Inherent Safety Analysis of a Cascade Biorefinery for the Utilization of Avocado Hard Waste from the South Colombian Region

Abstract

Cascade biorefineries have demonstrated potential due to their ability to further valorize a wide variety of waste, including agricultural residues such as avocado hard waste. The intrinsic safety aspects of these technologies have been scarcely studied. Therefore, an inherent safety analysis was applied to identify and assess the risks of an avocado cascade biorefinery in the Amazon region. Several available databases (online) were used to determine the safety data of the substances in the process, such as CameoChemicals, INCHEM, and NIOSH. Additionally, data from extended mass and energy balance (based on the literature) were collected to assess the process operating conditions. The results show that the process is slightly unsafe, with an overall inherent safety score of 25, and that it achieved a performance of 96% relative to the neutral operating point (24). Chemical risks represented the most critical challenges of the process, with a score of 16, with exothermic reactions, hazardous substances, and dangerous chemical interactions being the most significant sources of risks. On the other hand, the process safety indicator scored 9, indicating that these aspects are not a major source of risk, as the process had mostly low operating conditions (near-environment temperatures and pressures; low inventory), with equipment being the only significant risk factor. Nonetheless, the safety structure subindex for this process was 2, as no clear and recognizable risks existed (at least in the literature) for this type of scheme at the scale analyzed (small scale). This score needs to be studied to properly assess the risks in bioprocesses like cascade biorefineries. Finally, replacing acid hydrolysis with enzymatic hydrolysis, along with another method for bioactive extraction, is recommended to reduce the inherent risks.

1. Introduction

Bioeconomy presents a promising solution to global challenges such as pollution, climate change, and the depletion of natural resources—issues intensified by population growth and uncontrolled economic activities [1]. It integrates the principles of circular and ecological (biological) economics. While the first focuses on maximizing resource efficiency by transforming waste into raw materials, the latter aims to replace fossil-based resources with renewable, biologically sourced alternatives for energy, fuels, and products [2,3]. This approach involves shifting away from a traditional linear economy to a sustainable model that uses bioprocesses to convert biological waste into valuable products, contributing to a greener, low-carbon economy. Countries with abundant natural resources, such as Colombia, can particularly benefit from this transition, as it enhances material management, boosts economic performance, and supports vulnerable rural populations [4].

Biorefineries are envisioned to be a fundamental pillar of the bioeconomy. These are dynamic facilities that through the integration of conversion technologies (thermochemical, biochemical, etc.) and transform various forms of biomass into high-value-added products such as biofuels, bioplastics, biochemicals, and biomaterials, among others [5]. The production of these bioresources not only addresses concerns related to energy and environmental security but also enables better waste flow management. Furthermore, cascading processing is an innovative approach that increases the overall processing of these biomasses, as a greater proportion of residual streams are transformed through sequential processes [6]. The heterogeneous composition (starch, cellulose, hemicellulose, lignin, oil, starch, and proteins, etc.) of feedstock can be overcome, as it is harnessed into various functionalized bioproducts by using a combination of multiple processes [7].

Depending on the composition, biorefineries can perform physical pretreatment to reduce particle size, increase surface area, and improve the accessible surface for heat and mass transfer [8]. The extraction of valuable bioactive compounds (if present) uses different extraction methods, ranging from conventional solvent extraction to novel methods like microwave or ultrasonic, among others [9]. After this extraction, the depleted solid fraction can enter hydrothermal, thermochemical, physical, physicochemical, or biological pretreatment to fraction long structures into more easily processing compounds (short monomers) [10]. Then, these compounds are transformed through biochemical (microbial fermentation) and thermochemical technologies such as transesterification, liquefaction, torrefaction, pyrolysis, or gasification) that transform the treated compounds into commodities for further processing or valuable products where material products are prioritized. Subsequently, the waste of these subprocesses are transformed into energy, like biofuels or biochar, among others [11]. However, due to wide variability in biomass composition based on their source, type, and recalcitrance level, there is no standard method for the bioconversion of lignocellulosic biomass; sequencing and hierarchization are crucial for ensuring techno-economic feasibility [12]. Additionally, these processing systems follow the principles of green chemistry, as they must be technically and economically viable while posing no additional risk to human health and the environment [13].

The main raw materials include various types of waste materials, such as food waste, industrial residues (e.g., from the paper and pulp industry, among others), agro-industrial byproducts, forestry and agricultural waste, and lignocellulosic materials, as well as wastewater or sludge. Among all biomass resources, avocado has emerged as a particularly valuable raw material due to its chemical composition [14]. This fruit not only contains significant lignocellulosic material, but also has a high content of nutritionally valuable compounds, such as fatty acids and bioactive substances [15]. For Colombia, avocado represents a major economic opportunity, as the country is one of the leading producers in Latin America, ranking third in production. In 2021, Colombia produced approximately 61,853.6 tons of native avocado [16]. Moreover, the Colombian avocado oil market reached USD 484.6 million in 2020 [17].

Furthermore, the growing demand for avocado consumption has led to a significant increase in waste along the supply chain, including rejected avocado or its peel and seeds fractions, due to multiple reasons, ranging from not conforming to quality standards to sickness and transportation issues. The improper management of these residues represents a significant challenge, contributing to air and wastewater pollution, as well as having a negative impact on public health [18]. Considering this abundant raw material, integrating the utilization and valorization of avocado waste into the avocado value chain is justified through biorefinery schemes. This strategy fosters sustainability and supports the bioeconomic growth of the producing regions [19]. However, this alternative requires rigorous holistic analysis to maximize socioeconomic benefits while minimizing environmental impacts. This calls for a comprehensive evaluation of sustainability dimensions, including safety, to improve feasibility for industrial applications.

