Measurement area and vehicle
The data utilized in this study were obtained from two measurement campaigns, to achieve extended data ranges in different ambient particle concentration levels. Winter testing in China and summer testing in Sweden also represented Asian and European environments.
Gothenburg city is located on the west coast of Sweden. By year 2018, this second largest city in Sweden reached a population of around 571,900 people; the annual average atmospheric PM2.5 concentration was reported to 7.7 μg/m3 from the local environment monitoring station (Gothenburg Municipality 2018). In total, 250,483 motor vehicles were in use in year 2018, 76% of the total vehicles were passenger cars, and goods vehicles accounted for 9%; within passenger cars, 56% were petrol powered, 34% diesel, hybrid 4%, and pure electric 1%, while 91% of goods vehicles were diesel (Transport Analysis 2018).
The Lundby Tunnel located in Gothenburg was chosen as the test site, where the ambient particle level is elevated, compared with open city roads or highways in Gothenburg, since tunnels exhibit low dispersion and dilution of pollutants. Westbound and eastbound directions each have 2 lanes. According to the local traffic management institute, totally 26,562 vehicles passed the 2060-m-long tunnel in westbound direction in 24 h and 85% of them were personal cars driving in an average speed of 69 km/h, on April 26 in 2018 (Swedish Transport Adminstration 2018).
The measurements were performed during May 2018 until July 2018, between 08:00 and 14:00 in rain-free days, when relative humidity was lower than 70%, to maintain a suitable working environment for the measuring instruments. No major differences of the metrological parameters and no obvious traffic events or congestions were noted. Repetitions of the same test case were performed during different days with similar conditions. The test vehicle was standing inside the tunnel with engine and HVAC system on, at an uphill emergency parking spot in westbound direction, where passing vehicles can be expected to release elevated pollutants. This was to maintain long stable measurement periods for each test case, compared with driving through the tunnel which takes 2 min.
The winter measurements were performed during January 2019 driving on freeways and highways, with speeds ranging within 50–120 km/h, along the relatively polluted 760-km route from Linyi to Beijing, Northern China, as in Fig. 1. Monthly average ambient PM2.5 concentrations in major cities on the route at that time were, Beijing 52, Linyi 114 and Baoding 137 μg/m3 (CNEMC 2019). Beijing had a population of 21.54 million, and 6.08 million cars in 2018, of which 5.74 million were passenger cars (Survey Office 2018). Passenger cars accounted for around 89% of on-road vehicles in China, while goods vehicles were 11%, and 89% of all vehicles were petrol powered, 9% were diesel. Heavy goods vehicles were estimated to contribute to around 59.9% of total annual particle emission from all vehicles (MEEPRC 2018).
The measurements were carried out in two similar Volvo cars produced in 2018. The test vehicle used in Gothenburg was a PHEV (plug-in hybrid electric vehicle), Model Volvo XC90 (Table 1) with its original HVAC system. The test vehicle used in China was a diesel car, Model Volvo S90 (Table 1), with the same type of HVAC system as the PHEV used in Gothenburg.
The filter type used in both vehicles was a multi-layer electrostatically charged synthetic filter made of polypropylene and active carbon, with dimensions of 337 × 238 × 41 mm. One newly manufactured filter and one aged filter were used in both cars. The aged filter was aged in an HVAC test rig with ducts connected to outdoor air, in 2018 April at Shanghai, where monthly average outdoor PM2.5 level at time was 42 μg/m3 (CNEMC 2018). To simulate the actual usage of filter in customer driving, ventilation fan speed was set to low (1430 rpm, around 200 kg/h), no recirculation, and ageing time was 500 h, which represents around 1-year driving, the recommended filter service interval in China.
Furthermore, both cars were equipped with an ionizer unit to study the influence of ionization on air filtration. The unit is manufactured to fit the air inlet dimensions and is installed before the water separation unit in front of the HVAC, around 50 cm upstream of the filter. It was manufactured to fit to the inlet dimensions and worked with a voltage of − 7 ± 1 kV. The high-voltage bars with sharp edges form corona discharge tips, and continuously discharge unipolar ions. Particles contained in the incoming air thus changed polarity when colliding with ions.
