Compliance with regulations
In Taiwan, on average, each person drank 104 cups of coffee in 2020 according to annual coffee bean import statistics from the International coffee organization (ICO). Taiwan’s coffee shop density is the highest globally, and coffee chains Starbucks and Louisa have both exceeded 500 stores.37,38 People might have more chances to stay at coffee shops in urban area. This environment is not only a living and leisure space, but also a workplace. However, the Taiwan IAQ regulation does not regulate these shops up till now. Customers won’t carry any IAQ instruments usually. The aim of this study was to identify the observable factors that can be the significant indicators of IAQ concurrently. The time of spending in a coffee shop is ranged from 4-6 hours for working or studying to 15 minutes for grabbing a cup of coffee for to-go. So, we decided to inspect the 15-minute average concentrations to reflect the exposure of group with short staying periods and also an assurance for all groups. These time-weight averages were applied to check the compliance with the IAQ standards/guidelines. Among all investigated café, 11.2%, 18.2%, and 21.1% of the 15-minute averages of PM10, PM2.5, and CO2 accordingly did not meet the WHO guidelines and 8.3% of TVOCs exceeded Taiwan IAQ standard. However, these comparisons could only be a reference, as IAQ standard regulated the time-weighted concentration of 1 hour, 8 hours or 24 hours. Our results found the exceedances of short term (15-minutes) concentration remind the long-term time weighted average might underestimate the exposure of customers and employee at certain periods. The IAQ of café should be addressed as a workplace and/or a public environment to compliance with the regulations and to assure the healthy environments of people in these indoor spaces.
Cooking and bakery are the main sources of indoors’ PAHs. In coffee shop C and D, the p-PAHs were quantified in the range of 2.6 and 193.7 ng/m3 resulting from preparing the light meals. Abdullahi et al.39 reviewed the cooking emission studies and found the PAHs concentrations were variated with cooking styles, ranged from 0.2 - 1,590 ng/m3. Ielpo et al. 40 reported the mean concentration was 7.4 ng/m3 (range: 5 - 10 ng/m3) from bakery. All these were similar with our findings in p-PAHs. The difference might due to different cooking patterns and gaseous PAHs.
Significance of ranking
This study is the first one to rank the importance of indoor environmental factors and examine the relationships between factors and air pollutants in independent coffee shops as novel workplace. Results of this study showed that occupant density, indoor activities, CO2 concentration, and temperature can remind the concentration changes of PM10, PM2.5, TVOCs, p-PAHs, and CO. Limited researches have evaluated the determinants of IAP in coffee shops. In this study, occupant density was found being the most important determinant of the indoor concentrations of PM10, PM2.5, TVOCs, p-PAHs, and CO2. On the other hand, the occupant density was the third and fourth important determinant of the concentrations of CO and PM2.5-10. We found slightly positive dose-response relationship between occupant density and air pollutants, particularly in the increasing of CO2 concentration. This was also found in other peer studies.23,41 Previous study showed that the lower occupant density, the lower the PM concentration in preschools classrooms 16, but this study didn't find dose-response effects between occupant density and concentrations of PM10, PM2.5, or PM2.5-10. The higher occupant density, the higher the concentrations of specific VOC.15 We found the occupant density was the most important indicator for TVOCs. Coffee shops serve different functions for metropolitan people, such as social gatherings, studying, working and business meetings. Most of the shop design was the open-kitchen style, no significant segregation between dining and cooking areas. According the ANSI/ASHRAE standard 62.1, these investigated independent cafés were fitted in the occupancy category, cafeteria/fast-food dining and kitchen (cooking) both. While, the default occupant density with recommended ventilations of these two categories are different 35. So more empirical researches are recommended to assure the appropriate occupant density of these shops to compliance with ASHRAE recommended air class.
Indoor activities were the second important determinant of the concentrations of PM10, PM2.5, and CO. For TVOCs, p-PAHs, and CO2, the indoor activities were the fourth important determinant. Besides, this study further pointed out the major human activity associated IAQ was tobacco smoking. Previous studies showed that concentrations of indoor air pollutants, such as VOCs,1,9 PM2.5 ,1,11 p-PAHs,1 and CO,42 were associated with tobacco use. Only shop C allowed indoor smoking among the 4 investigated shops. The partial dependence effect of smoking in increasing the pollutant concentrations were stronger than other indoor activities by RF modeling with other environmental factors being controlled. Thus, it is important to ban the indoor smoking to reduce IAPs in coffee shops. Moreover, previous study pointed out the relationship between roasting coffee beans and concentration of CO,7 this study further showed that roasting beans was a more important human activity associated with indoor CO concentration than tobacco smoking. As due to a limited number of studied coffee shops, this result need further confirmed.
