Development and validation of a methodology to measure exhaled carbon dioxide (CO₂) and control indoor air renewal

2.1. General description of the methodology

A methodology based on three standard stages with specific goals was developed. They enumerate as follows:

1. Diagnosis. It is intended to acquire detailed knowledge on how CO₂ concentration levels vary within the physical limits of the shared space and what is its time evolution as a function of specific activities carried out in the shared space. The initial diagnosis includes a description of the space indicating the measures of the shared space (e.g., size, volume, and architectural geometry), other CO₂ sources (e.g., gas stove), the presence and ubication of windows and doors, as well as the availability of Heating, Ventilation, and Air Conditioning (HVAC) systems and hoods. During this process, it is necessary to identify the “person in charge” and perform minimal instruction about the project and concepts such as air renewal, ventilation, metabolic CO_2, and how to measure it.

2. Correction. According to the acquired knowledge in the diagnosis stage, it is possible to proceed with the design of customized protocols and procedures to improve the air renovation that will imply variations in the CO₂ concentration. The CO₂ levels will be monitored with a CO₂ meter to avoid trespassing a defined threshold considered as a low-risk level by the scientific community.

3. Monitoring, control, and surveillance (MCS): Continuous monitoring of CO₂ concentration is performed to guarantee that the shared space is always kept at lowrisk levels. In case of the CO₂ concentration levels reach the maximum level, an alert is issued so that people in charge of the shared space activate the protocols and procedures defined in the previous stage to fix the situation.

For each stage, achieving the primary goals depends on specific characteristics of the shared space, such as main business or social activity, geometrical properties, number and location of windows and doors, or a maximum number of people sharing the room at the same time. A case study and a practical application is presented in this article to explain this methodology.

2.2. Measurement of metabolic CO₂

Any time measuring metabolic CO₂ levels is required, Aranet CO₂ meters (Aranet Wireless Solutions España SL, Spain) and Signos CO₂ meters (Signos.io, Spain) will be used. The technical characteristics of these measuring devices are shown in Table 1.

2.3. The methodology explained in a case study

As a case study, the methodology for improving air quality was applied in a cultural center. The building has 1,200 m² divided in three floors (Figure 1a). A 132 m² room located in the basement was selected (Classroom 1, see Figure 1b).

• Download figure
• Open in new tab

Figure 1.

(a) General distribution of CO₂ meters in the education building and (b) classroom 1 detail.

1. Diagnosis: The shared spaces were analyzed for four days (February 24^th to 27^th, 2021) by placing 10 CO₂ meters (Aranet 4) in strategic locations, as shown in Figure 1a. A CO₂ measuring device was placed outdoors so that it was possible to calculate the increase in CO₂ (ΔCo₂) associated with human activity according to Equation 1; where (CO₂)_ext corresponds to CO₂ outside and (CO₂)_ind to CO₂ indoors. In the first approach, the data showed frequent high CO₂ concentration levels probably linked to specific activities. Subsequently, the shared spaces were carefully monitored for six more days (February 27^th to March 5^th, 2021) to confirm the hypotheses that could be made after the initial analysis. This process makes it possible to characterize the air renewal patterns in each space based on activity and capacity. At least one CO₂ meter was placed every 20 m² at a minimum height of 1.5 m. Ideally, they will be placed near exhaust fans (to measure a significant air sample) and/or in locations where air turnover is poor (e.g. corners), and far from windows and doors.

2. Correction: Based on the obtained results, an attempt was made to reduce the level of metabolic CO₂ below 800 ppm. One of the classrooms in the basement of the building (Classroom 1) showed inefficient ventilation patterns. Dry ice was used to saturate the room with CO₂ and find effective strategies to improve air renewal. According to Figure 1b, 10 CO₂ meters (Aranet 4) were placed in the classroom to analyze the improvement of changes in the HVAC and natural ventilation (through doors and windows opening or closing).

3. MCS: This stage was performed until May 15^th, 2022 using 10 CO₂ meters (Signos). Whenever the cultural center exceeds the 800 ppm level for more than 30 minutes, it receives a telematic alert or phone call to ask the “person in charge” to correct the situation.

