1. Introduction
The world’s population is projected to reach approximately 9.7 billion by 2050, with approximately 70% of people living in urban areas [
1]. This rapid population growth and urbanization are expected to create a surge in food demand. However, climate change has led to reduced crop yields in large expanses of the world [
2], while urban sprawl continues to diminish the land available for agriculture, with an expected loss of global croplands of 1.8%−2.4% by 2030 [
3]. Additionally, the overuse of chemical fertilizers has exacerbated soil degradation and inflicted damage on ecosystems [
4]. These challenges significantly threaten the food supply chain, making it imperative to explore innovative farming strategies to increase the security, stability, and sustainability of food production [
5,
6].
Vertical farming has emerged as a promising alternative to conventional farming techniques and common greenhouse systems [[
7], [
8], [
9], [
10]]. In a vertical farm, crops are grown in vertical stacks under controlled environmental conditions. This approach alleviates the pressure on land sources by expanding upward instead of outward [
11] and substantially improves crop yields per unit of land area [
12]. The isolation of crops from the external environment makes vertical farming resilient to climate change and extreme weather events, enabling year-round cultivation [
13]. Moreover, the possibility of deploying vertical farms in urban areas reduces the difficulty of food supply. Vertical farming can be implemented using hydroponics or soil cultivation. Hydroponics boasts high resource-utilization efficiency and has already seen widespread commercial application [
14,
15]. On the other hand, soil-based cultivation offers broader applicability and simpler implementation. Moreover, soil exhibits greater tolerance to fluctuations in water and nutrient levels, making management relatively straightforward. Successful commercial soil-based vertical farming systems have also been demonstrated [
16,
17]. In regard to soil-based techniques, the addition of biochar stands out as a promising strategy to enhance soil properties. Biochar is a carbon-rich material derived from the thermochemical conversion of biomass. The existing literature reports that biochar can increase the nutrient density and water retention capacity of soil, reduce the use of chemical fertilizers, and improve the microbial community [
18,
19]. More importantly, the agricultural application of biochar can bolster carbon sequestration in soil, playing a significant role in the mitigation of climate change [
20,
21]. While research into the mechanisms by which biochar improves soil is advancing, the assessment of biochar’s environmental benefits within agricultural contexts remains inadequate, underscoring the need for more targeted research.
Given the attractive features of vertical farming, it is crucial to conduct a thorough evaluation and design an appropriate system to fully leverage its advantages. In this context, the structure of a vertical farming system offers the potential for flexible adjustment of various cultivation parameters, including environmental control (i.e., regulation of lighting, temperature, humidity, and carbon dioxide (CO
2) concentration in the cultivation chamber), material usage (i.e., application of substances promoting crop growth, such as compost, fertilizer, and biochar), and crop species (i.e., inherent attributes of different target plants) [
22]. While plenty of research has investigated the individual impacts of these parameters [[
23], [
24], [
25], [
26]], practical farming is typically influenced by a multitude of factors. Therefore, it is essential to carry out comprehensive research on multiple parameters in order to optimize farming strategies. Moreover, despite the advantages of efficient resource utilization, vertical farming is still limited in its application, primarily due to its increased energy input and high operating costs [
27]. Analyses have revealed a tradeoff between the costs and benefits of vertical farming [
28,
29]. While high food productivity is attainable, vertical farming may result in a larger ecological footprint in comparison with conventional methods [[
30], [
31], [
32], [
33], [
34]]. Hence, there is a need to improve the system configuration, adopt low-cost and sustainable operating strategies, and conduct further quantitative assessments to evaluate the efficacy of vertical farming. In addition, although vertical farming is perceived as highly adaptable due to its compact and integrated features [
35], there is currently a lack of discussion on its modularity and scalability. Consequently, further exploration is required to reveal its potential for wide implementation.
To promote sustainable agriculture, this study introduces an eco-friendly, cost-effective, and modular vertical farming system (
Fig. 1). The system integrates key components including cultivation chambers (supporting both soil-based and hydroponic methods), a lighting unit, a direct air capture (DAC) unit, a power generation unit, and a heating, ventilation, and air conditioning (HVAC) unit. First, we constructed a soil-based experimental setup to identify productive cultivation strategies by regulating the lighting type, CO
2 concentration, biochar application, and plant species. The results showed that, compared with the control group, the fresh weight of the vegetables in the experimental groups increased by up to 57.5%. Second, to address current concerns regarding vertical farming’s high cost and unsustainability, a novel CO
2-enrichment strategy was proposed. A self-developed rotary adsorber was utilized to capture CO
2 from the air and deliver it into the cultivation chambers, and the stable and efficient CO
2-enrichment capability of the strategy was validated. We also developed a modular and highly scalable vertical farming scheme and evaluated its performance through a techno-economic analysis and life-cycle assessment (LCA). We found that, by integrating photovoltaic-thermal (PVT) and DAC units into the farming system, the net present value (NPV) increased by 157% and the carbon footprint (0.468 kg-CO
2 equivalent·kg
−1 (CO
2eq·kg
−1)-vegetable) decreased by an order of magnitude compared with those of conventional designs. Furthermore, we performed a sensitivity analysis and assessed the effectiveness of different cultivation methods and vegetable varieties. A global-scale examination of the implementation of this strategy revealed its wide adaptability to diverse local conditions. These findings highlight the application potential of vertical farming systems and offer insights into increasing food security through ecologically friendly practices.
