Indoor CO₂ concentration exceeding 1000 ppm may pose health risks to occupants, such as headaches, nausea, reduced cognitive function, and productivity (Jacobson et al. 2019). Levels above 10,000 ppm can be dangerous and even life-threatening (Figure 1a). In the absence of adequate ventilation, indoor CO₂ concentrations can reach values 3-7 times higher than those found outdoors. To address elevated indoor CO₂, HVAC systems typically increase ventilation rates, which substantially raises energy demand - especially when outdoor temperature and humidity differ markedly from indoor comfort standards. Conversely, reducing ventilation rates lowers energy use but does not effectively remove CO₂ from the indoor environment, resulting in a trade-off between maintaining high IAQ and minimizing building energy use (Figure 1b). To overcome this trade-off, a new strategy that allows for low ventilation rates while still sustaining safe indoor CO₂ levels is needed (Figure 1b).

This study introduces an indoor CO₂ capture (ICC) system that uses carbon capture devices to directly remove CO₂ from stale indoor air, recirculating CO₂-depleted air to replace the intake of outdoor air. This configuration enables the ICC system to maintain high IAQ at ultra-low ventilation rates, thereby reducing building energy demand. The captured CO₂ is concentrated for subsequent storage or utilization (Figure 1c). In this work, we evaluate and compare the energy-saving potential and net CO₂ emission reduction of various ICC methods, taking into account the effects of weather conditions, CO₂ capture rates, and the specific energy consumption of carbon capture methods. The specific energy consumption refers to the energy needed to regenerate each kilogram of CO₂ and is determined by carbon capture methods and materials used. This part of the energy requirement may lead to additional CO₂ emissions, potentially offsetting the overall benefits of carbon capture (Prajapati et al., 2022, Mertens et al., 2023), requiring nuanced analysis to ensure net carbon emission reduction.

This section describes the major modelling process for the proposed ICC system. The modelling framework is designed to quantify both energy savings and CO₂ emissions. All simulations are conducted under the condition that indoor CO₂ concentrations remain below 1000 ppm across various conditions. Notably, we assume that air filtration can be integrated into the ICC system to avoid the accumulation of potential pathogens and other air pollutants, such as volatile organic compounds. The real-time CO₂ concentration inside the room, , is calculated as follows:

represents the change in indoor CO₂ concentration at each time step, which is 30 seconds in this study. This variation is calculated based on the mass conservation of CO₂ flows, while also ensuring the maintenance of indoor air pressure.

The total energy loads of the HVAC system , in both conventional and proposed scenarios, comprises latent and sensible components, denoted as and , respectively. These heat loads are attributable to the newly introduced outdoor air and exclude potential internal heat gains from occupants or equipment, thereby isolating the energy-saving performance of ICC’s reduced ventilation rates. Specifically, the latent energy load , can be calculated as below:

represents the amount of water added or removed by the air-conditioning (AC) system for dehumidification or humidification. It is determined by the minimum difference in water content between the newly introduced air and the standard indoor air.

is the water content in newly introduced ambient air, which is the sum of ventilated air, β, and infiltration air, . is the water content generated by occupants inside (kg/s). is the water content in the standard indoor air.

is the humidity ratio of ambient air, defined as the mass of water vapor per kilogram of dry air. This value is determined by the temperature and ambient pressure, as well as outdoor relative humidity (RH).

is the humidity ratio of the indoor air, which is determined by Eq.(7).

Once the minimum value of is identified by the modelling, the corresponding value of that yields this minimum is determined. The value, varying between 40% to 60% for thermal comfort and minimum latent loads, can then be used to calculate .

denotes the specific heat capacity of air, and represents the temperature difference between the outdoor and indoor environments.

represents the percentage of the energy savings achieved by the new system compared to the conventional design.

denotes the total energy consumption of the proposed design, which includes both the HVAC energy consumption and the energy consumption of the ICC system.

represents the CO₂ emissions associated with the energy consumed by the HVAC system (kg); while denotes the penal carbon emissions from ICC operation (kg). In the conventional case, both and are zero. Notably, the exhaled CO₂, which constitutes a non-trivial yet often overlooked emission sector, is captured by ICC rather than being expelled into the atmosphere by conventional methods. However, not all of the respiratory CO₂ can be captured, and the actual amount of captured CO₂, (kg), can be calculated as below:

is the CO₂ capture rate (kg/s). The CO₂ emission factor used in this study, based on heat generation from natural gas and oil combustion, is 0.25 kg/kWh (Our World in Data 2023). We also conduct parameter analysis to investigate the impacts of CO₂ emission factor on the overall ICC performance.

