Climate-driven fuel-cost-based assessment of raw and torrefied biomass for greenhouse heating across different climatic regions of Korea Authors: Sunyong Park, Kwang Cheol Oh, DaeHyun Kim Published: 2026 ABSTRACT This study assessed the fuel requirement and annual fuel-based heating cost of conventional diesel (F0), raw biomass (F1), low-severity torrefied biomass (F2), and high-severity torrefied biomass (F3) for greenhouse heating across different climatic regions of Korea. Heating demand was estimated for seven major horticultural regions using climate-based data and crop-specific day/night setpoint temperatures and cultivation periods for strawberry, tomato, and paprika, which represent greenhouse crops with contrasting thermal requirements and production schedules. Fuel consumption and annual fuel-based heating cost were then compared across fuel options. Additional analyses included price sensitivity testing, an equal-efficiency scenario, transport-distance scenarios, break-even analysis, and decision-boundary mapping. Annual heating demand was lowest in Jeju and highest in Chuncheon for all crops, and increased in the order of strawberry, tomato, and paprika. Under the baseline assumptions, high-severity torrefied biomass (F3) showed the lowest estimated fuel requirement and annual fuel-based heating cost across all crops and regions. This ranking was maintained under the equal-efficiency scenario and under transport-distance scenarios from 0 to 200 km, in which both the fixed biomass logistics cost and the distance-dependent transport cost were included. In the decision-space analysis, F3 occupied the largest share (53.49%), followed by F1 (39.34%) and F2 (7.17%), indicating that F3 was not universally dominant across all price conditions. The break-even prices relative to raw biomass were 0.294 USD/kg for F2 and 0.442 USD/kg for F3 under the baseline transport condition. Overall, under the assumed fuel properties, delivered-price structure, transport-cost assumptions, and standardised greenhouse conditions, high-severity torrefied biomass showed favourable fuel-cost performance among the evaluated scenarios. However, its competitiveness was conditional on the interaction between raw biomass price, torrefaction premium, transport distance, and greenhouse heat-demand conditions. Therefore, the results should be interpreted as a screening-level fuel-cost comparison rather than as a direct prediction of farm-level heating bills.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-59325-w. Introduction Heating is an essential management practice in protected horticulture because it enables crop production during winter and other low-temperature periods1–3. At the same time, however, it represents one of the largest components of greenhouse annual heating fuel costs. Maintaining crop-specific thermal conditions requires substantial energy input, and the associated heating expenses directly affect both the economic viability and sustainability of greenhouse production systems. In countries such as Korea, where seasonal temperature variation is pronounced and climatic conditions differ considerably among regions, heating demand can vary substantially even under identical greenhouse structures and crop conditions4–6. These regional differences highlight the need for climate-responsive assessments of greenhouse heating demand that explicitly account for crop temperature requirements. Diesel and other fossil fuels are still widely used for greenhouse heating, but growing concerns over fuel price volatility and carbon neutrality have increased interest in alternative energy sources7–9. Among these, woody biomass has attracted considerable attention because of its potential to utilise domestic forest and rural resources as renewable heating fuel. Raw woody biomass offers the advantages of relatively low cost and broad availability, but its practical application is constrained by high moisture content, low energy density, and limited efficiency in storage and transport10–12. Torrefaction has been proposed as a thermochemical upgrading strategy to address these limitations, as torrefied biomass generally exhibits higher heating value, improved hydrophobicity, enhanced grindability, and better storage and transport characteristics than raw biomass13,14. However, torrefaction also introduces additional processing costs, and therefore improvements in fuel quality do not necessarily guarantee improved fuel-cost performance in greenhouse heating applications15,16. Previous studies on biomass and torrefied fuels have primarily focused on fuel property modification, reaction behaviour, combustion performance, or their application in power generation and industrial energy systems17–20. By contrast, relatively few studies have evaluated greenhouse heating demand in a way that directly links regional climate conditions, crop cultivation periods, and day/night temperature setpoints to fuel selection and heating cost. In many cases, existing studies have been limited to a single region, a narrow set of fuel options, or equipment-oriented performance analysis6,21,22. As a result, they have provided only limited guidance on a more practical question: under which climatic, price, and transport conditions is a given fuel option more economically favourable for greenhouse heating? Addressing this question requires more than simple fuel comparison; it requires an integrated fuel-cost-based assessment framework that considers climate, crop requirements, fuel efficiency, price structure, and logistics simultaneously23. In this study, greenhouse heating demand required to maintain crop-specific setpoint temperatures was estimated for major horticultural production regions in Korea using climate-based data. Based on the estimated heating demand, fuel requirement and annual fuel-based heating cost were compared for diesel, raw biomass, low-severity torrefied biomass, and high-severity torrefied biomass. The purpose of this study was not to reproduce dynamic indoor temperature changes or detailed boiler behaviour, but to quantify the heating demand needed to maintain target cultivation temperatures under different regional climatic conditions and to evaluate the fuel-cost implications of fuel choice under those conditions. Therefore, this study should be interpreted as a climate-driven, fuel-cost-based assessment of greenhouse heating rather than a full investment-level techno-economic analysis. Beyond a simple cost comparison, the effects of fuel