2. Materials and methods
2.1. Vegetable cultivation
An experimental setup was established to investigate the practical vertical farming process. The overall layout of the setup is depicted in Fig. S1 in Appendix A. The biochar used as the soil amendment was produced through the co-gasification of water hyacinth and woodchips (feedstock ratio 40 wt%:60 wt%). The water hyacinth was sourced from a Singapore local reservoir, while the woodchips were collected from local nurseries. Detailed information on the biochar preparation is provided in our previous study [
36]. The soil used in this experiment was purchased from Green Spade Pte. Ltd., Singapore [
37]. To analyze the effects of biochar application on vegetable growth, biochar weight percentages of 0 (control group), 3, 6, and 9 wt% were mixed well with soil to obtain potting substrates with a total weight of 950 g each, filling planters of about 2 L each (Table S1 in Appendix A). The applied biochar ratio was determined from our previous studies. Nine replicates were set for each vegetable and each biochar application ratio.
Pak choi (
Brassica rapa subsp.
chinensis) and Chinese kale (
Brassica oleracea) were chosen as the test crops, as both are common leafy vegetables consumed in Singapore. The cultivation experiments were conducted in a farm area in the outskirts of Singapore (1°24′58″N, 103°43′6″E). Four independent chambers were set to investigate the effects of environmental factors on vegetable growth. Chambers A, B, and C were enveloped with transparent polyethylene (PE) films to create an airtight environment, while Chamber D was covered with block-out cloth to create an airtight and dark environment. Each chamber was equipped with two fans—one for the air inlet and another for the air outlet—to facilitate ventilation and the regulation of temperature and humidity. The CO
2 concentrations in Chamber A, Chamber B, and Chamber D were regulated at approximately 1000, 700, and 700 parts per million (ppm), respectively. Sixteen light emitting diode (LED) lights (red to blue light ratio of 4:1, 30 W per light; Anhui Sanan Technology Co., Ltd., China) were installed in Chamber D to provide the light source for plant photosynthesis. A photoperiod of 16 h·d
−1was used based on previous research [
38]. Chamber C was set as the control group and was cultivated under natural sunlight, without CO
2 enrichment. The environmental conditions of the four chambers are summarized in Table S2 in Appendix A.
Each chamber contained a total of 72 pots of vegetables (i.e., both pak choi and kale, planted in substrates with four biochar application ratios, with nine replicates for each ratio). The pots were distributed vertically in a completely randomized design, with a 19 cm column distance and a 23 cm row distance. Sizes and photographs of the chambers, as well as labels and the positions of the pots, are provided in Figs. S2 and S3 in Appendix A. Irrigation was carried out through a system comprising suspended water pipes and nozzles above the vegetables, along with water pumps. Industrial-grade cloud-based Wi-Fi-data four-in-one loggers (UBiBot GS1-AETH1RS; GMM Technoworld Pte. Ltd., Singapore, detection accuracy: ±0.3 °C for temperature and ± 3% for humidity) were installed to monitor the CO
2 concentration, light intensity, temperature, and humidity in each chamber. The spectral distribution of light was measured using a portable agricultural spectrometer (OHSP-350UVP; Hangzhou Hopoo Light and Color Technology Co., Ltd., China). It should be noted that the components described above were specifically configured for experimental evaluation; for commercial deployment, different structural frameworks, construction materials, and enclosure systems would be necessary. Readers may refer to relevant design standards and technical manuals for more detailed specifications [[
39], [
40], [
41]].
The pak choi and kale seeds were directly germinated in the pots for 4 days (Days 1-4) in the absence of light. Starting on Day 5, the seeds were grown under natural sunlight (Chambers A, B, and C) or LED lights (Chamber D). The vegetables were watered once daily from the day they were sown. Fertilizer (Poly Fret 21-21-21; New Eastern 1971 Pte. Ltd., Singapore) was applied to each pot on Days 9, 26, and 39, with a recommended dose of nitrogen equivalent to field conditions. The pak choi and kale were harvested on Day 47. A detailed cultivation timeline is provided in Table S3 in Appendix A.
To evaluate vegetable growth, the leaf area was measured by photographing the laminae of each plant, followed by area determination using Image J v1.51 (National Institute of USA). On harvest day, the fresh weight of the harvested vegetables was measured using a precision electronic balance. Then the vegetables were dried in a convection oven at 70 °C for 4 days, after which they were weighed again to determine the dry weight.
2.2. CO2-enrichment strategy
A schematic diagram of the CO
2-enrichment strategy is provided in Fig. S4 in Appendix A. The CO
2 enrichment was conducted utilizing a DAC rotary adsorber that has been previously described in detail [
42]. Beyond the cultivation period, the adsorber was tested under various conditions (Table S4 in Appendix A) to study the parametric effects. Within the cultivation period, the rotary adsorber was connected to the cultivation chamber to examine its CO
2-enrichment capability. DAC is achieved through temperature swing [
42], so the majority of the energy consumed is thermal energy. For testing convenience, the rotary adsorber prototype was driven by electric heating. Therefore, in long-term cultivation experiments, to avoid high electricity costs, gas cylinders (Iwatani Corporation Pte. Ltd., Singapore) were also used to supply the CO
2 for cultivation experiments. The supplied CO
2 flow rate was controlled by the fan (for the rotary adsorber) or by regulating the rotameters (for cylinders). Air circulation within the chamber was facilitated by a fan (air flow rate: ∼100 m
3·h
−1), and the CO
2 concentration inside the chamber was continuously monitored using the UBiBot sensor mentioned above.
