A Review of Energy-Efficient Technologies and Decarbonating Solutions for Process Heat in the Food Industry

Abstract

Heat is involved in many processes in the food industry: drying, dissolving, centrifugation, extraction, cleaning, washing, and cooling. Heat generation encompasses nearly all processes. This review first presents two representative case studies in order to identify which processes rely on the major energy consumption and greenhouse gas (GHG) emissions. Energy-saving and decarbonating potential solutions are explored through a thorough review of technologies employed in refrigeration, heat generation, waste heat recovery, and thermal energy storage. Information from industrial plants is collected to show their performance under real conditions. The replacement of high-GWP (global warming potential) refrigerants by natural fluids in the refrigeration sector acts to lower GHG emissions. Being the greatest consumers, the heat generation technologies are compared using the levelized cost of heat (LCOH). This analysis shows that absorption heat transformers and high-temperature heat pumps are the most interesting technologies from the economic and decarbonation points of view, while waste heat recovery technologies present the shortest payback periods. In all sectors, energy efficiency improvements on components, storage technologies, polygeneration systems, the concept of smart industry, and the penetration of renewable energy sources appear as valuable pathways.

1. Introduction

The conclusions from the 6th Assessment Report from the IPCC have highlighted the most probable incapability of satisfying the Paris agreements signed in 2015 at the 21st Conference of Parties (COP 21) [1]. Whatever scenario is followed, the global temperature increase will exceed +1.5 °C in 2030 compared to preindustrial levels. Projections also show that remaining below +2.0 °C will only be possible if the emission reduction is further accelerated in all sectors to reach net-zero CO₂ and net-zero GHG emissions in the early 2070s. Among all, the industrial sector is responsible for over 30% of global CO₂ emissions and over 38% of global energy demand, according to the IEA [2,3], while in Europe, industrial energy consumption accounts for 26% of global energy demand [4,5]. Focusing on activities, the entire food chain also accounts for 30% of the EU’s CO₂ emissions (690 MtCO₂ per year) and 17% of the EU’s gross energy consumption (283 Mtoe), considering agriculture, processes and manufacturing, packaging, logistics, and transport, use, and end-of-life. The embedded energy in the food includes direct (agriculture motorization, process heating, electricity, etc.) and indirect contributions (fertilizer production, operation of irrigation systems, etc.). The energy consumed for processes and manufacturing is 62% consumed as heat and 38% as electricity; this latter is used in a large majority for cooling purposes [6]. To meet 2050s NZ (net-zero) scenarios, industries need to decrease their GHG emissions by 90%, while the food industry should decrease them between 80 and 95% [7]. Reaching those levels of emission reduction will require a PEST (political, economic, socio-cultural, and technological) analysis. Indeed, if technological improvements will not allow to cover the entire emissions reduction, its role will be crucial to reduce dependency on fossil fuels. Therefore, providing more efficient and low-carbon emissions solutions for industries is key for the road towards net-zero CO₂ and net-zero GHG, but it does not constitute a stand-alone solution as its impact without sobriety will not be sufficient.

In parallel with environmental and climatic considerations, the actual energy crisis also highlights the crucial role of energy efficiency in economic and social aspects. Li et al. [8] showed the importance of energy efficiency characterization with an in-depth analysis of efficiency in intensive industries and help direct sustainable developments. Regarding competitive aspects for the food processing industry in France, from 2015 to 2019, the average energy bill was about 2.35 billion € per year, which was largely dominated by electricity and gas supply. In 2022, this expense was expected to grow up to 5.6 billion € due to the energy crisis, and the price is expected to increase by +124% and +170%, respectively, for electricity and gas [9]. In the UK, according to a large survey in 2022 [10], the energy price increase was the main concern for nearly three out of five food and beverage companies, showing the crucial importance of energy efficiency measures to reduce financial pressure on operating actors in the sector. Figure 1 was built using data from the Eurostat database [11]. The prices of gas and electricity were averaged on a yearly basis depending on the amount of energy consumed by non-household users. It shows the dramatic increase of prices in 2021 due to the energy crisis and also points out the indexation of electricity on gas prices.

In terms of CO₂ emissions, a growing concern for industrials is related to the operational tax system evolution. Giving economic value to the CO₂ emissions will push industrials to address their emissions with greater incentives than only their brand image. This is achieved either by (i) directly taxing the emissions or (ii) organizing a financial market where CO₂ emissions quotas can be traded. This latter option, named Emission Trading System in Europe (EU-ETS), was put in place in 2005 and is currently going through its 4th phase (2021–2030) with an important carbon price increase that happened in 2021 and 2023 where it successively increased from 25 to 50 €/ton and from 50 to 100 €/ton [12]. It is expected to further increase from 100 to 167 €/ton in 2030 [13]. The downsides of this latter measure are the limited coverage of only 20% of all industrial sectors and its impact, which is suspected of enhancing energy inflation. For example, today’s agri-food industry is not concerned by the EU-ETS system. However, it is expected that complementary measures, like the recent EU’s Carbon Border Tax, will appear in the upcoming years with significant financial incentives that were put in place to push industries to drastically reduce their carbon emissions, either through sobriety, energy efficiency or the implementation of novel technologies. In 2016, Fluch et al. analyzed that most funding instruments were difficult to obtain because of complex procedures and limited budgets for programs or individual measures [14]. Today, Europe is engaged in a 2050 long-term strategy aiming at an economy with net-zero greenhouse gas emissions, consistent with Member States’ integrated National Energy and Climate Plans (NECPs). Six hundred million € are planned to be invested in the European Green Deal to finance environmental and ecological transition [15].

Regarding energy efficiency, the food industry is an interesting case study. Its processes are dragging a lot of attention because of their important energy consumption, reasonably low operating temperatures compared to other industry sectors, and potential improvement thanks to upcoming energy-efficient systems. In 2021, the total final energy demand in the industry was 2791 TWh [16] (Figure 2). The food and beverage industry consumed 11.6% of Europe’s final energy demand, which is among the top five energy-intensive industries with chemicals and petrochemicals (21.5%), non-metallic mineral (14.1%), paper pulp and printing (13.6%), and iron and steel (10.2%). If the quantity of energy is important, the process temperatures are relatively low, with 54.0% below 100 °C and 86.5% below 200 °C [17] (Figure 3). This makes it more easily accessible for emerging technologies compared to other sectors like chemicals or steel industries, where the majority of the issues are addressed above 500 °C. Finally, it has also been highlighted that current processes used in the food industry are relatively inefficient [18,19] and mainly rely on fossil fuels [5], leaving room for significant improvements in energy efficiency measures. Historically, within the whole farm-to-fork chain, energy for processes represents about 3% of total costs, leading to few incentives for metering, identifying inefficiency sources, and improving energy management. Today’s energy price increase is shifting the tipping point towards greater importance attributed to energy consumption.

Ricardo Energy [6] provided an insightful study illustrating the strong opportunities for reducing carbon emissions in the food and beverage industry, mainly through four factors: (i) accessibility in terms of processes operating temperatures, (ii) win/win situation coupling the economic and environmental interest of energy efficiency and carbon footprint improvements, (iii) important policy push and subsidies to come, and (iv) an increasing market pull from final food chain end-users and therefore securing market positions [20]. Nevertheless, according to Clairland et al. [21] and Acampora [22], the industry faces different challenges and barriers to retaining the development and spread of efficient and less carbon-intensive solutions on the market. Figure 4 presents an illustration of major drawbacks and barriers that do not allow a satisfying dynamic. They can be classified into five categories: regulatory, technological, economical, sociological, and environmental. The most numerous are in technological and economic aspects, but they can all be sufficiently restrictive to slow down or derail a project’s development.

The sociotechnical system of food and beverage is very large, meaning that the potential of decarbonation and energy efficiency can be multi-sourced. The solutions regarding process heat in the agri-food industry have never been collected and assessed in a review article. Global warming and global and European variations in energy costs are leading to changes in energy systems in the food industry. Energy-efficient technologies in food processing from all countries are presented in this bibliographical review article. The aspects of agriculture, retail and distribution, market, and end-users are not addressed. The perimeter of the study will first be defined to identify the areas with the highest energy consumption and GHG emissions. For that purpose, two case studies in high-consuming sectors will be detailed. Finally, an exhaustive review of applicable energy-efficient technologies and decarbonating solutions in food processing is performed. The main sectors addressed are (1) refrigeration, (2) heat generation, (3) waste heat recovery, and (4) thermal energy storage.

2. Perimeter Definition: Food Processes, Activities, and Temperature Levels

In terms of energy consumption, a report from FAO in 2013 [23] shows the large domination of cooking and of very diversified processing activities. This article separates them into six main categories: (i) materials reception and preparation, (ii) product processing technology, (iii) utility processes, (iv) separation techniques, (v) heat processing, and (vi) heat removal. To highlight which processes draw major attention in the research community, a count of articles depending on keywords was made. A combination of the word “food” and the process name was searched in the database of the Web of Science Core Collection from 1985 to the present [24]. This collection covers the world’s leading scholarly journals, books, and proceedings in the sciences, social sciences, and arts and humanities. The occurrences were counted, and a word cloud was built with a police size depending on the number of search results. The results are summarized in Figure 5. Thermal processes and, more specifically, “drying” appear to be the focus of the majority of research projects, as thermal processes represented 63.1% of the occurrences.

Cooling and heating applications also cover a wide range of temperatures from approximately −50 °C to 250 °C, with some specific applications reaching 300 °C. Figure 6 presents a general repartition of temperatures for the thermal processes listed in Figure 5. The range from −50 to 300 °C was established from eight references from the literature [24,25,26,27,28,29,30,31]. Two specific processes are analyzed: (i) sterilized milk powder (SMP) and (ii) deep-frozen fried potatoes, respectively chosen due to their energy intensity and extreme temperatures relative to the food sector. The dairy industry is one of the most representative and adapted sectors in the food industry for decarbonating potential. It contains many different processes, a wide selection of temperatures, and heating and cooling demands. Products such as sterilized milk powder, cheese, and other dairy products are part of the most mass-specific energy-intensive products [MJ/kg], except roasted coffee. The variety of processes (pasteurization, evaporation, drying) made sterilized milk powder a very interesting case study. In modern societies, a massive production is the one of deep-frozen fried potatoes. It was chosen as a second case study because it demands high amounts of heating and cooling energy.

2.1. Dairy Application: Sterilized Milk Powder

One of the most energy-intensive products within the agri-food industry is the production of sterilized milk powder (SMP) [32]. SMP production is subject to strict regulatory processes and temperatures. From the milk reception to the powder storage, the organic matter is subject to the steps illustrated in Figure 7 [33] and further described in Table 1.