The evaluation process starts with the input of weather conditions, occupancy profiles, and performance parameters of the ICC system. In the process flow diagram (see Figure 2), the conventional scenario is represented by grey lines, while the proposed ICC scenario is indicated by blue lines. The ventilation profile is determined based on occupancy and the satisfaction of IAQ requirements, that is, the indoor CO₂ concentration should be below 1000 ppm (Wang et al., 2018, ASHRAE 62.1, 2010). This design follows the dynamic regulation of demand control ventilation (Lu et al. 2011). For the ICC case, the new ventilation profile and the amount of captured CO₂ are established according to ICC performance and occupancy profile. Using the weather conditions and ventilation profile, the sensible and latent loads for both cases are calculated to maintain thermal comfort within the indoor environment. The standard indoor temperature for summer is 25℃, and 10 ~ 20℃ for winter, while the indoor RH is always maintained between 40% and 60% (Harrouz et al. 2022a), with the specific value determined by the real-time water vapor content of the outdoor air.

By comparing the HVAC loads of both scenarios, the energy savings of the HVAC system can be quantified. However, in the ICC case, there is an additional energy consumption sector associated with carbon capture, which is determined by the amount of captured CO₂ and the specific energy consumption required for the capture process.

Occupancy patterns have a direct impact on indoor CO₂ concentration and ventilation requirements. In this study, we selected the occupancy profile of a medium office setting, which is characterized by a typical pattern of clocking in and out (Figure 3a). Notably, in this study, we assume that the office room is a well-mixed room, in which the CO₂ and other air components are uniformly distributed throughout the space within each time step. The indoor CO₂ concentration fluctuates as a result of CO₂ generation (occupant exhalation, infiltration, ventilation) and removal (ventilation, carbon capture, exfiltration). In both the conventional (no capture) and ICC cases, the indoor CO₂ concentrations are maintained below 1000 ppm to comply with IAQ standards (Figure 3b). Figure 3c presents four operational cases comparing the dynamic curves of indoor CO₂ concentration, all of which exhibit exponential build-up and decay similar to findings from other studies (Lu et al., 2011, Youngmin Cho, 2013, Lee et al., 2013, Sinha et al., 2022, Grande, 2022, Tian et al., 2023, Harrouz et al., 2023). To prevent excessive CO₂ removal, an additional constraint was imposed: ICC operation is suspended if indoor CO₂ concentration falls below 350 ppm, the lower bound of the atmospheric CO₂ level (350~450 ppm) (Kim et al., 2015, Lee et al., 2013) (Figure 3b). This measure avoids unnecessary over-capture and balances the advantages of lower CO₂ levels against the increased energy consumption associated with capturing CO₂ at concentrations already satisfying healthy IAQ. With these constraints applied, the ventilation rates for each case were calculated, as shown in Figure 3c. While ventilation rates in both cases generally follow the occupancy profile shown in Figure 3a, a marked reduction in ventilation rates is observed in the ICC case, highlighting the system’s potential to significantly lower HVAC energy demand (Figure 3d).

In this section, we present the weather conditions in Xi’an during the hottest and coldest months, July and January, respectively, as well as their implications for building energy loads. Xi’an experiences hot and humid summers (Figure 4a), while in January, temperatures frequently and largely drop below 10℃ (Figure 4b), and the relative humidity (RH) is significantly lower compared to July. Consequently, the latent load is more prominent in July due to higher humidity, whereas in January, the sensible load – primarily for heating the space - is more prominent than that in July. To maintain indoor comfort, RH is kept between 40 and 60%. Seasonal indoor temperature settings are applied: heating is provided in winter when outdoor temperatures fall below 10℃, with an indoor setpoint of 20℃, and cooling is applied in summer when outdoor temperature exceeds 25℃. Regardless of the season, humidification and dehumidification are applied when outdoor RH is below 40% and above 60%, respectively. The indoor RH setpoint is selected between 40% ~ 60% and is determined by the minimum difference in water vapor content between indoor and outdoor air.

