Energy-Efficient Technologies and Strategies for Feasible and Sustainable Plant Factory Systems

Abstract

The challenge of meeting the increasing global food demand has driven a shift toward controlled-environment agriculture, particularly in plant factories. However, the high energy consumption associated with these systems raises concerns about their long-term sustainability and economic feasibility. A comprehensive review was conducted to identify existing and potential technologies and strategies that can enhance the energy efficiency of plant factories. Data regarding environmental conditions, energy efficiency, water efficiency, and space efficiency were also extracted to facilitate comparison across studies. Findings indicate that optimizing crop yields and reducing energy consumption are key to improving the efficiency of plant factories. These can be achieved by integrating advanced environmental control technologies, energy-efficient system designs, modular plant factory configurations tailored to local climatic conditions, and effective management practices. While adopting renewable energy alone is insufficient to meet total energy demands, it significantly reduces energy costs and carbon emissions. Furthermore, strategically integrating plant factories with other industries will promote the efficient use of residual resources, fostering a circular economy and enhancing resource efficiency within plant factory systems and the broader economic framework. The insights provided in this review will contribute to developing sustainable and economically viable plant factory systems, supporting future advancements in controlled-environment agriculture.

1. Introduction

There is a growing interest and investment in controlled-environment agriculture (CEA), particularly in plant factories (PFs) [1]. In PFs, crops are grown entirely enclosed and isolated from external conditions. Internal variables such as light, temperature, humidity, water, nutrients, and CO₂ concentrations are precisely controlled through automation [2,3,4,5,6,7]. This level of control is made possible through the application of artificial intelligence (AI) and the Internet of Things (IoT), which enable real-time, high-precision monitoring and intelligent adjustments to create optimal growing conditions.

Being protected from external conditions and having control over environmental factors in the growing chamber provides several benefits: it reduces disease risk, boosts crop growth potential, improves resource efficiency, shortens the growing cycle, and enables year-round production across seasons, leading to high annual yields and strengthening food security [8,9]. Additionally, by utilizing vertical farming techniques, the PF system enables high-density production in limited spaces, where crops are cultivated on multilayer platforms. This arrangement produces more biomass per unit of growing area than traditional open-field (OF) agriculture and greenhouse (GH) systems [8,9,10,11,12,13,14], making it ideal for regions with limited arable land. Furthermore, it can be recognized that urban areas are increasingly adopting this approach, enhancing food availability in cities, lowering transportation costs and carbon footprints for input materials and produce, and encouraging resource circularity for a more sustainable agricultural model [7,15,16,17]. PFs can be established in unused old buildings [18], integrated into residential areas [19,20], or incorporated into existing buildings of other industries [21,22] to promote the circular economy and create energy synergy, thereby reducing the environmental impact of the entire system. Additionally, urban farming through PFs can potentially engage younger generations in agriculture by offering a modern, technology-driven approach designed for urban settings.

Plant factories are highly efficient in terms of space and water. They require only 20 m⁻² of crop area and 2.4 m⁻³ of water to meet the annual protein needs of an adult, compared to 164 m⁻² and 111 m³ in traditional outdoor farming [12]. Despite the advantages of production performance and resource efficiency, one of the main limitations of plant factories is their high energy demand, arising from their isolation from free energy in the external environment. This significantly raises production costs, making it impractical for some crops [11,13,23,24,25]. Until now, crops grown in PFs have been primarily fast-growing leafy vegetables, especially lettuce (Lactuca sativa) [26,27,28,29,30], basil (Ocimum basilicum L.) [18,31,32,33,34,35], and crops under the families of Brassicaceae and Amaranthaceae [36,37]. Soybean was also grown in PFs, but it was not found to be cost-effective despite the increase in yield [12]. To diversify PF production and provide various nutrient sources, staple crops such as wheat [38] and potatoes [39] have been suggested for cultivation. However, based on the estimates of Kobayashi et al. [40], their economic feasibility faces limitations due to a long production cycle, leading to high energy requirements of at least 435 and 62 kWh kg⁻¹ (kilowatt-hour per kilogram), respectively, compared to 7 kWh kg⁻¹ for lettuce [40]. Also, staple crops have relatively low prices, making competition more challenging.

Accordingly, PFs use 2 to 20 times more energy than open-ventilated GHs [14], or requiring up to 251% more energy to produce per kg dry matter (DM) compared to semi-open GHs [24]. Nevertheless, PFs are considered more feasible than GHs in extremely dark and cold areas. Energy is the most significant cost contributor, accounting for up to 58% of the total production cost [5]. Numerous electricity-driven systems are necessary for regulating environmental conditions and monitoring crop health, making energy efficiency a critical challenge. The lighting system is the most energy-intensive operation, accounting for over 70% [5] of the total energy usage since artificial lighting is usually provided at least 11 or 16 h a day to optimize crop growth [5,7,29,41]. To tackle this issue, researchers are developing energy-efficient lighting solutions, particularly emphasizing light-emitting diode (LED) technology [42,43]. LEDs are highly durable and energy efficient, producing less heat than other artificial light sources, including incandescent, fluorescent, and high-pressure sodium lamps [44]. Additionally, unlike other light sources, LEDs provide a significant advantage in their capacity to be customized [43,45] to optimize the spectral composition, hereby enhancing plant productivity and system energy efficiency [46,47,48]. This can be achieved efficiently through deep learning techniques [49]. Nonetheless, additional efforts are needed to enhance the system’s energy efficiency and ecological sustainability. To increase the profitability of the PF, products can be marketed at a premium for their quality. However, agriculture fundamentally aims to provide high-quality food at affordable prices.

Several studies suggest that PF systems and other CEA farming methods are more sustainable than conventional farming because of their resource efficiency and smaller carbon footprint in the production area [10,24]. However, from a macroscopic perspective, PFs have a greater carbon footprint potential than PF agriculture, mainly due to their dependence on electricity, which is frequently generated from fossil fuels [7,18,50,51], and the energy-intensive production of materials used in PFs [15,18,52]. Furthermore, PFs can boost urban electricity demand, putting pressure on energy production systems and transferring carbon emissions from food production to electricity generation [16]. An assessment by Blom et al. [10] from cradle to grave found that vertical farming has a carbon footprint 16.7 times greater than conventional farming, measuring 8.177 kgCO_2-eq kg⁻¹ (kilogram CO₂ equivalent per kilogram) compared to 1.451 kgCO_2-eq kg⁻¹ in a hydroponic GH, 1.211 kgCO_2-eq kg⁻¹ in a soil-based GH, and 0.490 kgCO_2-eq kg⁻¹ in OF; 85% of these come from electricity production. This is concerning because greenhouse gas (GHG) emissions are associated with global farming, further complicating our food production [53,54].

As previously discussed, the current PF systems are facing challenges in providing sustainable solutions for growing food crops to meet global food demands due to their dependence on energy. Therefore, to enhance sustainability and economic feasibility, PFs must reduce energy consumption or improve energy use efficiency (EUE), or both, and explore alternative energy sources. This paper provides a comprehensive review of existing technologies, methodologies, and strategies to enhance the energy and resource efficiency of PFs. This review presents essential information to guide the development and planning of future PFs, focusing on maximizing economic benefits and improving sustainability. This paper is structured into eight sections: (1) Introduction, (2) Methodology, (3) Overview of Reviewed Studies, (4) Optimization of Crop Yield, (5) Optimization of the Energy System, (6) Energy-Efficient Management Practices, (7) Utilization of Renewable Energy and Residual Resources, and (8) Conclusion and Recommendations.

2. Methodology

A systematic review method was applied in this study. The search focused on research articles and conference proceedings published between 2014 and 2024 (access date: 16 October 2024) from four major online databases, including SCOPUS, Web of Science, IEEE Xplore, and Science Direct, based on the topic, title, abstract, and keywords. The following keywords were utilized: ‘Plant Factory’, ‘Vertical Farm’, ‘Indoor Farm’, ‘High-tech Greenhouse’, ‘Automatic Greenhouse’, ‘Urban Farm’, ‘Aeroponic’, ‘Hydroponic’, and ‘Skyscraper Farm’. Initially, keywords like ‘Energy’ and ‘Energy Efficiency’ were included in the search string. However, very few documents were retrieved, so these keywords were still considered during the screening process. The information from the retrieved documents was exported in CSV or Microsoft Excel format, then combined and saved in Microsoft Excel to facilitate the identification of relevant documents.

There were 2426 available scientific studies found in online databases. These were manually checked for duplication, document or article type, and relevance (Figure 1). Duplicate titles and DOIs were identified using a filter and equation and were removed. The compiling of the search was limited to research articles, case studies, conference proceedings, brief reports, and methodology articles. Review articles and books were excluded. The relevance screening was conducted based on the documents’ titles, keywords, and abstracts. Documents unrelated to ‘Energy’, ‘Energy efficiency’, and indoor farming, PFs, and vertical farming were excluded. Finally, the PDF copies of the remaining records were downloaded and reviewed in full text. Out of 2426 searched documents, 108 were included for comprehensive review.

3. Overview of Reviewed Studies

Most of the studies reviewed (75% = 81/108) were published between 2021 and 2024 (Figure 2a), demonstrating a growing interest among researchers worldwide in enhancing the energy efficiency of vertical or CEA farms. Almost half (48.15%) of the studies were primarily conducted in Europe (n = 52), particularly in Italy (n = 11) and the Netherlands (n = 10) (Figure 2b). This emphasizes the role of European nations as leaders in enforcing environmental regulations for sustainability. Furthermore, the Netherlands is acknowledged as the global leader in agricultural innovation [55]. We also noticed an increasing interest in Southeast Asia (n = 20), especially in Singapore (n = 7), compared to East Asia (n = 8). Despite having less arable land, Singapore’s aim to boost food self-sufficiency by 30% by 2030 may be the main factor driving this increase, leveraging CEA technologies [13]. Additionally, 17 studies (16.67%) were conducted in North America, including 11 from the US and 6 from Canada. We also noted a few studies from Oceania (Australia = 4), South America (Mexico = 1 and Chile = 1), and other regions (India = 2, Iran = 2, and Morocco = 1).

The primary solutions for reducing energy costs in vertical farming involve integrating renewable energy, enhancing lighting efficiency using advanced lighting technologies and effective lighting management, and implementing automation for environmental control. These solutions were addressed in over 90% of the studies reviewed (Figure 3). Additionally, the incorporation of vertical farms into urban systems and the use of residual waste from other industries to promote a circular economy were also discussed. However, a limitation of our review is that most of the study results (55.56%) were generated from simulations or modeling. Only 25.93% of the studies were experimental, primarily conducted in small-scale vertical farms or growth chambers, focusing mainly on lighting quality and management evaluations. The remaining studies presented proposals and prototypes of technologies deemed applicable to energy efficiency on vertical farms.