The analysis of this process has drawn a lot of research to improve historical procedures [34,35,36] in which all steps and energy provided were considered individually. Uçal et al. [37] analyzed a whole milk powder (WMP) production line principally based on natural gas for heating purposes and showed that pasteurization, evaporation, and drying represented 98.3% of the energy consumption and 95.7% of the CO₂ emissions—for a total of 10.2 MJ/kg_WMP and 0.588 kg_CO2/kg_WMP. Similar results are found in Ramirez et al. with 11.1 MJ/kg_WMP [32]. This shows the importance of focusing on energy efficiency and decarbonating solutions in those three steps. According to Walmsley et al. [38], current state-of-the-art solutions would allow the reduction of the energy intensity to 5.2 MJ/kg_WMP and 0.14 kg_CO2/kg_WMP, including different improvements from the process. The analysis has been applied to different plants, where energy utilities varied depending on the available resources, with geothermal, mechanical vapor recompression, heat pumps, and biomass boilers, among others.

From the process point of view, Moejes et al. [34] present different scenarios in which milk powder production can be reduced from 10.2 MJ/kg_WMP down to 5.4 MJ/kg_WMP if reverse osmosis, membrane distillation, and monodispersed drying with membrane contractor are used instead of conventional pre-heating, evaporation, and spray drying processes. This example shows that thermal processes are not always the best options, and other non-thermal processes need to be considered for future process lines. A more common process modification applies a pinch analysis methodology to maximize the potential heat recovery intra-process. Walmsley et al. [38] show that process heat recovery from evaporation and drying can cover up to 2.8 MJ/kg_WMP.

From the energy utility point of view, high-temperature heat pumps and solar thermal technologies present great potential. Solar thermal implementation has been studied by Camci [35] and Sobulska et al. [36], showing that the production of superheated steam with solar collectors leads to energy efficiency gains of 75% and 70% on the drying process, and Shah [39], showing a reduction of 60% of conventional boiler duties with solar thermal collectors’ integration at the process level in a dairy plant in Dubai. Regarding heat pumps, Bühler et al. [40] showed a potential energy savings of 65% for central heat pump systems and 56% for decentralized heat pumping systems, with a carbon emissions reduction impact of around 80% in the frame of Denmark’s electricity mix.

Table 1. Process description, temperature, and energy share for sterilized milk powder production [32,37,40].

Process	Description	Product Temperature [°C]	Energy Consumption Total: 10.2 MJ/kg
Chilling	Milk cooled down for optional storage before the process	4	-
Thermization	For 15 s: allows to postpone pasteurization treatment for hours/days	63–65	0.04 [0.4%]
Separation	Hot centrifugal separation to separate skimmed milk from cream	50–52	0.08 [0.8%]
Pasteurization	Continuous pasteurization for 15–30 s	72	2.1 [20.5%]
Chilling	If evaporation and pasteurization are not achieved inline	-	-
Evaporation/Concentration	Three-stage evaporation unit between 40 and 70 °C to increase solid content	70	2.7 [26.8%]
Homogenization	Decrease fat globule size by two-stage at different pressures for flavor homogenization	-	0.04 [0.4%]
Drying	Spray drying by atomization of concentrated milk at the top of the drying tower	120–200	5.2 [51%]

Table 1. Process description, temperature, and energy share for sterilized milk powder production [32,37,40].

Process	Description	Product Temperature [°C]	Energy Consumption Total: 10.2 MJ/kg
Chilling	Milk cooled down for optional storage before the process	4	-
Thermization	For 15 s: allows to postpone pasteurization treatment for hours/days	63–65	0.04 [0.4%]
Separation	Hot centrifugal separation to separate skimmed milk from cream	50–52	0.08 [0.8%]
Pasteurization	Continuous pasteurization for 15–30 s	72	2.1 [20.5%]
Chilling	If evaporation and pasteurization are not achieved inline	-	-
Evaporation/Concentration	Three-stage evaporation unit between 40 and 70 °C to increase solid content	70	2.7 [26.8%]
Homogenization	Decrease fat globule size by two-stage at different pressures for flavor homogenization	-	0.04 [0.4%]
Drying	Spray drying by atomization of concentrated milk at the top of the drying tower	120–200	5.2 [51%]

2.2. Frozen Catering: Frozen Potato Manufacture

Analyzing frozen manufacturing is also key to the market trend as it is foreseen to observe a cumulated annual growth rate (CAGR) of 8.37% volume increase for the frozen food market between 2022 and 2030 [41]. Among all frozen fruits and vegetable products, fried potatoes have the highest mass-specific energy intensity, around 3.3 MJ/kg, according to West et al. [42]. The analysis of the process is illustrated in Figure 8 and further described in

Table 2. Energy carrier, temperature, and energy share of different processes for frozen fried potatoes [42,43].

Process	Description	Product Temperature [°C]	Energy Consumption Total
			3.3 MJ/kg [42]	3.0 MJ/kg [43]
Washing	Removal of all dirt, stones, and other (non) organic elements issued from crops	Ambient	<1%	14.1%
Peeling	Usually achieved by steam peeling, where the skin is removed after short exposure to high-temperature steam and light mechanical pressure to remove the skin. Can also be made mechanically, which is less precise but much less energy-intensive (1/10)	180–200 °C	15%	<1%
Blanching	Exposure to hot water to reduce bacterial and enzymatic activity	60–90 °C	10%	12.7%
Frying	Potatoes are cooked in high-temperature oil. It can be performed multiple times or during variable times	160–190 °C	65%	25.7%
Cooling	This step is performed quickly to stop the cooking process but does not represent an important energy step	N.A.	2%	<1%
Freezing	Most of the refrigeration energy produced is used during the freezing phase, where sliced and fried potatoes are exposed to cold air	−15 to −40 °C	8%	46.3%

. Two sources have been selected as the literature presents different results for the same product, showing, respectively, 0.26 MJ/kg and 1.39 MJ/kg for the freezing process and 2.14 MJ/kg and 0.71 MJ/kg for the frying process.

From the process point of view, Masanet et al. [43] highlight the importance of removing surface water on potatoes before frying and freezing processes, ideally with vibratory systems. It also considers the important potential of waste heat recovery on combustion flue gases and on waste frying oil. Still considering heat recovery, Van Loon et al. [44] showed that integrating waste heat recovery after drying or blanching processes for frying oil pre-heating could contribute to reducing up to 23% of total energy consumption. Other research from the same group [45] focused on the entire process with temperature modifications, superheated steam, and vacuum cooling techniques but ended up with a deterioration of the product quality.

From the energy utility point of view, a specific focus is placed on reusing waste oil as fuel in biofuel boilers [42] as it provides energy efficiency and waste treatment solutions.

The food industry is actively seeking energy efficiency and decarbonating solutions. As seen previously, this can be achieved through processes or measures taken by energy utilities. Both are complementary and need to be addressed. However, including food quality, energy efficiency, decarbonation, economic, and technical feasibility is a complex task for each end-product and even more complicated when an industrial facility combines different products, fluxes, temperatures, and heating/cooling demands.

3. Energy Efficiency and Decarbonation Potential

In regards to the technological constraints listed in Figure 4 for energy efficiency and decarbonating solutions in food industry processes, this section presents an extensive review of existing and upcoming solutions for refrigeration, heat generation, heat removal, and thermal storage. The solutions presented are between research and ready-to-market stages.

3.1. Refrigeration

From a technological point of view, refrigeration is mainly produced by electricity-driven thermodynamic cycles using mechanical vapor compression. The most famous ones are the Carnot cycle, Linde cycle, Claude cycle (cryogenics), reverse Joule–Brayton cycle, and reverse Stirling cycle. All use different types of refrigerants that run temperature and pressure modifications to generate cooling duties and extract them through heat exchangers. If acoustic, magnetic, Peltier refrigeration and passive solar radiation can also provide cooling technologies, their cost, efficiency, maturity, and reliability are not sufficiently attractive for industrial implementation and are not discussed in this section. Figure 3 shows that refrigeration accounts for 23.9% of the total energy consumption of food processes. Its significant share motivated performance improvements in the basic refrigeration system by choosing new control strategies, implementing novel components, and using more efficient thermodynamic cycles. From the GHG emission point of view, as most of the systems are based on refrigeration cycles that consume electricity, energy efficiency in this sector will have a limited impact on global emissions for EU countries, where the carbon footprint of electricity is generally significantly lower than that of fossil fuels.

3.1.1. Refrigerant Alternatives

One of the innovation drivers in the refrigeration sector is the inflection of legislation towards natural refrigerants. The F-gas regulation accelerates the phase-out of HFCs towards HFOs and natural refrigerants, still considering interrogations regarding HFOs that need to be seen as transition fluids due to their potential impact on health and on the environment [46]. As an example, according to the French association AFCE, direct emissions due to refrigerant leaks for refrigeration in 2020 represented 85% of total fluorinated gases (HFC, PFC, SF₆, NF₃) leaks, which accounted in total for 3.2% of GHG emissions [47]. The food industry was responsible for approximately 42% of the total emissions. At the European level, the same tendency is observed. Figure 9 presents the evolution in million tons of CO_2-eq of emissions due to the use of HFCs in refrigeration and air conditioning units (blue line) and due to the industry of food processing, beverages, and tobacco. This graph was built using statistical data from the European Environment Agency [48]. The emissions due to older refrigerants, such as CFCs and HCFCs, were not reported in the database. It shows the beginning of the phase-down of HFCs and the impact on the reduction of total CO₂-equivalent emissions of the food, beverage, and tobacco industry in Europe.

In order to reduce this impact, industries have modified their processes, passing from direct expansion (DX) infrastructures where refrigerant was circulating in the entire facility to indirect expansion (IX) infrastructures where glycol–water mixtures and air are cooled in the machinery room and then distributed in the facility. This modification allows for the reduction of the refrigerant charge and leakages associated at the facility level. Even if refrigerants are concentrated in the machinery room, current restrictions on the CO₂-equivalent charge lead to a transition towards natural refrigerants (CO₂, ammonia, propane, (iso-)butane) that present great thermodynamic properties but also different characteristics demanding important innovations and security protocols regarding high pressure, toxicity, flammability, and explosivity.

3.1.2. Basic Refrigeration System

Figure 10 shows the P and ID (piping and instrumentation diagram) of a basic ammonia refrigeration system for the agri-food industry. Ammonia is the preferred refrigerant in the agri-food industry because of its very interesting thermodynamic properties, which enable high COPs. Ammonia is also an environmentally friendly refrigerant with zero ODP (Ozone Depletion Potential) and zero GWP (Global Warming Potential). Due to its toxicity, an ammonia concentration detector is installed in the machine room. Typically, the screw compressor has an oil separator with an electric heater to maintain the oil at a sufficient temperature (around 50 °C even when it is off) to ensure its low viscosity. A control board shows the pressure and temperature measurements at different points of the circuit around the compressor. A degassing safety valve is preset at 21 bars. During operation, the oil is circulated and cooled by a specific coil inside the evaporative cooler. The refrigerant is sent to the evaporative condenser and then expanded by one or several ball expanders. The number of expanders in parallel depends on the refrigerant flow rate, which is proportional to the cooling demand. The low-pressure reservoir collects the refrigerant and feeds the evaporator by gravity through the liquid column. The level of refrigerant inside the reservoir in usual operation is around one-third of the height, checked by the low-level bottle. A second safety valve is connected to the reservoir and preset at 12 bars. The oil bottle recovers the lubricant that was not trapped by the compressor’s oil separator, and that ends inside the reservoir. The quality of the oil collected at that component is also tested regularly. The flooded evaporator cools the glycol water that is delivered to the terminal units of the factory.