Since biorefineries are based on green chemistry principles, the safety of these processes must be addressed through the concept of inherently safe design. This approach focuses on eliminating, avoiding, or minimizing process risks prior to the implementation of control measures [20]. It aims to identify the potential hazards in equipment and processes at the early design stages and make modifications to eliminate them, leading to substantial reductions in cost and efforts [21]. There are various methods for analyzing the inherent safety of processes, such as index-based approaches, graphical methodologies, and computer-assisted techniques [22]. Index-based methods simplify the evaluation process by consolidating multiple hazard parameters into a quantitative factor that aids decision-making. These methods assess different process route alternatives by assigning scores and categorizing process routes from the safest to the most hazardous [23].

Among these methods, the Inherent Safety Index (ISI) is the ideal tool for comprehensive safety analysis during the conceptual stage. This method, proposed by Heikkilä, is capable of analyzing processes and systems with limited information [24]. Its holistic approach evaluates both chemical and process hazards by considering multiple risk factors (e.g., flammability, toxicity, temperature, etc.), based on the intrinsic properties of the process and its components (substances, equipment, operating conditions, etc.) [25]. Additionally, ISI consolidates these indicators into a single value, allowing for the comprehensive comparison and analysis of different process routes.

In the case of avocado waste biorefineries, the implementation and commercialization of the proposed processes remain limited. This is primarily due to challenges such as an irregular biomass supply chain, market uncertainties, and difficulties associated with scale-up—particularly the low technology readiness level and the lack of information regarding their viability and performance at an industrial scale [26]. Therefore, most of the literature addresses the analysis of rural applications, centered on technical feasibility at mostly the small or medium scales, technoeconomic potential, and environmental assessments using life-cycle-based analysis. Cardona et al. assessed the benefits of several food waste biorefinery processes, including avocado agricultural waste (300, 150, and 75 kg/h) proposed for rural populations. This analysis showcased the potential economic benefit of small-scale and low-technology applications in rural communities. The production of avocado oil, fertilizer, and biogas presented with positive maximum VPN values at 6 years for 75 kg/h (lowest at 3 years for 300 kg/h) [27]. The extraction of bioactive (from peel and seed) compounds using supercritical fluids increases the technological requirements and cost expenditure (more equipment required and cost to acquire and operate) of the biorefinery, but benefits the economic performance of the process; however, it is conditional on the composition of the feedstock, as peel presented more yield than the seed. Furthermore, the production of ethanol, syngas, and another energy vectors (steam) improved economic processes, as they reduced material and energy intake (from 21% to 47%) [28]. Moreover, Herrara et al. economically analyzed a biorefinery using oil extraction and biochar production from rejected avocado, with a higher scale of 10,500 t/y, and showed positive performance with a return on 6.67 years of investment. The process would be benefited if a cascade system were to be implemented, as several waste and byproducts are not considered in the economic analysis [29]. In contrast, Solarte-Toro et al. showcased the critical influence of the product cost and scale of biorefineries. Rural small-scale (2 t/h) biorefineries presented better economic performance (return on investment between 2 and 6 years) compared to high-scale (between 27 and 42 t/day) biorefineries (between 14 and 19 years); however, rural biorefineries offer low-cost high-volume products, whereas high-scale biorefineries offer specialty chemicals that are much more valuable [30].

In the case of environmental analyses, Solarte-Toro showcased that creole avocado biorefineries have low environmental impact, with a carbon footprint of 0.59 kg CO₂-eq/kg and a water footprint of 2.13 m³/kg. Among the two proposed biorefineries, Small-B1 (focused on avocado oil and animal feed) had significantly higher environmental impacts—8.99 kg CO₂-eq/kg and 6.63 m³/kg—while Small-B2 (guacamole production) showed much lower values at 0.72 kg CO₂-eq/kg and 1.38 m³/kg. The inclusion of further processing units for valorization generate a greater impact and need more water to operate, increasing environmental impact [31]. Furthermore, Piedrahita-Rodríguez showcased the importance of biorefinery location using an environmental analysis. Optimally, the facility should be close to both residue generation and product commercialization areas, as the latter have the highest emissions, around 60%–72% of the total CO₂ of the whole value chain. In rural areas, location strongly influences both fuel costs and environmental impact [32]. These studies highlight the limited attention given to the safety performance of these processes.

Table 1. Comparative findings on cascade and conventional biorefineries and safety assessment.

Biowaste	Cascade Biorefinery Approach	Inherent Safety Assessment	References
Spent coffee grounds	x		[6]
grape marc	x		[34]
Wheat bran	x		[35]
Banana rachis		x	[36]
Creole avocado		x	[33]
Avocado hard waste	x	x	This work

shows that cascade processing is still an emerging alternative to valorize biomass residues. Cascading systems have been used to process residues from coffee grounds, grape, and wheat. However, the application of safety assessments has been used to study more conventional biorefineries. According to the authors, very few studies exist where avocado biorefineries were analyzed. Some relevant studies on the safety performance on biorefineries include the work of Herrera-Rodríguez et al., who applied the Inherent Safety Index (ISI) to assess the risks of a biorefinery that extracts bio-oil, biopesticide, and chlorophyll from creole avocado. Their findings indicate an inherently safe process, with a score of 18. The main strengths, in terms of safety, were the process scale and the use of low to moderate operating conditions. However, the complexity level was low, as no chemical reactions were involved, with the primary safety concern being the use of solvents for metabolite extraction (methanol, ethanol, and acetone) [33].