For the designed scale-up system, in order to both ensure that the plants’ CO
2 requirements were met and reduce the loss of CO
2, we set the air-exchange rate (
N) at the standard recommended value of 0.83 time·h
−1 [
39]. Accordingly, the flow rate (
G, m
3·h
−1) of the supply air with elevated CO
2 concentration was determined by the following equation [
43]:
where
V is the volume of the cultivation chamber (m
3), and
ƞa is the air-exchange effectiveness. For the designed system, with the ventilation supply at the bottom and the return at the top,
ƞa was estimated to be 50% [
44]. The energy consumption for DAC comprises thermal energy consumption and electrical energy consumption, both of which were calculated using the equations detailed in Note S1 in Appendix A. Based on our previous research [
42], we adopted a thermal energy consumption of 1.98 kW·h·
and an electrical energy consumption of 0.24 kW·h·
in this work for the performance evaluation of the designed scale-up system.
2.3. Performance analysis of the system
We developed a vertical farming model (Fig. S5 in Appendix A) on which the techno-economic analysis and LCA were conducted. In this model, a substrate consisting of a blend of soil and biochar is utilized for vegetable cultivation. Throughout the cultivation process, inputs such as seeds, water, fertilizers, CO2, electricity, and heat are supplied to facilitate vegetable growth. The vegetables are packaged in PE film and transported to nearby retail stores, while the substrate is transported to a nearby location for sequestration. This farming system operates continuously throughout the year, completing multiple vegetable planting cycles annually. The system is assumed to have a lifespan of 20 years, with additional calculation assumptions detailed in Table S5 in Appendix A.
2.3.1. Economic analysis
The economic performance of the system was evaluated by calculating its expenditure and revenue. The expenditure consists of capital costs and operating costs, while the revenue stems from selling the vegetables. Given the system configuration and specific dimensions of the system, the material and energy flows can be calculated. Subsequently, the costs and revenue can be determined with the additional assumptions shown in Tables S6−S9 in Appendix A. The NPV of the system at the end of life and the discounted payback period were calculated to evaluate the economic performance, using the following equations:
$\begin{array}{l} \left.\mathrm{NPV}\right|_{\text {Year }=L}=\frac{(\mathrm{YR}-\mathrm{OPEX}) \times L}{\mathrm{CRF}}-\mathrm{CAPEX}\\ \text { Discounted payback period }=\text { Year }\left.\right|_{\mathrm{NPV}=0} \end{array}$
where YR is the yearly revenue, OPEX is the yearly operating costs, L is the system lifespan, CAPEX refers to the yearly capital costs, and CRF is the capital recovery factor.
2.3.2. Environmental impact assessment
The LCA of the vertical farming system was conducted according to the the International Organization for Standardization (ISO) 14040 LCA standard. Because this study compared options for production, the assessment followed a “cradle-to-gate” perspective; that is, the system boundary encompassed the upstream processes, vegetable production and packaging, and transportation to retail stores (Fig. S6 in Appendix A). The functional unit (FU) for the LCA was 1 kg of vegetable available to consumers. By referring to our experiments, literature data, and Ecoinvent v3 databases, as well as by combining the assumptions shown in Tables S10 and S11 in Appendix A, a life-cycle inventory was developed. The impact potentials were calculated using the ReCiPe 2016 (H) Midpoint and ReCiPE 2016 (H) Endpoint method, as it can reveal the environmental impacts across broad ranges while avoiding the transfer of environmental burdens between different categories [
45,
46]. Six impact categories—namely, global warming (GW), terrestrial acidification (TA), freshwater eutrophication (FE), human non-carcinogenic toxicity (HN), land use (LU), and water consumption (WC)—were examined because they are particularly relevant for agricultural production, as revealed by prior studies [
32,
47,
48]. The calculation were implemented using the SimaPro 9.5 software.
3. Results
3.1. Vegetable growth assessment
A vertical farming experimental setup was established in a farm area in the outskirts of Singapore. Four independent chambers (A-D) were set up to investigate the influence of environmental factors on the growth of two vegetables (kale and pak choi). In Chambers A, B, and C, vegetables were cultivated under natural sunlight (wrapped with PE films), with the indoor CO2 concentrations in Chambers A and B raised to approximately 1000 and 700 ppm, respectively. In Chamber C, no CO2 fertilization was applied (leaving the indoor CO2 concentration at ∼420 ppm). In Chamber D, the vegetables were cultivated under artificial lighting (wrapped with block-out cloth), and the indoor CO2 concentration was elevated to around 700 ppm. As a result, Chambers A, B, and C served as mutual controls for each other, and Chambers B and D also formed mutual contrasts (Table S2). Furthermore, this study incorporated biochar into the soil as an amendment, and the weight percentage of biochar in the biochar/soil mixture was varied to assess its effects (Table S1). More details about the cultivation experiment are provided in Section 2.