price, combustion efficiency, transport distance, break-even conditions, and fuel-selection boundaries were further examined through sensitivity analysis and decision mapping. In doing so, this study aimed to show that the economic competitiveness of biomass-based heating fuels is not fixed, but depends on the interaction among regional climate, crop-specific temperature requirements, fuel properties, delivered fuel price, and logistics conditions. By identifying the conditions under which raw biomass or torrefied biomass becomes more advantageous, this study provides a practical decision-support basis for fuel selection in greenhouse heating systems across major horticultural regions of Korea. Based on these considerations, we hypothesised that regional climate conditions and crop-specific thermal requirements would substantially alter annual greenhouse heating demand and thereby influence the economic competitiveness of biomass-based heating fuels. We further hypothesised that torrefied biomass would not be universally superior to raw biomass, but would become economically favourable only when the reduction in fuel requirement caused by higher energy density and combustion efficiency offsets the torrefaction-related price premium. Therefore, the suitability of raw and torrefied biomass was expected to depend on the combined effects of climate, crop setpoint, fuel properties, delivered fuel price, and logistics conditions. Materials and methods Study overview, regional selection, and crop selection This study quantified the greenhouse heating load required to maintain crop-specific setpoint temperatures in major horticultural production regions of Korea using climate-based data, and subsequently compared fuel requirements and annual fuel-based heating costs among fuel types. The study regions included Gangneung, Chuncheon, Hongseong, Sangju, Gimje, Miryang, and Jeju, while the target crops were strawberry, tomato, and paprika. The seven regions were selected to represent major horticultural production areas and contrasting climatic conditions in Korea. Gangneung was included as an eastern coastal region, Chuncheon as a cold inland region, Hongseong and Sangju as central inland agricultural regions, Gimje as a western plain region, Miryang as a southern inland region, and Jeju as a warm island region. This regional selection was intended to cover a broad range of outdoor temperature conditions relevant to greenhouse heating demand in Korea, rather than to represent a single local production case. Climate data for each region were obtained from the Meteorological Data Open Portal operated by the Korea Meteorological Administration (KMA)24. Rather than reproducing indoor temperature dynamics or boiler operation, this study focused on estimating the heat demand required to satisfy crop-specific temperature requirements under regional climatic conditions and on evaluating the economic implications of fuel choice3,21. Greenhouse and crop conditions Strawberry, tomato, and paprika were selected because they are representative greenhouse horticultural crops in Korea with distinct thermal requirements and cultivation schedules25,26. Strawberry represents a seasonal crop with relatively low night-time heating requirements, tomato represents an intermediate thermal-demand crop with an extended cultivation period, and paprika represents a high-temperature crop with year-round cultivation. Therefore, these three crops provide contrasting heating-demand scenarios for evaluating fuel requirement and annual fuel-based heating cost. The greenhouse considered in this study was assumed to be a multi-span glass greenhouse. The floor area was fixed at 1,000 m2, and the envelope-to-floor area ratio (r) was set to 1.4, resulting in an envelope area of 1,400 m227,28. The overall heat transfer coefficient (U) was adopted as 5.9 W/m2K based on previous literature, and the overall heat loss coefficient (UA) was determined as follows27,28. The U-value was treated as a fixed representative parameter to maintain a standardised greenhouse structure across all regions, crops, and fuel scenarios. This was done to isolate the effects of regional outdoor temperature, crop setpoint, fuel properties, delivered fuel price, and transport assumptions on greenhouse heating fuel-cost performance, rather than to assess the sensitivity of heating demand to greenhouse insulation performance. Accordingly, variations in U-value caused by greenhouse design, covering material, humidity, wind speed, material ageing, condensation, or thermal-screen operation were not dynamically modelled. Crop-specific day/night setpoint temperatures and cultivation periods were defined as follows: 20/8 ℃ for strawberry from October to June, 22/17 ℃ for tomato from August to June, and 26/18 ℃ for paprika throughout the year29,30.1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:UA=U\times\:{A}_{cover}=5.9\times\:1400=8260W/K$$\end{document} The crop setpoint temperatures and cultivation periods were used as representative scenario conditions rather than farm-level statistical samples. Therefore, the study does not statistically infer the behaviour of all farms cultivating the same crop species. Instead, it evaluates how commonly used crop-specific thermal requirements influence heating demand and fuel-cost outcomes under standardised greenhouse conditions. Climate data and heating load estimation Regional outdoor temperature conditions were constructed by extracting the month-day-time and air temperature data from the 10-year average sheet for each location. The ambient temperature data used for the heating-load calculation, together with the 10-year climatic summary including precipitation, wind speed, and relative humidity, are provided in Supplementary Dataset S1 to improve transparency and reproducibility. The reference year was standardised to 2024, a leap year, and sunrise and sunset times were calculated to distinguish daytime from night-time so that crop-specific setpoint temperatures could be applied accordingly. Time periods outside the cultivation season were excluded from the calculation. When the outdoor air temperature exceeded the crop-specific setpoint temperature, the heating load was set to zero through the max(Tset, t−Tout, t,0) term. Thus, high-temperature periods during which combustion-based heating was unnecessary were automatically excluded from the heating-energy calculation. Cooling demand, ventilation cooling, heat-pump operation, and electricity-based temperature control