The above discussion underscores the scarcity of information concerning safety issues in cascade biorefinery systems. While these configurations enhance the valorization of biomass by enabling the production of multiple valuable products (economic and technical), they also introduce heightened technological complexity due to the increased number of operations and substances involved. Therefore, for the first time, this study seeks to offer meaningful insights into the inherent risks associated with cascade biorefineries by identifying the key safety concerns and process vulnerabilities inherent to such systems. To this end, an inherent safety analysis is conducted. Data on the chemical properties of the compounds involved and the operational conditions are collected, and relevant chemical and process safety indicators are calculated to determine the overall inherent risk. The resulting scores are then analyzed and benchmarked against comparable processes to identify potential areas for improvement and critical bottlenecks.

2. Materials and Methods

2.1. Process Description

Figure 1 shows a block diagram of the proposed avocado hard waste biorefinery from the Amazon region. The scheme proposed is based on laboratory protocols and the literature available online. The biorefinery is divided into three main sections: (1) a pretreatment section, (2) a section for starch and starch-based bioplastic production, and (3) a section for polylactic acid (PLA) bioplastic production. The first section (blue) involves the reception and pretreatment of hard waste collected from the Amazon region. A hard waste flow of 27.4 kg/h enters this process; the amount selected was based on a fraction of the total avocado production of the region and a mass balance. Initially, the hard waste is washed with water to remove impurities and then dried in an oven at 50 °C using high-temperature air. Afterward, the hard waste (stream 4) is grounded in a crusher until it reaches a maximum size of 3 cm. The ground hard waste is then fractionated (streams 10 and 11) to be used as raw material and product in Sections 2 (red and green) and 3 (orange).

In Section 2, 60% of the ground hard waste is processed. It is mixed with a solution of water and 0.3% (v/v) sodium metabisulfite to extract the starch. This mixture is liquefied and filtered, separating a starch-rich liquid and a hard waste fraction rich in bioactive compounds. The starch-rich liquid is decanted to remove residual water containing traces of other compounds, including the solvent. The resulting solid starch fraction is then washed to eliminate the remaining impurities. Immediately after, the wet starch is dried at 60 °C using hot air. The dried starch (5.5 kg/h) is then divided into three fractions: pure starch (2.8 kg/h), starch for bioplastic production (2.4 kg/h), and a starch fraction as a film-forming agent (0.3 kg/h). The first fraction undergoes additional drying (60 °C) until it reaches a moisture content of less than 0.01%. The second fraction is gelatinized with cellulose and glycerol solutions at 5% and 4% (w/w), respectively, to form starch-based bioplastic. This film is then dried at 110 °C to remove residual water.

Finally, in Section 3, 40% of the ground hard waste is processed to obtain polylactic acid (PLA) bioplastic. The hard waste undergoes hydrolysis using a 5% diluted acid solution to extract fermentable substrates, which serve as raw materials for Lactobacillus fermentation to produce lactic acid. Additionally, a stream of pure calcium hydroxide (20 kg/h) is introduced to maintain the pH of the fermentation broth. After fermentation, the lactic acid suspension contains residual calcium hydroxide, which is neutralized using pure sulfuric acid (20 kg/h). The resulting salts are filtered out, and water is evaporated from the suspension. Immediately afterward, the lactic acid undergoes pre-polymerization into low-molecular-weight polylactic acid, which is then converted into lactides. These lactides are separated from water and residual lactic acid using flash separation and distillation. Light fractions are recirculated back to the pre-polymerization reactor, while purified lactides are polymerized via ring-opening polymerization to form polylactic acid. Finally, the obtained PLA is blended with a fraction of starch (0.3 kg/h) from Section 2 to produce a bioplastic film.

2.2. Inherent Safety Analysis

According to Heikkilä, the Total Inherent Safety Index ($I_{SI}$) can be calculated as the sum (1) of the Chemical Inherent Safety Index ($I_{CH}$) and the Process Inherent Safety Index ($I_{PI}$).

$$I_{ISI}=I_{CH}+I_{PI}$$

(1)

Moreover, these indicators are composed of several other sub-indicators that are based on the internal characteristics of the process, representing major factors that may affect inherent safety.

2.2.1. Inherent Chemical Subindex

The Chemical Inherent Safety Index contains chemical factors that affect the inherent safety of processes and is the sum (Equation (2)) of factors such as chemical reactivity, flammability, explosiveness, toxicity, and corrosivity of the chemicals involved in the process:

$$I_{CH}=I_{RM,max}+I_{RS,max}+I_{INT,max}+{\left(I_{FL}+I_{EX}+I_{TOX}\right)}_{max}+I_{COR,max}$$

(2)

where $I_{RM,max}$ and $I_{RS,max}$ are the index for chemical reactivity, $I_{INT,max}$ is the index for chemical interactions, ${\left(I_{FL}+I_{EX}+I_{TOX}\right)}_{max}$ is the index of hazardous substances, and $I_{COR,max}$ is the subindex of corrosivity. Chemical reactivity is calculated from the enthalpy of reaction ($∆H_f$) on the maximum values achieved for both the main (involved in the main product formation) and secondary reactions (not involved in the main product formation). These sub-indicators describe how exothermic a reaction is, with the score ranging from 0 for endothermic reactions ($∆H_f>0 j/g$) to 4 for extremely exothermic reactions ($∆H_f\le -3000 j/g$). Chemical interactions describe the unintentional reactions between the chemicals present in areas of the process outside the reactors; the score depends of the gravity of the potential interactions, such as fire (score 4), formation of flammable or toxic gas (score of 2 or 3), explosions (score 4), rapid polymerization (score of 2 or 3), release of heat (score 1 or 3), or whether there is none (score 0), among others. Flammability refers to the tendency of a substance to generate flame, determined by its flash point or boiling point (in °C), ranging from noninflammable, with a score of 0, to very inflammable, with a score of 4 (flash point lower than 0). Explosiveness describes a gas’ tendency to form an explosive mixture with air, measured by the difference between the upper and lower explosion limits (UEL-LEL %), with a score of 0 for a nonexplosive substance and up to 4 for a difference between 70% and 100% (very explosive). Toxicity refers to the potential health risks to humans, quantified using the threshold limit value (TLV) in ppm, which is the maximum concentration that workers can be safely exposed to in 8 h of worktime, the lower value being more dangerous, from 0 for a $TLV\ge \mathrm{10,000}$ to 6 for a $TLV\le 0.1$. Corrosivity indicates the risks associated with material degradation in equipment, determined by the material used for the equipment, with more resistant material having higher score, from 0 for carbon steel up to 2 for special materials (e.g., Hastelloy steel).

