into bioethanol and co-products, the environmental impacts were estimated with the IMPACT 2002+ methodology. This LCIA model covers 17 midpoint categories and four damage-oriented categories for evaluating environmental impacts Out of these, nine environmental midpoint impact categories were investigated in this study.

Table 2. Selected impact categories investigated in this study.

Impact Category	Unit	Abbreviation
Ozone layer depletion	kg CFC-11 into air-eq	OD
Respiratory (organics)	kg ethylene into air-eq	RO
Ionizing radiation	Bq Carbon-14 into air-eq	IR
Land occupation	m2 organic arable land-eq.y	LO
Global warming	kg CO2 into air-eq	GW
Non-renewable energy	MJ crude oil-eq	NRE
Mineral extraction	in MJ Iron-eq	ME
Terrestrial ecotoxicity	kg Triethylene glycol into soil-eq	TE
Aquatic ecotoxicity	kg Triethylene glycol into water-eq	AE

presents the selected impact categories. The selection of the categories was based on the main processes and reflecting environmental issues related to the biofuel production system.

IMPACT 2002+ is a comprehensive LCIA methodology that provides characterization factors, as well as normalization factors [34]. The environmental impacts were classified into different categories according to the midpoint indicators and then characterized into common equivalent units that reflected their contributions to the midpoint impacts. The characterized impacts were then normalized against the European inhabitant average impacts, so that the relative importance of the impacts in different categories could be considered. The normalized results were weighted and aggregated, to provide single-score LCA results, which were inclusive and convenient indicators for showing the final results. This method was chosen, as it reflects a European perspective on models and factors.

3. Results and Discussions

In this section, the LCIA results for the scenarios and sensitivity analysis are analyzed. The results are presented for the characterization and normalization phases, as well as the single-scored results.

3.1. Characterization Results

The results for the nine impact categories are shown in Figure 3. The first observation to be highlighted is that scenario A generated higher impacts in all categories compared with scenario B. The major driving factor of the environmental impacts in scenario A derived from the fat extraction process. The results show that the electricity consumption and condensate output from the dehydration process had a dominant contribution to all categories, in both scenarios. Electricity production accounted for the major contribution in all impact categories. This contribution in scenario A was 3.5% higher than in scenario B. Previous LCA studies at laboratory scale (Ang et al., 2020, and Pallas et al., 2020) concluded that electricity consumption is a main contributor to various impact categories, such as climate change [35,36]. In addition, an investigation of the burdens sourced from the dehydration process showed that condensate from the drying process had a major contribution in all impact categories in both scenarios, due to the VFA-rich effluent. Among the impact categories, AE with an amount of 1.30 × 10⁻¹ (kg TEG water) emerged as a significant impact from the dehydration process.

The utilization of enzyme, yeast, and n-hexane were the top-three contributors in the impact categories after electricity and VFAs. Since enzyme production requires chemicals, inorganic compounds, energy, and crops, it contributes to the amount of TE and AE, as the second-largest value. Terrestrial ecotoxicity and aquatic ecotoxicity values are affected by the enzyme used in the hydrolysis process, by 6.43 × 10⁻⁴ (kg TEG soil) and 4.56 × 10⁻⁴ (kg TEG water), respectively.

The environmental impacts associated with the yeast (sacharomyces cerevisiae) used in the fermentation process are attributed to the energy requirements in yeast production. Hence, yeast contributes to the NRE (1.11 × 10⁻⁴ MJ primary) and IR (3.27 × 10⁻⁵ Bq C-14 eq) categories, as the third largest contributor. Moreover, yeast provides environmental credit (as avoided impact) to terrestrial ecotoxicity (−1.07 × 10⁻⁵ kg TEG soil), due to the agricultural product saved by using molasses in yeast production. Some of the environmental impacts were caused by use of chemicals and nutrients in the processes. Therefore, optimizing the processes of bioethanol production, as well as the upstream processes (production of enzymes, yeast…), could help to reduce these effects. Taking into account the process optimization is more important when an industrial scale is applied, where the processes are not operating as in the well-controlled conditions in experiments.

Moreover, research on the materials produced from green sources to replace current materials, as well as the recycling and recuse of the chemicals/nutrients, can be considered in future studies, as a solution to mitigate potential environmental impacts.