4. Optimization of Crop Yield

One basic strategy to reduce production costs is to increase yield by maximizing a plant’s growth potential. This can be accomplished by providing optimal environmental factors essential for plant growth, such as light, temperature, humidity, water, nutrients, and CO₂ (Figure 4). Insufficient provision of these factors can restrict plant growth, leading to lower yields. Conversely, excessive provision is unnecessary, may not enhance growth, could have adverse effects, and will result in wasted resources, thereby increasing production costs.

4.1. Light Quality and Quantity

Plants are photoautotrophic, and light is a limiting factor in their growth. The light-dependent photosynthesis reactions occur in chloroplasts’ thylakoid membranes and depend on photosynthetically active radiation (PAR) that ranges from 400 to 700 nm [56]. The quantity of PAR, measured as photosynthetic photon flux density (PPFD), and the total amount of PAR received throughout the day, expressed as the daily light integral (DLI), are critical for maximizing photosynthetic activity [35]. However, photons in the PAR are not absorbed evenly in the thylakoid membrane; blue (400–500 nm) and red (600–700 nm) wavelengths are absorbed more, while the green (500–600 nm) wavelength is absorbed less [57,58,59]. Chlorophylls (a and b) and carotenoids are the most abundant pigments in most terrestrial plants and are critical for light harvesting [58]. Red light is absorbed efficiently by chlorophylls, providing steady energy for electron excitation crucial for sustaining photosynthesis. The photochemistry of photosystems I and II is highly excited at 700 and 680 nm [28]. Blue light, by contrast, enhances photosynthetic efficiency, supplying high-energy photons essential for electron excitation and influencing plants’ photomorphogenesis responses, including growth patterns, stomatal opening, and chloroplast movement [60,61,62]. PAR can be provided using white LEDs [36,37], which offer broad light spectrums that support various physiological processes in plants beyond photosynthesis, including flowering in saffron [63]. However, the spectral composition of white light varies depending on the types of LEDs or light sources [36,46,64], which may affect crop growth. However, LEDs can be customized to provide only the most important PARs [28,30,32] but must not be given solely for maximum efficiency and to prevent photo-oxidative stress at high levels. Therefore, the spectrum ratios must be considered to maximize photosynthetic efficiency while ensuring crop health.

The ratio of red to blue (R:B) used in the literature reviewed ranges from 0.18 to 10.75, depending on the crop species and growth stages; however, some studies maintained the same ratio throughout all stages (Table S1). Requirements for lighting during the germination period vary across different sources; some provide guidelines [18,27,28] while others do not [29,37]. Furthermore, Ptak et al. [29] suggested a ratio of 0.18 during the harvesting period. However, a wider ratio (higher red levels) should be provided during the growing period to enhance photosynthetic activity [27,28,29,30]. This is supported by the findings of Righini et al. [12], where soybean yield increased by 16% with R:B 3.18 compared to R:B 1, although crude protein increased by up to 13% in R:B 1. Similarly, lettuce yield improved by 34% at R:B 3 compared to R:B ≤ 1 [61] and maximum yield was observed at the same R:B ratio in chicory (Cichorium intybus L.) [18]. However, the maximum yield and lowest specific electricity consumption (kWh kg⁻¹) were observed at higher R:B (4) in rocket (Eruca sativa Mill.) but lower in basil at R:B 2 [18]. Narrow R:B ratios ≤ 1 during the growing period significantly reduced yield despite a significant increase in specific nutrient solution uptake [18]. The increased uptake of nutrient solutions may be attributed to higher transpiration rates induced by elevated stomatal conductance under intense blue light conditions [61], resulting in poor water and nutrient utilization efficiency. Transpiration is an essential process for cooling and transporting water and nutrients from roots to leaves and must be balanced to maximize growth and control humidity inside the building [24]. Furthermore, it should be noted that increasing blue light in proportion to red in the luminaire would proportionally increase energy consumption in watts/hour (W/h) since the conversion of electricity to blue light is low [47], also a critical factor to consider when optimizing spectral composition.

Far-red light (700–750 nm) is poorly absorbed by chlorophylls a and b, but it has significant physiological effects on plants, such as shade avoidance responses that increase leaf area and stem length [36], which can be used to manage plant architecture and induce flowering [65,66]. Adding 52 µmol m⁻² s⁻¹ far-red light on top of blue and red LED light or 19.26% of total PPFD increases leaf expansion, which improves light utilization efficiency (LUE) by 8–23% and plant dry weight by 46–77% [28]. Additionally, 15% to 35% far-red light increases leaf area by 95% to 173%, respectively, which is ideal for leafy vegetables like lettuce [27]. While all far-red levels boost crop yield, 35% far-red light diminishes chlorophyll and dry matter content. In the study of Boucher et al. [36], lettuce with 31% far-red light exhibited high water content. The reduction in DM content might be attributed to high water retention caused by low stomatal conductance, which led to reduced transpiration. This finding suggests that far-red light may counteract the effects of blue light on stomatal conductance. Furthermore, this finding implies that far-red light can be utilized to manage water usage and the dehumidification energy demand of the HVAC (heating, ventilation, and air conditioning) system by regulating transpiration.

Furthermore, ultraviolet (UV) light (100–400 nm) was shown to increase leaf area and biomass of plants [67], indicating its importance in plant biology. A deep learning model generated an optimal level of UV light of 23.96 µmol m⁻² s⁻¹ or 6.4% of the total PPFD (374.49 µmol m⁻² s⁻¹) for maximum growth and yield of choy sum (Brassica rapa var. parachinensis). Furthermore, UV light is being used to promote and increase secondary metabolite production in herbs and medicinal crops [68,69], which can also be used to enhance the flavor and quality of food crops [70]. However, the harmful effects of UV light on plants and humans must be carefully considered for its integration. Alternatively, increasing the switching frequency of blue LEDs to 850 kHz significantly enhanced the total antioxidant capacity of lettuce by boosting the levels of caftaric acids, chicoric acids, isoquercetin, and luteolin [71]. Pulsed LED lighting may not fully replace UV light needs, but it can serve as a safe and energy-efficient alternative for enhancing crop quality.

In addition to light quality, light quantity measures in PPFD and DLI are critical factors for maximizing photosynthetic activity for growth and yield. Lettuce yield increases with increasing PPFD from 100 to 300 µmol m⁻² s⁻¹ [35], and even up to 750 µmol m⁻² s⁻¹ [26]. This method is more effective than increasing CO₂ levels, resulting in a yield of 0.54% to 0.76% with a 1% increase in PPFD [5]. This indicates that light quantity is the most limiting factor for yield optimization. Increasing PPFD would also shorten the growing period needed to achieve a harvest weight of 250 g of lettuce from 26, 20, and 18 days at 200, 400, and 750 µmol m⁻² s⁻¹ [26]. It was also found that PPFD and temperature are crucial factors in maximizing crop productivity and minimizing energy consumption [26,72,73,74]. However, the increase in yield is not linear with increasing PFFD, despite CO₂ concentration being set at 1200 ppm [26], indicating that the plants are reaching a saturation point or that other factors are limiting their photosynthesis capacity. Kong et al. [64] found that light spectral composition (R:B ratio) is a significant predictor variable for yield. Furthermore, yield does not increase linearly beyond 500 µmol m⁻² s⁻¹; however, energy consumption increases linearly with rising PPFD [41], suggesting that careful lighting management is necessary to balance yield and costs. Section 6.1 discusses several lighting management strategies concerning energy efficiency.

Furthermore, DLI is the product of PPFD and photoperiod, and it is a more robust parameter influencing photosynthesis than PPFD alone [35]. Based on the model of Shasteen and Kacira [75], the DLI and yield show nearly linear relationships. Simulation studies showed that DLI levels of 30.24 [41] to 40.32 mol m⁻² day⁻¹ [72] were found to be optimal levels for yield and energy efficiency in lettuce. However, these findings contradict the results from empirical studies where 11.52 [26] and 14.40 mol m⁻² day⁻¹ [34,35] were found to be optimal, and this level range is typically used in the field studies shown in Table S1. Increasing DLI from 14.40 to 21.60 mol m⁻² day⁻¹ in lettuce, including basil, rocket (Eruca sativa Mill.), and chicory (Cichorium intybus L.), significantly reduced yield [34]. This adverse effect could be due to the absence of a dark period since the photoperiod was adjusted from 16 to 24 h instead of PPFD to increase DLI. Most plants require a dark period for normal physiological processes [76], and continuous light exposure at high intensity can lead to photo-oxidative damage [77], depending on cultivars and species [78]. Continuous lighting, even at 217.5 µmol m⁻² s⁻¹ PPFD (18.79 mol m⁻² day⁻¹), can cause severe leaf chlorosis in pak choi (Brassica rapa cv. ‘Chinensis) and arugula (Eruca sativa) [36]. Furthermore, we observed that crops’ response to DLI levels varies depending on spectral composition, which was not considered in the simulation studies. Pennisi et al. [34,35] utilized red–blue LED with a 3.04 ratio and showed higher optimal DLI compared to Carotti et al. [26], who utilized broad-spectrum light. Additionally, although it could be a species-specific response, milkweed (Euphorbia peplus) had the highest yield at 32.40 compared to 9.72 and 16.20 mol m⁻² day⁻¹ under LED light with 52.5% red, 20.8% blue, 22.7% green, and 4% far-red proportions [79]. Therefore, light quality, which is species- and growth stage-specific, must be adjusted first before increasing DLI to improve yield. Otherwise, the increase in energy costs will be higher than the increase in yield. Moreover, PPFD and photoperiod must be adjusted together rather than separately to meet DLI requirements and maintain the health of the crops.

The photoperiod can be extended up to 20 h, or even 24 h, while maintaining the DLI level without negatively impacting the photosynthetic capacity of crops. This can lead to significant yield improvements through lighting segmentation (for example, a 5 h photoperiod followed by a 1 h dark period per cycle) or by replacing the dark period with low light intensity (20 µmol m⁻² s⁻¹), also known as the light compensation point (LCP) [36]. Moreover, increasing the photoperiod would improve the nutritional quality of the produce. Microgreens such as amaranth (Amaranthus tricolor), collard (Brassica oleracea var. viridis), and basil exposed to continuous lighting had high levels of antioxidant and secondary metabolites such as phenolics and anthocyanins [37]. Similarly, increasing the photoperiod from 12 h to 16 h significantly increased the total phenolic content, total betalains, betacyanins, betaxantins, and total antioxidant capacity of beet (Beta vulgaris L. ssp. vulgaris) microgreens [80]. Therefore, these findings highlighted the importance of light quality and quantity for improving crop yield and quality under a controlled environment.