In this basic refrigeration system, many potential improvements can be spotted. At the compressor level, a heat pump could replace the electric heater used to ensure sufficiently viscous lubricant. Upgrading components and control strategies seems to be one of the most efficient ways to improve the performance of current systems. Compressors, heat exchangers, valves [49,50,51], changes from asynchronous towards brushless motors [51,52], implementation of variable speed drivers for pumps and fans [53], energy monitoring, and data management [54]. The use of a floating condensing control can significantly reduce the overall electricity consumption of compressors and fans [55]. Over 100 technical articles, including artificial intelligence, were reviewed by Ahmed et al. [56]. They concluded that stochastic methods were the most efficient in solving optimization problems in the refrigeration sector. Refrigerant leak detection can also benefit from machine learning [57]. Secondary loop refrigerants can also be the subject of performance improvement and decarbonation [58]. Their implementation usually reduces the amount of refrigerant inside the networks. Moreover, instead of cooling a mixture of glycol and water, other single-phase or two-phase fluids or mixtures can be employed, such as CO₂ or water–ammonia mixtures. Finally, the evaporative condenser in Figure 10 could be assisted or replaced by a useful heat exchanger and a low-grade heat storage system for satisfying other heat demands such as washing. In that case, the lubricant would also need to be cooled by another secondary, storable, and useful fluid.

The refrigeration system shown in Figure 10 works with a flooded evaporator following the P-h diagram presented in Figure 11, drawn with EES software V11.823 [59]. The entire latent heat is exploited in that typical cycle. The evaporating and condensing temperatures are –12 °C and 50 °C, respectively, for cold chambers maintained at −5 °C. With an isentropic compression, the refrigerant discharge temperature is very high, around 140 °C. This level of temperature can ruin pieces of the compressor, diminishing its lifetime or increasing the number of maintenance operations. When the evaporating temperature is lower for cold chambers at −20 °C, for instance, the discharge temperature becomes even higher. Therefore, enhanced thermodynamic cycles (presented in the following section) were developed to address this issue.

3.1.3. Enhanced Refrigeration Cycles

Enhanced refrigeration cycles were developed to improve the coefficient of performance (COP) and to decrease the discharge temperature. Two-stage cascade systems combine two thermodynamic cycles. Therefore, two compressions with lower pressure ratios enable the reduction of the discharge temperature at the high-temperature cycle and increase the overall COP. Some common cascaded systems use refrigerant couples such as R134a-R407C, R134a-CO₂, Ammonia–CO₂, Propane–CO₂, R449A-CO₂, transcritical CO₂-subcritical CO₂…CO₂ is often chosen as the low-temperature fluid because of its interesting thermodynamic properties in the subcritical cycle and its relatively low environmental impact. Hu et al. reviewed 15 articles, tested six combinations of fluids, and confirmed this tendency [60]: CO₂ becomes a very frequent fluid in cascade systems.

Cooler refrigerant can be injected during the compression process. The refrigerant can come from an open or a closed intercooler. An economizer or a flash tank can be implemented. Khan and Bradshaw investigated the cycle performance of a flash tank economized system (FTES), an oil flooded system (OFS), and a multiple refrigerant injection system (MRIS). They found COP improvements compared to the single-stage cycle of 21%, 31%, and 38%, respectively [61].

Mota-Babiloni et al. draw an overview of the 2015 status of commercial refrigeration [62]. The refrigeration cycle improvements reported are additional subcooling with a secondary thermodynamic cycle, a dedicated subcooler [63], and the ejector cycle. Rostamzadeh et al. studied the role of ejectors in Ammonia–CO₂ or Propane–CO₂ cascaded refrigeration systems [64]. They concluded that the COP was improved by more than 50%. However, the quantitative risk assessment was degraded. The expansion device can be replaced by an expander for work recovery. The performance improvements are between 6% [65] and 12% [66] with R134a, 10% to 15% with CO₂ [67], and reported up to 30% with a CO₂ transcritical cycle [68]. The ejector technology was implemented as a booster for an R1234yf cycle with a maximum of 22% COP improvement [69] and a 38% increase in second-law efficiency [70]. The CO₂ booster cycle is currently being massively introduced in the supermarket sector. The booster compression and parallel compressor technologies were studied experimentally and numerically in China [71], in the UK [72], in Chile [73], in Spain [74], in India [75], in Turkey [76,77], in France [78] and for all climates [79,80,81]. The internal heat exchanger position for a CO₂ booster refrigeration system was investigated by Liu et al. [71]. Nano-refrigerants were even implemented in booster-assisted ejector expansion refrigeration systems by Aktemur and Ozturk [76,77].

Finally, heat recovery and thermal storage are expected to help in decarbonating food retail buildings [82]. Recovery heat exchangers should be systematically implemented on every refrigerating machine for possible washing purposes, supplying hot water for the heating floor of cold chambers, and, more generally, for any process requiring low-grade heat. For example, heat recovery from a supermarket refrigeration system for district heating achieved significant energy and cost savings, 93% and 41%, respectively [83]. Moreover, heat recovery is highly desirable because heat generation, which is the subject of the next section, is the major energy consumption share in the agri-food industry, as shown in Figure 3.

3.1.4. Summary of Solutions for Refrigeration Systems

Besides the ongoing fluid transition, most energy-efficient and decarbonating solutions are related to using improved components and storage systems in single-stage and two-stage cycles.

Table 3. List and details of industrial implementation of refrigeration units.

Location, Company	Application	Previous Technology	Novel Technology	Power	COP and Energy Savings [- and %]	Annual CO2 Emissions Savings [% and tCO2/Year]	Payback Period [Years]	Year	Ref.
United States, Lowes Food	Climatic chamber for food storage	Conventional R404A DX system	Secondary propylene glycol secondary circuit for commercial refrigeration (and VFD control on compressors)	319 kW	- 5.2% (and 10.2%)	-	-	2010	[84]
United Kingdom, -	Refrigeration in food retail	DX CO2 transcritical booster with an air-cooled condenser	DX CO2 transcritical booster with a hybrid condenser (water-cooled and air-cooled condensers in parallel)	-	2.4–2.9 40 6%	11 -	4.8	2021	[85]
Denmark, McDonald’s	Refrigeration in food retail	HFC refrigeration units with R404a, R134a, R407C	Natural refrigerant units with R290, R600a, and R744	-	- - 12%	- 27%	-	2003	[86,87]
Scotland, Youngs Seafood Group	Slamon processing	Two refrigeration units working with R22	Air-cooled low-charge ammonia refrigeration unit	111–650	- - 15–20%	453–815 -	-	-	[88]
United Kingdom, Blakemans	Sausages and meat-based products manufacturer	Six separate units working with R404A	Low-charge ammonia refrigeration units	290	- - 77.3%	350 -	-	-	[89]

reports the industrial implementation of energy-efficient and decarbonating solutions for refrigeration plants available on the Internet. The efficiency improvement related to thermal storage is described in Section 3.4.

3.2. Heat Generation

In this section, technologies available to either provide energy efficiency or decarbonating solutions of heating processes in the food industry are addressed. Other non-thermal solutions, such as high-pressure processing, membrane or pulsed electric fields, are not considered. They can sometimes be more efficient and relevant than thermal solutions depending on the process and its texture, and nutrient properties objectives.

3.2.1. Combined Heat and Power (CHP) Systems

One of the first steps for energy efficiency and decarbonation of industrial processes using conventional boilers is CHP systems, also called co-/trigeneration systems. In cogeneration systems, fuel combustion generates steam for heating and potential energy in the form of pressure for electricity generation. Each system is then defined by its overall efficiency (η_g), the sum of the electrical (η_e) and thermal efficiencies (η_th). In trigeneration systems, excess waste heat is also used to regenerate the sorbant and an absorption chiller to simultaneously produce heat, cold, and power. As most of the carbon footprint gain is issued from heat processes, it is important to consider cold generation as a secondary output and not oversizing heat generation for refrigeration purposes. Figure 12 shows the operating principles of cogeneration and trigeneration.

Cogeneration implementation can be either (i) topping, meaning that the main output is electricity and thermal energy is produced by recovering the waste heat produced from the power generation plant, or (ii) bottoming, meaning that the main output is thermal energy and electricity is produced using waste heat from the thermal process [91]. In this review, bottoming technologies are relevant. Steam generated from the boiler’s waste heat is driven through a steam turbine to produce electricity as a by-product that can be either consumed or discharged to the grid. Bottoming cycles are usually installed in high-temperature industries (glass, cement) where the important waste heat temperature allows for important pressure lifts and, therefore, higher efficiencies but higher costs for the steam turbine [92]. Depending on the heat-to-electricity ratio required by the installation, different cogeneration technologies can be implemented. For example, an internal combustion engine (ICE) will be beneficial for installation with a low heat-to-electricity ratio, while a steam power plant with a backpressure turbine will be more beneficial for installation with high heat-to-electricity ratio [93].

Vellini et al. [93] focus on the food industry and on an extensive experimental campaign, where cogeneration ICEs, steam power plants with a backpressure turbine (SPP-BPT), with a condensing turbine (SPP-CT), and a gas turbine (GT) perform primary energy savings (PES) between 3.9 and 17.2%. Bianco et al. [94] studied the implementation of a cogeneration system in a bottling plant. The parametrical study showed PES between 10.2 and 11.3% depending on the compression ratio obtained in the turbine. More efficient performances are technologically feasible but not profitable from the economic point of view due to higher expenses. In economically optimized cases and with a beneficial gas-to-electricity price ratio, payback periods can reach down to 2.4 years. There is also a strong benefit for CHP plants of being close to an electricity consumption hub to provide flexibility to the industrial consumption and production profiles. In the case of British Sugar CHP plants of 2.2 MW_e and 2.8 MW_th, the fluctuating excess electrical load is furnished to 8000 households [95]. Figure 13 shows a photograph of a cogeneration unit in a milk and dairy factory [96].

For trigeneration systems, Freschi et al. [97] analyzed the case study of fruit conservation and showed the sensitivity related to economic, environmental, or energetical objectives. First, CO₂ emissions outcomes are not always beneficial and are very sensitive to the installation size and match with the user consumption profiles. The best results show PES of 9% without and 12% with thermal storage. From the GHG emissions point of view, as natural gas usually has a higher carbon footprint than electricity, if the design implies producing a larger amount of excess heat to satisfy the refrigeration load, then the CHP implementation will have a negative effect on CO₂ emissions. The second outcome shows that, from the energy savings point of view, trigeneration systems are sensitive to power loads and thermal storage inclusion. Lower power requirements and thermal storage allow an increase in energy savings from 5% to 10–15%. Finally, regarding energy consumption, Tassou et al. [98] show the impact of the absorption chiller COP and internal system efficiencies on the operation of a (micro)turbine. More generally, Popov et al. [99] show that trigeneration systems can be extremely useful for the integration of renewable energy production in polygeneration energy facilities as they provide flexibility increase between heating, cooling, and power fluxes.