were outside the scope of this study because the objective was to compare fuel-based heating options during heating-demand periods. The heating load at time t, Qheat, t, was calculated as follows:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Q}_{heat,t}=UA\times\:max({T}_{set,t}-{T}_{out,t},0)$$\end{document} where, Qheat, t is the hourly heating load (W), Tset, t is the crop-specific setpoint temperature at time t (℃), Tout, t is the outdoor air temperature at time t (℃), and UA is the overall heat loss coefficient of the greenhouse (W/K). Total heating energy was obtained by integrating the hourly heating load over time with a time step of Δt = 1 h. In this study, dynamic indoor temperature variation was not considered. In addition, solar heat gain, ventilation, thermal curtains, heat storage, and internal heat generation were not included in the analysis.3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{E}_{heat}={\sum\:}_{\mathrm{t}=1}^{\mathrm{n}}{Q}_{heat,t}\varDelta\:t$$\end{document} The heating-load model used in this study represents a theoretical envelope heat-loss demand required to maintain crop-specific setpoint temperatures under outdoor temperature forcing. Therefore, the calculated heating demand should be interpreted as an upper-bound or conservative estimate of heating demand rather than as a direct prediction of metered greenhouse energy consumption. In commercial greenhouses, thermal screens, solar heat gain, curtain operation, heat storage, ventilation control, and indoor thermal inertia can substantially reduce the net heating requirement. These factors were not explicitly modelled because the purpose of this study was to compare relative fuel-cost performance under standardised regional climate and crop-setpoint conditions, rather than to reproduce farm-specific indoor temperature dynamics or boiler operation. Fuel scenarios and cost calculation The fuel scenarios considered in this study comprised conventional diesel (F0), raw biomass (F1), low-severity torrefied biomass (F2), and high-severity torrefied biomass (F3). All monetary values were converted from Korean won (KRW) to United States dollars (USD) using a rounded exchange rate of 1 USD = 1,350 KRW. This rounded rate was adopted to reflect the approximate average USD/KRW exchange rate in 2024. USD values are used as the primary monetary unit in the revised manuscript, while the original KRW values are provided in parentheses where necessary to maintain traceability to Korean market-price assumptions. The diesel price was set at 1.333 USD/L based on March 2026, while the tax-exempt agricultural diesel price was set at 0.852 USD/L for comparison. The lower heating value (LHV) of diesel was assumed to be 35.4 MJ/L with a density of 0.835 kg/L, corresponding to a mass-based LHV of 42.40 MJ/kg. For biomass-based fuels, the LHVs were set at 12.87 MJ/kg for F1, 20.12 MJ/kg for F2, and 26.45 MJ/kg for F331. The baseline combustion efficiencies were assumed to be 0.80, 0.82, and 0.84 for F1, F2, and F3, respectively. These values were used as scenario assumptions reflecting the expected improvement in effective combustion performance with decreasing moisture content, increasing heating value, and improved fuel quality after torrefaction. Because actual boiler efficiency depends on boiler design, fuel feeding system, part-load operation, combustion control, moisture variation, and fuel particle characteristics, an equal-efficiency scenario in which F1, F2, and F3 were all fixed at 0.80 was additionally analysed to avoid over-reliance on the baseline efficiency gradient32,33. Because commercial price data for torrefied biomass chips and pellets are not yet widely available in Korea, the biomass fuel prices used in this study were treated as reference-informed scenario assumptions rather than directly observed market prices. Previous studies and reports on biomass and torrefied pellet production were used as proxy references to account for the additional cost associated with drying, thermal treatment, mass loss, and densification34,35. Biomass fuel prices were defined using the raw biomass price (Praw) and torrefaction premium (Ppremium). Under the baseline condition, Praw = 0.133 USD/kg and Ppremium = 0.059 USD/kg. The purchase prices of F1, F2, and F3 were calculated as follows:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{F1}={P}_{raw}$$\end{document}5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{F2}={P}_{raw}+1.5{P}_{premium}$$\end{document}6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{F3}={P}_{raw}+3.0{P}_{premium}$$\end{document} These equations resulted in purchase prices of 0.133, 0.222, and 0.311 USD/kg for F1, F2, and F3, respectively34,36. These values were intended to represent severity-dependent torrefaction price scenarios, not fixed commercial quotations. For solid biomass fuels, the delivered price was calculated as the sum of the purchase price, a fixed logistics component, and a distance-dependent transport component. The fixed logistics component was set at 0.133 USD/kg, and was applied uniformly to F1–F3 to represent collection, handling, storage, and basic delivery-related costs. In the transport-distance scenario, an additional distance-dependent transport cost was calculated using a ton-kilometre cost of 0.244 USD/ton-km. Therefore, the delivered biomass price at transport distance d was expressed as follows:7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{i,delivered,d}={P}_{i,purchase}+{P}_{fixed\:logistics}+\frac{d\times\:{c}_{tkm}}{1000}\:\:\:\:\:(i\in\:\left\{F1,F2,F3\right\})$$\end{document} where Pi, purchase is the purchase price of biomass fuel i, Pfixed logistics is the fixed biomass logistics cost of 0.133 USD/kg, d is the transport distance in km, and ctkm is the unit transport cost of 0.244 USD/ton-km. The factor 1000 converts the ton-based transport cost to a kg-based fuel price. The fuel requirement for fuel i was calculated using the total heating energy demand, lower heating value, and combustion efficiency as follows:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{m}_{i}=\frac{{E}_{heat}}{LH{V}_{i}\times\:{\eta\:}_{i}}$$\end{document} where mi is the required fuel mass of fuel i (kg), Eheat is the total heating energy demand (MJ), LHVi is the lower heating value of fuel i (MJ/kg), and ηi is the combustion efficiency of fuel i. The annual fuel-based heating cost was then estimated by multiplying the required fuel mass by the delivered fuel price:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{i,d}={m}_{i}\times\:{P}_{i,delivered,d}$$\end{document} where Ci, d is the annual fuel-based heating cost of fuel i at transport distance d (USD), and Pi, delivered, d is the delivered fuel price at transport distance d (USD/kg). For liquid fuels, volumetric prices were converted to mass-based prices using fuel density before fuel-cost calculation. The distance-dependent transport cost was calculated using the ton-kilometre approach:10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{distance,i,d}={m}_{i}\times\:d\left(km\right)\times\:{c}_{tkm}$$\end{document} where Cdistance, i,d is the distance-dependent transport cost of fuel i at distance d, mi is the transported fuel mass in tons, d is the transport distance in km, and ctkm is the transport cost of 0.244 USD/ton-km. The fuel properties and cost assumptions used in the fuel-cost calculation are summarised in Table 1. Table 1Fuel properties and cost assumptions used in the fuel-cost calculation.FuelLHV [MJ/kg]Efficiency [-]Purchase priceFixed logisticsDistance-dependent transportDelivered price at 0 kmBasisF0, conventional diesel42.400.851.333 USD/L—\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{m}_{i}d\times\:0.244$$\end{document}USD—Literature/official priceF0, tax-exempt diesel42.400.850.852 USD/L—\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{m}_{i}d\times\:0.244$$\end{document}USD—Official price/scenarioF1, raw biomass12.870.800.133 USD/kg0.133 USD/kg\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d\times\:0.244}{1000}\:$$\end{document}USD/kg0.267 USD/kgLiterature-informed scenarioF2, low-severity torrefied biomass20.120.820.222 USD/kg0.133 USD/kg\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d\times\:0.244}{1000}\:$$\end{document}USD/kg0.356 USD/kgLiterature-informed scenarioF3, high-severity torrefied biomass26.450.840.311 USD/kg0.133 USD/kg\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d\times\:0.244}{1000}\:$$\end{document}USD/kg0.444 USD/kgLiterature-informed scenario Table 1Fuel properties and cost assumptions used in the fuel-cost calculation.FuelLHV [MJ/kg]Efficiency [-]Purchase priceFixed logisticsDistance-dependent transportDelivered price at 0 kmBasisF0, conventional diesel42.400.851.333 USD/L—\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{m}_{i}d\times\:0.244$$\end{document}USD—Literature/official priceF0, tax-exempt diesel42.400.850.852 USD/L—\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{m}_{i}d\times\:0.244$$\end{document}USD—Official price/scenarioF1, raw biomass12.870.800.133 USD/kg0.133 USD/kg\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d\times\:0.244}{1000}\:$$\end{document}USD/kg0.267 USD/kgLiterature-informed scenarioF2, low-severity torrefied biomass20.120.820.222 USD/kg0.133 USD/kg\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d\times\:0.244}{1000}\:$$\end{document}USD/kg0.356 USD/kgLiterature-informed scenarioF3, high-severity torrefied biomass26.450.840.311 USD/kg0.133 USD/kg\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d\times\:0.244}{1000}\:$$\end{document}USD/kg0.444 USD/kgLiterature-informed scenario Fuel properties and cost assumptions used in the fuel-cost calculation. For solid biomass fuels, the delivered price shown in the table corresponds to the 0 km transport-distance condition. In the transport-distance scenarios, an additional distance-dependent transport cost of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:(d\times\:0.244)/1000\:$$\end{document} USD/kg was added to the fixed logistics component. Sensitivity, break-even, and decision-boundary analysis Sensitivity analysis was conducted using three structures. In Sensitivity A, the prices of F2 and F3 were varied within ± 30% of their baseline values. In Sensitivity B, a price structure linking the raw biomass price and torrefaction premium was applied. In Sensitivity C, the prices of conventional diesel and tax-exempt agricultural diesel were varied within ± 30% of their baseline values. In addition, under the equal-efficiency scenario, the combustion efficiencies of F1, F2, and F3 were all fixed at 0.80 to isolate the effect of fuel quality from that of efficiency. Transport-distance scenarios were set at 0, 25, 50, 75, 100, 150, and 200 km. In these scenarios, the distance-dependent transport cost was added to both liquid and solid biomass fuels using the same ton-kilometre cost. For solid biomass fuels, this distance-dependent component was added in addition to the fixed biomass logistics component of 0.133 USD/kg. The break-even point was defined as the delivered fuel price at which the total annual fuel-based heating cost of torrefied biomass (F2 or F3) became equal to that of raw biomass (F1). Accordingly, the break-even delivered price of torrefied fuel k, Pk, BEP, was calculated as follows:11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{k,BE}=\frac{{m}_{F1}}{{m}_{k}}{P}_{F1,delivered}$$\end{document} where mF1 is the required mass of raw biomass (kg), mk is the required mass of torrefied fuel k (kg), and PF1,delivered is the delivered fuel price of F1 (USD/kg). For the decision-boundary analysis, the raw biomass price and torrefaction premium were varied simultaneously, and the fuel showing the lowest annual fuel-based heating cost under each price combination was identified. The winning fuel was defined as:12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:Winner=\mathrm{arg}\underset{i\in\:\left\{F1,F2,F3\right\}}{\mathrm{min}}{C}_{i}\:$$\end{document} This analysis allowed the price regions in which F1, F2, and F3 were economically preferred to be distinguished. Results and discussion Comparison of greenhouse heating demand by region and crop condition Annual greenhouse heating demand was estimated by incorporating the 10-year average outdoor temperature conditions for each region, together with crop-specific day/night setpoint temperatures and cultivation periods. The results showed that, for all crops, Jeju had the lowest heating demand, whereas Chuncheon had the highest (Fig. 1). For strawberry, the annual heating demand was 956.19 GJ/year in Gangneung, 1,446.86 GJ/year in Chuncheon, and 554.52 GJ/year in Jeju. For tomato, the corresponding values were 1,720.26, 2,296.71, and 1,185.37 GJ/year, respectively, while for paprika they were 2,197.46, 2,773.27, and 1,634.65 GJ/year. These results indicate that, even under the same greenhouse structure, the required heating energy can vary substantially depending on regional climatic conditions. In