4.2. Temperature and Humidity

The enzyme Rubisco plays a central role in the Calvin cycle of photosynthesis, and its activity is sensitive to temperature [81]. Its effectiveness increases with temperature up to a certain point, which promotes the fixation of CO₂ into sugars [82]. The rate of photosynthesis increases until it reaches the optimal range; however, beyond that point, the rate of photorespiration increases. This process is wasteful as it uses energy and releases CO₂ [24]. When temperature is maintained, yield and CO₂ concentrations are positively correlated [75]. Most crops grown in the PFs are C3 plants sensitive to high temperatures. Therefore, it is essential to maintain optimal growing temperatures to balance photosynthesis and photorespiration rates to maximize crop productivity.

Most of the literature discusses the importance of maintaining air temperature to optimize crop yield. However, Carotti et al. [26] showed that maintaining nutrient solution temperature in a hydroponic system to maintain root temperature is also important. The optimal temperature for crops varies depending on the species, growth stage, and diurnal cycle [29]. The typical temperature settings range from 20 °C to 26 °C, as shown in Table S1. A 1% rise in temperature above the optimal level will lead to a 0.32% decrease in yield and an undesirable increase in electricity consumption [5]. Moreover, a simulation analysis in lettuce showed that annual yield decreases by more than 70% if leaf temperature exceeds 26 °C [72]. This finding aligns with the experiment conducted by Carotti et al. [26]. They observed a significant reduction in yield at air temperatures of 28 °C and 32 °C, with the lowest yield and highest incidence of tip burn occurring at 32 °C. Additionally, they identified optimal air and root temperatures for lettuce at 24 °C and 28 °C, respectively. However, the difference in root temperature effects between 24 °C and 28 °C was minimal, indicating that heating the nutrient solution may be unnecessary for energy saving. Furthermore, a simulation study by Graamans et al. [24] showed a 62% increase in dry matter production by lowering the production temperature to 19 °C. However, the energy requirement for cooling would increase and might not be economically beneficial. In soybeans, reducing the ambient temperature from 24/22 °C to 18/16 °C (photoperiod/dark period) 42 days after sowing significantly decreased yield by 28% [12].

Diurnal variations in temperature, humidity, and vapor pressure deficit (VPD) are typically adjusted to support physiological processes during both the photoperiod and dark periods [12,24,83,84]. During the photoperiod, temperatures are usually kept at elevated levels to enhance photosynthetic activity and the enzymatic processes that promote growth and development [85]. Relative humidity during this period is balanced to facilitate transpiration, which supports water and nutrient uptake and cooling. Excessive transpiration, however, increases water loss and energy demand for dehumidification [24,83]. VPD measures the difference between the amount of moisture in the air and the maximum moisture the air can hold and it is influenced by both temperature and humidity [12,84]. At high temperatures, humidity can be increased to normalize the VPD, and vice versa [41]. An optimal VPD typically ranges from 0.5 to 1.50 kPa depending on the crop [86], which aligns with most studies reviewed. VPD is crucial for achieving balance. A higher VPD can increase transpiration rates and potentially lead to water stress, while a lower VPD may reduce transpiration, impacting nutrient transport and causing diseases due to excess moisture on leaf surfaces [87]. In contrast, the environmental conditions are adjusted during the dark period to align with plants’ reduced metabolic activity. Temperatures are lowered to minimize nocturnal respiration to conserve energy and minimize the depletion of stored energy produced during the photoperiod [88], critical for optimizing yield. Simultaneously, humidity levels are often increased to minimize water loss through residual transpiration [89,90]. This adjustment in temperature and humidity lowers the VPD, reducing the air’s evaporative demand and helping to prevent dehydration and nutrient imbalances. However, overly high humidity levels must be avoided to prevent the development of fungal diseases. Most of the studies reviewed kept the humidity at levels not more than 90%, except in Ptak et al. [29], as levels beyond this make plants prone to fungal infection. Maintaining a day–night cycle with appropriate regulation of temperature, humidity, and VPD not only mimics natural conditions but also promotes healthy growth, enhances crop quality, and improves resource efficiency in controlled environments.

4.3. Carbon Dioxide Concentrations

The CO₂ concentrations in CEA farms typically range from 450 to 1200 ppm, and the requirements depend on the crop species, cultivars, growth stages, and diurnal periods, as shown in the Table S1. CO₂ is one of the inputs and limiting factors for photosynthesis [7,24]. Increasing concentrations of CO₂ promote crop growth until a saturation point is reached, at which concentration levels provide no further improvement in growth. Studies have shown that the response of plants to CO₂ enrichment is age-dependent, with a higher response in the later stages than in the earlier stages, suggesting that dynamic enrichment control can influence both yield and cost management [75].

Furthermore, increasing CO₂ concentrations increases the saturation point of other environmental parameters and vice versa. Increasing PPFD up to 300 µmol m⁻² s⁻¹ [35] or even increasing photoperiod from 16 to 24 h [34] would not significantly improve yield and resource efficiency. This contradicts the findings of Carotti et al. [26], which indicated that lettuce yield increased at 750 µmol m⁻² s⁻¹. These discrepancies may be attributed to differences in CO₂ concentrations used: 450 ppm [35] versus 1200 ppm [26]. The 450 ppm CO₂ concentration resembles the open-field conditions reported by Li et al. [7], as this level restricts the photosynthetic capacity of lettuce, even with optimized environmental conditions. Increasing PPFD from 100 to 600 µmol m⁻² s⁻¹ and a 1% increase in CO₂ concentrations would increase yield by 0.33% [5]. On the other hand, specific energy costs (energy per kg yield) and specific costs (costs per kg yield) would increase if PPFD is increased without increasing CO₂ concentrations. Therefore, optimal CO₂ concentration should be maintained and distributed evenly to maximize yield. Although the enrichment of 1200 ppm CO₂ is higher than in most CEA farms, this strategy ensures that crop growth is not hindered by CO₂ deficiency [41]. However, the cost of CO₂ enrichment constitutes the limitation that accounts for 22% of the production cost reported by Keyvan & Roshandel et al. [5]. This strategy may be more economical if exhaust air is recirculated instead of bringing in fresh air from outside since the latter results in excess CO₂ being wasted to maintain temperature and humidity. Additionally, CO₂ requirements are significantly influenced by ventilation needs in open and semi-open ventilation systems [24]. Alternatively, using CO₂ emissions from other industries, like the electricity provider [91], can decrease production costs while enhancing carbon sequestration in the areas.

4.4. Water and Nutrients

Water consumption in PFs largely depends on crop evapotranspiration, which is affected by factors such as crop species, temperature, humidity, photoperiod, and spectral composition [7,14,27,32,36,61,92,93]. Additionally, water usage is influenced by the types of irrigation and HVAC systems, where water loss is minimized in closed-loop systems [24] (see Section 5.4 and Section 5.5). Our review does not address each nutrient’s optimal combinations and concentrations for plant yield and energy efficiency. However, ensuring a continuous supply of water and nutrients in balanced concentrations at the appropriate time through the irrigation system will not hinder crop growth.

Fertilizers, available as nutrient solutions with macro- and micronutrients, are often bought commercially and mixed with water in various ratios depending on the desired nutrient concentrations [6,30,32]. Alternatively, nutrients can be sourced from organic waste, which can reduce fertilization costs [7,17,19]. Delivering nutrients via irrigation systems, commonly called fertigation, is highly effective and efficient in PFs, especially when combined with automation. Since nutrients are dissolved in water, their concentrations and availability are monitored and controlled through measurements of electrical conductivity (EC) and pH [8,28]. EC measures the total concentration of dissolved ions in the nutrient solution, and it varies by crop species for as low as 0.2 decisiemens per meter (dS m⁻¹) for milkweed (Euphorbia peplus) [79], 1.2 dS m⁻¹ for soybean [12], from 1.6 to 2.3 dS m⁻¹ for lettuce and other leafy vegetables [18,26,27,28,34,35,61,75], and as high as 2.4 dS m⁻¹ for basil [31]. However, EC only provides estimates of the overall concentration of nutrients in the nutrient solution but does not specify the concentrations of individual nutrients [94]. An irrigation system that features dosing individual nutrients into the nutrient solution would be more beneficial for improving the precision of nutrient delivery and yield.

Furthermore, certain nutrients are only soluble and available to plant roots within a specific pH range, and maintaining the correct pH ensures that all nutrients in the water solution are readily available for plant uptake [95]. The optimal pH of the nutrient solution ranges from 5.5 to 6.5 [12,18,26,27,28,31,34,35,61,75,79]. Plant roots can maintain electroneutrality in the rhizosphere by releasing hydroxide ions (OH⁻), bicarbonate ions (HCO₃⁻), and hydrogen ions (H⁺) depending on the nutrient ions absorbed [96,97,98,99]. However, releasing such ions would change the pH of the rhizosphere or the plant roots’ surroundings and the nutrient solution. This change is primarily influenced by nitrogen uptake, which is taken up in large amounts. Nitrogen is typically provided in the forms of ammonium (NH₄⁺) and nitrate (NO₃⁻), which exert opposing effects on the pH. When NH₄⁺ is taken up, H⁺ ions are released by the roots, acidifying its surroundings. In contrast, OH⁻ or HCO₃⁻ are released once NO₃⁻ is taken up, resulting in increased rhizosphere pH [98,99]. These changes in the pH can either enhance or limit the uptake of other nutrients. This highlights the importance of the composition of nutrient solutions. For better nutrient utilization, the nutrient solution should be balanced for nutrient concentrations, cations (positively charged), and anions (negatively charged).

Substrates are used to replace soil for the plants. Although they may not contain nutrients, their characteristics are essential to support the growth of plants by providing anchorage for the root system, a medium for water and nutrient exchange, and air spaces for root aeration [79,100,101]. Air spaces are crucial for providing adequate oxygen (O₂) to the root system. Furthermore, the dissolved O₂ in the nutrient solution is monitored and enhanced for the same purpose [26,75]. The commonly used substrates are coco coir or fiber, rockwool, vermiculite, perlite, and clay pebbles [30,32,61,79,100]. Vermiculite, perlite, and clay pebbles provide good aeration but offer less water retention than organic media [79,101]. This may necessitate more frequent irrigation, potentially raising energy costs. Combining these substrates with high-water-retention materials like coco coir and peat would balance this limitation [30,61]. However, substrates of organic origin may generally be cheaper and more sustainable, but they cannot be reused many times as they degrade and may attract microbial growth, which can also lead to increased maintenance needs in irrigation systems [79]. Therefore, it is essential to consider the cost-effectiveness of the substrate not only for yield but also for the overall energy efficiency of the entire system.