As previously stated, the fat extraction process added undesirable effects to scenario A. The electricity supply and n-hexane are the main factors behind the difference between the scenarios, with those contributing about 99% to the total impacts of the fat extraction process. The n-hexane, as an organic compound, led to higher impact category values in scenario A than in scenario B. This result is in line with Lam et al. (2018), who claimed that the production of co-solvent performed least favorably in all impact categories [11].

N-hexane was an important contributor to the AE and NRE categories, with the amounts of 3.33 × 10⁻⁴ kg TEG water and 1.25 × 10⁻⁴ MJ primary, respectively. The consumption of n-hexane ranked as the third largest contributor to five of the impact categories in scenario A. The reason for this was the usage of water and naphtha (produced from petroleum mixtures through refinery-based processes) in n-hexane production. To lessen the impacts caused by the electricity supply in the laboratory scale processes, deploying an electricity grid from renewable sources would reduce these impacts, but this investigation has to be country specific, due to the differences in energy supply from one country to another, as the location for carrying out the experiments. However, a complete elimination of n-hexane use may not be possible in the fat extraction process, but it is suggested that solvent recovery can be used as much as possible. Furthermore, the possibility of using green solvents, recycling and recusing of the chemicals/nutrients for future research, before application of the system at full scale, could be considered.

Avoided Burdens

In the processes CO₂ of dehydration, fat extraction, and distillation, the avoided impact of each approach was considered. IMPACT 2002+ characterizes the total impacts calculated through the avoided burden approach for the different scenarios, with reference to a functional unit of 1 kg of waste biomass.

In both scenarios, the produced condensate of the dehydration process contained VFAs, which are planned to be used as an input for anaerobic digestion, as volatile fatty acids play an important role in controlling anaerobic digestion and increasing the efficiency of the process. Using effluent for dehydration will avoid the production of VFAs and water, which requires the input of energy and material. Furthermore, in scenario A in the case of the fat extraction process, the yield of bio-oil and recovered n-hexane avoid the environmental impacts of the production of virgin biodiesel and n-hexane. The benefits derived from the avoided products in the dehydration and fat extraction processes are relevant in all impact categories.

Another output in scenarios A and B that is considered an avoided product is the stillage yielded from the distillation process. This stillage is a nutrient-rich medium, to enhance the performance of anaerobic digestion. Hence, the avoidance of the production of stillage from another source allows savings in all the impact categories, except for the terrestrial ecotoxicity due to the emissions into the soil through agricultural activities, as well as the use of energy and material during the production of food crops. In scenario A, ozone layer depletion is the impact category that mainly benefits from avoiding the considered products, with savings of about −2.49 × 10⁻¹⁰, −1.54 × 10⁻¹⁰, and −4.86 × 10⁻¹¹ kg CFC-11 eq for dehydration, fat extraction, and distillation, respectively. The same credit of OLD in the dehydration and distillation processes resulted in scenario B for the dehydration and distillation processes. However, in general, the compensation of the environmental burdens in the distillation process of scenario B is greater than in scenario A, due to the lower direct impacts in scenario B. Figure 4 depicts the credit in avoided products and environmental burdens in the relevant processes in both scenarios. It is worthwhile to note that the value of direct burdens of the processes is higher than the avoided impact values. Hence, subtracting the avoided burdens from the main impacts shows that the application of avoidance approach would lead to partially mitigating the burdens, but not a complete erasure of the negative environmental consequence.

3.2. Normalization Results

The idea of normalization is to present the respective share of each impact on the mid-point and overall damage of the considered categories. By comparing the characterized results with a reference value, this phase helps in understanding which categories are worthy of further investigation [37]. The aforementioned potential impacts in the previous sections were normalized with the IMPACT 2002 methodology, using European normalization references, which represent European emissions and resource consumption per average European inhabitant during one year [34]. In

Table 3. Normalized values for each environmental damage category.