This section investigates the net energy-saving performance of the ICC system, which is defined as the difference between the reduction in building HVAC energy consumption and the energy consumed to operate the ICC device. Energy loads are evaluated on hourly, daily, and monthly timescales to capture dynamic variations. As illustrated in Figure 5, the sensible and latent loads for July – calculated based on the medium office occupancy profile and the satisfaction of indoor CO₂ concentrations below 1000 ppm, demonstrate a clear reduction in the ICC case compared to the conventional scenario.

Figure 6 evaluates the impact of ICC implementation on mean hourly sensible and latent energy loads during the peak summer and winter months. The analysis compares scenarios with and without ICC for both July and January, revealing that the ICC case consistently achieves lower energy loads across both periods. The shaded areas in the subfigures highlight the magnitude of energy savings achieved by ICC. For instance, in Figure 6a, the daily total sensible load in July is 12.7 kWh for the conventional case and 1.6 kWh for the ICC case. In Figure 6b, the daily total latent load in July is 22.9 kWh for the conventional case, and 4.9 kWh for the ICC case. In January, the daily total of sensible load is 41.4 kWh for the conventional case, and 5.3 kWh for the ICC case (Figure 6c), while the daily total latent load is 17.3 kWh for the conventional case and 1.3 kWh for the ICC case (Figure 6d). These comparative results underscore the substantial energy-saving potential of ICC in months with high HVAC loads.

Figure 7 presents the monthly variation in daily total energy loads under ICC operation, using a CO₂ capture rate of 0.18 g/s. The specific energy consumption for this section of the case study is assumed to be 6 MJ/kg, which is consistent with values reported in DAC studies that typically range from 5.25 to 6.7 MJ/kg for atmospheric CO₂ capture (Zhu et al., 2021, Deutz and Bardow, 2021, Baciocchi et al., 2006, Keith et al., 2018, Sabatino et al., 2020). Notably, the minimum separation work of direct air capture methods is 19~21 kJ/mol (0.43~0.45 MJ/kg) (Siegelman et al. 2021). To facilitate comparison between different capture methods, we assume that the CO₂ capture rates are constants for the same occupancy profiles, without periodic fluctuations from capture-release cycles, such as those seen in membrane-based capture methods. For capture methods involving capture-release cycles, stable capture rates can be achieved through capture device designs, such as multi-chamber configurations and active capture control. Figure 7a presents the daily totals for each of the 31 days in July and January. The results indicate that, in July, more than half of the HVAC energy consumption is attributed to overcoming the latent load due to the high ambient RH in July in Xi’an. In contrast, in January, the dry conditions mean that latent energy is primarily used for humidification, while a greater portion of HVAC energy is consumed to overcome the sensible heat loads of the ambient air. Overall, the ICC system achieves substantial energy savings, reducing total energy consumption by 67.5% in July and 46.5% in January, after accounting for the energy required to operate the ICC device, as shown in Figure 7b. These findings highlight the significant seasonal variability in the energy-saving potential of ICC, with benefits more pronounced in periods when outdoor RH and temperature deviate substantially from indoor comfort standards.

As shown in Figure 8a-b, the energy-saving potential of ICC is higher during periods of high original energy demand. For instance, the average energy-saving potential from December to February reaches 65.6%, while in July and August, it is 37.9%. Conversely, in months with low HVAC energy demands, the energy-saving potential of ICC diminishes or becomes negative due to the additional energy consumed for CO₂ capture. For example, in April, the conventional case consumes 7.5 kWh daily, whereas the ICC case requires 13.25 kWh/day (0.72 kWh for HVAC and 12.53 kWh for CO₂ capture). In contrast, January exhibits the highest energy savings, with the conventional case consuming 59.6 kWh/day, and the ICC case consuming 19.2 kWh/day in total, resulting in a reduction of 67.8%. Despite negative savings in certain months, the annual average energy-saving potential remains substantial at 28.5%, further complemented by the carbon capture benefits provided by ICC.

To further investigate the impacts of different CO₂ capture methods on net CO₂ emission reductions, we evaluate and compare the corresponding CO₂ emissions based on the energy loads and the CO₂ emission factor. For the conventional case, CO₂ emissions are solely attributed to HVAC operation, calculated as . With a CO₂ emission factor, δ, of 0.25 kg/kWh, a peak emission of 14.9 kg/day is reached in January, while the lowest emission is 0.15 kg/day in October. Notably, in May, September, and October, the daily CO₂ emissions for the conventional case fall below 1 kg/day due to low HVAC energy demands. In the ICC case, total CO₂ emission from both HVAC and ICC operation averaged around 4 kg/day across months. Although this value exceeds that of the conventional case in certain months, it is offset by the amount of CO₂ captured , which is approximately 7.52 kg/day. As a result, the net CO₂ emission for the ICC case, , is negative (as shown in Figure 9), demonstrating the strong decarbonization potential of the ICC system.