Co/trigeneration units are interesting when compared with traditional energy systems in separate production modes, used to provide simultaneous heat and electricity [100]. Compared to such systems, using cogeneration allows for the provision of PES between 10 and 30% [93]. However, considering modern electricity production methods, the interest in cogeneration systems decreases until they are counter-efficient. Respectively, with 49% and 34% of power production efficiency in separate modes, the trigeneration system allows a 9.1% and 31% improvement. As soon as the power production efficiency is higher than 60% or renewable energy production is sufficiently important in the mix, trigeneration is counter-efficient. A study in Portugal has shown that CHP units are inefficient when renewable electricity production overpasses 19% [101,102]. Hasan et al. reveal that future electricity mixes are expected to be significantly more efficient with the progressive development of renewable energy sources [103], which implies that cogeneration units will present less interest in the future.

As heat and electricity are produced together, flexibility is also affected by cogeneration units, which either leads to installations designed to cover part of the demands, or to implementation of important thermal/electricity energy storage systems.

Overall, in CHP systems, cogeneration is largely implemented due to its technical simplicity, lower payback periods, and the ability to control the heating and cooling demand separately from the cogeneration and refrigeration units (

Table 4. List and details of industrial implementation of cogeneration units.

Location, Company	Application	Previous Technology	Novel Technology	Power [MWe]	Efficiency and Primary Energy Savings [% and %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
Sicily, Italy -	Wheat pasta drying	Separate production with grid and natural gas	CHP gas turbine operating with natural gas working in (dis)continuous mode	0.8	70% (th: 46 e: 24) 5–9%	271–481 6–9%	14.8–16.1	2007	[106]
Italy	Soft drinks bottling	Separate production with grid and natural gas	CHP gas turbine operating with natural gas	1.8–2.2	- 10.2–11.3%	- -	2.4–5.7	2016	[94]
Belgium	Aquatic center	Separate production with grid and natural gas	CHP-ICE operating with natural gas	0.2	85% (th: 51 e: 34) - -	- -	3–10	-	[100]
Spain	Olive sludge—hot water and drying process	Separate production with grid and natural gas	CHP-ICE operating with natural gas	-	- 16%	753 -	4.5	2006	[105]
Australia, Midfield	Beef, lamb, veal, and mutton sterilization	-	CHP-reciprocating gas engine operating with biogas and natural gas co-combustion	1.5 MWe	89% (th: 47 e: 42) -	- -	+/−5	2009	[107]

). Trigeneration systems are detailed in

Table 5. List and details of industrial implementation of trigeneration units.

Location, Company	Application	Previous Technology	Novel Technology	Power [MWe]	Efficiency and Primary Energy Savings [% and %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
Italy, Brunel University	University test for food applications	Electricity from the grid and natural gas boiler	Microturbine, gas absorption chiller, and natural gas boiler	0.294	88.1% (th: 50.5 e: 37.6) -	- 0–12.5%	4.51	2007	[98]
United Kingdom	Brewery	Oil boiler and diesel generator	Biogas boiler from brewery waste and cattle slurry, water–ammonia absorption chiller	-	88.4% (th: 69.4 e: 30.6)	- -	5.4	2012	[108]
Brazil, -	Soluble coffee manufacturing	Coffee ground and firewood biomass boiler	Biomass boiler, water–ammonia absorption chiller, steam turbine	5.3	- -	- -	>2	2001	[109]
France, Nestlé	Starch, wheat product manufacturing	-	Turbine, absorption chiller, and biomass boiler	16	- -	90,000 -	-	2016	[110]

. Those systems are still considered temporary solutions, as their energy savings and decarbonation potential are largely under the objectives set by the food industry perspectives. In future polygeneration energy networks, with high-efficiency and low carbon intensity of electricity and thermal productions, only biomass cogeneration units, able to valorize on-site wastes, will remain economically and environmentally relevant [104,105].

3.2.2. Biofuels and Hydrogen

Using biofuels or hydrogen for process heating in replacement or partial substitution of fossil fuels is a widely investigated way for industrial heat decarbonation. This can be achieved either by purchasing novel biomass boilers/burners, retrofitting conventional ones, or by co-combustion, which is a progressive mix of different fuels. The main advantage from the GHG emissions point of view relies on the lower carbon intensity of fuels per kWh_th produced. Natural gas emissions are estimated between 266.9 and 586.2 g_CO2/kWh compared to (i) 16.8 to 93.4 g_CO2/kWh for pulverized solid biomass depending on its size [111], (ii) 23.4 to 44 g_CO2/kWh biogas from methanation [112], (iii) 303.0 to 393.9 g_CO2/kWh for grey hydrogen, and (iv) 60.6 g_CO2/kWh for yellow hydrogen [113] (

Table 6. Carbon intensity of different fuel inputs [110,111,112].

Fuel	Carbon Intensity [gCO2/kWh]	Fuel	Carbon Intensity [gCO2/kWh]
Natural gas	266.9–586.2	Grey hydrogen	303.0–393.9
Pulverized solid biomass	16.8–93.4	Yellow hydrogen	60.6
Biogas from methanisation	23.4–44

). Grey hydrogen is produced by natural gas, and naptha steam–methane reforming represents 80% of 2022 H₂ production. Yellow hydrogen is produced by electrolysis using electricity from the grid (40 g_CO2/kWh_e) and represents 0.07% in 2022, with an increase of 35% compared to 2021. Current trends and objectives of decarbonation envision reaching 90.9 and 30.3 g_CO2/kWh, respectively, in 2050s APS (Announced Pledges Scenario) and NZ (Net Zero) scenarios. In addition to decarbonation potential, biomass also addresses economic considerations as, most of the time, in the food industry, an important amount of valuable biomass is available on-site. When biomass is not directly available, large-scale heating plants need complementary incentives (carbon taxes, subsidies) to be competitive [114,115]. The usage of hydrogen is still widely reserved for very high-temperature processes found in steel, cement, or glass due to its prohibitive price compared to other technologies. The levelized cost of heat for hydrogen can also be found in Section 4. For now, only research pilot projects like the development of a flexible hydrogen/natural gas burner developed at West Virginia University [116] are under early-stage development for the food industry. Future increases in low-carbon hydrogen availability and cost decreases could promote its use for food processes.

Biofuels are a wide and heterogeneous scope of biomass-based fuels. Depending on their nature, treatment, and quality, a large scope of products can be obtained. Clairland et al. [21] present different ways of producing thermal energy and electricity from food waste through direct combustion of wastes or biofuels. Biofuels are separated into three categories: (i) gaseous fuels such as biogas via anaerobic digestion, syngas via gasification, or pyrolysis and ethanol via fermentation, (ii) liquid fuels such as biodiesel via esterification, ethanol via fermentation, vegetable oil via extraction, and pyrolysis oil via pyrolysis, and (iii) solid fuel, like char obtained via pyrolysis or torrefaction.

Two case studies from IEA Bioenergy on wood chips and composting residues in direct combustion for the potato processing industry [117] (Figure 14) and wood chips and grain residues in direct combustion for bakery processing. Nussbaumer [111] studied the impact of biomass industrial heating. This task contribution showed the relevance of using wastes as low-grade biomass for thermal needs, especially when no other higher-value application can be obtained.

Overall, biofuels are a promising solution for the food industry, as they allow the generation of significant carbon emissions reduction and keep similar installations from a technical point of view. However, understanding the problem not only from the technological point of view but also from the biomass availability and fuel logistics is extremely important. For hydrogen, it is widely accepted that its use for such low process temperatures is not relevant before a significant take-off of green hydrogen production. Its usage needs to be first reserved for higher-constrained industries and mobility applications.

Table 7. List and details of industrial implementations of biomass/biofuel boilers.

Location, Company	Application	Previous Technology	Novel Technology	Power [MWth]	Efficiency and Energy Savings [% and % or m3]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
Netherlands, ProBiomass BV	Potato industry steam generation	Natural gas boilers	Woodchip and composting residues boiler	10	- 8.2 × 106 m3	14,500 -	-	2020	[117]
Australia, Simplot	Vegetable processing for cleaning, blanching, defrosting	Natural gas boiler	Wood waste or woodchips biomass boiler + heat recovery	5	- -	- 33%	-	2020	[118]
France, Tereos	Sugar manufacturing and distillation	Natural gas boiler	Biogas and natural gas co-combustion boiler	8.3	- 5 × 106 m3	11,200 -	<5.0	2013	[119]
United Kingdom, British Sugar	Sugar manufacturing	-	Biogas and biomethane CHP	2 × 2.8	86% (th: 48.2 e: 37.8) -	- -	-	2016	[95]
Spain, Spain	Wet pet food manufacturing for steam generation	Natural gas boiler 2 × 5.4 MWth	Solid-biomass fuelled boiler	2.05	-	- 42%	-	2019	[120]
			Biomethane fuelled boiler		-	- 27%	-
France, Nestlé	Sugar refining evaporation process	Natural gas boiler	Solid biomass boiler	20 + 5	- 8%	40,000 -	-	2013	[121]
Switzerland, Coop Group	Large bakery process	-	Woodchips and grain residue pellets biomass boiler	2.9	- -	4000 -	-	2015	[111]
- -	Rice processing plant	Coal boiler	100% Rice husk, 50% Rice husk and wood, 100% wood combustion	-	- 19–22%	11,612 98	1.6–2.0		[122]

presents the industrial implementations of biomass and biofuel boilers.

3.2.3. Electro-Heating Technologies

As mentioned in the IEA inter-task project on industrial process heat 2021 [2], the electrification of processes is a key milestone to provide significant process heat decarbonation. Atuonwu et al. [123] also highlight features and advantages of electricity-driven processing techniques in the industry, including reactivity, ease of automation/control of novel solutions, heating selectivity, and uniformity. On the other hand, the important price ratio between electricity and other fuels leads to a significant competitive disadvantage for such technologies compared to conventional installations. Zakeri et al. [124] highlight that fossil fuels, and mainly natural gas, set electricity prices in 58% of cases and are the primary electricity price setter in Europe, as 34% of the electricity is issued from their usage. The electricity-to-gas price ratio is often used as an indicator of electro-heating market attractiveness. Going towards electricity-driven technologies will be possible if electro-heating technologies provide an efficient gain sufficient to compensate for the economic gap or if, before reaching such efficiencies, tax support allows them to be competitive.

Electric Immersion Heater

The most famous and still most widely used electric technology in the industry is the electric immersion heater, which has a straightforward principle based on Joule’s effect, which is transmitted by conduction and convection to the immerged medium. Usually, the most efficient heaters have an energy efficiency from 80 to 95%. According to Gruber et al. [125], using an electric immersion heater is a very straightforward and easy-to-implement solution for processes with temperatures below 240 °C, which covers almost all of the food industry range. However, due to the low thermal conductivity of the heated medium, especially for solid medium, the increase in temperature is usually extremely unbalanced, leading to food quality and texture deterioration. Naji et al. [126] showed that using porous resistive material allows for a decrease in the heat exchange thermal resistance and allows higher electrical power within a heater.

