Particulate matters emitted by an firewood cook stove

3. Methods and Materials

The sampling kit included one Plantower PMS7003 as the dust sensor, one Bosch BME280 for the temperature and relative humidity. The data from the sensors were recorded using Python scripts and saved into SD card in .csv or .txt files in a Raspberry Pi. The sampling interval was one minute. The kit was equipped with a 4S-lithium battery to improve the mobility of sampling.

The primary location was at a residential home in an agriculture village. No industrial combustion was found in a 2-km radius. The provincial road is about 500-m to the North with dense trees in between. The traffic was considered as light to moderate which mostly motorbikes, passenger car and light struck operated. The wind direction during sampling was mixed with no distinguishable advective flow. The second location was a residential home in Hong Linh town with the national highway (A1) was about 100m to the South. The sampling points in both places for the background concentration (BC) were at the front yard with no active fan and no stove nearby.

sensor setup
Fig. 1: Configuration of sampling sensors including one Plantower PMS7003 for PM(s) and one Bosch BME280 for temperature and relative humidity. Front yard with no active fan was selected as a background sampling.

For sampling the PM(s) emitted by a firewood stove, the kit was placed about 50-cm away from the fireplace and at the same height as the base of stove. The stove was used to boil water and no others sources such as frying meats were occurred during this campaign. The stove was a simple setup with two steel bars bridging across a base made by concrete. No active fan was installed for accelerating burning as might found in other stoves. Firewood used was small trees and bushes gathered around. The cookstove was placed relatively open with only two sides adjacent to walls and the other sides were open to the garden.

measure stove
Fig. 2: Setting sensors to record PM(s) emitted by firewood cook stove.

For a comparison with a gas stove, one-time sampling was carried around 8:30 on April 30.

measure stove
Fig. 3: Setting sensors to record PM(s) emitted by gas (LPG) cook stove.

Data recorded into .csv and .txt files were processed using Jupyter Notebook editor with Python language. The graphs were produced using Matplotlib and seaborn libraries. The data was first cleaned abnormal peaks recognized as single-peak pulsing with the value larger than 500 (µg/m3) by filtering the dataset by Pandas library. The changing of PM(s) concentration were evaluated by comparing to the average concentration while cooking and the immediate background concentration (IBC) and to the average background concentration (ABC). The IBC is the average concentration of two-hours before and after the cooking period. The ABC is the average concentration for each day. The sampling kit was moved from the wood stove to the front yard after 10 to 30 minutes after the active cooking. Each cooking period lasted between 1.5-2 hours.