particular, heating demand increased markedly in colder inland regions. A consistent pattern was also observed among crops. In all regions, annual heating demand increased in the order of strawberry, tomato, and paprika. Strawberry exhibited the lowest heating demand because of its relatively low night-time setpoint temperature and limited cultivation period. Tomato required more heat owing to its higher day/night setpoint temperatures. Paprika showed the greatest heating demand among the three crops because it combined the highest setpoint temperatures with year-round cultivation. For example, in Chuncheon, the heating demand was 1,446.86 GJ/year for strawberry, 2,296.71 GJ/year for tomato, and 2,773.27 GJ/year for paprika, indicating that crop choice itself structurally affects the level of greenhouse heating demand. These findings show that greenhouse heating demand cannot be explained simply by differences in average regional temperature. Rather, the heating load reflects the cumulative difference between outdoor temperature and crop-specific setpoint temperature over time, together with the duration of the heating period. Accordingly, crops with lower night-time setpoint temperatures and shorter cultivation periods tend to have lower heating demand, whereas crops with higher temperature requirements and longer cultivation periods generate substantially greater heat demand even under the same regional climate. In this respect, the regional and crop-dependent differences shown in Fig. 1 provide the basis for interpreting subsequent differences in fuel requirement and annual fuel-based heating cost. In addition, the pattern of minimum heating demand in Jeju and maximum heating demand in Chuncheon was consistently repeated across all crops. This suggests that the influence of regional climatic variation on greenhouse heating demand was stable and systematic rather than crop-specific. This confirms that the selected regional set captured a meaningful climatic gradient for greenhouse heating assessment, while the selected crops provided distinct heating-demand levels associated with their cultivation periods and setpoint temperatures. Fig. 1Annual heating demand according to region and crop. Annual heating demand according to region and crop. Variation in annual fuel requirement according to fuel properties The annual fuel requirement needed to satisfy the same heating demand varied substantially depending on the lower heating value and combustion efficiency of each fuel (Fig. 2). Across all regions and crops, high-severity torrefied biomass (F3) consistently showed the lowest fuel requirement, whereas raw biomass (F1) showed the highest. Under the Chuncheon condition, for example, the annual fuel requirement for strawberry was 140.54 ton/year for F1, 87.69 ton/year for F2, and 65.13 ton/year for F3. The corresponding values for tomato were 223.10, 139.19, and 103.38 ton/year, respectively, while those for paprika were 269.39, 168.07, and 124.83 ton/year. These results indicate that the required fuel mass decreased markedly with increasing torrefaction severity. Figure 2 further demonstrates that heating demand and actual fuel consumption should not be regarded as identical. Although Fig. 1 showed clear differences in heating demand among regions and crops, Fig. 2 reveals that the mass of fuel required to meet that demand also depends strongly on fuel properties. Because F1 had the lowest energy density, it consistently required the greatest mass input, whereas F3 supplied the same heat demand with the smallest mass because of its higher heating value and greater combustion efficiency. This result suggests that the benefits of torrefaction should be considered not only in terms of fuel price, but also in terms of reduced fuel requirement and the associated implications for storage, transport, and handling. These differences became more pronounced under high-demand crop conditions, particularly for paprika. As shown in Fig. 2c, under the Chuncheon–paprika condition, the difference in annual fuel requirement between F1 and F3 reached approximately 144.56 ton/year, and similar trends were observed in Gangneung, Hongseong, Sangju, Gimje, and Miryang. This indicates that as heating demand increases, differences in fuel characteristics are amplified in terms of actual fuel consumption. Therefore, under colder regional conditions or for crops requiring greater heating input, fuel energy density becomes a practically important factor affecting storage and transport burden rather than merely a fuel property indicator. Although diesel and tax-exempt agricultural diesel can show lower mass-based fuel requirements than biomass because of their high heating values, this reflects energy density alone and does not necessarily indicate economic superiority. Nevertheless, Figs. 1 and 2 together suggest that the substantial reduction in fuel requirement achieved by F3 under high-demand conditions is likely to contribute to lower overall heating costs. In this respect, the results in Fig. 2 provide the physical basis for interpreting the subsequent economic comparison among fuel options. Fig. 2Annual fuel requirement according to region, crop, and fuel scenario: (a) strawberry, (b) tomato, and (c) paprika. Panel-specific y-axis ranges were retained to improve the visibility of within-crop regional and fuel-scenario differences. Annual fuel requirement according to region, crop, and fuel scenario: (a) strawberry, (b) tomato, and (c) paprika. Panel-specific y-axis ranges were retained to improve the visibility of within-crop regional and fuel-scenario differences. Comparison of annual fuel-based heating costs under the baseline condition and economic interpretation Under the baseline condition, high-severity torrefied biomass (F3) consistently resulted in the lowest annual fuel-based heating cost across all regions and crops. For strawberry, the annual fuel-based heating cost of F3 was 19.13 thousand USD/year in Gangneung, 28.95 thousand USD/year in Chuncheon, and 11.10 thousand USD/year in Jeju. For tomato, the corresponding values were 34.41, 45.95, and 23.71 thousand USD/year, respectively, while for paprika they were 43.96, 55.48, and 32.70 thousand USD/year. In contrast, F1 always showed a higher heating cost than F3, and both tax-exempt agricultural diesel and conventional diesel were more expensive than all biomass-based fuels. Overall, the cost ranking under the baseline condition was consistently F3 < F2 < F1 < tax-exempt diesel < conventional diesel. This result indicates that fuel economics cannot be assessed on the basis of purchase price alone. Although F3 had a higher unit price per kilogram than F1, it required substantially less fuel mass, which more than compensated for the higher unit cost. Together with its higher heating value and combustion efficiency, this led to the lowest total heating cost among the fuels considered. Across regions, the absolute magnitude of annual fuel-based heating cost closely followed the heating-demand pattern shown in Fig. 1, with Jeju showing the lowest values and Chuncheon the highest. Across crops, paprika produced the highest heating cost and strawberry the lowest, reflecting their different heating-demand levels. However, the relative ranking among fuels remained unchanged, indicating that heat demand determined the overall cost level, whereas fuel properties determined the relative economic competitiveness of the fuel options. The cost advantage of F3 was also maintained under the equal-efficiency scenario and across transport distances of 0–200 km. The comparison between the differentiated-efficiency assumption and the equal-efficiency assumption is provided in Supplementary Table S1. The useful-energy cost ranking remained unchanged as F3 < F2 < F1 under both assumptions, confirming that the main economic conclusion was not driven solely by the assumed combustion-efficiency gradient. In the revised transport scenario, the delivered cost of solid biomass fuels included both the fixed logistics component of 0.133 USD/kg and an additional distance-dependent transport cost. Although increasing transport distance raised the annual fuel-based heating cost of all transported fuels, F3 remained the lowest-cost option across all regions, crops, and transport distances examined. This indicates that its economic advantage was not dependent solely on the baseline efficiency assumption and was relatively robust to changes in transport conditions within the range examined in this study. Accordingly, high-severity torrefied biomass showed favourable fuel-cost performance among the evaluated scenarios. Fuel-selection boundaries and break-even points under changing price structures Although F3 showed the lowest estimated annual fuel-based heating cost under the baseline condition, the broader pricing landscape revealed a more differentiated fuel-selection structure. Figure 3 presents the decision map under the baseline logistics condition, defining the fuel with the lowest annual fuel-based heating cost across combinations of raw biomass price and torrefaction premium. Each coloured region corresponds to F1, F2, or F3. F3 was preferred under low torrefaction-premium conditions, whereas the dominant region of F1 expanded as the premium increased. In contrast, F2 appeared only within a narrow band between the F1- and F3-dominant regions, indicating that it functioned primarily as a transitional option rather than a broadly competitive fuel. This result qualifies the baseline conclusion. Under the reference price structure used in this study, F3 showed the lowest estimated annual fuel-based heating cost; however, when the torrefaction premium increased sufficiently relative to the raw biomass price, F1 became more favourable. Fuel selection for greenhouse heating should therefore not be interpreted as a fixed ranking under a single price assumption, but rather as a boundary problem governed by the interaction between raw biomass price and torrefaction premium. The average share of the decision space further supports this interpretation. As summarised in Table 2, F3 occupied the largest proportion of the price space (53.49%), whereas F1 also retained a substantial share (39.34%). By contrast, F2 accounted for only 7.17% of the decision space, supporting the view that low-severity torrefied biomass represents a narrow transition zone between the raw-biomass and high-severity torrefaction systems rather than an independently dominant option. Table 2Share of decision space for each fuel in the raw biomass price–torrefaction premium map.FuelShare of decision space (%)F139.34F27.17F353.49 Table 2Share of decision space for each fuel in the raw biomass price–torrefaction premium map.FuelShare of decision space (%)F139.34F27.17F353.49 Share of decision space for each fuel in the raw biomass price–torrefaction premium map. A similar pattern was observed in the break-even analysis. As shown in Table 3, the break-even prices were 0.294 USD/kg for F2 and 0.442 USD/kg for F3, both of which were higher than their current purchase prices of 0.222 and 0.311 USD/kg, respectively. This indicates that both torrefied fuels remained economically competitive relative to raw biomass under the present condition. The allowable price margin was 0.072 USD/kg for F2 and 0.131 USD/kg for F3, indicating that F3 retained economic competitiveness over a wider allowable price range and was therefore more robust against price escalation. Table 3Break-even purchase price of torrefied fuels relative to raw biomass under the baseline and 200 km transport-distance conditions.FuelCurrent purchase price [USD/kg]Break-even price at 0 km [USD/kg]Break-even price at 200 km [USD/kg]Allowable margin at 0 km [USD/kg]F20.2220.2940.3240.072F30.3110.4420.4990.131 Table 3Break-even purchase price of torrefied fuels relative to raw biomass under the baseline and 200 km transport-distance conditions.FuelCurrent purchase price [USD/kg]Break-even price at 0 km [USD/kg]Break-even price at 200 km [USD/kg]Allowable margin at 0 km [USD/kg]F20.2220.2940.3240.072F30.3110.4420.4990.131 Break-even purchase price of torrefied fuels relative to raw biomass under the baseline and 200 km transport-distance conditions. In addition to the baseline break-even analysis, the effect of transport distance on the break-even price was examined using the distance-dependent transport-cost formulation. When the distance-dependent transport component was included, the break-even prices increased with transport distance because the logistics term was applied on a mass basis and raw biomass required a larger transported mass than torrefied biomass. At 0 km, the break-even prices were 0.294 USD/kg for F2 and 0.442 USD/kg for F3. At 200 km, these values increased to 0.324 USD/kg and 0.499 USD/kg, respectively. This indicates that longer transport distance can increase the allowable price range for torrefied biomass relative to raw biomass because its higher useful energy density reduces the transported mass