CO2 is a global indicator of IAQ and a rough indicator of the effectiveness of ventilation. High CO2 level implies the possibilities of indoor IAPs accumulations,20,23-25 our results showed that the concentrations of PM, TVOCs, p-PAHs, CO were proportional to the concentration of CO2. This confirmed the findings of other studies.20,43 Besides, we found that the CO2 were the top three important predictors of the concentrations of multi-size PM, TVOCs, p-PAHs, and CO. The partial dependence plots indicated that the concentrations of IAPs increased proportional to the CO2 concentration once it exceeded 1,000 ppm (Taiwan IAQ standard of 8-h average CO2). So, we recommend the low-cost CO2 monitor shall be installed in coffee shops to monitor CO2 and alert the IAQ. CO2 was affected by the occupants. However, RF modeling can overcome the collinearity of CO2 and occupant to ranking the importance of these two determinants.
Temperature and RH are the important factors of thermal comfort. Previous studies reported that temperature and RH were related to the IAPs levels positively.44-47 We found that the temperature was the top four important predictors of the concentrations of multi-size PM, TVOCs, p-PAHs, CO, and CO2. Relationships between indoor pollutants concentrations and temperature were complicated, and the partial dependence plots of RF models in our study show that the IAP concentrations were consistently increased as the temperature increased within a certain range. The air conditioners were turned on during the business hours of these investigated shops. So, the temperature was kept constant with small variation (range 19.9 - 25.8°C). The results were complied with the IAQ standards/guidelines of major Asian countries. Therefore, if temperature was selected to be an indicator to alert the levels of IAQ, future research should include the indoor temperature with big variation and be caution of the non-linear relationship with the IAP concentrations.
RH is less important than temperature to be an indicator of IAQ, and our partial dependence plots results showed the complicated non-linear relationships between IAPs and RH. We observed the RH was the most important predictor of the PM2.5-10 concentrations. PM2.5-10 decreased slightly as the RH increasing. Oliveira et al. found the concentrations of CO2 and PM2.5 were affected by the RH inversely in the kindergartens.43 Some studies reported the positive correlations between RH and IAPs (e.g.CO2, HCHO).20,23 The relationships were inconsistent among different researches. As we found, the relationship between RH and IAPs was complicated and nonlinear. RH is not an appropriate indicator for IAPs. Still, RH is relevant on the IAQ study, because it affects perceived IAQ comfort, synergistic effects may occur with air pollutants as well.48 High RH provides the optimal condition for bacteria, fungi and viruses proliferation.49 Indoor RH is not easy to control in Taiwan’s subtropical climate. Our monitoring data of coffee shops resulted in 42.6% of the RH over 70%. Currently, the IAQ standards or guidelines of Asian and European countries for temperature and humidity criteria are different.35 The humidity (RH) of coffee shops shall be maintained within a comfort range according to the climate conditions.
The factors, weekday, ventilation status, shop’s pattern, and locating on the main street were less important in predicting the IAPs as they were not listed by the rfcv module in RF model analysis. However, it did not mean that they had no effects of IAQ. The possible reasons were due to the small variations of the four investigated shops. For example, only three situations of the ventilation status were observed (table 2), so the significance of these factors can’t be identified by the statistics analysis.
Limitations and strength
This study was limited by the scale variations of the café, but we proved that RF was able to deal the limited numbers of data and build models to examine relatively complex and non-linear relationship between IAPs and determinant variables. Meanwhile, the RF model calculation considered all environmental factors simultaneously and provided insight in potential causal relationship between air pollutants and environmental factors, particularly the temperature and RH. The total picture and interrelationships between different enironmental parameters were illustrated. On the other hand, our RF model identified the most important determinat of CO2 concentraiotn was occupant density (Figure 5(A)) and the most important indoor activity in affecting PM10 and PM2.5 levels was occurred with “smoking”(Figure 2(C), (H)). A Rome’s study reported the tobacco smoking increased the indoor PM2.5 concentrations by two to three times of the non-smoking sites.1 This was consistent with the present study and proved the reliablility of the RF analysis results.
Recommendations
Although the strength of this study is that it used RF models to examine indoor environmental indicators for reminding levels of IAQ in novel independent coffee shops, there still were some limitations. First, some studies showed that ventilation could influence IAQ, but this study didn’t measure outdoor air change rate in these participated shops. However, previous studies showed indoor ventilation was non-significantly associated with concentration of CO and PM.50,51 Moreover, previous findings showed natural ventilation, such as window opening, and outdoor air pollutants would influence IAQ.17,50 Therefore, it is suggested that future researchers could consider outdoor air pollution when investigating IAQ if there are natural ventilation in coffee shops. In addition, this study did not include chain coffee shops. In Taiwan, most of the chain coffee shops are located at the commercial buildings with central air conditioning systems which are different from the independent café with individual air-conditioners of this study. More coffee shops with different air-conditioning designs will be included in our future studies to validate and extend the applicability of the results of this study. Second, only four independent coffee shops participated in this study. Hence the findings cannot be inferred to other types of coffee shops. The importance ranking of determinants that affect IAPs may be changed due to the large variations of environmental factors if various types of coffee shops are included in future study. Last, our sampling time did not include the summer season, future study should evaluate potential seasonal variations and their influence.