The cultivation experiment in Singapore was conducted from April to May in 2023. The environmental data of the four chambers during the cultivation period is shown in Figs. S7 and S8 in Appendix A. The growth status of the vegetables during cultivation is depicted in
Fig. 2. The significance of the data was analyzed using the analysis of variance (ANOVA) and
t-test methods, and the results are presented in Tables S12 and S13 in Appendix A. The fresh weight and dry weight of the vegetables in different chambers are presented in
Figs. 2(a) and
(b), respectively.
The findings reveal a significant sequential increase in both fresh and dry weight from Chamber C to B to A for both vegetable types, indicating that higher CO2 concentrations facilitated the conversion of CO2 into biomass and promoted plant growth. Compared with the control group, the kale grown under CO2 concentrations of 700 and 1000 ppm showed a weight increase of 30.9% and 57.5%, respectively. For the pak choi, these percentages were 25.2% and 53.5%, respectively. A comparison of the vegetables grown in Chambers B and D showed that, for pak choi, both the fresh and dry weights in Chamber D slightly surpassed those in Chamber B, while this trend was reversed for kale. However, according to the statistical analysis, these differences were not significant at the 0.05 confidence level (Table S13). Therefore, we conclude that the growth statuses of the crops under LED light and natural sunlight were nearly identical, demonstrating the feasibility of using LED lights for crop cultivation.
Figs. 2(c) and
(d) depicts the leaf area of the vegetables across different biochar application ratios during cultivation. It is noticeable that the leaf area of both vegetables initially increased and then decreased with higher biochar ratios, and most of the data presented significant differences. The peak leaf area was observed at a biochar ratio of 6 wt% for both vegetables. These findings suggest that a moderate inclusion of biochar in the soil facilitates leaf growth, possibly due to the increased availability of nutrients in the substrate, which aligns with prior research [
49,
50]. However, even though some changes in vegetable weights were observed at different biochar application ratios, these differences were not significant, as shown in
Figs. 2(e) and
(f) and Tables S12 and S13. Overall, these results indicate that the application of biochar has a beneficial effect (with an optimal ratio of 6 wt%) in the present case because the leaves of the vegetables studied here are the primary edible part for humans.
3.2. CO2 enrichment via DAC rotary adsorber
In this study, CO
2 enrichment was achieved using a DAC rotary adsorber. The function of this device is to continuously capture CO
2 from the air and deliver it to the chamber through the rotation of the wheel-shaped adsorbent. More information on the rotary adsorber is available in our previous work [
42], and the method used for the CO
2-enrichment experiment is provided in Section 2.
Fig. 3 illustrates the CO
2-enrichment effect of the rotary adsorber. First, the performance of the rotary adsorber was evaluated under different operational conditions, as shown in Table S4.
Figs. 3(a) and
(b) presents the CO
2 concentrations at both the inlet and outlet of the adsorption section of the adsorber for Cases 1-6. By integrating these concentration curves and multiplying the process air-flow rate, the amount of captured CO
2 was determined. As shown in
Fig. 3(c), by setting the regeneration air-flow rate to 100 m
3·h
−1 and varying the regeneration temperature (Cases 1-3), the amount of captured CO
2 increased from 0.037 to 0.067 mol·min
−1. When the regeneration temperature was set to 70 °C and the regeneration air-flow rate was adjusted (Cases 4-6), the amount of captured CO
2 increased from 0.030 to 0.048 mol·min
−1. As shown in
Fig. 3(d), when the adsorber’s regeneration section outlet was connected to the cultivation chamber, the CO
2 concentration inside the chamber rapidly increased and eventually stabilized. In Cases 1-3, which involved varying the regeneration temperature of the adsorber, the indoor CO
2 concentration escalated from around 700 to 1000 ppm. This is because, when a higher regeneration temperature was used, the rotary adsorber delivered a greater amount of CO
2 into the chamber per unit time. It took approximately 90 min for the concentration to achieve a steady status, indicating a relatively swift regulatory response. Additionally,
Fig. 3(e) demonstrates the calculated CO
2 inflow based on the cultivation chamber inlet air-flow rate and indoor CO
2 concentration, which closely matches the CO
2 captured on the adsorption section of the adsorber (
Fig. 3(c)). This result indicates that the CO
2 captured by the adsorber was adequately released into the chamber without notable gas leakage.
Based on a regeneration temperature of 70 °C and a supply air-flow rate of 100 m
3·h
−1 (the same operating conditions as those in Case 1),
Fig. 3(f) depicts the continuous operation of the adsorber for 1 day. The CO
2 concentration inside the chamber initially reaches around 750 ppm at 6:00 AM before sunrise and gradually decreases thereafter. Subsequently, as the light intensity increases, the CO
2 concentration reaches its lowest point (∼600 ppm) at around 1:00 PM and then gradually increases. These concentration fluctuations can likely be attributed to variations in the net photosynthesis rate of the vegetables, which was affected by the light intensity. Overall, compared with the chamber without CO
2 enrichment (indoor concentration of 300-500 ppm), the treated chamber exhibited an average increase in CO
2 concentration of approximately 300 ppm due to the effect of the rotary adsorber. These results demonstrate the efficacy of the rotary adsorber for CO
2 enrichment.