Here, we examine the combined impacts of climate conditions and CO₂ capture speed, , on buildings’ net energy savings and CO₂ emissions reduction. A capture rate of 0 ( = 0) represents the conventional case without CO₂ capture. During months with high HVAC energy consumption - such as January, February, July, August, and December – all non-zero cases exhibit lower daily total energy loads compared to the conventional case. During these periods, increasing the CO₂ capture rates leads to greater energy savings, primarily due to reduced ventilation requirements; however, the incremental benefits gradually level off as continue to increase. This occurs because increasing further reduces the HVAC subsystem’s ventilation demand and associated energy use, but also raises the ICC’s own energy consumption. When becomes excessive, the additional energy consumed by the ICC may outweigh the energy savings in the HVAC subsystem, thereby diminishing or even negating the overall energy-saving benefit. For instance, the annual average daily energy consumption decreases from 22.11 to 17.03 kWh/day, and 15.81 when increases from 0 to 0.12 and 0.18 g/s, respectively. However, further increasing to a larger value, i.e., 0.24 g/s, the daily energy demand rises again to 16.39 kWh/day (see Table 1). This rebound occurs because, while a higher reduces HVAC loads, they also increase the energy consumption of the ICC system, partially offsetting the overall energy savings (Figure 10a).

In months with low HVAC energy demand, such as April, May, and October, the daily energy consumption of the conventional case, , is actually lower than that of the ICC case, . This is because the ICC system continues to consume energy for capturing indoor CO₂, even when HVAC loads are minimal (see Figure 10a). Despite the higher energy use in these months, the net CO₂ emission of the ICC, , remain significantly lower - or even negative - compared to the conventional case (Figure 10b). For instance, for the case = 0.12 g/s, the net CO₂ emission in January is 1.7 kg/day for the ICC case, compared to 14.9 kg/day for the conventional case. When increases to 0.18 g/s, the net CO₂ mission drops further to -2.7 kg/day. This demonstrate that, even in months where energy savings are limited or negative, the ICC system still offers substantial benefits in terms of CO₂ emission reduction.

To investigate the impacts of different CO₂ capture methods on energy-saving performance, we identify a specific CO₂ capture method using two key indexes: specific energy consumption and CO₂ capture rate. In Figure 11, we examine the combined impacts of the specific energy consumption of CO₂ capture () and on yearly total energy consumption () and net CO₂ emissions (). In this study, we design to span from 0 to 4.17 kWh/kg based on literature-reported results (Goeppert et al., 2012, Zhu et al., 2021, Keith et al., 2018, Li et al., 2020a, Gu et al., 2022, Deutz and Bardow, 2021, Baciocchi et al., 2006, Sabatino et al., 2020) to cover as many capture methods and materials as possible. For the conventional case, the average daily total energy consumption is 22.11 kWh/day, with a corresponding net CO₂ emission of 5.5 kg/day. For the ICC system, generally increases as increases, but decreases as rises. Notably, for high cases, such as those higher than 2.5 kWh/kg, actually increases with higher (Figure 11a). For example, when is 1.1 kWh/kg, decreases from 20.3 to 11.6 kWh per day as increases from 0.02 g/s to 0.2 kg/s. In contrast, when = 3.34 kWh/kg, increases from 22.5 to 24.2 kWh/day over the same range of This trend occurs because, at low increasing leads to significant reduction in HVAC ventilation rates, resulting in substantial energy savings and a decrease in . However, when the ICC method relies on processes with high specific energy consumption, the energy required for ICC outweighs the savings from reduced HVAC loads, causing to rise. Thus, the effectiveness of ICC in reducing depends critically on the specific energy consumption of the chosen CO₂ capture method.