Alternative Thermal Processing

Food possesses dielectric and conductive properties, allowing electro-heating methods such as (i) infrared heating, (ii) induction heating, (iii) radiofrequency or dielectric, (iv) ohmic or direct resistive, and (v) microwave (

Table 8. List and details of different electro-heating technologies.

Electro-Heating Technology	Description	Efficiency	References
Infrared (IR)	Bring electric heaters at very high temperatures (650–1200 °C) to produce radiation for uniform and oriented surface heating Drying, baking, roasting	40–70%	[129,130]
Induction (IH)	Heating pipe or magnetic element in contact with food product with a magnetic field generated by an electrical coil All processes, including liquid or slurry heating	70–95%	[127]
Radiofrequency (RF)	Placing food between two electrodes and generating an electromagnetic field through capacitors at low frequency (10–50 MHz)	60–65%	[131]
Ohmic (OH)	Placing conductive food in direct contact with two electrodes leading to alternating electric current passing through the volume at low frequency (50 or 60 MHz)	90%	[128,132]
Microwave (MW)	Applying an electromagnetic field at high frequency (0.3–300 GHz) to polar molecules leads to friction/rotation converted into heat	10–85%	[133,134]

). Those treatments are generally much faster than conventional thermal treatments, making them extremely interesting for many processes such as sterilization, pasteurization, or food disinfection. However, they are non-applicable to baking processes where chemical reactions need longer time spans. If those technologies appear to be promising, their penetration in the food industry is still extremely marginal. With longer payback periods compared to immersion heaters due to higher installation costs and lower specific efficiencies, those technologies do not represent a significant axis of solution for energy efficiency or decarbonation. Nevertheless, they sometimes have more uniform and oriented heating methods and can, therefore, decrease heating temperatures for volumetric heating. For example, Yilmaz et al. [127] and Balthazar et al. [128], respectively, showed an energy efficiency increase of 19.5% and 40–72% for induction and ohmic heating compared to immersion heaters in pasteurization processes of tomato and milk. Despite similar technological efficiencies, pasteurization processes can be simplified and require less energy with located, uniform, and oriented heating techniques.

3.2.4. Solar Thermal

Generation of heating and cooling can be achieved by radiant to thermal energy conversion. With a very broad range of temperatures, up to 1500 °C, solar thermal has a strong potential for energy efficiency and decarbonation in food industry processes below 250 °C. Based on radiation concentration, different temperatures are reached depending on the geographical localization, the mirrors and collectors’ arrangements shown in Figure 15, the presence of a sun-tracking system, and the heat transfer fluid (HTF). All factors have an important impact on the economic viability of an installation [33]. Today, the most widely implemented technology is flat plate collectors that have, according to Figure 16, a maximal temperature output of around 80 °C, while for concentration technologies, the most commonly used ones are linear Fresnel and parabolic trough collectors.

If the IEA Task 49 solar heating and cooling program agrees on the important potential of solar thermal, its installed capacity today is still limited. Paya et al. [135] highlight the non-pilotability and need to be complemented by supplementary equipment (i.e., biomass boiler), having a direct impact on the CAPEX and the technology attractivity. Saini et al. [136] show the impact of the incoming irradiance and solar ratio on the levelized cost of heat (LCOH). Feedback from existing projects usually presents a solar ratio (effective thermal output compared to installed capacity) between 15 and 35%. Sharma et al. [137] also highlight the principal drawbacks, which are higher installation costs, intermittent heat streams, and fluctuating temperatures. Coupled with the lack of optimal, compact, and cheap thermal storage technologies and the lack of appropriate incentives, solar thermal is still a marginal heating technology at the industrial level. A survey among industrials [138] highlighted that payback periods between 6 and 15 years, when the food industry sector expects payback periods between 3 and 5 years, were too long, which was the main drawback for the significant take-off of the technology. Anyhow, improvements in thermal storage and fossil fuel price increases might reduce payback periods and bring attractivity and competitiveness for solar thermal projects. Finally, another benefit of solar thermal systems is their easy recyclability with low energy consumption compared to solar photovoltaic panels.

From a technical point of view, solar thermal energy can be installed at the process or supply level. The supply level is generally found, as it requires less installation and rearrangement at the plant level. Then, heating can be either (i) directly transmitted to the process, meaning that the HTF is the same as the process fluid, usually steam, or (ii) indirectly transmitted to the process, meaning that the HTF will exchange heat with the process fluid through a heat exchanger. Indirect installations are interesting when temperatures increase over 100 °C and thermal oil is required to keep a liquid HTF within the collectors. Although this technology is extremely interesting, its non-pilotability and full dependency on weather conditions push current installations to cover additional systems that need to be able to provide 100% of the required load at any time. Therefore, the CAPEX of such systems can be seen as additional and not self-sufficient.

In Europe, 28 projects are currently running in the food, meat, vegetables, food products, dairy and beverages industries. Among them, in Switzerland, a dairy plant replaced its fuel oil water boiler by a 439 kW direct-steam system equipped with parabolic trough collectors, storing steam at 120 °C required at the process supply allowing to cover over 50% of heating demand [139].

Table 9. List and details of industrial implementation of solar thermal installations.

Location, Company	Application	Previous Technology	Novel Technology	Power, Footprint and Solar Irradiance [MWth, m2, W/m2]	Energy Produced and Energy Savings [MWhth and %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
Austria, Goess Heineken	Hot water and steam for drying. Cleaning, pasteurizing, and mashing	Natural gas boiler	Standard flat plate solar collectors + biomass boiler for complement	1.0 1500 666	471 30%	500 -	-	2013	[140]
France, Lactalis Group	Milk powder manufacturing	Natural gas boiler	Standard flat plate solar collector	10.7 14,843 713	8000 11%	2000 7%	-	2023	[141]
Switzerland, Emmi AG	Dairy plant for steam generation	Fuel oil boiler	Parabolic trough solar collectors and linear Fresnel reflectors	0.44 627 700	255 -	- -	-	2012	[33]
Greece, Colgate Palmolive	Household products manufacturing	Natural gas boiler	Parabolic trough solar collectors with rotation mechanism	- - -	163 -	- 39%	-	2018	[142]
Italy, Nuova Sarda	Dairy, cheese plant for steam generation	Fuel oil steam generation	Linear Fresnel reflectors	0.47 995 474	500 -	- -	-	2015	[33]
Netherlands, Tesselaar Greenhouse	Flower greenhouse heating for hot water production	Natural gas boilers	Flat plate solar collectors	6.5 9300 698	5000 -	CO2 neutral -	-	2019	[143]
France, SARL Cavet	Cheese maturing for hot water production for heating and cleaning	-	Flat plate solar collectors	- 90 450	57 33%	- 16%	6	2005	[119]
Italy, La Felicetti Pasta Factory	Pasta drying for	Natural gas boiler and cogenerator	Parabolic collectors, troughs, reflectors	- - 600–900	- -	99 -	9	-	[144]
Egypt, Sana Foods	Sweets manufacturing	Natural gas steam boiler	Solar flat plate collectors	- 110 -	189.7 3.0%	51 -	5.4	2021	[145]

reports the list and details of industrial implementation of solar thermal installations found in the literature.

3.2.5. Geothermal

Geothermal installations can be found in direct-use (DU) applications to provide heat at temperatures usually lower than 150 °C [146] or in indirect-use (IU) applications to supply an electrical demand through an additional heat-to-power (ORC or steam turbine) installation requiring temperatures between 150 °C and 500 °C. Modern technologies such as binary flash cycles and enhanced geothermal systems (EGS) allow us to envision electricity production from 70 °C geothermal outputs. The Lindal diagram in Figure 17 summarizes potential applications for DU and IU, including agriculture and food industry processes, as listed by Dickson et al. [147]. Food processes seem extremely appropriate for geothermal technology as the temperatures are appropriate, and geothermal heating can provide efficiencies up to 320–370% [148]. According to a report from the FAO [146], low to medium enthalpy geothermal resources are the best option for food drying and present strong interest for pasteurization, pre-heating, evaporation, distillation, blanching, and sterilization processes.

If geothermal facilities are widely spread in agriculture, especially for greenhouse heating, their implementation in the industry is still very limited. According to a report from the IPCC [149], geothermal power plants in 2009 represented 10.7 GW_e, while direct uses were 50.6 GW_th. According to Lund et al. [150], DU capacity increased to 70 GW_th in 2015 but is largely dominated by ground source heat pumps (GHP) with a share of 70.9% and tertiary heating with a share of 23.6%. Greenhouse heating, aquaculture pond heating, agricultural drying, and industrial uses, respectively, represent 1.97, 0.70, 0.16, and 0.61 GW_th, which all combined represent 3.4%. Dalla Longa et al. [151] suggest that its use remains low in industry due to important costs linked to exploration and in-depth drilling, which adds to the geographical correspondence required between the existing plant and the optimal hole drilling point. For novel implementations, especially in developing countries, the reversed approach, where the plant is located according to the drilling point, would contribute to a significant uptake of geothermal DU installation [150]. Recent innovations regarding drilling technologies developed by companies, such as Canopus [152] or Celsius Energy [153], work on cost and risk reduction of drilling phases. Such innovations are still limited to shallow geothermal resources applied to tertiary heating applications, but if those proof-of-concepts are validated, the same approaches could be developed for deeper applications. Such development could then reduce the impact of the drilling phase, which is today the principal barrier of geothermal uptake, with around 50–75% of total costs, according to Robins et al. [154].

Table 10. List and details of industrial implementation of geothermal energy production.

Location, Company	Application	Previous Technology	Novel Technology	Power, Depth [MWth, m]	Energy Produced and Energy Savings [GWhth, %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
France, Roquette Frères	Starch extraction EGS	Wood biomass and natural gas boiler	Geothermal double borehole	27 2500	180 25%	41,000 -	-	2017	[155]
Australia, Midfields Meats	Beef, lamb, veal, and mutton sterilization	Natural gas boiler	Geothermal borehole	- -	7.7 -	1800 -	-	2009	[130]
New Zealand, Rogue Bore Brewery	Brewery	-	-	3 -	- -	- -	-	2020	[156]
Costa Rica	Vegetables and grain drying	-	-	1 -	5.8 -	- -	-	2005	[150]
Kenya, GDC	Dairy pasteurization	-	Geothermal borehole	-	- 40%	- -	-	2021	[157,158]
New England, -	Greenhouse heating	-	Geothermal borehole	- 75–500	- -	- -	< 10	2008	[159]

reports the industrial implementations of geothermal energy production, some of them achieving substantial energy savings and carbon emission reductions.

3.2.6. Vapor Compression Heat Pumps

Among all electrification solutions, specific attention is brought to heat pumps (Figure 18). In the agri-food industry, investing in heat pumps for low- and high-grade heat production is a strong tendency due to the higher costs of natural gas and the necessity of decarbonation for climate change mitigation. Important improvements have been made in recent years. Integration of heat pumps for processes below 250 °C is particularly relevant. Heat pump performance can be measured with (i) the 1st law energy efficiency (COP) that measures the ratio between the useful output thermal energy (Q_h) and the electrical input energy (W_e), and the 2nd law exergy efficiency (η_ex) that measures the ratio between the effective performance COP and the ideal cycle performance COP_id. If the COP is the common indicator used to characterize a heat pump, its value strongly depends on the operational environment and heat source temperatures. To compare technologies together, the η_ex is preferred as it depends only on the machine itself, quantifying the losses and irreversibilities caused by all components.