5. Optimization of the Energy System

Understanding the effects and interrelations of various environmental factors on crop biology is a crucial foundation for enhancing crop productivity. However, the levels of environmental factors considered optimal for maximizing growth or yield do not always result in the best financial returns due to the energy-intensive nature of PFs, where environmental parameters are carefully managed and controlled using various equipment. Therefore, it is essential to understand the interactions between environmental conditions, plant biology, and plant factory systems (such as lighting, HVAC, etc.), as shown in Figure 4. This section presents an examination of different technologies and methods aimed at energy conservation.

5.1. Structural Designs

The internal and external designs of PFs significantly impact energy efficiency by affecting energy consumption related to lighting and air quality control. The building’s interior should be designed to reduce energy loss, while the exterior (facades and roof) should be designed to minimize the impact of the external environment on the internal conditions [102]. However, the external environment, when favorable, can sometimes be utilized to save energy. Incorporating natural sunlight along with artificial lighting in hybrid PFs with transparent facades can help reduce energy needs for artificial lighting [13] and has been found to be more feasible than PFs with opaque facades due to a significant reduction in artificial lighting compared to the increase in heating and cooling demands [102]. The latter findings, however, result from simulations and are limited to regions with cold (Sweden and Netherlands) and arid (United Arab Emirates) climates. Therefore, more investigations are recommended considering varying climatic conditions. In contrast, Li et al. [7] stated that fully closed PFs are more feasible in tropical cities because of the pronounced increase in cooling demand caused by high solar heat gain in GH systems. Various technologies can be employed to reduce solar heat gain while maximizing natural light, including energy-efficient screens that also control heat during colder periods [24,74,103], insulated glass [102], and SG film ULR-80 (Solar Gard; Saint-Gobain Performance Plastics) that can block heat-generating spectrums (88% of infrared and far-infrared, and >99% of ultraviolet) [104]. However, the SG film ULR-80 also blocks 26% of red light, which can reduce yield. Incorporating transparent facades with transparent solar photovoltaic (PV) cells or dye-sensitized solar cells (DSSCs) offers a promising alternative by reducing the need for artificial lighting while generating electricity [105,106]. Nevertheless, the structural design of PFs should be adapted to the local climate to achieve optimal energy efficiency.

External conditions have a minimal impact on the internal conditions of PF systems compared to GH systems [24,93]. Nonetheless, the external temperature continues to be one of the most significant factors influencing the internal environment of PFs and determines the amount of energy required for heating and cooling for maintenance [14,72,102,107]. It is also one of the factors influencing the seasonal and daily energy consumption of HVAC systems [5]. However, the degree of variation is influenced by the types of technologies used in the HVAC system. Additionally, well-insulated buildings with low U-values exhibit better internal temperature stability, which reduces the workload on the HVAC system and lowers energy costs [14,25,102]. Increasing the U-value of facades from 0.05 to 5.75 W m⁻² K⁻¹ could reduce total energy demand by up to 12.5%, with a pronounced reduction in cooling demand by up to 31.6% [102]. The reduction in cooling demand using facades with high albedo or light reflectance is significant in hot climates. In contrast, lower albedo values are more advantageous in colder climates, as they help decrease heating needs by harnessing solar heat gain. Consequently, the building’s orientation will depend on whether it avoids (east–west) [108,109] or utilizes (north–south) [102] free energy from the sun. Additionally, the size of PFs is essential for energy efficiency since smaller sizes have a higher surface area and undergo more significant heat loss or gain, necessitating more energy to maintain optimal conditions [73]. However, building size has a lesser impact in hotter climates with high insulation quality [102]. Nevertheless, size must be considered when planning to establish PFs, paying attention to lessening operational costs.

5.2. High-Density and Multi-Cropping Operations

PFs are cost-intensive systems for growing crops because of the provision of artificial environments. PFs with smaller sizes have higher specific costs compared to larger sizes [73]. Therefore, increasing yield is a strategy to dilute the operation costs. Increasing plant density can directly increase the farm yield [24,27]. However, several factors need to be considered when exploring this approach. Light availability and distribution are crucial elements that must be considered. Canopy-intercepted PPFD (CI PPFD) measures the PPFD absorbed at the canopy level, directly correlating with photosynthesis rates and light use efficiency (LUE) [110,111]. At low plant densities, the leaf area index or LAI (the ratio of the total leaf area to the ground area) is low, and the CI PPFD is low, indicating that more light reaches the ground and is wasted (photon loss) instead of being utilized by the plants [28]. Increasing plant density results in more leaves available to capture light, which raises LAI and enhances CI-PPFD by improving light interception. However, very high plant densities can cause self-shading, which reduces light distribution and intensifies competition for other resources such as water, nutrients, and CO₂, leading to low individual weight and poor uniformity [27]. Moreover, increasing density would proportionally raise operational costs for humidity control, ventilation to avoid microclimate build-up, and resource delivery (nutrient solution and CO₂) and elevate the risk of diseases. Consequently, the design and capacity of the PF and its environmental control systems must be considered to prevent overloading that could reduce efficiency.

Utilizing modular towers can enhance PFs’ production capacity. With mobile towers, aisles for human access become unnecessary, maximizing production space. Ptak et al. [29] proposed a design featuring rotating towers suspended from the ceiling with sliding mechanisms that transport plants between two chambers with distinct environmental conditions: a ‘day’ chamber and a ‘night’ chamber. The day chamber maintains fixed temperature and humidity levels, with the lighting system operating continuously for 24 h. In contrast, the night chamber has a fixed temperature, variable humidity, and no lighting system. Multiple chambers can be added to accommodate the needs of crops at different growth stages. This system ensures that the lighting and HVAC systems operate at a constant load, minimizing inefficiencies and reducing wear on system components caused by frequent adjustments to variable settings. Additionally, rotating the towers three times a day, based on a 16 h photoperiod (16 h in the ‘day’ chamber and 8 h in the ‘night’ chamber), helps prevent mold growth, reduces shading, facilitates more manageable plant condition monitoring through computer vision, and enables the potential automation of harvesting. However, one possible limitation of the system is its flexibility regarding the cultivation cycle, as the number of towers is fixed based on the predefined cultivation cycle. For instance, in a cycle with a 16 h photoperiod, the ratio of towers and the capacities of the ‘day’ and ‘night’ chambers are 2:1. Adjusting the photoperiod or the dark period would be impossible unless the chamber sizes are modular.

Furthermore, most CEA farms with vertical systems utilize fixed shelves. Bagnerini et al. [38] proposed an adaptive vertical farm where the height of the shelves adjusts automatically as plants grow taller. The primary benefits of this system include reduced space requirements for environmental conditioning during the early growth stages, which conserves energy, and enhanced space efficiency by maximizing the number of shelves in vertical areas. Production is proportional to the total number of shelves; however, the number of shelves is constrained by the crops’ maximum harvest height and the building’s height. Nonetheless, overall production can be optimized using multistage cropping techniques or by cultivating plants at various growth stages across different shelf levels, supported by automated scheduling for both sowing and harvesting, ensuring a continuous supply of food to the market. According to the simulation, yields increased by 86% for wheat and 108% for lettuce with multistage cropping techniques that feature adaptive shelves compared to fixed shelves. Cultivating various crops with different height requirements also serves as an advantage of this technology.

5.3. Lighting System

Lighting systems account for 42% to 85% of the energy load in indoor farms, while energy costs constitute 58% of total production costs, driving researchers to explore solutions for improving lighting efficiency and LUE [5,29,30,41,72]. LEDs are considered highly energy-efficient, capable of converting electrical energy into photosynthetic light with photosynthetic photon flux efficacy (PPFE) ranging from 2.1 to 3.5 μmol J⁻¹ as reported by Weidner et al. [14], but can be further improved up to 2.91 μmol J⁻¹ for blue LED, 4.15 μmol J⁻¹ for red LED [48], and 2.98 for white LED μmol J⁻¹ [46]. Generally, red LEDs are more efficient than other spectra of LEDs. However, Gao et al. [63] reported an inefficient red LED with only 1.65 μmol J⁻¹, lower than blue and white LEDs with 2.74 μmol J⁻¹ and 2.75 μmol J⁻¹, respectively. These differences could be due to differences in phosphors used, which influence energy conversion efficiency and thermal stability [44], and efficiency decreases with increasing temperature [112]. The PPFE metric is crucial for LED selection because it is positively associated with LED efficiency, and the latter is directly correlated with lighting energy consumption and costs [14,72]. Accordingly, an improvement in LED efficiency from 59% (2.7 μmol J⁻¹) to 75% (3.5 μmol J⁻¹) would reduce specific energy by 21.4% [14]. Nevertheless, Graamans et al. [24] stated that increasing LED efficiency even up to 100% would still not be enough to reduce the electricity requirements of PFs to the level of GHs. This suggests exploring other approaches to enhance the EUE. Furthermore, LED efficiency can affect the energy demand for the HVAC system, as less efficient LEDs generate more heat, and more energy for cooling is needed [14,24,72]. Accordingly, LEDs generate heat equivalent to 70–80% [113] or as high as 90% [114] of their energy consumption, making the lighting system the most sensible heat contributor to internal climate.

White LEDs vary in spectral composition [36,46], which may not meet crop requirements, and generally have higher energy consumption than other LED color combinations [47]. When using commercial luminaires, Kong et al. [64] recommended employing LED luminaires with a higher R:B ratio for optimal yield, as discussed in Section 4.1. Additionally, thermal generation is influenced by the spectral composition of luminaires. An increased proportion of the red spectrum, particularly closer to the far-red and infrared regions, especially in less efficient luminaires, can result in a higher thermal load within the building [63]. An improvement in yield would not be enough to compensate for the increase in cooling requirements. Thus, the luminaire design should be balanced to enhance crop productivity and the system’s energy efficiency. A deep learning model can produce the best spectral composition for optimal plant growth and yield [49]. Leveraging this technology to optimize spectral composition can be more efficient than the traditional trial-and-error research methods. Furthermore, integrating water-cooling into LED luminaires is an effective method for directly removing the heat generated by the lighting system, which can enhance temperature distribution on the shelves [14,63]. Increasing input power to increase light intensity would decrease the PPFE of LEDs due to increasing temperature [112]. However, it could be improved using water-cooling by effectively removing heat in the LED fixture [63] with assumptive efficiency by up to 85%, which improves environmental control requirements by 349.9% [14]. Furthermore, the LED water-cooling system can be integrated into the HVAC system for centralized cooling and ease of management. However, this technology is expensive, and there is a lack of empirical data on its cost-effectiveness in the CEA.