2.2.2. Inherent Process Subindex

The Process Inherent Safety Index expresses the safety of the process itself. It is the sum (Equation (3)) of the subindices for inventory, process temperature and pressure, equipment safety, and the safety of the process structure,

$$I_{PI}=I_I+I_{T,max}+I_{P,max}+I_{EQ,max}+I_{ST,max}$$

(3)

which includes the subindexes of the inventory ($I_I$), temperature ($I_{T,max}$), and pressure ($I_{T,max}$) subindices of the process, equipment safety ($I_{EQ,max}$), and safe process structure ($I_{ST,max}$). The inventory subindex indicates the risks associated with maintaining certain quantities of substances. This is performed for the internal or inside battery limit (ISBL) and the external or outside battery limit (OSBL), which are geographical areas that delineate the boundaries of specific processes. The ISBL encompasses the core processes that are critical for the transformation of raw materials into products, characterized by dense equipment and piping within a confined space. In contrast, the OSBL includes auxiliary systems such as storage and utilities, typically occupying more open areas. It is determined by the capacities of the equipment or flow over a given time period (tons per hour), with the ISBL being 0 in an inventory of less than 1 t/h up to 5 for 500 t/h or more; for the OSBL, it is 0 for 10 t/h or lower while it is 5 for 5000 t/h or more (requiring more than the ISBL). Temperature serves as an indicator of the thermal energy in the system and is calculated based on the highest temperature of the process, ranging from 1 for temperatures lower than 0 °C (subzero), 0 for a temperature between 0 °C and 70 °C, and up to 4 for temperatures higher than 600 °C. Pressure is used to determine the risks related to leakage rates in the case of loss due to potential energy, which affects containment in the equipment; for the scoring, it includes the vacuum pressure of 1 (between 0.5 and 0 bar), going to 0 for pressures between 0.5 and 5 bar, up to 4 for pressures higher than 200 bar. The equipment safety subindex measures the risk associated with the use of equipment. This is determined by both the ISBL and the OSBL, with values assigned to the equipment based on the risks they present, ranging from 0 for equipment handling nonflammable and nontoxic materials up to 4 (for furnaces, fired heaters) and 3 (for flares, boilers, and furnaces) for the ISBL and OSBL, respectively. The safe process structure subindex describes the risks of an operation from a system engineering perspective. It shows how well the system functions together (configuration and sequence), and it is the most complex indicator to score, as it does not depend on a physical or chemical property, but on past data (if available) of same or similar processes. Like safety audits and inspection reports, maintenance data, and accident reports, among others, the scoring ranges from 0 for recommended (processes for which there is extensive knowledge of their possible structural configurations, and the recommended configuration has been adopted) up to 5 major for accidents (arrangements known with certainty to have a high probability of minor and major accidents).

On the other hand, it is important to clarify that the calculations of the Inherent Safety Index ($I_{SI}$) are based on the worst-case scenario. This approach involves identifying the most hazardous conditions present in the process [37]. Consequently, a low index value ($I_{SI}$ lower than 24) represents an intrinsically safer process [24]. For the analysis, an inventory of data is collected for materials, equipment, and operative conditions, depending on the indicator data from various sources (available), including data extracted from simulations, online databases, safety reports, the relevant literature, etc. For example, commercial safety data sheets are used to determine the intrinsic properties of the substances involved in the process. After gathering the inventory, each indicator is scored based on the magnitude of the risk, with the highest score, i.e., the most hazardous, being taken when there are multiple risks for a single indicator (e.g., multiple elevated temperatures). In

Table 2. Summary of the inherent safety subindices and their score ranges.

ICH	Symbol	Variable	Score
Chemical reactivity (Main reaction)	$I_{RM,max}$	J/g	0–4
Chemical reactivity (secondary reaction)	$I_{RS,max}$	J/g	0–4
Chemical interaction	$I_{INT,max}$	-	0–4
Flammability	$I_{FL}$	℃	0–4
Explosiveness	$I_{EX}$	%	0–4
Toxicity	$I_{TOX}$	ppm	0–6
Corrosivity	$I_{COR,max}$	-	0–2
IPI	Symbol		Score
Inventory	$I_I$	t/h	0–5 (ISBL, OSBL)
Temperature	$I_{T,max}$	℃	0–4
Pressure	$I_{P,max}$	bar	0–4
Safe equipment	$I_{EQ,max}$	-	0–4 (ISBL); 0–3 (OSBL)
Process safe structure	$I_{ST,max}$	-	0–5

, the different scores considered for the subindices assessed in the method are shown. The range between low and high values represents the severity of the parameter measured or quantified. For example, higher process temperatures present more risks, leading to a higher score. A Supplementary Material is included containing the detailed scoring ranges for each sub indicator.