3.3. Analysis of scale-up systems
The experiments described above verified the feasibility of increasing crop growth using the rotary adsorber, LED lights, and biochar. They also provided the practical data necessary for the further design of the farming system (i.e., the layout pattern of the planters inside the chamber, the growth status of crops, etc.). Based on these results, this section describes our design of a modular vertical farming system and our evaluation of its performance. To fulfill this goal, a vertical farming model was established, as shown in Fig. S5. The core of the model is the minimized cultivation module, within which multiple planters are vertically stacked. When several minimized modules are combined and connected with auxiliary facilities (i.e., power-generation, HVAC, and DAC units, etc.), they form a vertical farming system with tailored characteristics. Multiple variables must be considered when designing the system. For instance, connecting modules with different components results in varied system configurations, and combining different numbers of minimized modules results in systems of different sizes. Different crop types can be cultivated, and the system can be deployed in different regions. Changes in these variables may affect the system efficacy. Therefore, we examined systems with different characteristics in order to comprehensively assess their application potential.
3.3.1. Identifying the optimized system configuration
Developing an effective design for a farming system is essential in order to maximize production efficiency. However, the lack of international standards for specifying parameters in vertical farming systems makes it challenging to directly establish an appropriate scheme. To this end, we referred to the existing standard for plant factories [
39] to outline our system and developed an exemplary design. We envisioned this system as being composed of multiple minimized modules. The design of these modules was then identified; specific parameters are detailed in Note S2 and Fig. S9 in Appendix A. In contrast to traditional agriculture, our proposed design boosts planting efficiency by ensuring more consistent indoor environmental control. Furthermore, the vertical stacking of crops maximizes spatial utilization, thereby yielding a higher crop output per unit area. Moreover, by utilizing renewable energy and efficient CO
2-enrichment techniques, the issue of high energy consumption and the associated large carbon footprint can be mitigated.
We selected pak choi as the target crop. Six minimized modules were stacked together to form a medium-sized vertical farming cultivation chamber (Fig. S10 in Appendix A). The system was assumed to be constructed in Shanghai (which has a subtropical monsoon climate) because the climate there undergoes significant changes throughout the year, making it more representative from an analytical perspective. For this design, four system configurations were compared (Fig. S11 in Appendix A). In Configuration 1, the farming system consumes electricity from the power grid, and no DAC unit is attached. In Configuration 2, the farming system consumes electricity generated by a photovoltaic (PV) unit, and no DAC unit is attached. In Configuration 3, the farming system consumes electricity generated by a PV unit; a DAC unit is attached, which consumes heat supplied by the heat pump. In Configuration 4, the farming system consumes electricity generated by a PVT unit; a DAC unit is attached, which consumes heat supplied by the PVT unit. In these configurations, LED lights were employed, a 6 wt% biochar application rate was adopted, and the indoor CO2 concentration was set as 1000 ppm. More detailed information is provided in Table S5. Based on this information, we calculated the economic benefits and environmental impacts of the systems with different configurations. The calculation methods and results are provided in Section 2 and in Tables S7−S11, S14, S15 in Appendix A.
Fig. 4 presents the main outcomes generated by these calculations. As shown in
Fig. 4(a), Configuration 1 had the lowest capital expenses, since it comprised only the basic components. Configuration 2 had higher capital costs due to the investment in the PV unit. Configuration 3 emerged with the highest capital costs because of its additional components, including the PV unit, heat pump, and rotary adsorber. Configuration 4 adopted a simpler design compared with Configuration 3 by omitting the heat pump unit, which resulted in lower capital costs. As shown in
Fig. 4(b), Configuration 1 had the highest operating costs, as its reliance on grid electricity resulted in substantial expenses. By contrast, Configuration 2 boasted the lowest operating costs because all the power was derived from the off-grid PV unit. Configurations 3 and 4 had to pay for additional adsorbent costs due to the use of the rotary adsorber. Moreover, they had higher maintenance and labor fees because these systems were more complex.
Considering both expenditure and revenue, we calculated the NPV at the end of the system’s lifespan and the discounted payback period. As shown in
Fig. 4(c), due to its low capital costs, Configuration 1 had the shortest payback period. However, its NPV was also the lowest. Compared with Configuration 1, Configuration 2 had increased capital costs but significantly reduced operating costs, thereby gaining stronger profitability. Compared with those of Configuration 2, both the capital and operating costs of Configuration 3 increased. However, due to the greater yield of vegetables under CO
2 enrichment, the NPV increased, and the payback period was shortened. Configuration 4 had lower capital operating costs compared with Configuration 3; thus, the NPV was further improved and the payback period was further shortened for Configuration 4. These findings suggest that Configuration 4, which incorporates a PVT unit and a rotary adsorber, presented the best economic outcomes. Compared with the conventional design (Configuration 1), the NPV of Configuration 4 increased by 157%. Configuration 4 also outperformed the design with a PV unit (Configuration 2), achieving a comparative 113% increase in NPV and a 27% decrease in the payback period.