Improving lighting control can improve LUE by minimizing photon loss. A lighting system with a dimming capacity allows adjustment of light intensity to optimal levels and prevents oversupply in specific periods [35,37,43]. Additionally, integrating artificial intelligence to regulate light intensity based on crop requirements dynamically can significantly enhance crop yield and LUE [115,116]. Some photons emitted by LEDs are naturally wasted on the edge of the shelves due to their beam spread, which decreases light distribution and growth uniformity [79]. Typically, LEDs have a beam spread of 180°, but installing a primary optic narrows it to 120° [117] and can be further reduced as much as 8° using a HEIDI-RS total internal reflection lens (LEDil Oy, Salo, Finland) [118]. When applied strategically, it would minimize photon loss. The distance of the light source from the crops also proportionally affects photon loss. Research indicates that decreasing the light source distance to 15 cm can significantly reduce photon loss, enhance light uniformity, improve PPFD on crops, and double energy efficiency compared to the standard distance of 40–50 cm [30,48]. Additionally, reducing distance alongside dimmed lighting lowers energy needs at the same crop yield [30]. The photon loss can also be reduced by using reflectors within the shelf platform [119], such as reflective curtains [30], or adjustable lampshade-type reflectors [120] to redirect escaped photons back to the crops, increasing light uniformity by more than 90% [48]. For the latter, the width (10 cm) and angle (32°) of the reflectors should be optimized to achieve optimal light reflection without obstructing the airflow to maintain temperature and humidity distribution [120]. This reflector improves the fresh weight and dry matter yield of choy sum (Brassica rapa var. parachinensis) by up to 14% and 18%, respectively. Moreover, the design features an adjustable angle to optimize light reflection based on the distance of the crops from the light sources. Automating these adjustments and integration into modular shelves proposed by Bagnerini et al. [38] could maximize crop-specific illumination and enhance energy efficiency. Additionally, a promising approach involves automated lighting systems equipped with multi-segmented luminaires. This luminaire dynamically adjusts light emission based on canopy surface areas monitored through computer vision, potentially reducing energy consumption to less than half that of conventional lighting systems [118].

Alternative lighting system technologies that utilize free energy from the sun to reduce energy demand are discussed in Section 7.1.

5.4. HVAC System

The HVAC system is a crucial element of PFs for controlling and distributing temperature and humidity within the building, and it is the second-largest energy consumer and could reach over 50% of total energy needs (

Table 1. Differences between different crops in the efficiency of energy, water, and space from various studies.

Crops Species	Data	Areas	Systems	Contribution to Energy Usage (%)	SEC (kWh kgFW−1)	SWC (L kgFW−1)	Space Efficiency (kgFW m−3)	References
Lettuce	M	Riyadh, Naples and Stockholm	Lighting	65–85	20.37–37.41		120.19	[72]
			Cooling	15–20
			Heating	10–15
			Circulating Fans	<1
	M	Istanbul, Hong Kong and Oslo	Lighting	29.35–76.57	26.68–66.97		9.83	[25]
			Cooling	17.34–70.58
			Heating	0.00–3.85
			Others	0.07–2.25
			Lighting	47.67–85.37	37.91–67.90		9.83
			Cooling	17.34–52.33
			Heating	0.00–1.39
			Others
	M	Iran	Lighting	77.50	11.34		56.37	[5]
			Cooling	3.00
			Dehumidification	18.50
			Others	1.00
	E	Poland	Lighting	49.90	7.63	5.12	320.36	[29]
			HVAC	40.40
			Automation and Irrigation	9.70
	M	Canada	Lighting	42–50	10.28–32.72		38.57–112.33	[41]
			Cooling	23–35
			Dehumidification	14–26
			Heating	1–9
	M	Italy	Lighting	89–93	4.74–4.95			[73]
			Cooling	4.00
			Heating	3–7
	M	Hungary	Lighting	68.60				[91]
			HVAC	17.88
			Others	13.51
	E	USA			1.2–2.2			[30]
	E	Italy			8.13	18.55		[71]
					11.56	18.32
	M	Sweden, Netherlands, and UAE	Lighting	50.00				[102]
			Cooling	14.00
			Heating	2.00
			Dehumidification	34.00
	E	Italy				14.7–21.3		[61]
	M	USA	Lighting	44.00				[108]
			Heating	38.00
			Cooling	18.00
		USA	Lighting	50.00
			Heating	40.00
			Cooling	10.00
Lettuce	E	Italy	Lighting	>77	14.80–25.50	12.90–21.30		[18]
Basil					34.00–50.60	22.70–26.20
Rocket					35.80–48.50	38.30–55.60
Chicory					53.40–63.00	39.40–49.60
Lettuce, Kale, Herbs (Basil, Parsley, Oregano, Rosemary), Tomatoes, Beans, and Strawberries	M	Canada	Lighting	88.55				[92]
			Cooling	4.86
			Heating	6.08
			Irrigation	0.51
Lettuce, Tomato, Broccoli, Bell Pepper, Spinach, Zucchini, etc.	M	Reykjavik Region			15.60			[14]
		Stockholm Region			16.00
		Tasmania			16.00
		Massachusetts			16.30
		Tokyo			17.20
		Santiago			17.30
		Gauteng			17.50
		Maricopa			18.90
		Singapore			20.40
		UAE			20.10
Soybean	E	Netherland			292–505	30–59		[12]

SEC: specific energy consumption; SWC: specific water consumption; FW: fresh weight; M: modeling; E: experimental.

). However, energy consumption is influenced by climate conditions, the types of HVAC systems, the internal design, the set environmental parameters for crops, and the management practices [5,121].

Since PFs are fully enclosed, especially with opaque facades, excess energy in the form of sensible and latent heat must be removed mechanically. The sources of sensible heat include the lighting system, which is the main contributor and varies based on light efficiency, spectral composition, and lighting management; walls and ceilings, which are affected by the types of facades and the quality of insulation; and other electrical equipment [5,7,24,25,41,72,92,93,107]. People working on the farm also generate and absorb heat, but their influence on the energy dynamics in the system is negligible [92]. Nevertheless, sensible heat influences plant biology by altering the transpiration rate and dissipating latent heat as water vapor [83]. This excess vapor or humidity must be eliminated from the system to ensure crop health, which will increase electricity demands. Therefore, balancing the energy dynamics within the building is essential for crop productivity and system efficiency.

An HVAC system can be either a non-recirculating or a recirculating (closed-loop) system. In a non-recirculating system, fresh air is continuously drawn into the HVAC system, conditioned (cooled, dehumidified, or heated), and distributed throughout the building; then, the humid and warmer air is vented outside. Typically, heating, cooling, and humidification or dehumidification occur inside the building, where HVAC units operate independently [5,26,107,122]. Fluctuations in environmental conditions are challenging, leading to high seasonal variations in the HVAC system’s internal environment and energy workload. However, this can be mitigated using a dehumidifier rather than adjusting the ventilation rate [5]. The drawbacks of a non-circulating HVAC system include significant waste of CO₂ and moisture in the air, although moisture can be recovered from the humidifier [26] but the CO₂ is not. Additionally, the HVAC system’s energy workload is high due to the constant conditioning of new outdoor air unless the outdoor air conditions closely match the internal air requirements. In contrast, the exhaust air is reconditioned—either cooled, dehumidified, heated, or all—and recirculated back to the growing chamber in a closed-loop HVAC system [14,24,72,102,108,114,123]. Exhaust air recirculation reduces CO₂ enrichment, water usage, and HVAC system workload. Nevertheless, the HVAC system’s energy performance varies depending on the design.

Several HVAC system designs have been reviewed, but in this study, we created a closed-loop design mainly inspired by Blom et al. [114,123], and incorporating other energy-efficient ideas (Figure 5). The HVAC system utilizes several heat exchangers (HEs) for indirect reconditioning of exhaust air. Their design and size must be optimized for efficient heat transfer [14]. Furthermore, the system has modular features that undergo economy mode [25,108] which utilizes favorable outdoor air (dry and cooler or close to set growing temperature) and mixes it with exhaust air in air-mixing unit 1 (AMU1). Additionally, the system supports cooling-only, reheating-only, or cooling–dehumidification–reheating processes, depending on the conditions of the exhaust air, which are facilitated by bypass mechanisms [14]. Finally, the reconditioned air will go to air-mixing unit 2 (AMU2) for CO₂ enrichment and will then be evenly distributed throughout the growing chamber, ensuring uniform airflow across the entire area. Dehumidification is achieved by cooling exhaust air below its dew point to condense moisture in the air [14,114,123] and this is taken in either by air-to-air heat exchanger 1 (HE1) or air-to-water heat exchanger 2 (HE2). The water collected in HE1 and HE2 is to be repurposed for irrigation. During cold seasons, outdoor air is used for cooling only or cooling and dehumidification in HE1, leveraging the proposed methodology by Pacak et al. [124] and Jurga et al. [125] for harvesting water from exhaust air. To ensure effective condensation and prevent frosting, the heat exchanger’s temperature must be maintained at 2.5 °C below the exhaust air temperature [124], but can still be further reduced to maximize cooling and condensation as long as it remains above the frost formation point. When outdoor conditions are inefficient for cooling and dehumidification, a water-to-water heat pump (HP) will supplement the process [14,114,123]. The evaporator of the HP is used to cool the exhaust air in the HE2, and the condenser is used to reheat the air coming from either AMU1, HE1, or HE2 in water-to-air heat exchanger 3 (HE3). Additionally, the system can utilize residual heat or an external heat source to reduce condenser workload. This can be added to the heat buffer tank or directly used in HE3 through heat exchanger 4, depending on the temperature requirement and the recovered temperature. HE4 can be either air-to-water or water-to-water, depending on the residual heat source available in the location. Cold and heat buffer tanks are added to the system with mixing values to allow temperature variations within the system while maintaining uniform internal conditions [114]. The excess heat produced by the HP during low heat demand is expelled through water-to-air heat exchanger 5 (HE5) to the outside of the building or repurposed for heating other facilities for energy synergy [114]. Additionally, aquifer thermal energy storage (ATES) can be integrated to store the residual heat from the HP during summer and retrieve it to utilize during winter [114]. However, its efficiency depends on geological and hydrogeological conditions and requires a high initial cost, which limits its application. The HVAC system we designed has complex features that require complex control. However, integrating AI into the control system would be more effective and efficient. Additionally, optimizing the design is recommended to make the HVAC system more compact and space-saving, and it should be tailored to the local environment. In addition to precise control of environmental conditions, the uniformity of airflow distribution must be considered when designing PFs. The ventilation design needs to be optimized to minimize variations in air conditions between upper and lower tiers and areas that are farther or closer to walls to enhance uniformity and crop growth rates while reducing energy consumption [121].