The ISI methodology estimates safety performance based on key hazard-related parameters. Although it does not conduct a fully probabilistic risk assessment, it incorporates several implicit assumptions to approximate potential risk in early design stages. These assumptions include the use of hazardous properties as proxies for risk, given that ISI relies on the intrinsic characteristics of substances—such as flammability, explosiveness, toxicity, and corrosiveness—to represent potential hazards. It assumes that more hazardous properties correspond to higher risk levels, without explicitly quantifying the probability of occurrence. Additionally, the methodology typically scores each hazard category based on the most hazardous substance present in the process, assuming that the presence of a single high-risk material defines the risk profile for that category. The total ISI is calculated as the sum of several partial indices (chemical and process-related), under the assumption that the contributions of each hazard are independent and can be linearly aggregated. The ISI also assumes that increases in temperature, pressure, or inventory lead to proportional increases in risk, without accounting for potential nonlinear escalation effects, such as those associated with runaway reactions or critical thresholds. Furthermore, the methodology focuses exclusively on inherent safety, meaning that it does not consider the presence or effectiveness of engineered safety systems, alarms, or procedural controls. As a result, the risk is assessed assuming no external mitigation. Finally, the ISI is based on steady-state operating conditions and does not take into account dynamic operational scenarios, such as start-up, shutdown, or transient disturbances, which may involve different risk profiles.

2.3. Process Performance in Comparison to Neutral Operation

Authors such as Meramo-Hurtado et al. proposed an equation to analyze the deviation between the global performance and the recommended minimum safety point of 24 [38]. This metric converts the value quantified for a process into a scale from 0 to 100%.

$$\%sfn=1-\left({\displaystyle \frac{sfn-s{f}_i}{sfn}}\right)\ast 100\%$$

(4)

where $\%sfn$ is the relation between the process actual safety performance $s{f}_i$ and the neutral safety point $sfn$. This indicator allows us to understand how far the process deviates from a neutral operation and to analyze the potential effects of different improvements or strategies aimed at achieving it.

3. Results and Discussion

3.1. Determination of Chemical Inherent Safety of the Process

To estimate the inherent safety of a biorefinery, the first step is to quantify the score of the main and secondary reaction indicators. These are determined by the heat of the reaction between the reactants and products of each reaction. In

Table 3. Reactions and heat of reactions of the biorefinery.

Reaction	Type of Reaction	$∆{\mathit{H}}_{\mathit{f}}$ [J/g]	Classification
Hydrolysis	Main	15,877.8	Endothermic
	C6H10O5 + H2O → C6H12O6
Fermentation	Main	−7921.9	Exothermic
	C6H12O6 → 2 C3H6O3
Neutralization	Secondary	13,276.4	Endothermic
	H2SO4 + Ca(OH)2 → CaSO4·2 H2O
Pre-polymerization	Main	−89,053.7	Exothermic
	8 C3H6O3 → 3.6 C6H8O4 + 0.4 C6H8O4 + 8 H2O
Polymerization	Main	−3297.6	Exothermic
	3.6 C6H8O4 + 0.4 C6H8O4 → (C3H4O2)n

, it is shown that, for the studied process, there are five reactions, all of which occur in the acid polylactic acid film production section. The reactions of hydrolysis, fermentation, pre-polymerization, and polymerization are categorized as main reactions, while the sulfuric acid neutralization reaction is categorized as secondary. The most exothermic (lowest) heat of each reaction was quantified for the pre-polymerization reaction with −89 kJ/g, which categorizes it as highly exothermic. According to this, for the main reaction safety subindex, a score of 4 was assigned. For the secondary reaction indicator, a score of 0 was given, as the sulfuric acid neutralization reaction is an endothermic reaction with heat above zero.

Chemical interaction subindex analyzes the chemical reactivity of the substances in the process, including water or air. For this indicator, CameoChemicals (3.1.0 rev 2) from the National Oceanic and Atmospheric Administration of the United States of America (NOAA) was used to identify chemical reactivity; the worst chemical interaction, among all the possible between all the compounds present, is pinpoint [39].

Table 4. Summary of main chemical interactions possible in a cascade biorefinery.

Compounds	Lactic Acid	Sulfuric Acid	Water	Calcium Hydroxide	Ethanol	Sodium Metabisulfite
Lactic acid	-	Explosive; flammable; generate heat and gas; intense or explosive reaction; toxic	Compatible	Generate heat and gas	Flammable; generate heat and gas; intense or explosive reaction	Generates heat and gas; toxic
Sulfuric acid	Explosive; flammable; generate heat and gas; intense or explosive reaction; toxic	-	Compatible		Explosive; flammable; generate heat and gas; intense or explosive reaction, toxic	Generates heat and gas; intense or explosive reaction
Water	Compatible	Generate heat and gas; corrosive; toxic	-		Compatible	Compatible
Calcium hydroxide	Generate heat and gas	Generate gas; intense or explosive reaction		-	Compatible	Compatible
Ethanol	Flammable; generate heat and gas; intense or explosive reaction	Explosive; flammable; generates heat and gas; intense or explosive reaction; toxic	Compatible		-	Compatible
Sodium metabisulfite	Generate heat and gas; toxic	Generates heat and gas; intense or explosive reaction	Compatible	Compatible	Compatible	-

shows a graph of the different possible chemical interactions between the most dangerous substances in the process. It was observed that the worst chemical interaction possible is the potential for fire and explosion between sulfuric acid and calcium hydroxide, ethanol, or sodium metabisulfite. As both fire and explosion have the same score of 4, for the subindex for a cascade biorefinery, a score of 4 is assigned. It is important to clarify that, of these possible interactions, two of them occur frequently: sulfuric acid with lactic acid and calcium hydroxide.