2.4. Broader validation in commercial spaces

Subsequently, the method was applied to 40 businesses / retail stores spaces located in Zaragoza’s city center (Spain). These included: one optician, twelve clothing stores, two kids’ clothing stores, five shoe stores, two jewelry stores, four food stores, two restaurants, five home stores, two consultancies, two dietary and health centers, and two florists. The selection of commercial spaces was made to cover different typologies. For example, customers spend long periods in restaurants, while their exposure to indoor air in flower retailers is minor. Two CO₂ meters were placed in each establishment, except for a kid’s clothing store, where only one was placed, and another children’s clothing store, where three were placed.

Interventions in shared spaces were carried out following the three phases: diagnosis, correction, and MCS.

1. Diagnostic. Two CO₂ meters (Aranet 4) devices were placed at strategic points for 2 or 3 days in each business. These meters were chosen because they record the CO₂ level every minute. The CO₂ curves were analyzed during the business hours of each establishment. This phase allowed us to know if the sensors took representative measurements of the total space and have a global image of the air quality in the volume area. The diagnostic process might include the use of risk calculators or methods for simulating disease propagation via aerosols based on the standard Wells-Riley model for infectious respiratory disease transmission by airborne particles.^71,72,76

2. Correction. In spaces where it was necessary (typically presenting CO₂ levels >1,000 ppm for more than 30 minutes), a correction phase was carried out where preventive measures were implemented to improve indoor air renewal. In most cases, this phase was superimposed on monitoring to determine the advantage obtained at the time.

3. MCS. The monitoring stage occurred during January and February 2022. For this stage, CO2 (Signos) meters were installed in each establishment. These CO₂ meters, include IoT technology, enabled remote access to streaming data.

3.1. The diagnostic stage to allow detection of inefficient air renewals

The diagnosis must be performed on a long enough time interval to cover the specific and regular casuistry of the establishment. For example, in the base case study (cultural center), the classrooms usually are occupied for daily classes. Therefore, a selection of days where the classrooms are occupied is sufficient. In this case, four days were used for the preliminary analysis of the space. Assuming a proper maximum CO₂ level of 800 ppm, a maximum increase of 400-450 ppm compared to fresh outdoor air was considered acceptable.

As shown in Figure 2, there are peaks in the increment of CO₂ associated with the activity of each room. Knowing the activity and the number of occupants is essential to interpret the CO₂ measurements. Determining the ppm/person ratio can be helpful, as we proposed in previous work (see ⁶¹). This value incorporates the occupancy variable and allows a more precise definition of the efficiency of air renewal, although it assumes homogeneity in exhalation between individuals. In addition, it is crucial to request the usual activity in the space. Those responsible must not consciously increase ventilation (e.g., opening windows) to avoid underestimating CO₂ measurements. A typical table with essential information is shown in Table 2. The method to fill in the table must be explained in the training phase since it is of utmost importance to interpret the data properly later. In the base case, the room had no open windows, except for the Hall and the Art Classroom. In contrast, all doors were open during diagnostic measurements.

• Download figure
• Open in new tab

Figure 2.

Increment of CO₂ values recorded in the diagnostic period.

Comparing the ppm/person indicator in similar activities in different spaces makes it possible to determine the relative efficiency of air renewal in each space. For instance, in English classes (not oral), 9.2 ± 16.4, 13.1 ± 21.7, and 18.1 ± 21.3 ppm/person were achieved in Classroom 1, the Auditorium, and the Multipurpose Room, respectively. It indicates that air renewal is similar in all rooms, although slightly improved in Classroom 1. On the contrary, disparate ratios were obtained in the Cambridge exams in Classroom 1 and the Informatics Classroom. While in Classroom 1, the ratio was 12.2 ± 17.3 ppm/person, in the Informatics Classroom, a 46.9 ± 107.3 ppm/person was found. Again, this indicates more efficient air renewal inside the basement room.

Some gaps may go unnoticed during the diagnostic phase. Due to the atmospheric conditions or the activity during the diagnosis, the CO₂ levels may be decreased. Although we recommend analyzing 3-4 critical days, subsequent days (e.g., 5-6 days) should be closely monitored to ensure representative results. For example, in the base case, on the monitoring days after diagnosis, higher CO₂ concentrations were recorded in Classroom 1 than in other areas, even in similar conditions of capacity and activity.