required to satisfy the same heating demand. Overall, although F3 showed the lowest estimated cost under the baseline condition, the combined interpretation of Fig. 3; Table 2, and Table 3 shows that a clear economic boundary exists between F1 and F3 across the wider price space. Therefore, the applicability of torrefied biomass for greenhouse heating should be evaluated not only by average annual fuel-based heating cost, but also by the extent of economically favourable regions and the allowable price margin relative to raw biomass. Fig. 3Decision boundary for the lowest-cost greenhouse heating fuel according to raw biomass price and torrefaction premium under the baseline logistics condition. The coloured regions indicate the fuel with the lowest annual fuel-based heating cost: raw biomass (F1), low-severity torrefied biomass (F2), and high-severity torrefied biomass (F3). Decision boundary for the lowest-cost greenhouse heating fuel according to raw biomass price and torrefaction premium under the baseline logistics condition. The coloured regions indicate the fuel with the lowest annual fuel-based heating cost: raw biomass (F1), low-severity torrefied biomass (F2), and high-severity torrefied biomass (F3). Practical implications and limitations of the fuel-cost-based assessment The results of this study provide practical implications for fuel selection in greenhouse heating systems under different climatic and crop-production conditions in Korea. Heating demand consistently increased from Jeju to Chuncheon and from strawberry to paprika, indicating that greenhouse fuel selection should not be determined solely by unit fuel price. Instead, it should be evaluated together with regional heat demand, crop-specific setpoint temperature, cultivation period, fuel energy density, transport distance, and delivered fuel cost. Under the baseline assumptions, F3 required the lowest fuel mass and resulted in the lowest annual fuel-based heating cost, suggesting that the improved fuel quality of high-severity torrefied biomass can offset its higher purchase price when heating demand is sufficiently high. However, the economic superiority of torrefied biomass should not be interpreted as unconditional. The decision-boundary analysis showed that raw biomass can become more favourable when the torrefaction premium increases relative to the raw biomass price. Therefore, the adoption of torrefied biomass for greenhouse heating should be assessed based on delivered fuel prices, local biomass availability, transport distance, and the additional price premium associated with torrefaction. This consideration is particularly important for farms located far from biomass supply chains or torrefaction facilities, where logistics costs may reduce the economic advantage of upgraded fuels. The selected regions and crop scenarios were intended to represent contrasting greenhouse-heating conditions in Korea rather than to provide farm-level statistical estimates for all greenhouse operations. The crop setpoint temperatures and cultivation periods were used as standardised scenario assumptions based on commonly reported cultivation conditions. Although the selected regions and crops cover a broad range of greenhouse-heating conditions, they cannot fully represent cultivar-specific differences, farm-level management practices, greenhouse structures, heating-system configurations, or local microclimates. Therefore, the results should be interpreted as standardised comparative estimates for evaluating relative fuel competitiveness, rather than as direct predictions for every greenhouse farm in Korea. This study should also be regarded as a climate-driven, fuel-cost-based assessment rather than a full investment-level techno-economic analysis. The analysis focused on annual heating demand, fuel requirement, delivered fuel price, transport assumptions, break-even fuel price, and fuel-selection boundaries. Capital investment for boilers and fuel-handling systems, depreciation, financing structure, labour cost, maintenance cost, storage infrastructure, and farm-specific operational constraints were not included. A full farm-level techno-economic analysis would require site-specific information on boiler capacity, installation cost, equipment lifetime, maintenance cost, fuel storage and feeding systems, discount rate, and actual farm operation schedules. Future studies should integrate these investment and operational factors with measured greenhouse energy-use data to further evaluate the feasibility of biomass-based greenhouse heating systems. A major limitation of the present heating-load estimation is that the absolute heating-demand values may be overestimated relative to commercial greenhouses equipped with thermal screens. Because thermal screens reduce night-time heat loss through the greenhouse envelope, their exclusion represents a no-screen theoretical baseline. Accordingly, the calculated heating demand and annual fuel-cost values should not be interpreted as direct estimates of actual farm-level heating bills. Instead, they represent standardised comparative values under identical greenhouse, crop-setpoint, and climate-data assumptions. This limitation affects the absolute magnitude of heating demand, but it is less likely to change the relative fuel ranking because the same heat demand was consistently applied to all fuel scenarios. Fuel-specific storage costs and storage-related quality changes were not explicitly included in the cost calculation. Raw biomass may require larger storage volume and more intensive ventilation management because of its lower energy density, higher moisture content, and greater susceptibility to moisture re-absorption. These factors can reduce effective heating value and combustion efficiency during long-term storage. In contrast, torrefied biomass may offer practical advantages in storage stability and hydrophobicity. However, storage-related costs depend strongly on farm-level storage infrastructure, local humidity, storage duration, fuel form, and boiler feeding system. Therefore, these factors were treated as practical limitations rather than explicitly parameterised in the present scenario model. The transport-distance scenario should be interpreted as a standardised mass-based logistics sensitivity rather than as a farm-specific supply-chain optimisation model. The same ton-kilometre unit cost was applied across fuels to isolate the effect of transport distance on delivered fuel cost. In practice, actual logistics costs