Particle mass and number concentration measurements were performed with Grimm MiniWRAS (Mini Wide Range Aerosol Spectrometer) model 1.371, with a log interval of 1 min. The instrument measures particles of aerodynamic diameter from 10 nm to 35 μm, distributed into 41 channels. Light scattering and electric mobility detection methods are jointly adopted, for size ranges of 0.253 to 35 μm, and 10 to 193 nm, respectively. Within a measurable mass concentration range 0.1 μg/m3 to 100 mg/m3, reproducibility of mass concentration is ± 2 μg, and ± 3% of count values (GRIMM 2019). Thus, mass and number concentration of all size channels are acquired, including PM2.5, UFP counts from 10 to 100 nm. Annual calibration was performed by a supplier and automatic self-test done by instrument at each startup.
Two inter-calibrated MiniWRAS were measuring simultaneously outside and inside the cabin. The outside sampling tube was placed immediately outside of the HVAC air intake below the wind shield, which measured exactly the air at HVAC upstream. The inside sampling tube was placed above the middle armrest between the front seats, as in Fig. 2, as recommended by Abi-Esber and El-Fadel (2013). This position was chosen to measure the particle concentration in the well-mixed in-cabin air, rather than air samples at HVAC direct outlets. Two instruments started at the same time.
In addition, to investigate the possibility of ozone generation from the ionizer, 1B Technologies Model 205 Dual Beam Monitor (UV-absorption principle) was used to monitor the in-cabin ozone concentration, with a measurement frequency of 0.5 Hz and accuracy of 2% of reading above the 2.0-ppbv detection limit (1B Technologies 2019). Ozone was measured at the same place as the in-cabin particles.
Temperature, relative humidity (Rotronic Hygroflex HF534) and solar intensity sensors were mounted both inside and outside the cabin. A CO2 meter Vaisala Darbocap GM70 in the cabin, with the sampling head mounted on the back side of the co-pilot seat pointing to the centre of the cabin, was also used in the cabin. The testing personnel kept distance to all sampling heads throughout the measurement period, to refrain from direct breath influence.
Measurement content and setup
To represent common driving conditions, as well as investigate PM2.5 concentration and UFP counts inside and outside the car under different circumstances, several climate parameters were varied and combined. Each combination of parameters, including different ventilation mode, airflow, filter, and with or without ionization, was defined as a test case, and ensured to be performed at least three times in different days. Totally, 28 test cases (explained in Table 2) equally comprise 14 baseline (no ionization) and 14 ionization cases. Considering repeated data collections, 127 of all 134 data collections were valid, which is presented in Table 3 as number of samples.
The baseline measurement was firstly measured as follows: the ventilation airflow rate was set to low, there was no recirculation, and both new and 500-h-aged filters were tested. Simultaneous inside and outside particle and ozone concentration levels were recorded. The climate settings during all testings were windows closed, AC on, and desired temperature of 22 °C, as well as a constant ratio of airflow at the panel and floor vents. Smoking was forbidden, and 2–3 persons sat in the vehicle. All these parameters were controlled by a software connected to the vehicle control unit, to maintain a similar environment in all measurements.
Furthermore, 3 other mechanical ventilation airflow levels (extra low (Xlow), medium, high) were also tested, to investigate the influence of the ventilation rate (estimated airflow rates at these 4 levels are around 100, 180, 260, and 380 kg/h, respectively). Three recirculation (RC) levels (30%, 50%, and 70% of total ventilation air comes from recirculation) were compared as well. Recirculation air from the vehicle cabin was mixed with incoming air and then filtered. Test cases of 4 airflow rates were always done in sequence, similar to 4 recirculation degrees, to gain similar ambient conditions when relevant cases were compared. These parameters were also controlled by the aforementioned software.
Apart from baseline testing, pre-ionization was added as another varying parameter. The ionizer was turned on, after each corresponding test case, and then the same data collection procedures were repeated. The aim was to maintain as close ambient pollutant and metrological conditions as possible for comparisons of with and without ionizer.
All the instruments were turned on at least 30 min in advance for warming up and stabilization. The ventilation parameters were varied firstly, and when a stable in-cabin air quality was achieved, a data collection interval of around 5–10 min started.
Average inside and outside PM2.5 concentrations and UFP counts were calculated for each data collection interval firstly, by averaging the 1-min data, and then the general average was calculated for each test case (based on all data collection repetitions). The indoor to outdoor ratio (I/O ratio) of PM2.5 mass concentration and UFP counts were analysed afterwards, to evaluate the filtration performance regardless of ambient pollution level.