Local climatic conditions largely determine the equipment used for the HVAC system. In areas without winter or cold seasons, heating systems are not required, and the focus on energy reduction shifts to cooling and dehumidification [5,7]. Moreover, the total energy demand in warmer climates is higher compared to colder climates [24,25,72,92,102]. In regions with particularly hot climates, especially near the equator, cooling requirements are needed year-round. In contrast, in colder areas, mechanical cooling is only necessary seasonally, and the heat generated by the lighting system can offset the heating energy needs [92]. Various types of cooling systems are used, including air conditioners, air-source heat pumps (ASHPs), ground-based heat pumps (GBHPs), and chillers [5,14,72,73,104,108]. ASHPs and BGHPs are effective for small to medium-scale PFs, serving both to cool (by expelling heat from inside the building outdoors or underground through a refrigerant) and to heat through the reverse process [14], ideal for regions with heating requirements. However, while GBHPs are more sustainable than ASHPs, their adoption remains low due to installation complexities, a shortage of technical experts, high initial costs, and, depending on the country, the need for permits or approvals to ensure groundwater protection before installation [126,127]. On the other hand, chillers solely offer cooling, yet they are frequently utilized in larger facilities because of their ability to deliver large-scale cooling [14]. Moreover, energy-efficient absorption chillers are available, and utilizing this technology in PFs would enhance energy efficiency and promote synergetic integration with other industries. Unlike air-cooled and water-cooled chillers and the previously mentioned cooling systems, absorption chillers do not depend on mechanical compressors but can harness waste or residual heat to compress and drive the refrigeration cycle [128,129,130]. Furthermore, the utilization of natural cold water, including deep seawater [131], groundwater, and surface water [126], is a sustainable and energy-efficient alternative method for cooling. Additionally, fogging and misting systems can be supplemented in the cooling systems by pre-cooling incoming air through evaporative cooling, thereby reducing the workload on primary cooling systems [7,107]. However, managing humidity can be challenging with this approach, as it raises moisture levels in the air. The effectiveness of evaporative cooling diminishes when the humidity of the incoming air is high [132]. Consequently, fogging systems are more effective and appropriate in hot, dry climates where additional humidity is advantageous rather than harmful.

Conversely, a heating system is crucial in colder climates where outdoor temperatures can fall below 0 °C [24,102]. Heating may also be required in warmer climates when exhaust air is cooled for dehumidification [114,123]. The heating system can operate on either electricity or natural gas. Direct air heating can be achieved by burning natural gas, which is commonly carried out in GHs [15]. Alternatively, natural gas is burned to heat the boiler for heating the building through floor heating or by radiant pipes [104,107]. However, achieving uniform heat distribution can be challenging with these heating methods unless the boiler is used as a heat source to recondition the exhaust air before it is recirculated [72]. As discussed, ASHPs and GBHP are the mechanical methods for cooling and heating [14,126,127]. The primary factors in choosing or combining energy sources for heating are the prices of electricity and natural gas in the region. Additionally, the environmental impact of using natural gas is a significant concern; however, it can be alleviated when CO₂ emissions from combustion are captured and repurposed for CO₂ enrichment [5]. Alternatively, geothermal water and residual heat from other industries can be recovered and utilized for heating (see Section 7.2).

5.5. Irrigation Systems

The energy requirements for irrigation systems are much lower than for lighting and HVAC systems [29]. However, the irrigation system plays a crucial role in maintaining pH levels and nutrient concentrations and delivering them promptly to the root systems to support crop growth. When the pH of the nutrient solution changes due to imbalanced ion uptake, correction is achieved by adding pH buffers, hydrochloric acid, or nitric acid, which are integrated into the system [8,75]. Furthermore, the frequency of irrigation is critical for the timely delivery of water and nutrients; it can be set at regular intervals [27] or automated based on moisture content and nutrient concentrations in the rhizosphere [100,122]. The latter is ideal because it would prevent water and nutrient deficiency and improve yield. However, the typical irrigation system supports the dosing of concentrated nutrient solution, formulated to contain macro- and micronutrients for specific crops, to meet the EC setpoint of the circulating nutrient solution. Moreover, plants’ requirement for individual nutrients varies by growth stages and environmental conditions [133,134], which cannot be precisely delivered through this system. Incorporating individualized nutrient dosing systems can optimize plant nutrition by accurately replenishing specific nutrients depleted in the nutrient solution. This approach may also enhance the energy efficiency of the plant production facility by increasing yields. However, it could result in higher initial costs and require a more complex control system. Implementing real-time monitoring and automation, supported by IoT and AI technologies, can significantly improve the system’s effectiveness.

The two main types of irrigation systems used in PFs are hydroponics and aeroponics (Figure 6). In hydroponics, plant roots are submerged in nutrient solutions constantly in the case of a deep-water culture [26,75], partially in the case of nutrient film techniques (NFTs), or in regular intervals in the case of an ebb-and-flow system [27,122]. The disadvantage of the deep-water culture is the depletion of dissolved oxygen in the nutrient solution for the root system [27], requiring a continuous monitoring of dissolved oxygen [26,75] and aeration [61] which would add energy demands for the process. Aeration is better in NFTs than in the deep-water culture because most of the root system is exposed to air, and only the tips are exposed to a continuously flowing thin layer of nutrient solution [135]. However, supporting a continuous flow of nutrient solutions requires more energy. On the other hand, in ebb-and-flow systems, the root system is periodically flooded with a nutrient solution and then drained [27], which enhances oxygen availability and reduces irrigation demand. However, substrates have varying water and nutrient retention capacities [79,101]. Therefore, selecting the right substrates is crucial for ebb-and-flow systems, as they affect irrigation frequency and maintenance—factors that can be managed to reduce energy usage.

Compared to traditional hydroponics, monitoring of dissolved oxygen and aeration of nutrient solutions are unnecessary in aeroponics since crop roots are naturally exposed to the air, making for healthier root systems and better growth [135]. Furthermore, the nutrient solution is directly delivered as a fine mist to the roots [136], which reduces the water requirement, an essential factor for regions with scarce water supply. Moreover, in combination with hydroponics, aeroponics offers an opportunity to diversify the crops that can be cultivated in the PFs [39,92]. Aeroponics can be ideal for crops sensitive to root rot in the case of root crops [39]. Nevertheless, factors to consider in aeroponics applications include the high initial cost, the need for precision and maintenance of nutrient solution delivery (such as mist size and pressure), frequent irrigation to keep the roots moist since they are exposed to air, and the additional HVAC workload required to remove excess humidity from misting [27]. Furthermore, Carotti et al. [27] found that aeroponics had no significant improvement in growth performance compared to ebb-and-flow systems, suggesting unnecessary investment when growing certain crops. Moreover, the energy needed for irrigation in the ebb-and-flow system can be minimized by utilizing gravitational energy: first, pump the nutrient solution to the upper shelves, and then allow it to flow down to the lower shelves [136].

Furthermore, the energy needs for irrigation are affected by crop species; for example, kale requires frequent watering [92]. Moreover, as discussed in Section 4, growing environmental conditions also influence the water requirements of crops; thus, the energy requirement for irrigation is also affected. Managing these environmental factors would save energy, water, and nutrient resources. The water–energy nexus is an increasingly important topic nowadays because these resources are interconnected and are vital for our daily lives [137,138,139]. The increasing global population, a growing economy, and climate change place additional pressure on these resources. Therefore, minimizing water wastage not only cuts the cost of purchasing water from external sources but also lowers the energy requirements for pumping and treatment within the PF and the entire economic systems. A closed-loop irrigation system is recommended as it can reduce water wastage [14,24] and can save up to 95% of water compared to conventional farming [140], and this could be further increased when water from exhaust air is recovered [124]. Furthermore, harvesting water from exhaust air can improve water sufficiency by 67.12% [124] and this can be further improved by up to 90.4% in combination with rain-water harvesting [125]. Pacak et al. [124] and Jurga et al. [125] highlighted that low-energy water harvesting from exhausted air is possible by utilizing natural cold air to cool HE for condensation. However, this methodology is ineffective during summer months when the external temperature is insufficient to cool down the HE plates to at least 2.5 °C lower than the exhaust air temperature, a threshold for effective condensation. In this case, a cooling or dehumidification system can be used for harvesting, as discussed in Section 5.4. Furthermore, Yang et al. [141] proposed a new water harvesting method using a hygroscopic copper(II)–ethanolamine complex (Cu-complex), with a maximum water uptake of up to 300% and a water production rate of 2.24 g g⁻¹ h⁻¹. This technology can potentially be integrated into the HVAC system for alternative low-energy dehumidification or moisture harvesting systems. However, a complex design would be needed because the Cu complex must undergo a heating process to liberate the entrapped water, facilitating its subsequent reutilization for moisture absorption. In this case, natural heat from sunlight [141] or residual heat from the condenser of a cooling system or other sources can be used.

5.6. Decision Support System and Automation

Several models have been developed and these can be utilized to determine optimal environmental conditions for productivity and profitability [5,7,14,41,72], to evaluate the energy performance of indoor farming facilities and energy-saving strategies [7,14,24,107], and to optimize ecological benefits [142], which can be modified based on local conditions for the planning and development of PFs. However, the effectiveness of the models heavily relies on the accuracy of the estimates used for the assumptions; for example, not being limited to the crop transpiration coefficient or the fraction of the radiation load dissipated by the crop as latent heat [83] and computational fluid dynamics in airflow distribution control [121] are crucial for estimating the energy balance of the PF system.

PFs with flexible operational schedules can capitalize on daily fluctuations in electricity prices, a capability enhanced by integrating an energy management system (EMS) [32,107,143,144]. An EMS enables real-time monitoring and control of energy consumption, reducing overall energy costs and improving efficiency by shifting power demand to periods when grid energy prices are lower, prioritizing renewable energy use during peak periods, and prioritizing energy storage during low grid prices when renewable energy sources are integrated. Moreover, it can be enhanced by incorporating other input prices to determine optimal operational strategies that may reduce electricity costs by up to 40% [91]. However, depending on input pricing for operational optimization highlights the need for precise price forecasting. The involvement of PF in the demand response program of a smart grid facilitates the identification of peak electricity prices and allows for operational adjustments, potentially resulting in savings of up to 23% in lighting electricity costs [145] or up to 30% of the total energy costs [146]. Active distribution networks (ADNs), which allow PFs to function as prosumers by modifying their energy use and selling excess energy produced from on-site renewable sources back to the grid, are another advantage of smart power grids that EMSs can take advantage of [16]. Compared to PFs without flexible scheduling, PFs with EMSs and flexible operation schedules within ADNs can cut carbon emissions from the agriculture and electrical sectors by up to 70% and lower electricity network operating costs by 10%.