The indicators of toxicity, explosiveness, and flammability were scored through an analysis of the properties of hazardous substances. Several datasheets were reviewed from databases such as CameoChemicals from NOAA, INCHEM from the World Health Organization (WHO), and the National Institute for Occupational Safety and Health (NIOSH).

Table 5. Properties of the substances in the process and their assigned score.

Substance	TLV [ppm]	Score	$\mathit{U}\mathit{E}\mathit{L}\mathbf{\%}-\mathit{L}\mathit{E}\mathit{L}\mathbf{\%}$	Score	Flash Point [°C]	Score
Sodium metabisulfite	0.59	6	-	0	-	0
Ethanol	1000	2	24.6%	2	12	3
Sulfuric acid	0.05	6	-	0	-	0
Calcium hydroxide	1.51	4	-	0	-	0
Hydrated calcium sulfate	1.30	4	-	0	-	0
Lactic acid	-	0	-	0	110	1

presents a selection of high-risk substances and their intrinsic properties. It is noted that ethanol is the most dangerous substance compared to sulfuric acid and sodium hydroxide, with a combined score of 7. This score is due to the risks posed by ethanol in terms of explosiveness and flammability, properties with higher values than the other substances, as they do not present the same problem. In contrast, regarding toxicity, ethanol is the safest, with a TLV of 1000, while sodium metabisulfite, sulfuric acid, and calcium hydroxide have TLVs around 1 (resulting in a score of 6).

Compatibility tables were used to determine the materials used for the different substances present in the process (available online website of manufacturers were accessed like Alleima and Cole-Parmer). In

Table 6. Recommended materials for key compounds. The table showcase a potential challenge of the process substances on the material selected.

Substance	Material
Lactic acid (other organic acids)	Stainless steel, special materials
Sulfuric acid (c < 10%)	Stainless steel, special materials
Sodium hydroxide	Carbon steel, stainless steel, special materials
Ethanol	Carbon steel, stainless steel, special materials
Sodium metabisulfite	Stainless steel
Water	Carbon steel, stainless steel, special materials

, selected substances are shown for their corrosive capacity and recommended materials. The most complex compound is sulfuric acid when selecting materials, as its corrosive effect varies depending on the conditions that it is used (like concentration, temperature, and the presence of moisture) [40]. Fortunately, for the ISBL of the process, it is only present in a diluted concentration (1% p/p), and stainless steel is sufficient to withstand the corrosion of the acidic solution. However, temperature could be an issue in the starch hydrolysis stage (higher temperature induce more corrosion), requiring more resistant material [41]. In the case of other substances, such as calcium hydroxide and lactic acid, while they are corrosive to less resistant materials like carbon steel, they are more flexible because there are no issues with temperature fluctuations. Additionally, for organic compounds like ethanol, any material can be chosen. Therefore, the corrosion subindex score is 1, with stainless steel as the selected material. However, the use of only one material is fixed, as several materials can be used for different operations due to other less technical requirements. Additionally, other more special material could be used for other reasons, for example, based on the required product quality conditions (such as transparency, color, among others).

In Figure 2, the chemical safety sub-indicators of the avocado hard waste biorefinery from the Amazon are shown. A hazardous substance sub-indicator of 7 is observed, with ethanol being the source of this score. On the other hand, the hazardous reaction sub-indicator scored 4, as there are strong exothermic reactions to produce polylactic acid.

Similarly, the chemical interaction sub-indicator reached a score of 4, as there is a possibility of risky interactions during the same stage due to sulfuric acid and sodium hydroxide. The corrosion sub-indicator scored 1 due to the presence of various corrosive substances and thermal conditions that may exacerbate them. Finally, the secondary reaction sub-indicator reached a score of 0, as the only neutralization reaction is endothermic.

3.2. Determination of the Process Inherent Safety

To evaluate the inherent safety sub-indicators of the process, information was gathered about the technical aspects of the process such as flows, temperature, pressure, type of equipment, and process structure.

Table 7. Summary of the process equipment and operating conditions for each section. The process equipment showcase low to moderate operating conditions.

Section	Equipment	Mass Flow [t/y]	Temperature [Celsius]	Pressure [Bar]
Hard waste pre-treatment	Wash	929.8	25	1.01
	Dryer	6297.2	50	1.01
	Grinder	218.4	18.5	1.01
Starch production	Extraction	498.9	25	1.01
	Grinder 2	498.9	25	1.01
	Filter	498.9	25	1.01
	Decanter	103.3	25	1.01
	Starch wash	306.5	25	1.01
	Dryer	6093.1	60	1.01
	Dryer 2	6046.9	60	1.01
Starch biofilm production	Gelatinization	137.8	27	1.01
	Heating	137.8	70	1.01
	Dryer	6077.8	110	1.01
Polylactic acid biofilm	Hydrolysis	942.7	121	1.37
	Fermentation	1102.2	37	1
	Exchanger 1	1102.2	70	1
	Neutralization	1264.6	25	1
	Filter	1264.6	25	1
	Flash 1	875.4	100	0.5
	Mixer 1	51.5	76	1
	Exchanger 2	51.5	160	1
	Pre-polymerization	51.5	240	1
	Flash 2	51.5	100	0.5
	Distillation	30.2	253	1
	Polymerization	25.1	200	2
	Mixer 2	27.2	180	1
	Flash 3	21.3	28	1
	Mixer 3	9.8	28	1

presents the operational information of the process, considering the different sections of the process. For the temperature and pressure sub-indicators, the maximum values for both variables were identified. The maximum temperature was recorded in the polylactic acid biofilm production section, within the distillation unit, with a value of 254 °C, for which a score of 2 was given (moderate according to Heikkila’s work). For pressure, the maximum found was 10 bars, with a score of 1 assigned (low according to Heikkila’s work).