The review provided by Arpagaus et al. [161] shows that current state-of-the-art exergy efficiencies are between 35 and 60%, even if on-market products are generally more around 35 to 40%. Royo et al. [162] present the study of an experimental set-up working with an exergy efficiency of 38.4% and breaks down the exergy losses at the compressor (50.6%), the expansion valve (22.5%), the condenser (14.2%), the evaporator (6.9%), and the IHX (5.6%). Similar results were obtained by Byrne et al. [163], assessing the exergy losses at the compressor (54.4%), the expansion valve (14.7%), the condenser (15.0%), and the evaporator (15.7%). Technological improvements have been developed to decrease exergy destruction rates and, therefore, increase the provided COP. Main sources of improvements have been focused on compression segments with multi-stage or parallel compressions [74,161,164,165,166], the inclusion of ejectors to recover part of the discharge pressure [64,66,76,77,79], the gas cooler and evaporator performances [49,74,75]. Nelissen et al. [167] separate technologies into four categories: closed systems with Ab/Adsorption Heat Pumps (AHP), Vapor Compression Heat Pumps (VCHP), and open systems with Thermal Vapor Recompression (TVR) and Mechanical Vapor Recompression (MVR). The focus of this section is on systems that use refrigerants in a closed loop to take waste heat or heat from the environment and upgrade it to higher temperatures transmitted to a heat transfer fluid, usually water or steam used for the processes or to a thermal storage pit. Figure 19 shows the evolution of COP depending on source and sink temperatures for an HTHP with an exergy efficiency of 40%.

The review provided by the IEA Annex 58 for HTHP [168] presents different technologies able to provide COP around 2.5 to 4 for temperatures between 120 and 160 °C. This places heat pumps as the most valuable technology to provide electrified, efficient, decarbonized, and economically viable solutions compared to fossil fuels and can cover most of the food industry’s processes. However, the low exergy efficiency limits the temperature lift between the heat source and the heat sink. Therefore, to reach temperatures above 90 °C, current technologies are all based on waste heat recovery sources required around 60 to 100 °C (

Table 11. List and details of industrial implementation of regular heat pumps and high temperature heat pumps.

Location, Company	Application	Previous Technology	Novel Technology	Power, COP, Tsource/Tsink [MWth, -, °C/°C]	Energy Produced and Energy Savings [MWhth, %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
Norway, TINE SA	Dairy products for cold, hot water, and steam	Natural gas boilers	HTHP cascade R290 (LT) and R600 (HT)	0.30 2.5 20/115	- 62%	- <94%	2.5	2021	[25,170,171]
Austria, Agrana	Starch drying	-	HTHP twin cycle R1336mzz	0.37 2.8–3.2 80/160	2200 20–80%	660 40–90%	-	2020	[172]
Ireland, Ahascragh	Distillery whisky	-	HTHP	1.0 5 60/120	- 50%	736 70%	<3	2023	[167]
Denmark, Arlan Arinco	Natural gas boiler	Milk powder drying and air pre-heating	HTHP hybrid R717 R744	1.25 4.6 45/85	7200 -	1400 -	1.5	2012	[25]
Switzerland, Slaughter-house Zurich	Hot water generation in a slaughterhouse	Fossil fuel boiler	HTHP R744	0.80 3.4 30/90	2590 75%	510 30%	-	2011	[25]
France, Lesaffre	Yeast production	Waste heat recovery for performance increase in yeast production	HTHP	19 - 28/90	- 70	30,000 70%	-	2025	[173]
France, Actalia	Pasteurization for milk, cream, and serum manufacturing	Natural gas	HP single stage	0.64 5.6 28/50	4310 -	890 -	2.5–2.9	2014	[174]

). As shown in Figure 19, with a targeted COP of 2, heat sources need to be adjusted depending on the process temperature. If the available waste heat does not allow to satisfy the design, it is also possible to design a hybrid system with solar thermal [50] or geothermal pits [169]. Alternative solutions to provide further efficient systems are to rework the heat pump design and sufficiently increase its exergy efficiency to enable good performances with greater temperature lift. If the first objective can be evaporating at the environment temperature, further exergy efficiencies could contribute to evaporating at an even lower temperature and provide an HTHP heat pump that is able to simultaneously provide heat for steam production and cold for cooling and freezing applications. Such solutions are being envisioned but are in the early stage of development and have very low technological readiness levels (TRL).

3.2.7. Ab/Adsorption Heat Pump (AHP) and Ab/Adsorption Heat Transformer (AHT)

Instead of using electricity to drive a compressor, it is also possible to use thermal energy to run similar cycles known as ab/adsorption cycles. Absorption cycles work with a liquid sorbant, and adsorption cycles work with a solid sorbant. AHP is based on the Osenbrück cycle: (i) an endothermic desorption–charging phase during which heat Q_g is provided to the saturated generator to release working fluid in the vapor phase that is sent to the condenser, where heat Q_c is released until the sorbant is dried and (ii) an exothermic ad/absorption–discharging phase during which the liquid working fluid is evaporated by receiving heat Q_e from the low-temperature source and then captured by the absorber Q_a until its saturation. The cycle works reversely for AHT. Depending on the heat sources and heat sinks available, ab/adsorption systems can be of different types and be arranged according to different heat sources/sinks, as presented in Figure 20, Figure 21 and Figure 22. AHPs can be seen as systems allowing the merging of two heat streams at different temperature levels to be combined in a single medium heat stream without energy nor exergy losses in ideal cases. AHTs can be seen as heat splitters where medium heat is divided into two with higher and lower temperatures. The scientific literature also shows the great importance of absorption chillers (AC) that are AHP designed to generate thermally driven cold streams. Two key aspects of those systems are to (i) determine the adsorbent/adsorbate working pair and (ii) determine the operational arrangement of the system depending on the driving and sink temperatures [175]. Regarding the energy balance, it is necessary to first look at the environment and boundary conditions of the system. Xie et al. [176] present a theoretical model that allows us to determine AHP and AHT performances. Their study shows that depending on the temperature lift, the COP of such systems is, respectively, ceiled at 1 and 0.5 for ideal single-stage AHP and AHT. Conversely, the literature also reveals coefficients of performance for AHP up to 1.6–1.7 [177,178], where the low-temperature heat source was the ambience and, therefore, not considered in the energy balance.

Riaz et al. [178] present an extensive review of AHPs used for heating applications, presenting different adsorbant–adsorbate working pairs such as zeolite/water, silica gel/water, activated carbon/methanol, and activated carbon/ammonia. If expected COPs vary generally between 1.3 and 1.6 depending on the working pair and, therefore, temperature levels, this study also highlights the COP deterioration of about 18% between commercial or experimental plants and theoretical studies. This study also provides insights into the durability and hysteresis potential of some working pairs. Cudok et al. [179] present a review of single and double-effect AHT systems. Double-effect AHTs allow us to envision higher temperature lifts but with a COP decreasing to around 0.3 compared to 0.5 for single-effect ones. Currently, many research studies are focusing on finding novel working pairs as historical ones (H₂O/LiBr and NH₃/H₂O) present corrosion problems.

A solar thermally-driven pilot plant with an ammonia–water absorption chiller of 12 kW was tested by Döll et al. [180]. An integrated cooling tower, two ice storages of 52 kWh latent storage capacity, and a cold store of 100 m³ were connected to the chiller. The electrical EER (energy efficiency ratio) of the system was measured at 9.5 ± 0.6 without considering the cold distribution. Some potential improvements were noticed: changing the refrigerant in the secondary circuit, using a variable speed pump and fans, optimizing operating temperatures, increasing ice storage capacity, and avoiding shocks when switching modes. An EER of 12 was achieved by Weber et al. [181]. The thermally driven ammonia–water absorption chiller was supplied by a heat source at 200 °C coming from a linear concentrating Fresnel collector.

Hybridization between mechanical and thermal heat pumps is often a way to justify the use of ab/adsorption systems from a techno-economical point of view. An open compression absorption heat pump using a CaCl₂/H₂O solution to recover water and heat from flue gas was studied by Wang et al. [182]. The simulation results show the highest water recovery efficiency of 82.4% and maximum COP values of around 1.6. A hybrid compression–adsorption heat pump cycle is proposed by Shamim et al. [183]. The calculated COP was found between 2.64 and 6.37 depending on adsorber and desorber operating pressures. Mohammed et al. studied an integrated adsorption–absorption system driven by transient heat sources for cooling and desalination by evaporation at low pressure [184]. The low exergy efficiency of the absorption cycle (15%) was compensated by the higher exergy efficiency of the adsorption cycle (43%). The combination of absorption and adsorption cycles also enhanced water production by 30%.

Table 12. List and details of industrial implementation of ab/adsorption heat pumps.

Location, Company	Application	Previous Technology	Novel Technology	Power COP [kWth, -]	Energy Produced and Energy Savings [MWhth, %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
- -	Poultry processing plant	-	Ammonia/water absorption for cooling/heating demand	350 -	- -	- -	-	2006	[185]
Spain, -	Edible oils processing	CHP operating with natural gas	Trigeneration based on absorption chiller	2400 -	- -	- -	1.3	-	[186]
Egypt, Sana Foods	Sweets manufacturing	Natural gas steam boiler	Furnace exhaust heat recovery absorption chiller	- -	113.6 4%	30 -	2.4–3.2		[144]
Japan, Alcohol industry	Alcohol industry	-	Single-stage absorption heat transformer (H20/LiBr)	2475 0.45	- -	- -	-	-	[187]
Italy, Baronia Flumeri	Pasta dough pre-heating and superheated water production	CHP operating with natural gas engines	Single-stage absorption heat transformer (H2O/LiBr)	1202 0.46	- -	- -	-	2013	[187,188]

reports the details of industrial implementations of AHP and AHT found in the literature. The majority of identified projects are based on absorption chillers, which today are mainly based on trigeneration principles that are listed in Table 5. This table only gathers absorption chillers driven with renewable heat, AHPs, and AHTs. Unfortunately, information related to case studies of AHPs or AHTs in food industries is not common. Among all projects reported by Cudok et al. [179], only 9.3% of the projects were applied to the food industry, while 62.8% were applied to chemical applications.

3.3. Waste Heat Recovery

Waste heat recovery is an important source of energy consumption reduction. According to the ADEME, industrial waste heat in France represents a total of 109.5 TWh, which corresponds to approximately 22.4% of the global industrial energy demand [189]. In the UK, Simeone et al. show the importance of waste heat recovery in manufacturing sectors, as thermal processes represent 72% of the energy consumption [190]. If its potential is important in terms of energy, food processing waste heat can be difficult to recover due to low temperatures. Luberti et al. [191] show that the large majority of waste heat issued from food industries is at ultra-low temperatures under 140 °C. Only baking, roasting, and frying processes can issue waste heat at higher temperature levels. For example, most of the waste heat temperature levels for grain milling are at 65 °C, dairy processes are at 80 °C, and refining processes are at 95 °C. More important temperatures can be found in frying, roasting, and baking processes or in combustion flue gases issued from biomass or fossil fuel combustion [192]. Depending on the temperature, heat carrier, and heat destination, the most appropriate heat recovery technology can be deduced to provide the shortest payback periods.