The environmental impacts calculated using the Endpoint method are shown in
Fig. 4(d). It is evident that, among these categories, the system’s most significant impact factor pertains to human health, followed by ecosystems and resources. Meanwhile, consistent trends emerge within each category: Configuration 1 exhibits the highest weighted environmental impact point (Pt), followed by Configurations 2, 3, and 4. The environmental impacts calculated using the Midpoint method are shown in
Figs. 4(e) and
(f).
Fig. 4(e) shows that the carbon emissions per functional unit are of the same order as the environmental impact calculated using the Endpoint method. Configuration 1 has the highest emissions due to its reliance on grid electricity, with a higher carbon footprint and lower vegetable yields. Although Configuration 3 produces more vegetables, its electricity and heat consumption are higher because of the additional components, which results in a higher carbon footprint. Configuration 2 has lower vegetable yields but consumes less energy, so its carbon footprint is lower. Configuration 4 combines the advantages of a simplified design and improved yields, resulting in the lowest impacts. More specifically, the carbon emissions of Configuration 4 were 0.468 kg-CO
2eq·kg
−1-vegetable, which is an order of magnitude lower than that of the traditional grid-electricity-powered system (Configuration 1).
Fig. 4(f) displays a breakdown of the environmental impact of Configuration 4 in terms of six categories particularly relevant for agricultural production: GW, TA, FE, HN, LU, and WC. The results for other categories are provided in Table S15. It is evident that, among five of the six categories (except for WC), energy consumption (electricity and heat) contributes the most to environmental impacts, accounting for over 50%. In terms of carbon footprint, the use of adsorbent also has a large influence. The impact of soil usage (excavation and transportation) is high as well. However, the use of biochar effectively sequesters carbon, contributing negatively with an offset of approximately 10%. Transportation and packaging also make a notable contribution toward GW, while irrigation is responsible for over 60% of WC. Overall, to further reduce the environmental footprint of the system, the most important strategy would be to increase the equipment efficiency and decrease the energy consumption. In addition, more environmentally friendly synthesis methods for the adsorbent and packaging materials are needed. Soil preparation should also be improved by increasing the soil excavation efficiency and reducing transportation distances. Moreover, the irrigation should be further optimized to reduce WC.
3.3.2. Sensitivity analysis and potential for wider deployments
Yield and selling price. The evaluation described above highlights Configuration 4 as excelling in both economic and environmental aspects, prompting a further analysis using this design. Considering that the yield and selling price of vegetables may fluctuate with different temporal and spatial conditions in practice, we conducted a sensitivity analysis to assess their impacts on the NPV. As shown in
Fig. 5(a), within the pak choi yield range of 0.025-0.043 kg·plant
−1, the selling price should be 0.7-1.25 USD·kg
−1 to make the system achieve break-even. This price range falls within the common fluctuation range of pak choi prices, indicating that the system has strong profitability against variations in vegetable yields [
51]. If the vegetable quality can be improved through technological advancements without additional costs and sold at prices of 1.2-2.0 USD·kg
−1, the NPV can be greatly increased to 340 000-1 140 000 USD.
System size. The minimized modules are uniformly designed (featuring consistent dimensions, internal architecture, and assembly techniques) and can be stacked to form a system in a manner similar to building with blocks (Fig. S12 in Appendix A). This modular nature confers benefits such as reduced labor requirements, cost savings, and expedited construction timelines. Additionally, the geometric structure of the system can be easily adjusted by changing the quantity of these minimized modules [
52]. Given the potential for diverse scenarios necessitating varying system scales, we investigated the impact of system size. In addition to the design of a medium-sized system with six minimized modules mentioned above, we further evaluated small (one minimized module) and large (12 minimized modules) vertical farming systems, as shown in Figs. S13 and S14 in Appendix A. In
Fig. 5(b), it can be seen that, as the system size increases, the NPV of the system improves, while the payback period shortens. This result indicates that a larger system generates greater economic benefits. The economic results are also listed in Table S16 in Appendix A. Additionally, as shown in
Fig. 5(c), the carbon emissions per functional unit decrease with an increase in system size. Other environmental impacts are shown in Table S17 in Appendix A. Overall, these results indicate that a larger system presents a better overall performance because, with increasing size, the increase in the benefits of the vertical farming system surpasses the increase in cost. Therefore, promoting the system to expand its layout will lead to more efficient food production. In our design, each tier contains five vertically stacked layers of crops. Different sizes of vertical farm systems can accommodate 1-3 tiers, resulting in a total of 5-15 vegetable layers. Depending on site-specific conditions, our system allows for the additional vertical stacking of floors to achieve even higher planting densities per unit area, demonstrating the advantage of a modular design.
Vegetable species. The choice of vegetable type may influence the system’s efficacy due to differences in cultivation methods, growth cycles, and vegetable sales. To further explore the potential of the system, we compared the effects of cultivating four different types of vegetables (lettuce, pak choi, kale, and tomato). The systems were again assumed to be located in Shanghai. A method similar to the above design was used to calculate the performance of the systems. One difference was that we adjusted the structure of the minimized module according to the characteristics of the vegetables. The input data was also changed based on the growth information of each crop. The data for kale was taken from the experiment in this work, while that for lettuce [
53] and tomato [
54] came from the literature. The results are provided in Tables S18 and S19 in Appendix A. As shown in
Fig. 5(d), lettuce cultivation had the highest NPV, the shortest payback period, and the lowest carbon footprint. Pak choi cultivation yielded slightly lower benefits and had a larger carbon footprint, while kale cultivation had the lowest outcome and an even larger carbon footprint. This is because lettuce has a higher selling price and yield, while kale shows the opposite trend. Growing tomatoes brought good economic benefits but had the highest carbon emissions per functional unit, due to the difference in yield between fruit vegetables and leafy vegetables. Thus, when deciding the target vegetable species, it is necessary to consider consumer demand and the cultivation conditions. Moreover, it is essential to take into account both economic and environmental factors in order to achieve the best comprehensive benefits.