3.2. Interpretation of CO₂ measurements, methods to correct gaps, and monitoring

After detecting spaces that can be improved, it is necessary to implement efficient corrective measures that guarantee adequate air renewal. First, it is crucial to interpret the measurements and begin to rule out possibilities. In the case of Classroom 1, the following hypotheses were prepared:

• Hypothesis 1. The activities studied in the initial diagnosis were not representative, and the high activity in the classroom, so more CO₂ is exhaled. Then, the following 12 English classes in Classroom 1, comprising the months of April to June, were analyzed. In the studied activities, no significant increase in CO₂ was observed, except for the classes held in June, where an increase of 522 ppm was reached. Figure 3 shows the evolution of average CO₂ from April to June, showing an increase in concentration in the latter month. However, CO₂ exhalation in the records studied is similar, so this hypothesis is ruled out.

• Hypothesis 2. The weather conditions of the initial diagnosis were favorable for air renewal ratios. The weather influence on the ventilation ratios was evaluated by analyzing the incidences (values above 800 ppm for more than 30 minutes) recorded from April to July (regardless of the activity carried out). Of the 13 incidences found, 46 % (6) came from July, while 31 % (4) came from June, and the remaining 23% (3) from May. The climatic variation has affected the ventilation performance of the space. Working on days with less favorable weather will be necessary to test the corrective measures.

• Testing corrective measures. An experiment where the space was saturated with CO₂ using dry ice was conducted, stabilized as a function of the amount evaporated. Two scenarios were evaluated with an activated HVAC system: with and without opening doors. Figure 4 shows the results obtained. With doors closed, it takes ∼1,600 ppm ∼37 minutes to reduce the CO₂ concentration to ∼800 ppm, at 22 ppm/minute. While, with doors open, the reduction from ∼1,600 ppm to ∼800 ppm occurs in ∼17 minutes, at 47 ppm/minute. The opening of doors with the HVAC system active supposes a remarkable improvement of the ventilation in this classroom.

• Download figure
• Open in new tab

Figure 3.

Evolution of average ΔCO₂ in English classes from April to June.

• Download figure
• Open in new tab

Figure 4.

Results of the experiment with dry ice in Classroom 1, with and without open doors.

In addition, this method helps calculate actual air changes per hour (ACH) in space. The HVAC flow in Classroom 1 is 750 m³/s. For a volume of 363 m³ corresponding to the space, a theoretical ACH of 2.06 (door closed) is estimated. An ACH of 4.42 (with opened doors) and 2.03 (closed doors) has been determined by dry ice tests. It suggests that the tests have been carried out correctly and that the opening of the doors represents a considerable improvement in air renewal compared to the closed space.

Sometimes, the subsequent monitoring phase can be used to assess the benefit of the corrective actions imposed. In the base case, opening doors in Classroom 1 was enough to achieve adequate air renewal, even in the summer months. Ideally, during monitoring, it would be necessary to implement a system that alerts those responsible for the space if the established CO₂ limits are exceeded. This strategy allows for preventing possible incidents and correcting CO₂ levels exceeded in streaming, thus, guaranteeing adequate air quality.

3.3. Applying and validating the method in local businesses

To perform a broader application of the methodology to diagnose, correct, and monitor CO₂ we performed a study in 40 retail stores. To this end we modified the phases to meet the pre-established requirements, and the absolute CO₂ concentration was measured (instead of the outdoor-indoor CO₂ increase). In addition, the ppm/person indicator could not be used since the businesses could not provide a continuous record of the occupancy. Initially, a preliminary characterization (diagnosis) of the participant businesses was conducted. 23 businesses with CO₂ values above the maximum recommended (800 ppm) were identified. Eight of them did not require specific follow-up interventions as they were minimal and discrete increases. The other 15 participants were subjected to an exhaustive evaluation that served to design and implement a series of prevention measures that had to be implemented to improve air recirculation in the commercial spaces.

The implementation of corrective measures was modified from the original approach. Instead of in-situ experimenting, it was decided to assess the improvement of the proposed measures throughout the MCS stages. Thus, we evaluate the air renewal patterns from week to week. In our case, it was enough to agree on the terms with each person in charge through fortnight reports and phone calls. The presented strategy allowed us to reduce the costs of the project and to be able to evaluate several stores at the same time.