As discussed in the previous sections, environmental conditions are critical for optimizing the yield and energy performance of PFs, which require precise monitoring and control. An intelligent environment controller allows for the adjustment of environmental conditions based on crop requirements through the integration of plant sensors, making it more effective and beneficial than a traditional controller that only adjusts based on a fixed schedule [122,147]. Controlling environmental variables based on crop needs through growth rate monitoring enhances crop growth and shortens growing cycles, resulting in higher annual yields and reduced specific costs [23]. Machine and deep learning models can be used to predict crop growth using environmental factors as predictive variables [49]. Furthermore, computer vision would be more effective because it may be utilized for a variety of purposes, such as harvest prediction [6,75] and disease identification and monitoring [29]. Tang et al. [39] and Shateen and Kacira [75] proposed the integration of computer vision into the intelligent environment controller, which monitors crop growth and delivers feedback to the controller to align with crop environmental needs. To precisely control and distribute air, the ventilation system can also be automated using a prediction model that forecasts environmental conditions based on crucial variables, including volume flow, air supply temperature, and internal relative humidity [93]. The IoT enables real-time, wireless monitoring and control of various systems, including HVAC, irrigation, lighting, and more [6,39,44,136,148,149,150,151,152]. Additionally, the IoT enables easier integration of all systems into intelligent environmental control. However, for automated precise control to be effective, each unit’s mechanical system must be managed appropriately [153]. For example, if the ventilation damper is not controlled correctly, it may open repeatedly, unintentionally allowing air from the building to escape. By minimizing excess inputs and system workload, precise environmental control will improve growth performance and resource efficiency [107,152].

Although automation adds energy requirements, its economic benefits are much more significant since precision is crucial in PF, particularly in water and nutrient delivery, lighting intensity and spectral composition, and environmental conditions. Furthermore, the automation of economizer control in the HVAC system is also crucial for the timely utilization of favorable outdoor air to reduce the HVAC system workload [25,108]. Moreover, leveraging AI to predict harvest periods can help prevent the overuse of resources as crops near harvest [6,23]. Moreover, this would facilitate efficient scheduling of activities, allowing for extra production cycles and ensuring that the PF operates at nearly full capacity throughout the year. Furthermore, PFs with modular shelves [38], towers [29], or elevators that facilitate movement for automated harvesting and other purposes would increase production efficiency with little labor. By optimizing algorithms, improving decision-making processes regarding operating schedules, and identifying the most efficient low-energy route in the elevator case, the energy needed for automation can be reduced [154]. Since the PF system relies on sensors, adding more would improve the accuracy of the measured parameters but would also lead to increased energy consumption. Although sensors generally use less energy, determining the optimal number of sensors and their placement without compromising accuracy and precision would help reduce energy needs. Moreover, there are numerous potential applications for computer vision in PFs, especially for monitoring crop conditions. Consequently, it could serve as a replacement for other plant sensors or possibly as a light-quality sensor. Additionally, the system can be set up to gather and monitor data in an energy-saving mode. In this mode, certain sensors operate at regular intervals instead of continuously. It is hypothesized that constant monitoring may not be necessary. By monitoring at the optimal intervals, both energy and storage can be conserved.

6. Energy-Efficient Management Practices

The primary purpose of the PF systems is to maintain optimal environmental conditions inside the building to maximize plant growth potential. However, sustaining these conditions requires significant energy, which can increase production costs. We have identified several management techniques that could enhance the energy efficiency of the PF systems.

6.1. Lighting Management

The lighting system naturally produces heat, which might affect how much energy the HVAC system uses. Since the light system creates less heat at low PPFD, more energy will be required for heating [72]. Conversely, for high PPFD and longer photoperiods, more cooling and/or dehumidification is needed due to more significant heat generation [14,41]. These principles can be leveraged to control the HVAC system’s energy demand and raise the system’s overall energy efficiency. Switching the lighting operation at night would reduce the cooling requirements during the day and heating requirements at night [14,102]. This is ideal in regions with warmer temperatures or in the summer when the daytime temperature is generally higher than the set growing temperature. Alternatively, heat generation is reduced by intermittent lighting [73] and lighting segmentation [36]. Additionally, due to fluctuations in outdoor temperatures during the winter, less heating is required during the day compared to the night. Therefore, by utilizing the natural heat produced by the lighting system, shifting the photoperiod to nighttime throughout the winter would minimize the need for heating at night [1]. With a seasonal change in outdoor temperatures, seasonal operation schedule setpoints may also aid in lowering energy requirements [73]. Energy demand for heating, cooling, and dehumidification has been observed to decrease when PPFD and photoperiod are adjusted based on climate conditions [26,41]. In winter, extending the photoperiod while maintaining the DLI would decrease the need for heating [92]. Conversely, a shorter photoperiod in summer would lessen the need for cooling.

Jayalath et al. [155] investigated whether high DLI due to light abundance from the previous day could carry over to the succeeding day when light quality is low, a typical condition in the open-field and GH farms. They found that the growth and yield of lettuce are not significantly affected when the average DLI between the days is comparable to the requirement of plants (15 mol m⁻² day⁻¹) and the fluctuation is no more than 10.5 mol m⁻² day⁻¹ between days. The investigated DLI fluctuation was only 2 days, but they suggested that it can be extended up to 5 days based on the study of Mayorga-Gomez and Iersel [156]. These results imply that when plants received more light than their requirement in the preceding days, artificial lighting could be reduced for the following days. This method could be feasible to lower lighting requirements in GH and PF with transparent facades or optical fiber daylighting (OFD) (see Section 7.1). Furthermore, by raising DLI during periods of low electricity costs and decreasing DLI during periods of high electricity prices, the DLI carryover may also be relevant in PFs with opaque facades.

Photoperiod reduction is a straightforward way to cut energy use but may also affect crop productivity. However, studies on basil have shown that yield and quality are not significantly affected by reducing the photoperiod from 16 to 14 h with intermittent lighting (4 h of continuous light, three times per 24 h, and 10 min of light every hour of darkness) [32]. However, reducing the photoperiod (from 16 to 14 h) and increasing the PPFD (from 200 to 228 μmol m⁻² s⁻¹) to meet DLI requirements boost basil production and decrease energy costs. These studies, however, were performed in a small chamber over a short duration (23 to 37 days); therefore, a long-term economic assessment of intermittent lighting is necessary, as frequent switching of the lighting system could strain the power supply or LED driver, leading to a shorter lifespan [47]. Additionally, we suggest replicating the methodologies used in fast-growing crops like lettuce to verify the findings. When intermittent lighting is strategically employed, such as during times of lower energy costs, yields increase, and energy expenditures are significantly reduced [122]. This is possible when energy prices can be predicted in advance, as seen in the Nordic countries, using an EMS or within a smart grid, as discussed in Section 5.6 and Section 7.1.

Furthermore, a pilot study by Song et al. [157] on pulsed LED lighting with a 10 s pulse interval and maintained DLI revealed no adverse effects on the growth of seedlings of Red Russian Kale (Brassica napus), Purple Top Turnip (Brassica rapa), and Ruby Queen beet (Beta vulgaris) while saving over 50% of energy. Similarly, it was investigated in lettuce and found that low switching frequency (293 kHz for blue LED and 443 kHz for red LED) significantly improved EUE by 42% while having no significant adverse effects on yield compared to high switching frequency (850 kHz for blue LED and 437 kHz for red LED) [71]. Additionally, exposure to pulsed light at a frequency of 100 Hz and various light spectra did not negatively impact the photochemical efficiency of photosystem II of chili pepper plants (Capsicum annuum var. Serrano) [47], suggesting a further decrease in pulsing frequency to save more energy. According to a mathematical model developed by Olver-Gonzalez et al. [47], the luminaire’s spectrum composition, duty cycle, and PPFD needs all affect the ideal frequencies for energy use. Furthermore, it must be noted that the design and electronic components of the lighting system play a significant role in attaining energy savings with pulsed LED light applications. Additionally, the lower LED life spans could be a drawback of pulsed illumination. It is advised, therefore, to initiate a cost-effectiveness study and ascertain the ideal switching frequency and how it affects the PF’s energy dynamics.

6.2. Crop Management

The ideal temperature range for various crop species or cultivars varies. With this, changing crop cultivars seasonally while modifying environmental setpoints can lessen the HVAC system load and increase PF energy efficiency. According to a modeling study, growing mild-temperature cultivars and lowering the setpoint temperature in the winter can cut HVAC electricity use by 50% [14]. Additionally, wide temperature setpoints can decrease the HVAC workload, reducing the consumption of electricity and natural gas [73,107].

A multi-tier cropping system is a type of multi-cropping system that involves planting multiple crop species with varying heights to enhance resource utilization in open-field farms [158]. An important consideration when using this method is crop compatibility. The secondary crop should require less light, space, and nutrients than the primary crop to minimize resource competition. This principle can be applied in the PF system. The wasted light can be utilized by introducing a secondary crop that thrives in low-light conditions, potentially increasing the PF’s output and efficiency. Furthermore, cultivating multiple crop species simultaneously can enhance yield diversity and satisfy a range of market demands [39]. One of the difficulties of multi-cropping is handling the requirements of several crops growing in the same space. Furthermore, considering the small variety of crop species usually cultivated in plant factories, choosing appropriate crop combinations to prevent competition is challenging.

7. Utilization of Renewable Energy and Residual Resources

7.1. Renewable Energy

Energy production, which usually depends on fossil fuels, is the primary source of GHG emissions from PF systems [7,18,50]. However, the integration of renewable energy sources can decrease water usage and emissions by as much as 40% [45]. Solar technologies, particularly PVs, are most widely used in agriculture because of their growing availability, the abundance of sunlight, and affordability. PV systems can entirely replace grid energy to power irrigation and lighting systems in small indoor farms [105] and even a 117 m² agrotunnel [92]. Nevertheless, a PV array with 17% efficiency in a 10,000 m² PF can only supply 2.71% of the annual electricity requirements [24], and it would take a large land area to cover all the requirements [7]. However, PV systems are modified to improve the efficiency of energy collection. This includes using AI for modular PVs to increase efficiency by up to 20.05% [6], integrating solar trackers to optimize sunlight capture, which increases power output by up to 22.58% [105,159], and coupling PV panels with thermoelectric generators to generate electricity from both light and heat from the sun, which could add up to 3.4% of electricity production [160,161,162]. Such efficiency gains can reduce the land area required for PV systems. However, the PV system can still be substantially enhanced. Accordingly, the multi-junction cells have a theoretical efficiency of 50% [40].