For the inventory sub-indicator, only the internal battery of the process (ISBL) was considered, as OSBL is not properly defined (it cannot be, as that area of the process is performed in more advance design stages, like the detailed engineering phase). The units analyzed in the process belong to this category. Additionally, for this subindex, the mass contained in any process equipment (tanks, reactors, mixers, and others) was measured during a hydraulic retention time of 1 h. Using the mass flow information from Table 6, the process inventory was calculated to be 10.7 tons per hour, placing it in the lowest range, with a score of 1. For the equipment safety sub-indicator, the type of equipment used for the different operations performed must be identified. For the studied cascade biorefinery, reactors, pumps, heat exchangers, towers, and separation drums are present. Among these, reactors represent the highest risk, especially the one where acid hydrolysis of starch occurs and the neutralization reactors, both of which are highly hazardous due to the presence of sulfuric acid and calcium hydroxide. Consequently, a score of 3 is assigned.

The process safety structure subindex is the most complex subindex to determine. The scoring is based on engineering knowledge about units, systems, and how they operate together in terms of safety. The index quantification relies on data based on experience, such as legal and industry standards, design recommendations, and accident reports. For the process studied, there is a significant lack of information about cascade biorefinery systems. The wide range of processes for valorizing biomass and its residues complicates definitive assessment, as the heterogeneous composition of feedstocks results in numerous alternatives. Furthermore, the novelty of these emerging schemes makes it difficult to evaluate their safety in the absence of reliable on-site data. This represents a relevant limitation in the analysis of such technologies. However, given that the processes remain at the design stage, this lack of data is a recognized bottleneck. Nonetheless, important highlights exist, as biofuel commercialization in the last decade has generated a significant amount of data regarding accidents (property damage, injury, and even deaths) in biofuels facilities (bioethanol and biodiesel even biogas) [42]. The most significant risks are related to potential fire and explosion due to the nature of the chemicals within the process. Nonetheless, studies confirm that biotechnological processes (emerging technologies) intrinsically have a lower risk factor than the traditional chemical plants [43].

Overall risks in these facilities are conditioned by plant characteristics such as size, location, capacity, production technology; plant lifetime—including plant life stage—and operating time, as well as substances involved [44]. The records also showcase that small-scale facilities have presented more accidents, as it is normally assumed that it is safer to operate these processes due to the small quantity of hazardous chemicals to be handled. Nevertheless, the same underlying causes found in major full-scale plant accidents affect small-scale processes [45]. Danzi found that tanks are by far the most affected (at least in biodiesel plants) and, surprisingly, process units (ISBL) are rarely primarily involved in accidents [44]. Nonetheless, units, systems, or equipment where conditions can provoke a mixture of volatile compounds, glycerol, and sulfuric acid, or the present of violent exothermic reaction due to excess amount of acid in the neutralization step or bad mixing operations in batch processing, particularly at small scales, are inherently more unsafe [44,46]. This information pinpoints the PLA process section, as it uses hazardous compounds for pretreatment and neutralization (calcium hydroxide and sulfuric acid), having batch reactors with exothermic reactions such as neutralization. At the same time, the biorefinery of avocado waste barely uses light organic compounds, at least in the PLA section and with conditions that propitiate gas mixtures. Furthermore, the overall conditions of the process are mild and, combined with a small inventory, make the process safer. Still, the lack of information on avocado biorefineries is prevalent, so for a cautious estimate, a score of 2 is assigned.

Figure 3 compiles the scores obtained for the process safety sub-indicators of the cascade biorefinery of avocado hard waste from the Amazon region. The equipment safety subindex was the highest, with a score of 3, followed by the temperature and safe structure sub-indicators, both with a score of 2.

The lowest scores were achieved for the inventory and pressure sub-indicators, with scores of 1. This indicates that, in terms of process safety, biorefineries present risks mainly due to unsafe equipment, while the small inventory and low to moderate operational conditions do not represent significant risks.

3.3. Inherent Safety Performance of the Cascade Biorefinery

Figure 4 showcases the inherent safety performance of the cascade biorefinery. The cascade biorefinery of avocado hard waste obtained an ISI of 25, classifying the process as unsafe. The chemical aspects contributed the most, with a score of 16, accounting for about 60% of the total value. In contrast, the process safety indicators only reached a score of 9, which is approximately 40% of the total. This proportion indicates that reducing the inherent chemical risks should be prioritized, if the goal is to improve the safety of the process before implementation. Equation (4) allows us to analyze how far the process is from a neutral process, with the process ISI being 25, resulting in 96% performance compared to the neutral operation, confirming that, although the process is above 24, it still relatively safe. Moreover, this performance suggests that any optimization efforts will help bring the process into the neutrality point, reducing its impact on its overall safety performance.

In Figure 5, the comparison between the inherent safety performance of the studied cascade biorefinery and other bioprocesses is shown. This comparison is particularly important due to the lack of safety assessments on cascade biorefineries. The bioprocesses selected can point out relevant information about the safety of factors, such as the scale of the process, the complexity of operations, and the range of products offered, which are crucial for implementing productive projects in avocado-producing regions like the Amazon region [30]. The proposed biorefinery has a high complexity and small scale, with products such as bioplastics, requiring significant biocomponent conversion processes and separation techniques, reflected in a higher inherent risk compared to bioactive extraction processes, like those studied by Herrera and collaborators, with scores of 17 and 18 [33,47]. The reactions substantially increase the risk, from an $I_{ch}$ of 8 to 16. Reactions for transforming polysaccharide compounds, such as fermentation and hydrolysis, are scored as risky, as they have high heat of reaction. This is observed in processes for obtaining bioethanol [48] and modified chitosan microparticles [38], which presented main reactions scores of 4.