In the period studied (January-February 2022), the average CO₂ of all businesses was 547 ± 87 ppm in the first fortnight of January, 559 ± 86 ppm in the second fortnight of January, 531 ± 82 ppm in the first fortnight of February, and 529 ± 89 in the second half of January (Figure 5a). Substantial differences were observed according to the type of store, as shown in Figure 5b. Specifically, the highest CO₂ peaks were observed in Kids’ clothing stores (691 ± 209 ppm with maximums of 2,297 ppm). It may be associated with a limited opening of doors or windows to maintain the thermal comfort of the children. In contrast, florists reported the lowest CO₂ levels (mean of 466 ± 47 ppm with a maximum of 874 ppm), possibly due to customers’ reduced time during service.

• Download figure
• Open in new tab

Figure 5.

(a) Mean of CO₂ value (inbox) of all businesses based on the fortnight and (b) mean of CO₂ and maximum measurement depending on the type of business where the mean value (blue) and the maximum values (pink label) are shown.

A total of 15 businesses were selected for exhaustive monitoring. Among them, only five agreed to implement the corrective measures. The following sections describe the evolution of 3 establishments under the three typical scenarios:

1. Retail stores with good starting air quality. Initially, 17 shops showed a renewal of the air appropriate to the architecture of the space and the type of activity carried out. Three cases are shown below as an example.

   Example 1) In a shoe store, the limit of 800 ppm was not exceeded during the recorded period (Figure 6a). Specifically, it obtained an average CO₂ value of 441 ± 42 ppm, reaching a maximum value of 776 ppm.

   Example 2) A food store was kept below 800 ppm (Figure 6b). Despite the large influx of customers, the levels maintained an average of 454 ± 39 ppm, reaching a maximum value of 782 ppm

   Example 3) In one of the clothing stores, the maximum established by the scientific community of 800 ppm was not exceeded (Figure 6c). On average, it reached a value of 469 ± 33 ppm, reaching a maximum value of 783 ppm.

2. Businesses with improvable air quality that did not implement corrective actions. Of the 23 stores that showed levels above 800 ppm during the first week, only 15 of them did so continuously and with values above 900 ppm. The remaining eight stores were recommended to maximize the frequency of opening doors and windows. This section shows the evolution of CO₂ in a business that refused to establish the proposed corrective measures.

   Example 1) On average, a clothing store reached a value of 709 ± 295 ppm, reaching a maximum of 2956 ppm (Figure 6d).

   Example 2) A restaurant establishment reached an average value of 589 ± 139 ppm, reaching a maximum value of 1976 ppm (Figure 6e).

   Example 3) The mean of CO₂ in a children’s clothing store was 762 ± 229 ppm, reaching a maximum of 2297 ppm (Figure 6f).

3. Spaces with improvable air quality that implemented corrective actions. After implementing the prevention measures (especially the reinforcement of natural ventilation), the reduction of CO₂ in the establishments was substantial. Three cases are described below. The values of the first weeks concerning the final weeks are comparable, given that similar trends and averages are observed in the previous section.

   Example 1) In a clothing shop, the mean of CO₂ was reduced from 605 ± 78 ppm (maximum of 1100 ppm) to 529 ± 78 ppm (745 ppm) after trying different measures. The increase in the frequency of door opening (Figure 6g) was more effective in the fourth fortnight. In this period, no values over 800 ppm were registered.

   Example 2) CO₂ levels on average were reduced from 536 ± 89 ppm (maximum of 1001 ppm) to 485 ± 51 ppm (714 ppm) from the third fortnight in a food store, as shown in Figure 6h.

   Example 3) In a shoe store, the mean of CO₂ was reduced from 528 ± 67 ppm (maximum of 946 ppm) to 473 ± 45 ppm (764 ppm). Although, at one point, 800 ppm was exceeded (Figure 6i).

• Download figure
• Open in new tab

Figure 6.

Evolution of CO₂ levels in the monitoring period (January-February 2022) in (a-c) businesses with good starting air quality, (d-f) businesses with improvable air quality that did not implement corrective measures, and (g-i) businesses with improvable air quality that implemented corrective actions. The pink dots indicate the CO₂ curves of stores with poor air renewal, while the blue dots represent CO2 curves of establishments with good air renewal.