Damage Category	Unit	Scenario A	Scenario B
Human health	DALY	8.75 × 10−7	8.43 × 10−7
Ecosystem quality	PDF.m2.y	2.73 × 10−5	2.63 × 10−5
Climate change	Kg CO2 eq	5.73 × 10−4	5.52 × 10−4
Resources	MJ primary	5.81 × 10−4	5.60 × 10−4

DALY = disability-adjusted life years, PDF.m².y = potentially disappeared fraction of species over a certain amount of m² during a certain amount of years, Kg CO₂ eq = equivalent to the effect of one kg of CO₂ emissions, MJ primary = measures the amount of energy extracted or required to extract the resource.

, the normalized results are displayed, based on damage categories, including human health, ecosystem quality, climate change, and resources.

As shown, the resources and human health categories have the highest and lowest scores, respectively, in both scenarios. Regarding the normalization results, the five substances that contribute to the potential impact to the highest extent are lignite, natural gas, crude oil, uranium, and coal. These are the sources of electricity production in Greece.

The normalized results in the midpoint impact categories indicated a higher magnitude of non-renewable energy and global warming compared to the other categories. The most significant contributors to those impact categories are the production of electricity and chemicals. In Figure 5, the normalized results are displayed for selected midpoint categories.

3.3. Single Score LCIA Results

The normalized results were weighted and aggregated to provide a single score in milli-points (mPt) using the IMPACT 2002+ assessment method. To provide an overview of the impacts of the scenarios, Figure 6 presents LCIA results expressed as a single score.

The environmental impacts were categorized into human health, ecosystem quality, climate change, and resources. The single score results show that resources and climate change are the most affected by the activity of both scenarios. Those two categories contribute around 49% and 48.5% to the overall impacts, respectively. Resource depletion is a result of extracting raw materials and resources for energy, organic solvent, yeast, and enzyme production. The LCA study on food waste valorization into hydroxymethylfurfural (HMF) conducted by Lam et al. (2018) revealed that the impacts from the resource aspects were the highest, due to the limited natural resources of tin to produce catalysts [11].

From Figure 6, it can be seen that the value in the climate change impact category is slightly lower than the value in the resources category. The high environmental damage in the climate change category is attributed to the electricity consumption in laboratory units and the nutrient/chemical production. Electrical power is mainly generated from natural gas, lignite, and oil in the Greek electricity network. The high value of CC is associated with the amount of carbon dioxide and nitrogen oxides originating from the electricity consumption in experimental procedures, as well as from yeast and chemical production.

The environmental stress on the ecosystem quality and human health categories is much lower than the potential impact on CC and resources. Both scenarios have a notably low influence on ecosystem quality, which expresses the sum of aquatic acidification/eutrophication, aquatic/terrestrial ecotoxicity, terrestrial acidification/nutrification, land occupation, and water consumption. However, the production of electricity causes adverse impacts on human health by emitting particulates < 2.5 μm, sulfur dioxide, and nitrogen oxides. It has the lowest score out of the four categories, since the conversion processes and materials have a low footprint associated with human health.

3.4. Sensitivity Analysis

A sensitivity analysis was performed to investigate the influence of conversion components on the LCIA results. An analysis of the results shows that electricity consumption, enzyme, yeast, and n-hexane are the main contributors to the background parameters of the environmental impacts. According to the available data, two parameters out of those above were subjected to sensitivity analysis, to check their effects on the impact indicator values. Subsequently, a description of the analyses was performed to assess the robustness of the obtained outcomes.

3.4.1. Enzyme Loading

Although there are different enzyme strategies implemented in the hydrolysis process, which depends on the type and concentration of feedstock and retention time, enzyme loading is an important parameter that can be an environmental hotspot in the system. Determining the optimal amount of enzyme consumption is very important, for two reasons: it can directly optimize the enzymatic hydrolysis process, and indirectly save resources in the enzyme production chain, as an energy-consuming industry.

Enzymatic hydrolysis of defatted FW was performed using different dosages of amylase, to identify the optimum enzyme loading. Based on the conducted experiments, the dosage of 15 μL was considered the optimum dosage for the amylase product formulation NS22109. To examine the sensitivity of the environmental performance of the bioethanol production, different enzyme dosages of 3, 30, and 108 μL were tested in both scenarios. The findings (Figure 7) of the sensitivity analysis revealed that with increasing enzyme consumption, the impact categories followed the same path as the environmental impact contribution. Aquatic ecotoxicity and non-renewable resources showed the biggest sensitivity to enzyme loading. However, this does not mean that the minimum dosage should be selected to receive the lowest environmental burdens.