Additionally, the use of unsafe chemical agents in bioprocesses opens the possibility for adopting other technologies to mitigate risks. For example, the chemical hydrolysis (acid or base) can potentially be replaced with enzyme-based methods [49]. This change could reduce the total inherent safety score by at least two points, minimizing the use of substances like sulfuric acid and calcium hydroxide. However, this switch of technologies would involve other types of risks that should be analyzed, as well as techno-economic issues. As shown by Singh et al., the enzymatic has a lower capital and operative expenditure than the acid hydrolysis, but the method has low efficiency and a time-consuming process; at the same time, it requires less material and energy intake, and presents less severe conditions [50]. Nonetheless, the biorefinery presents a performance similar to high-complexity systems, as mentioned earlier, ranking above bioethanol production (21) and below chitosan microparticle production (28), despite lower processing capacity. Additionally, the use of solvents such as hexane, methanol, and ethanol represents a risk, with safety scores for these substances ranging between 6 and 8, particularly due to properties like flammability and explosiveness. It is recommended to replace conventional extraction methods with alternatives that do not involve solvents, such as eutectic solvents (DES and NADES), supercritical fluids, and enzymatic methods, among others [51,52]. These new methods have advantages and drawbacks, as they present, at least in the literature, higher yields with lower extraction temperatures, faster kinetics, solvent recycling, and operative costs; however, as they are only used in lab scales, the high viscosity of solvents and the difficult removal of extract from the solvent are bottlenecks for these technologies [52]. On the other hand, the inventory of a biorefinery can become a potential risk factor, as it is closely linked to the scale of the process. The scale of biorefineries is a multifaceted aspect, influenced by factors such as feedstock availability, technology readiness level, commercial logistics, and local conditions. In the case of a small-scale process, such as the cascade biorefinery, a relatively low inventory was observed—an annual capacity of approximately 220 tons per year (corresponding to an inventory of 11 t/h)—which contributed positively to its safety performance, reflected in a low Inherent Safety Index (ISI) score of 1. Conversely, a larger-scale process, like chitosan microbead production, with an annual capacity of 2033 tons and an inventory of 159 t/h, exhibited a significantly higher risk, with an ISI score of 3. This indicates a fourteen-fold increase in inventory compared to the cascade system. The comparison highlights the influence of scale on process safety: larger-scale operations require greater quantities of materials (including hazardous solvents), more complex equipment, more frequent exposure to intense operating conditions, and, consequently, a higher likelihood and severity of potential failures. As an illustrative example, increasing the inventory from 10 t/h to 15 t/h could raise the ISI score by one additional point. However, the process scheme and potential changes must be rigorously studied in terms of technical and economic feasibility, ensuring acceptable product yields and monetary benefits (which are not part of the scope of this paper). The safety criteria alone is not enough to not implement these types of bioprocesses; however, at the design stage, it can pinpoint opportunities to explore under a holistic approach.

The process structure is a complex index to quantify, as required, engineering historical process data (including reports from accidents). The multiproduct biorefineries are still emerging technologies that are not yet in the full commercial stage, especially in cascading systems (which are very dependent on hierarchy, sequencing, and integration) [53]. For the hard waste cascade avocado biorefinery, we can highlight issues with high-temperature equipment being close to equipment handling gases and volatile substances, alongside reactors with exothermic reactions and batch processing, particularly in the PLA section. The level of complexity of a biorefinery depends on factors such as the commercial readiness of the technology deployed [54]. In contrast, bioethanol production achieves a score of 1, being a sound engineering practice, as multiple facilities worldwide exist in several capacities (including large scale). In contrast to the latter, chitosan-microbead- and creole-extractive-based biorefineries presented a score above 2, as safety issues were found (especially related to heat and fire), but no relevant information on structural risks was available. Remarkably, avocado oil production, a well-established process, had a score of 3. These scores showcase the underlying necessity for the establishment of databases of these new emerging technologies, as well as more detailed safety assessments.

4. Conclusions

Biorefineries are emerging as a key technology for the socio-economic development of countries like Colombia. They have demonstrated significant potential due to the wide range of valuable bioproducts they can generate from agricultural waste. However, it is necessary to conduct an in-depth analysis of them in regard to sustainability criteria, especially focusing on inherent safety. For this study, an inherent safety analysis was applied to identify and assess the risks of an avocado hard waste cascade biorefinery in the Amazon region. A global safety score of 25 was obtained, with a performance of 96% compared to the neutral operating point, meaning that the process is slightly unsafe. The chemical aspects represented the most significant risk of the process, with a score of 16, with exothermic reactions, hazardous substances, and chemical interactions being the primary risk factors. On the other hand, the process safety indicator scored 9, indicating that these aspects do not present significant risks, with equipment being the only major risk. However, compared to other bioprocesses, the cascade biorefinery is a relatively safe option for the valorization of avocado hard waste. Furthermore, a significant challenge is present, as a lack of data on systems and structures has been identified. Finally, alternatives can be proposed, such as replacing acid hydrolysis with enzymatic hydrolysis, along with alternative methods for extracting bioactive compounds, which are recommended to reduce inherent risks, but need to be assessed amongst other factors.