Heat recovery technologies can be regrouped into three main categories: (i) heat-to-heat or direct heat recovery, (ii) heat-to-power, or (iii) thermoelectrical technologies. The latter is not addressed in this paper as its apparition in food industries is very rare.

3.3.1. Direct Heat Recovery

Largely dominated by firing technologies, the food industry has developed many solutions to improve combustion processes and save fuel. As highlighted by Zuberi et al. [193], in the US, the food and beverage sector has the highest rate of boiler combustion usage, with 40% of total energy demand. The boiler’s efficiency can be improved by recovering heat in the flue gases after combustion. Recovering this heat through heat exchangers to (pre-)heat water or air is the source of efficiency. Based on this principle, regenerative and recuperative burners, waste heat boilers, economizers, and air pre-heaters can provide between 3 and 20% energy savings. Table 13 presents the main technologies and improvements available. Nevertheless, this path aims at pursuing industrial heat generation through fossil fuels, which makes them easy to implement but temporary solutions.

Waste heat can also be found on other utilities, such as refrigeration systems and process outflows. It can then be interesting to recover energy targets with lower costs as facility energy production does not need to be changed. The major complexity of such a procedure is to identify the (sub-)optimal arrangements that allow the lowest payback periods. The pinch analysis methodology allows industrials to optimize heat recovery integration with heat exchanger networks when considering continuous complex and large amounts of energy fluxes during energy audits. The complexity of this exercise relies on defining the optimal arrangement from the energetical point of view, considering on-site constraints such as distance between processes, piping, and space available. This analysis is based on energy balances and heat exchanges. Researchers worked on optimizing this methodology to integrate batch process integration, energy storage, and in situ and industrial constraints. More recent research aims to develop similar tools for exergy balances and combine both energy and exergy analysis during audits [194].

Overall, heat recovery presents low payback periods compared to novel technologies with the aim of consuming less input energy. In global decarbonation and energy efficiency strategies, after doing on-site energy consumption mapping, energy and exergy optimization through heat exchanger network development are necessary before investigating novel heat generation technologies. This strategy is relevant from the economic, energetic, and environmental points of view. Table 13 and

Table 14. List and details of direct heat recovery and efficiency improvement on refrigeration units and processes.

Location, Company	Application	Technology Modification	Energy Recovered and Energy Savings [MWhth/Year and %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
Waste heat recovery on refrigeration units
France, Haagen Dazs	Pasteurization processes for ice cream manufacturing	Heat recovery compressor’s oil and flue gases	51 -	-	-	2013	[119]
France, Actalia	Pasteurization processes for milk, cream, and serum manufacturing	Heat recovery on condensation waste heat recovery of refrigeration unit	3450	710 -	0.07–0.28	2014	[174]
Waste heat recovery on processes (pinch analysis)
Danemark, Star Food Ltd.	Paté production facility	Recovery of used cleaning water	19 -		0.42	2018	[200]
Iceland, MS dairy plant	Cream and processing milk facility	Heat exchanger network construction after pinch analysis	4.4 -	- -	-	2010	[201]

report the lists and details of direct heat recovery and efficiency improvement on conventional burners/boilers and on refrigeration units and processes.

Table 13. List and details of direct heat recovery and efficiency improvement on conventional burners/boilers.

Location, Company	Application	Technology Modification	Energy Recovered and Energy Savings [MWhth/Year and %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
Regenerative and recuperative burners: Novel designs of burner nozzles
-	Biscuit, chocolate, cake, and wafer manufacturing	Furnace heat recovery on hot exhaust flue gases	44.9 -	- -	2.15	-	[195]
-	Wafer manufacturing	Baking oven heat recovery	232 4.0%	43 4.0%	-	2017	[196]
Egypt, Sana Foods	Sweets manufacturing	Heat recovery from the burner’s exhaust	113.9 3.0%	30 -	1.6	2021	[144]
Economisers and waste heat boilers: Pre-heating boiler feedwater
-	Biscuit, chocolate, cake, and wafer manufacturing	Furnace heat recovery on hot exhaust flue gases	136 -	- -	1.13	-	[195]
-	Biscuit and crackers baking	Heat recovery from natural gas kilns for hot water and steam production	5627 30%	1’128 -	1.5	2017	[197]
-	-	Heat recovery from hot exhaust flue gases for air treatment water pre-heating	53 -	1100 -	1.3	-	[198]
-	Canning applications	Heat recovery from natural gas boiler flue gases	4395 14%	894 -	2.5	2019	[197]
Air pre-heaters:
United Kingdom	Wafer baking	Single oven air pre-heating	385 6.7%	71 -	5.13	-	[199]
		Multiple oven air pre-heating	1927 34%	356 -	1.57

Table 13. List and details of direct heat recovery and efficiency improvement on conventional burners/boilers.

Location, Company	Application	Technology Modification	Energy Recovered and Energy Savings [MWhth/Year and %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
Regenerative and recuperative burners: Novel designs of burner nozzles
-	Biscuit, chocolate, cake, and wafer manufacturing	Furnace heat recovery on hot exhaust flue gases	44.9 -	- -	2.15	-	[195]
-	Wafer manufacturing	Baking oven heat recovery	232 4.0%	43 4.0%	-	2017	[196]
Egypt, Sana Foods	Sweets manufacturing	Heat recovery from the burner’s exhaust	113.9 3.0%	30 -	1.6	2021	[144]
Economisers and waste heat boilers: Pre-heating boiler feedwater
-	Biscuit, chocolate, cake, and wafer manufacturing	Furnace heat recovery on hot exhaust flue gases	136 -	- -	1.13	-	[195]
-	Biscuit and crackers baking	Heat recovery from natural gas kilns for hot water and steam production	5627 30%	1’128 -	1.5	2017	[197]
-	-	Heat recovery from hot exhaust flue gases for air treatment water pre-heating	53 -	1100 -	1.3	-	[198]
-	Canning applications	Heat recovery from natural gas boiler flue gases	4395 14%	894 -	2.5	2019	[197]
Air pre-heaters:
United Kingdom	Wafer baking	Single oven air pre-heating	385 6.7%	71 -	5.13	-	[199]
		Multiple oven air pre-heating	1927 34%	356 -	1.57

3.3.2. Heat-to-Power Thermodynamic Cycles

Due to limited waste heat temperatures, heat-to-power technologies are not the most promising solutions for the food industry. The basic principle is the opposite of heat pumps, as heat is consumed to produce potential energy in the form of pressure to drive a turbine in a closed loop circuit illustrated in Figure 23. Unlike conventional power plants that issue pressure from steam production based on fossil fuel combustion, thermodynamic cycles are used to recover waste heat from processes and convert it into electricity. The most common cycles are the Organic Rankine Cycles (ORC), the Kalina cycle, and the steam generation cycles. Their performances are based on the reverse principle of heat pumps working between heat sinks and sources, inducing that their efficiency will further increase when the temperature difference between both temperature sources is high. Figure 24 shows the relatively limited performance of such systems, especially in standard conditions where the heat sink is the environment at 25 °C and waste heat recovery temperature is around 90 °C, giving an efficiency of 7.1% in optimistic conditions. In comparison, using similar waste heat recovery to feed the heat source of a HTHP for steam production at 150 °C would increase its COP from 1.35 to 2.82.

It is more current to find such heat-to-power conversion units in industries where process temperatures are over 500 °C. Its standalone operation in the food industry does not allow sufficient profitability to be installed. Nevertheless, the higher potential is highlighted for reversible HTHP to ORC systems as their operability only depends on the capacity of inverting the compressor-as-turbine and having an ORC pump in by-pass of the discharge valve as presented in Mateu-Royo et al. [202]. Such an arrangement can provide the ability of industrials to increase their load-shifting capability to eliminate peak hour electricity consumption and use the machine in ORC mode, if economic conditions are interesting, to generate additional revenue. It allows for an increase in installation profitability and to increase energy grid synergies and flexibility.

Figure 23. Equations and ORC module integration principle [203].

Figure 23. Equations and ORC module integration principle [203].

Figure 24. Efficiency evolution as a function of waste heat temperature and heat sink temperatures for an ORC with 40% exergy efficiency.

Figure 24. Efficiency evolution as a function of waste heat temperature and heat sink temperatures for an ORC with 40% exergy efficiency.

Industrial processes represent 33% of the available heat recovery potential, while recovery on primary engines represents 64% of this potential. Among industrial processes, the food industry is not present in the top 10 applications where ORCs are the most installed. Nevertheless, the market dynamic is interesting as since 2016, the number of plants and installed capacity, respectively, increased by +200% and +36% [204].

Table 15. List and details of industrial implementations of heat-to-power solutions.

Location, Company	Application	Previous Technology	Novel Technology	Power, Efficiency, Tsource/Tsink [kWe, -]	Electricity Produced and Electricity Savings [MWhth, %]	Annual CO2 Emissions Savings [tCO2/Year and %]	Payback Period [Years]	Year	Ref.
Italy, Cereal Docks	Heat recovery from diesel engines used in grain and oilseed processing	-	ORC	600 - -	- -	-	-	2012	[205]
United Kingdom, Pilot scale unit	Heat recovery on wafer baking oven	-	Regenerative ORC working with R245fa	23 13.8 165/-	- 6	-	15.7	-	[199]
France, Laiterie Saint Père	Dairy industry, implementation pre-study	Fossil fuel (natural gas, fuel oil, propane) boilers	ORC module coupled with novel wood biomass boiler	500 - -/-	- 16	4140 42%	10–15	2021	[206]

shows some industrial implementations of heat-to-power solutions.

3.4. Thermal Energy Storage (TES)

Providing solutions for efficient and cost-effective thermal energy storage (TES) is a key milestone for energy efficiency and decarbonation of the industry. As electricity storage is fundamental for the electrification of our societies, renewable heating, and cooling storage methods are also fundamental for the energy transition. Storage capacity, charge and discharge power, losses, and cost per unit of storage are fundamental to characterizing storage technologies. They can be regrouped into three different categories illustrated in Figure 25 and further explained hereafter. The interest in thermal storage relies on load-shifting and renewable energy integration, which both have environmental and economic interests, as during peak electricity demands, energy is expensive and intensive in terms of carbon footprint.