Location is a critical factor when integrating solar technologies into PFs, as the electrical output is influenced by solar radiation and temperatures [5,92,161]. Solar technology, particularly concentrating solar power (CSP), requires direct sunlight to concentrate solar energy and is ideal in arid areas [163]. Meanwhile, areas with shorter daylight hours may struggle to meet energy demands, especially during cold seasons. Additionally, PFs are often situated in urban environments, where shadows from tall buildings can limit PV and CSP systems’ output. Therefore, in addition to the efficiency of the current PV systems, weather conditions limit the supply of energy requirements of the PFs. Integration with other renewable energy sources could mitigate this limitation, such as solar–wind hybrids that can supply energy during periods of limited sunlight [40,164]. Wind turbines can be installed on the roof, or the building facade, but the design must be optimized based on the design of the building for maximum electricity generation [165]. Electricity generation through wind energy is also insufficient and would require more areas than PV to meet demands [40]. Integrating PFs into other structures, such as residential buildings, and utilizing the roofs and facades of the buildings for PV installation or other technologies is another potential way to boost the production of renewable energy in urban settings [19,20]. In effect, food availability within the community is improved. Another potential renewable energy source for PFs that can be utilized for both electricity generation and direct heating is hydrothermal energy. However, factors such as installation costs, energy production capacity, and local grid energy pricing influence the economic viability of renewable energy systems [5].

An alternative method for utilizing sunlight in PFs is using OFD systems, where sunlight is concentrated using convex lenses, Fresnel lenses, parabolic concentrators, etc., and distributed inside the building through optical fibers for illumination [25,166,167]. Compared to PV systems, OFD systems are more efficient for lighting, as they directly utilize sunlight, whereas PV systems first convert sunlight into electricity before converting it into light. However, the sun’s broad spectrum could add thermal load to the building. This thermal load depends on the spectral absorption range of the materials used for the lens and the length of the optical fiber [167]. The OFD can be modified by incorporating infrared filters into the parabolic concentrators to reduce the thermal load [168] and adding fluorescent pigments (K₂SiF₆:Mn₄⁺) embedded in the parabolic concentrator reflectors to convert the least absorbed light spectrum (green) to red to improve light quality [169]. However, the OFD system alone is insufficient to meet the lighting needs of the PF. However, by integrating it with LEDs in a hybrid luminaire featuring AI modules that automatically adjust light intensity based on the available light from the OFD system, lighting energy consumption can be reduced by over 50%, and total energy consumption can decrease by up to 29%, depending on the system type, geographical location, and weather conditions [25,167]. Longer daylight hours near the equator allow for greater lighting energy savings. However, PFs at high latitudes have the most significant decrease in overall energy use. On the other hand, if OFD systems lack infrared filters, the thermal load redirects energy consumption toward cooling, leading to minimal overall energy savings in lower latitudes [25]. Despite the benefits, the primary obstacle to adopting OFD systems is their high initial investment costs, particularly regarding parabolic concentrators. However, a static OFD system designed with geometry that can capture sunlight without requiring automation for sun-tracking, and utilizing Fresnel lenses as sunlight concentrators, could serve as a cost-effective alternative [166,167].

Urban areas generate tons of organic waste, which can be used in aerobic digesters to produce methane-rich biogas, considered a cleaner energy source than fossil fuels [17,19,50,91,170]. Biogas contains 43% of CO₂ and, if separated, could provide quality methane (CH₄) as an energy source and CO₂ for enrichment [10]. Moreover, nutrient solutions for the crops in the PFs can be produced by processing the waste from biogas systems to extract nutrients [17,19].

It is challenging to meet energy demand in large-scale PFs using only renewable energy sources [171]. This is only feasible if the land area saved by growing in PFs is used for renewable energy production, such as wind turbines and PV systems, which is sufficient for lettuce but insufficient for other crops like wheat, potatoes, and tomatoes [40]. Additionally, the concerns of increasing dependency on renewable energy are the availability of the critical materials that would limit production, such as silver, copper, and rare metals [40], as well as the destruction of the environment by mining such materials. Consequently, it is more practical to combine grid electricity with renewable energy and integration of EMSs. PFs can maximize the economic benefits of this integration by employing the EMS in the system [32,91,143,144]. As discussed in Section 5.6, the EMS enables efficient energy prioritization based on demand, renewable resource availability, and grid prices. Integrating weather forecasting further enhances the decision support system for operational adjustments. Furthermore, when incorporated into smart electricity grids, grid–renewable hybrid allows PFs to function as prosumers [16].

7.2. Residual Resources

Residual resources are materials, energy, or by-products left over after a primary process or production activity, which can still be repurposed or utilized instead of being discarded as waste [172]. Waste heat or residual heat, organic waste, CO₂, and wastewater are examples of residual resources that PFs can utilize. Integration of PF into or in the proximity of other industries’ facilities, such as but not limited to residential buildings [19,114,173], data centers [21], offices [114,174], restaurants [114], city incinerators [103], district heating and cooling providers [91], and manufacturers like breweries [7,19,22], allows access to these resources, which would reduce energy, water, CO₂, and nutrients from commercial sources, therefore reducing operation costs. This integration would promote a circular economy that could reduce GHG emissions of the whole system by over 60% [19,22]. Additionally, the air quality of the host building will be improved, and the ventilation requirements will be reduced when the exhaust air of the PF is integrated [21,174]. Although poultry and livestock farms are situated in rural areas, there is a high potential for PF integration, which must be explored. For instance, the CO₂ concentrations in a swine house could reach more than 800 ppm [175,176], almost double the typical enrichment concentrations (Table S1). Repurposing this exhaust air for CO₂ enrichment in the PF could reduce enrichment costs, lower carbon emissions from livestock farms, and improve animal air quality. However, specialized facilities may be necessary to process the exhaust air and remove pollutants, including dust, ammonia, and hydrogen sulfide.

The aforementioned facilities typically generate low-grade waste heat as heated water or exhaust air, which is inefficient for electricity generation. Integrating them into the HVAC system would decrease the heating needs of the PFs [21,103,114]. However, the amount of recoverable residual heat varies based on the types and sizes of the sources, as well as the distance from them, which inversely affects the recovery potential, making integration in the same facilities ideal [21]; 90% of the electricity consumption in the data centers is converted into low-grade heat, with exhaust air temperatures between 25 °C and 35 °C, of which 55 to 68% can be recovered [21]. This amounts to 0.11 to 26.73 MJ or 50.64 to 742.51 J m⁻² of energy from Irish data centers, providing an average saving of 36.6% for heating 1000 m³ PFs. Moreover, urban businesses such as saunas, food and beverage production, and steel and metal manufacturing could serve as significant sources of high-quality residual heat to satisfy the heating needs of larger PFs. Their potential synergistic benefits with PFs have yet to be explored.

Furthermore, PFs also generate residual heat, mainly from the lighting system and other electrical devices [45,114]. Using a closed-loop HVAC system can be an efficient method to recover this heat and reduce the need for heating incoming external air [14,72]. Another possible and effective way to recover heat generated by the lighting system is by integrating LED water-cooling into the system [14,63], which can be integrated into the HVAC system. Furthermore, composting within the PFs is another potential source of residual heat, which can generate 1.94–2.78 kWh/kg of green waste, depending on the materials used [177]. Accordingly, utilizing this excess heat in the PF would not only save energy demand but also extend the lifespan of production site structures and equipment by preventing excessive heat buildup [45]. Furthermore, the synergies between PFs and apartments, offices, restaurants, supermarkets, or swimming pools in the bidirectional exchange of residual energy were evaluated [114]. Blom et al. found that this synergy could reduce the annual energy requirements of the entire system by at least 12% to 51%. The authors emphasized that the synergy is most effective when the energy demand of the host building is compared to the PF.

The amount of recoverable heat and the PF heating need, which change depending on building size, crop requirements, and external temperature, are important considerations for utilizing residual heat. However, the application of residual heat in PFs is constrained in regions where heating is required unless this can be integrated with other systems, such as absorption chillers [128,129,130]. Furthermore, there is a lack of empirical evidence on using residual heat in PFs. All the data used in the analysis were generated through modeling.

8. Conclusions and Recommendations

In conclusion, environmental variables such as light, temperature, humidity, water, nutrients, and CO₂ serve as limiting factors for crop growth and yield. Various systems, including lighting, HVAC, and irrigation, in PFs critically regulate these factors. However, these environmental variables must be balanced to maximize production while decreasing operational expenses to achieve profitable operations. This balance can be achieved by carefully regulating the interaction of crop biology with the PF systems. However, caution must be taken when applying the environmental factor levels presented in this paper, as most of the evaluated data were generated through simulations. We recommend that further large-scale empirical research be conducted to validate theoretical models and simulations, particularly for non-traditional crops.

Energy is the main cost driver and contributor to carbon emission in PFs, making energy efficiency a key factor in developing cost-effective and sustainable operations. Two primary solutions were identified: yield optimization and energy usage reduction. These can be achieved by precisely controlling optimal environmental conditions, designing modular PFs suited to local climates, and integrating advanced and energy-efficient systems designs powered by IoT and AI technologies. Furthermore, various management practices were identified to improve productivity and energy efficiency. Leveraging renewable energy sources would further reduce the environmental impact by reducing dependency on grid energy. Finally, integrating PFs into other industries offers a promising approach to creating a circular and resource-efficient economic model.

Nevertheless, PF systems require a higher initial investment than other controlled CEA systems. Therefore, government initiatives and support are crucial for the widespread adoption and advancement of this technology. Governments should implement supportive policies, including subsidies, low-interest loans, and tax incentives, to assist farmers and companies investing in PF systems, particularly those adopting energy-efficient and sustainable technologies. To accelerate innovation, governments should prioritize funding for research into cost-effective and energy-efficient PF designs and technologies. This should include exploring alternative applications for residual heat beyond mere heating, which could benefit regions with minimal or no heating needs. We also recommend that plant breeders develop cultivars specifically suited to indoor farming conditions. Furthermore, governments should update land-use policies and integrate PFs into urban planning strategies. This will enhance food security, reduce carbon footprints, and promote a circular economy by optimizing resource use and minimizing waste.