According to the experiments, a low enzyme content (under 15 μL) did not result in an efficient saccharification performance; on the other hand, the increase of enzyme loading (over 15 μL) slightly improved the efficiency of the process; while in the latter, the environmental burdens caused by enzyme production were high. Hence, the optimum dosage (15 μL) of the enzyme, along with a proper performance of the process, showed a better environmental performance than the other amounts tested.

3.4.2. Solvent Type in the Fat Extraction Process

In scenario A, the fat extraction process from DFW was conducted using the Soxhlet method, with n-hexane 95% as a solvent. To conduct a sensitivity analysis of the environmental performance, replacement with 99.9% methanol was considered instead of the n-hexane in the process. Both solvents have the potential to be recovered and reused in the process, although the recovery potential for methanol is four percent lower than n-hexane. Within the optimum conditions of the experiments for both solvents, methanol substitution led to a lower amount of extracted oil and a lower efficiency. Furthermore, oil produced with n-hexane usage presented a low acidity as well as permissible content of moisture, which makes it an ideal substrate to use for biodiesel production. As shown in Figure 8, the sensitivity modeling with methanol added demonstrated that most of the categories, except OLD, LO, and ME, were sensitive to methanol consumption in the process. Those excepted impact categories showed the least sensitivity to solvent substitution. In terms of the overall analysis, the environmental savings attributed to methanol use in oil extraction will decrease due to a lower recovery potential and a higher value of environmental burdens.

3.5. Limitations of the Study

The main limitation of this study are the efficiency of the laboratory scale, in which inventories provided from experiments contain an inherent measurement uncertainty. Bioethanol and the co-products yielded at lab scale can be different from at real scale, because of carrying out experiments in well-controlled conditions. Preparation of the best experimental conditions for the samples may not be provided at an industrial scale. In addition, a part of this uncertainty might be derived from technology or material replacements in a real industrial plant. These changes cannot be captured in the early stage of an LCA. Accordingly, any major changes in the production system should be reflected in the LCA model and a new analysis needs to be performed. However, sensitivity analysis of different scenarios is a method used to deal with uncertainties, but it had to be done based on the available experimental data in the considered system boundaries of this research. Hence, technologies emerging in the future and other sensitivities were not investigated within the scope of this study.

4. Conclusions

In the present work, an early stage LCA was carried out to evaluate the sustainability of bioethanol production, by covering the usage of co-products and closing the loop. This assessment showed the potential role of LCA in process improvement, based on the outcomes of the environmental hotspot analysis, and to achieve improved sustainability of bioethanol production. Decisions made early in the process of designing or improvements influence the environmental performance of future technologies.

The LCA conducted in this study assessed two different scenarios of restaurant food waste valorization. Electricity consumption, condensate containing VFAs, enzymes, yeast, and n-hexane were the main contributors to environmental burdens in all the selected impact categories, in both scenarios. Regarding single score results, resource depletion results in extracting raw materials and fossil fuels for energy, organic solvent, yeast, and enzyme production. Moreover, the high environmental damage in the climate change category was attributed to the emission of CO₂ from electricity and nutrient/chemical production. Environmental savings resulted from the yield of condensate containing VFAs, while bio-oils and recovery of n-hexane slightly minimized the total environmental impact of food waste valorization. The conducted sensitivity analysis showed that the hydrolysis and fat extraction processes have a potential influence on the environmental performance of bioethanol production. Accordingly, the optimum amounts for enzyme loading provided by experiments, and n-hexane application instead of methanol, are the best options to improve the environmental performance, along with the process efficiency.

To conclude, performing laboratory-scale LCA for advanced biofuel production may allow the forecasting of environmental hotspots at a pilot or industrial scale. It will be necessary to conduct industrial-scale LCAs and a techno-economic assessment in the future. As shown above, the footprint of electricity consumption in both scenarios was inevitable. Hence, the consideration marginal energy sources is expected in the future, when the development of a large-scale valorization system is available and information is more readily accessible. These factors will provide more comprehensive results based on environmental and techno-economic performance.