may vary depending on road conditions, loading capacity, bulk density, delivery frequency, storage infrastructure, and regional supply-chain organisation. Auxiliary liquid ignition fuels such as diesel or gasoline were not included in the fuel-cost calculation. In many wood pellet or wood-chip boiler systems, start-up is electrically assisted and, after ignition, combustion is sustained by the solid fuel itself. Therefore, continuous auxiliary liquid-fuel input was not assumed in this study. However, start-up energy demand and ignition-system design may vary among boiler types and were outside the scope of the present annual fuel-cost comparison. Electric heating may also be considered as an alternative greenhouse heating option because electricity is already involved in the start-up and control of some solid-fuel boiler systems. Electric heating has several potential advantages, including local zero direct combustion emissions, relatively simple controllability, and possible compatibility with renewable electricity. However, its economic performance in large-scale greenhouse heating strongly depends on electricity price, peak-load charges, grid capacity, and the carbon intensity of the supplied electricity. In addition, the present study was designed to compare fuel-based heating scenarios using biomass-derived solid fuels and diesel under a standardised heating-load framework. Therefore, a full comparison with electric heating, heat pumps, or renewable-electricity-coupled heating systems was beyond the scope of this study. Future studies should extend the present fuel-cost-based framework to include electric heating and heat-pump scenarios using region-specific electricity tariffs, coefficient of performance, peak-demand charges, and grid-emission factors. Conclusion This study estimated the greenhouse heating demand required to maintain crop-specific setpoint temperatures in major horticultural production regions of Korea using climate-based data, and compared the fuel requirement and fuel-cost performance of conventional diesel, raw biomass, low-severity torrefied biomass, and high-severity torrefied biomass. The results showed that annual heating demand was lowest in Jeju and highest in Chuncheon for all crops, while the crop-wise order was strawberry, tomato, and paprika. These findings indicate that greenhouse heating load is jointly determined by regional climate, crop-specific temperature requirements, and cultivation period. The required fuel mass to satisfy the same heating demand differed markedly according to fuel properties. Across all regions and crops, high-severity torrefied biomass (F3) consistently showed the lowest fuel requirement, whereas raw biomass (F1) showed the highest, reflecting the increase in energy density and effective heat supply with increasing torrefaction severity. This pattern was directly linked to annual fuel-based heating cost. Under the baseline condition, F3 resulted in the lowest heating cost across all regions and crops. Moreover, the preferred fuel remained unchanged under the equal-efficiency scenario and across transport distances of 0–200 km, with F3 maintaining the lowest estimated annual fuel-based heating cost among the evaluated fuel scenarios under the assumptions examined in this study. This indicates that the relative advantage of F3 was not solely dependent on the baseline combustion-efficiency gradient, but was also associated with its higher useful energy density and lower required transport mass. The extended price-structure analysis provided a more important insight than the baseline comparison alone. Although F3 showed the lowest estimated cost under the reference price condition, the decision map based on raw biomass price and torrefaction premium revealed a clear economic boundary between F1 and F3. The shares of the decision space were 39.34% for F1, 7.17% for F2, and 53.49% for F3, indicating that F3 occupied the largest economically favourable region but did not dominate the entire price space. In contrast, F2 was selected only within a narrow transitional zone, suggesting a limited role under the price structure examined in this study. The break-even analysis also supported the relative advantage of F3. The break-even prices relative to raw biomass were 0.294 USD/kg for F2 and 0.442 USD/kg for F3, while the current prices were 0.222 and 0.311 USD/kg, respectively, indicating that both torrefied fuels remained economically competitive under the present condition. In particular, F3 showed a larger allowable price margin, suggesting greater robustness against price increase. The preferred fuel ranking remained unchanged under the equal-efficiency scenario and across transport distances of 0–200 km, even when distance-dependent transport costs were added to the fixed biomass logistics component. F3 maintained the lowest estimated annual fuel-based heating cost among the evaluated scenarios because its higher useful energy density reduced the required fuel mass and partly offset its higher purchase price. The transport-distance analysis further showed that longer transport distance can increase the allowable price range for torrefied biomass relative to raw biomass, because the mass-based logistics penalty is lower for fuels with higher useful energy density. Overall, under the assumed fuel properties, delivered-price structure, transport-cost assumptions, and standardised greenhouse conditions, high-severity torrefied biomass showed favourable fuel-cost performance among the evaluated greenhouse heating fuel scenarios. However, this advantage was not absolute under all price conditions, and a competitive boundary with raw biomass remained depending on the combination of raw biomass price and torrefaction premium. Therefore, the applicability of torrefied fuel for greenhouse heating should be evaluated not only by average cost comparison, but also by considering regional heat-demand levels, delivered-price structure, transport-distance assumptions, economically favourable regions across the price space, and break-even ranges. Because this study used a standardised heating-load model and fixed greenhouse parameters, the results should be interpreted as a screening-level fuel-cost comparison rather than as direct predictions of farm-level heating bills. In this respect, the present study provides a fuel-cost-based decision-support basis for fuel selection in greenhouse heating under Korean production conditions. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2