In Figure 11b, we discuss the relationships between net CO₂ emissions, and . demonstrates a negative correlation with and a positive correlation with . For instance, at = 1.67 kWh/kg, net CO₂ emissions decrease from 4.0 to -4.1 kg/day as increases from 0.02 to 0.2 g/s. Conversely, for = 0.10 g/s, increases from -2.15 to 2.6 kg per day when rises from 0.28 to 4.17 kWh/kg. In Figure 11b, the “ = 0” line marks the threshold for negative carbon emissions; regions below this line represent net carbon removal. However, the decarbonization benefit plateaus as increases, evidenced by the gradually flattening curves in Figure 11b. This trend mirrors the pattern observed in Figure 11a, and results from the interplay between reduced HVAC energy demand and increased ICC energy consumption. Therefore, it is crucial to identify the optimal that maximizes both energy savings and carbon emission reductions, balancing the benefits of reduced ventilation loads against the energy requirements of the ICC process.

The impacts of the maximum occupancy levels, on ICC’s energy savings and emission reduction are also investigated. When a lower is applied, for both the conventional and ICC cases, are reduced. For instance, the minimum and maximum for = 0.2 is 1.1 and 15.3 kWh/day, while for = 0.8, these values increase to 8.5 and 52.3 kWh/day. Regarding net CO₂ emission, increasing results in a wider range of net CO₂ emission values, depending on the operating conditions of and Despite these variations, the ICC system consistently achieves greater energy savings and lower CO₂ emissions compared to the conventional case across all scenarios (as shown in Table 2). Specifically, compared to the conventional case, the maximum CO₂ emission reduction amount for = 0.2 is 5.3 kg/day, increasing to 13.3 kg/day for = 0.5 and reaching 17.2 kg/day for = 0.8.

In Figure 12, we analyse the impact of regional CO₂ emission factors φ on the net CO₂ emission performance of the ICC system. φ represents the amount of CO₂ emitted per unit of electricity generated by a power grid, reflecting both the carbon intensity and the energy mix of the electricity supply. For instance, power grids with a high penetration of renewable energy, such as those in Sichuan, Yunnan, and Qinghai in China, can have φ as low as 0.1~0.16 kg/kWh (Ministry of Ecology and Environment of the People's Republic of China 2024). In contrast, coal-dominated power grids exhibit φ values as high as around 0.73 kg/kWh, such as those in Hebei Province, China. Net CO₂ emissions of the entire system increases with higher φ across all months, due to the greater carbon intensity of the energy consumed by both the HVAC subsystem and ICC operation. For instance, in January, net CO₂ emissions rise from -5.6 kg/day to 11.7 kg/day as φ increases from 0.1 kg/kWh to 1.0 kg/kWh. On an annual basis, net CO₂ emissions increase from -6 kg/day to 8.3 kg/day.

To identify feasible operating conditions for the ICC system, we applied two criteria for the selection of and : 1) the total energy consumption of the new system () being lower than the HVAC energy consumption of the conventional system (), and 2) net CO₂ emission, , being less than 0. Figure 13 presents phase maps illustrating the combinations of and that satisfy these two criteria under different occupancy profiles and CO₂ emissions factors. In these maps, the region where < is shaded red, while the region where < 0 is shaded blue. The overlap of these regions indicates the feasible operating domain for the ICC system, providing valuable design guidance. The phase maps are investigated for three different crowdedness scenarios, represented by varying maximum occupancy during the day: = 0.2 (less crowded), = 0.5 (crowded) and = 0.8 (extremely crowded). For the objective of achieving < , the feasible range of for the case φ = 0.25 kg/kWh is similar across all three cases, with values below 2.78 kWh/kg. This is the critical threshold for the ICC system to achieve lower HVAC energy consumption than its conventional counterpart. When > 2.78 kWh/kg, the energy required for CO₂ capture becomes dominant, and the ICC system is no longer able to reduce total energy consumption compared to the conventional system.

For the objective of < 0, a lower occupancy level (smaller ) broadens the feasible region where both objectives can be met. For example, at = 0.2, the feasible values for and range from 0.02 ~ 0.24 g/s and 0 ~ 2.5 kWh/kg, respectively (Figure 13a). In contrast, for = 0.5. the feasible ranges for narrows down to 0.06~0.24 g/s (Figure 13b), and for = 0.8, the feasible ranges of and narrow further to 0.10~0.24 g/s (Figure 13c). Notably, for = 0.2, increasing beyond 0.08 g/s does not further expand the feasible range for , suggesting an optimal value for that minimizes ICC’s energy consumption, while still yielding a negative CO₂ emissions. Compared to that of φ = 0.65 kg/kWh, the viable range, specifically the < 0 of the φ = 0.65 is narrower. These phase maps highlight the importance of selecting ICC methods for occupancy levels to maximize both energy savings and decarbonization benefits.