Soil-based and hydroponic cultivation. In the above analysis, we primarily evaluated cultivation approaches using biochar-amended soil as the growing media. At present, hydroponic technology is widely adopted in vertical farming. Therefore, we also assessed the potential effects of integrating hydroponics into this system using data extracted from the literature [[
55], [
56], [
57]]. The results (Table S20 in Appendix A) indicate that, for a hydroponic system, the initial equipment investments are higher compared with those for soil-based cultivation, while the operational costs are lower. Furthermore, hydroponics yields higher vegetable production. Correspondingly, the hydroponic system exhibits a higher NPV and a shorter payback period (
Fig. 5(e)). In terms of environmental impact (
Fig. 5(f); Table S21 in Appendix A), the hydroponic system results in a larger carbon footprint and lower LU. Its effects on human non-carcinogenic toxicity are comparable to those of soil-based agriculture, while its impact on TA is smaller. However, the hydroponic system contributes significantly more to FE. On the other hand, its WC is notably lower. These characteristics may be attributed to the continuous filtration and processing of nutrition solution and the highly efficient recycling of water in the hydroponic system. Overall, these findings suggest that hydroponic systems may present higher initial economic barriers but can generate greater added value over long-term operation. Hydroponic and soil-based cultivation exhibit distinct environmental impact profiles, indicating that the optimal choice depends on local conditions.
Deployment location. Another important consideration for the design is the location of the system. As such, we selected seven cities (Shanghai, Singapore, Cairo, Barcelona, Brasilia, New York City, and Sydney) worldwide with diverse climatic features to deploy the system cultivating pak choi and evaluated the system’s economic and environmental performance in each location. A calculation method similar to that used in the above investigations was employed; however, in this case, we changed the input—including solar energy intensity, energy consumption for indoor environment regulation, vegetable selling price, transportation, and so forth—to account for the varying deployment locations of the system. The data for solar energy resources [
58], climate conditions [
59], and sale prices [
60] were collected from the Internet. The results are shown in Tables S22 and S23 and Fig. S15 in Appendix A. Local solar irradiance had a significant impact on system performance, as the system was designed to be powered by solar energy. Therefore, the systems deployed in Singapore, Barcelona, and Sydney, where there is ample sunlight, had higher NPVs while generating smaller carbon footprints. Among these locations, Singapore has a relatively consistent climate throughout the year, with a photoperiod of around 12 h for nearly all clear days, making it possible to use natural sunlight to cultivate the vegetables and significantly reducing the energy consumption of the lighting unit [
33]. Moreover, the vegetable selling price is high in Singapore. Due to these factors, the Singapore system exhibited the best overall performance. Shanghai and New York City both have milder climates, but New York City has more abundant solar energy than Shanghai, resulting in better performance of the system in New York City. Cairo has the strongest sunlight but also has higher temperatures throughout the year; thus, the Cairo system showed increased energy consumption for indoor environmental control and consequently had a larger carbon footprint. The climate in Brasilia is suitable for vegetable cultivation, leading to a smaller carbon footprint for that system. However, the vegetable selling price in Brasilia is lower, resulting in a relatively reduced NPV. Overall, the proposed system can generate desirable economic and environmental benefits when deployed in different locations, indicating its potential for widespread adoption.
3.3.3. Comparison with current studies
To gain a deeper understanding of the results, we conducted a comparison of our research with the existing literature. To compare indicators within the same scope, we selected studies with a “cradle-to-gate” system boundary. Given the lack of a unified standard for evaluating farming systems [
61], it is important to acknowledge differences in methodologies across various studies when making comparisons. Consequently, the main assumptions from different studies are summarized in Table S24 in Appendix A for reference [[
62], [
63], [
64], [
65], [
66], [
67], [
68], [
69], [
70], [
71], [
72]].
Fig. 6 shows the crop production costs, electricity consumption, and carbon footprints of farming systems as reported in current studies. It indicates that, except for a few cases, the indicators across different studies are relatively close (with fluctuations remaining within an order of magnitude).
We first compared our system with conventional greenhouse cultivation systems, which possess a simpler structure and have a lower requirement for indoor environmental control. Natural sunlight is typically utilized for vegetable cultivation. As a result, such conventional systems present a low production cost (Study A in
Fig. 6(a)) and reduced electricity consumption (Study H in
Fig. 6(b)). However, greenhouses consume fossil fuels for heating, so their carbon footprints are not correspondingly small (Study H in
Fig. 6(c)). Our system has a production cost comparable to that of a greenhouse. Although our system consumes more electricity, it is powered by renewable energy, thereby maintaining a small carbon footprint. Most importantly, our system employs a vertical stacking arrangement for vegetable pots, thus achieving a yield per unit area of 147 kg·m
−2·a
−1, which is significantly higher than that of a conventional greenhouse (21.5 kg·m
−2·a
−1 [
62]). This finding highlights the compact and productive features of the vertical farming system.