3.4.1. Sensible Heat Storage

Sensible heat can be stored in liquid or solid materials. It usually offers the cheapest and easiest form of storage. As illustrated in Figure 25, the storage capacity is proportional to the temperature difference obtained in the thermal storage medium. Therefore, industries need sensible storage either with important volumes or important temperature differences. Available technologies can be regrouped into four categories: (i) water/steam tank, (ii) solid-state or packed beds, (iii) molten salts, and (iv) underground thermal energy storage. The principle relies on storing heat as a difference in temperature in a medium that can be either a fluid or a solid. Biglia et al. [207] show the key role played by a sensible water thermal storage tank for multisource energy production in a chocolate industrial facility. In Hungary, Kall Ingredients is about to install, in 2024, a 56 MWh molten salt sensible heat storage capacity to provide 30 GWh annually. This will allow the corn processing manufacturer to avoid emissions of 7400 tons of CO₂ annually by allowing the replacement of part of the natural gas load by intermittent renewable electricity production [208].

For sensible heat storage below 100 °C, water tanks and underground thermal energy storage are the best technologies available and can be found in large stores, ground stores, and acquifer stores. From 100 °C up to 300 °C, steam accumulators, thermal oil, and molten salts are more common [209]. Seyitini et al. [210] present a review of solid-state storage materials characterized by several characteristics, such as their thermal diffusivity, conductivity, specific heat, density, and stability. Those systems are well demonstrated even at large scale, are relatively inexpensive, and can cover the largest range of temperatures. The main concern about sensible heat storage is its low density. As the energy stored is proportional to temperature difference and specific heat, extracting low-temperature differences to keep high efficiencies at the heat exchangers implies using important storage volumes. To increase the storage density and thermal properties of heat exchanges, some technologies include solid particles in suspension in the liquid medium. Another concern regards the non-constant temperature of the storage during the charging and discharging phases.

3.4.2. Latent Heat Storage

The principle of latent heat thermal storage relies on using a phase change material (PCM) to store or deliver heat at a constant temperature. Latent heat is usually charged in a PCM storage from solid to liquid and is discharged from liquid to solid. Latent heat can use organic or inorganic material and offers a much denser form of storage. Water latent heat absorbs 334 kJ/kg while its isobaric specific heat absorbs 4.19 kJ/(kg.K). It is, therefore, five times more compact to store heat in such a storage than in a sensible storage with a 15 K temperature difference.

PCMs are classified according to four characteristics determined by their (i) chemical nature, namely organic, inorganic, or a eutectic mixture; (ii) phase change temperature; (iii) phase transition type, namely solid/liquid, liquid/gas, solid/gas; and (iv) shape stabilization at nano, micro, or macro scale [211]. Fallahi et al. [212] provided an extensive review of PCMs, classifying them by their temperature range and volumetric phase change enthalpy, presented in Figure 26. For cold thermal storage, water–salt/glycol and clathrates are available materials, while for hot storage, up to 200 °C, paraffins, salt hydrates, sugar alcohols, polymers, organics, and organo-metallics materials can be selected. Besides the temperature and heat storage density, PCMs are characterized by other aspects such as their thermal conductivity to enable fast charging and discharging phases, their reversible freezing and melting abilities, their non-corrosive properties in regards to storage tanks and heat exchangers, their safety, and non-toxicity [211].

In addition to PCM choice, the heat exchanger design is particularly important in latent heat storage due to the usually low thermal conductivity of PCMs and the apparition and thickness increase of the solid phase on the heat exchanger surface. The thermal resistance induced by the solid layer thickness leads to novel techniques to enhance the heat transfer or detach the solid layer of the heat exchanger surface. Different techniques have been adopted to counter this phenomenon, with finned tube bundles heat exchangers [213,214], accumulators [215,216,217], composite PCMs coupled with graphite for thermal conductivity enhancement [218,219], metal foam-based composites [220,221], or macro encapsulation [222,223].

Another possibility to keep low thermal resistance is to remove the solid layer at the heat exchanger surface when its thickness starts to be penalizing. It can be achieved by mechanical scrapping, hydro-jets, chemical surface treatment (anti-icing), or hot stream counter-flow punctual de-icing [224].

3.4.3. Thermochemical Heat Storage

Thermochemical storage transforms heat into a material chemical bond that depends on the temperature required. Like latent heat storage, it has the advantage of storing heat at constant temperature through the operation of coupled endo/exothermic chemical reactions. Like in ab/adsorption systems, working pairs exist and allow for the achievement of more compact and durable thermal storage. Unlike latent and sensible heat storage, the potential energy is chemically stored and is not degraded with time. However, thermochemical thermal storage is a more expensive technology than latent and sensible heat storage [225].

More specifically to the food industry, the durability capacity of thermochemical energy storage is not a key aspect and does not justify the price increase, as storage duration never exceeds two or three days, which remains a quite short time span that would not affect the performance of a sensible or latent heat storage.

4. Discussion

If all the technologies presented in this review can have a positive impact on energy efficiency and decarbonation of process heat in the food industry, their implementation in facilities depends on their economic attractiveness. For this, the Levelized Cost Of Heat (LCOH) metric can furnish essential information to choose the most appropriate solution for a specific process and working environment (Figure 27). Based on many different techno-economical parameters, it is complex to gather uniformized information for all processes and solutions listed before. Figure 27 presents the equation of LCOH and the parameters and variables involved in the calculation.

Table 16. Levelized Cost of Heat for steam generation at 120 °C.

	LCOH [€/MWhth]	References
Gas boilers	105–341	[50,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224,225,226]
CHP—Co/Trigeneration units	85–102	[226,227]
Biofuels	153–162	[227]
Green Hydrogen boilers	Not found	Not found
Electric immersion heaters	247	[135]
Solar thermal	119–309	[135]
Geothermal	N.A.	N.A.
HTHP 60 °C	75–190	[135]
HTHP 80 °C	56–143	[135]
HTHP 100 °C	38–89	[135]
AHT 80 °C	25–42	[135]
AHT 100 °C	21–36	[135]

and Figure 28 show a collection of LCOH values reported for steam production at 120 °C. Among all technologies reported, AHTs have the lowest LCOH and are also less dependent on the waste heat temperature level than HTHP. The only disadvantage of AHTs is the low flexibility in terms of temperature and the non-feasibility of providing heat at 120 °C if waste heat temperature is lower than 80 °C. For HTHP, the LCOH is very different depending on the country’s gas-to-electricity ratio. The calculations made by Paya et al. [135] include waste heat sources at 100 °C, even if those levels of waste heat temperatures are not common. Then, solar thermal technologies also provide a wide range of LCOH between 119 and 309 €/MWh_th. This range is mainly due to geographical locations and their respective variations of DNI and solar ratio. These are the parameters that affect the economic attractiveness of solar thermal technologies. For combustion processes, CHP units are the most interesting, with a LCOH between 85 and 102 €/MWh_th. Depending on the country considered, gas boilers or biofuels are more interesting, with respective LCOH values between 105 and 341 MWh_th for natural gas boilers and between 153 and 162 MWh_th for biofuel units. Finally, for solar thermal and green hydrogen boilers, the literature did not provide comparable situations to include the technologies in the analysis. An in-depth study should be carried out to calculate all LCOH based on similar scenarios.

A great answer to this technological issue is to provide hybrid solutions. The concept of renewable polygeneration systems has gained popularity in the scientific literature. Such research integrates different forms of energy production and conversion methods to cover heating, refrigeration, electricity, and water demands with multiple inputs and technologies. This multiplicity of inputs and outputs highlights the major role that needs to play thermal storage systems to reduce the installation complexity and improve its reliability. This concept is further developed under “smart industries” with the inclusion of advanced control/command tools that are able to predict all non-pilotable energy production sources, adjust different kinds of energy storage, and even adapt the production capacity. Giodarno et al. [228] present a numerical study evaluating the complementarity of solar, geothermal, and biomass technologies through different configurations and storage options. The ENOUGH Demo 2 project [229,230] is also a great example of such advanced implementations in a dairy process in Norway.

Still, to be reminded, this paper only focused its attention on food manufacturing steps. When considering the entire food chain from “farm-to-fork”, the dietary patterns are not to be neglected as they provide the most significant impact on energy efficiency and decarbonation potential. Burke et al. [231] reviewed five dietary patterns in Europe and North America, considering all the alimentation life-cycle analyses. It showed that shifting from high-meat and omnivorous diets towards vegetarian and vegan diets would decrease high-income countries’ GHG emissions by between 42 and 86%. Therefore, working on both diet patterns and technological improvement is a necessary combination to reach today’s objectives of remaining in an environment suited for our human societies.

5. Conclusions

Agri-food industries are expected to take action regarding their energy efficiency and decarbonation planification. From an economic point of view, with incoming legislation, high energy prices, and brand image, this planification needs to be included in a comprehensive approach that includes reducing needs, improving the efficiency of processes, and adopting novel technologies. The following conclusions can be addressed regarding the refrigeration:

• The refrigeration sector presents an important consensus on technological options with refrigeration units and absorption chillers;
• Actions taken on the refrigerant replacement have been a great success with the successive and still ongoing phase-down of CFC and HFC fluids;
• Ammonia, CO₂, (iso-)butane, and propane refrigerants seem to be the most promising fluids for tomorrow’s refrigeration systems;
• Improved leakage detection systems will significantly decrease direct emissions in the upcoming years;
• Energy efficiency improvements on unit’s components and penetration of renewable sources on the electricity mix will keep decreasing indirect emissions of refrigeration units;
• More advanced decarbonation options are still in the very early stages of development, such as radiation and passive cooling could provide increased performances, but their complex industrialization and economic interest could limit their market penetration.

From the heating point of view, this study showed a large panel of solutions available for industrials. If it is admitted that fossil fuels need to be progressively replaced, all the solutions mentioned provide different interests from the complexity, efficiency, decarbonation, or economic point of view. Regarding heat generation, waste heat recovery, and thermal storage, the following conclusions can be drawn:

• Biomass represents a great interest due to its operation similarity with fossil fuels and acceptable decarbonation potential and payback periods, but it only provides solutions when on-site biomass is made available by the product process;
• Solar thermal systems also show great potential for decarbonation due to their renewable aspect but suffer from long payback periods and a lack of reliability that requires them to be implemented in addition to existing systems;
• The most promising technologies for the agri-food sector are heat pumps and high-temperature heat pumps. With short payback periods and important decarbonation potential, their market penetration should be important in the upcoming years, thanks to their recent technological maturity;
• The only limitation highlighted in the reviewed articles is the requirement of waste-heat sources to provide sufficient COP and ensure interesting payback periods. Temperature lifts between 40 and 80 K limit today’s technologies;
• The development of heat pumps with greater efficiencies reaching temperature lifts of 120 to 160 K could provide an extremely interesting option to industrials, as they could cover the major temperature range between −20 °C and 140 °C;
• The LCOH analysis comes in addition to the review of payback periods, as it allows us to compare technologies on similar applications. This analysis presented AHTs and HTHPs as the most interesting technologies;
• The solutions with shorter payback periods are waste heat recovery applications. However, their decarbonation potential remains very limited;
• Storage technologies, possibly coupled with the emerging concepts of “polygeneration systems” and “smart industries”, illustrate the need for complementarity of existing technological solutions;
• Even if technological solutions present a great improvement potential from the economic and emissions points of view, considering all steps, from farm to fork, it appears necessary to redesign the agri-food industry, especially the society diets that provide an incomparable impact on the entire food chain GHG emissions.