Comparisons with other farming systems provided further insights. The high production cost in Study A can be attributed to the assumption of higher labor and management costs in the analysis. The broad range of carbon emissions in Study G results from its comparison between scenarios involving electricity consumption with a high proportion of fossil fuels and those utilizing clean renewable energy sources. The present work sets the unit cost of vegetable production as ranging from 0.71 to 1.06 USD·kg−1, electricity consumption as ranging from 2.35 to 3.95 (kW·h)·kg−1, and carbon footprints as ranging from 0.358 to 0.562 kg-CO2eq·kg−1-vegetable. These values are relatively low compared with the indicators reported in existing studies, which to some extent reflects the favorable economic and environmental performance of this system.
4. Discussion
In this study, an economical, environmentally friendly, and modular vertical farming system was proposed. Through experiments, this study examined the combined effects of multiple parameters on vegetable growth. The experimental results showed an increase of 57.5% in vegetable yield under regulated cultivation conditions. Additionally, a self-developed rotary adsorber was employed to achieve CO2 enrichment. Tests demonstrated the device’s capability to continuously capture CO2 from the air and deliver it to the cultivation chamber, maintaining a stable CO2 concentration. A techno-economic analysis and an LCA were conducted to evaluate the system’s performance. Four distinct configurations were compared. The results show that the vertical farming system equipped with a CO2-enrichment unit and powered by a PVT unit achieved a 157% increase in NPV compared with the conventional design. Meanwhile, a small carbon footprint (0.468 kg-CO2eq·kg−1-vegetable) was maintained in the vegetable production process of the system. These findings highlight the promising prospects of the design.
The proposed system features modular characteristics: It is composed of minimally sized building units with identical geometric structures that serve as independent elements for forming scalable constructions. The uniformity of each module simplifies the process of scaling up the system, akin to adding more building blocks, which in turn significantly diminishes the complexity of construction. Auxiliary components such as the rotary adsorber and solar panels also possess scalable properties and can be scaled up accordingly based on the dimensions of the system. The minimized modules can be prefabricated in factories and conveniently transported. The onsite assembly process is streamlined to simply consist of stacking and integration. This approach not only expedites construction time but also reduces costs, as the assembly process is less challenging and requires less labor compared with conventional construction methods.
In future work, the dimensions of the minimal modules and the layout of the internal planters can be refined according to the characteristics of the deployment area and the cultivated crops. Judicious selection of the locations for ventilation ports, CO2 release, and sensor measurement is crucial to ensure uniformity of the indoor environment. Subsequent research can integrate fluid mechanics principles to construct mathematical models, thereby ascertaining an optimized design that increases plant production. Furthermore, research into optimal stacking configurations such as cylindrical, linear, or pyramidal arrangements will be necessary. Later studies can explore favorable combinations to achieve a certain level of solar utilization for crop cultivation while minimizing the land footprint. Improving stacking methods to increase the material and energy flow between modules could also facilitate indoor environmental control. Together, these strategies can contribute to reducing the overall system costs and environmental impact.
Further technological advances in the farming system can be achieved by optimizing the supply chain [
73]. Onsite production and sale are a desirable scenario to minimize the cost and carbon footprint related to transportation. If this is not possible, optimizations can be performed to reduce transportation loads between raw material sources, the farming system, and sale points. Additionally, it is encouraged to develop a variety of energy sources to power the farming system [
74]. In addition to solar energy, wind power, hydropower, geothermal energy, and industrial waste heat can be explored. Research should focus on how to arrange, utilize, and recycle these energy sources to increase sustainability. With the emergence of various technological routes, it is also important to develop a set of standardized design specifications. Correspondingly, there is a need to establish a set of unified agricultural system evaluation methods that are as universal as possible, so as to more objectively compare the performance of different systems. Furthermore, by combining state-of-art information technology and leveraging artificial intelligence for reinforced learning, comprehensive and precise data collection related to crops can be achieved, enabling efficient response control of indoor environments [
75]. This approach holds potential to increase both crop yield and quality while consuming less energy. Employing these strategies together is expected to further improve the efficiency of vertical farming, thereby increasing food security and achieving more sustainable societal development.
CRediT authorship contribution statement
Junye Wu: Writing - review & editing, Writing - original draft, Resources, Methodology, Investigation. Yoke Wang Cheng: Writing - review & editing, Resources, Investigation. Guiying Lin: Writing - review & editing, Resources, Investigation. Dequan Xu: Investigation. Yiying Wang: Resources. Clive Chong: Resources, Methodology. Yanjun Dai: Funding acquisition. Chi-Hwa Wang: Writing - review & editing, Supervision, Funding acquisition, Conceptualization. Tianshu Ge: Writing - review & editing, Supervision, Funding acquisition, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This research was financed by the National Research Foundation (NRF), Prime Minister’s Office, Singapore, under its Campus for Research Excellence and Technological Enterprise (CREATE) program (A-0001032-01-00), and the National Natural Science Foundation of China (52376011).
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2025.07.043.
PDF
(2420KB)
