Natxo García López 2017
Fine particle emissions from biomass cookstoves
Evaluation of a new laboratory setup and comparison of three appliances
Natxo García López
Fine particle emissions from biomass cookstoves
Evaluation of a new laboratory setup and comparison of three appliances
Natxo García López
Umeå University
Supervisor: Christoffer Boman
Co-supervisors: Robert Lindgren and Ricardo Carvalho
Abstract
It is estimated that around three billion people globally rely on traditional usage of biomass to cover their daily energy needs, which causes health and social inequality problems and contributes to global warming. Thus, the study of particle emissions from cookstoves provides important information that can help improve global welfare.
This study aims to (a) evaluate a new laboratory setup for measurement of particle emissions from cookstoves and (b) use this setup to compare the particle emissions from three cookstove appliances that cover the whole spectra of used technologies, namely a 3-stone fire, an improved cookstove and a gasifier stove.
Emissions of total suspended particles (TSP), fine particles (≤ 2500 nm) and other emission components such as carbon dioxide were measured.
Results from this study show that the new laboratory setup is appropriate to measure and investigate fine particle emissions from cookstoves as well as cookstove efficiency. Further, it also shows that the 3-stone fire was the cookstove with the highest emission factor of all, followed by the rocket stove and the gasifier stove respectively. The analysis of the data obtained from the transient particle measurement provided some information on the particle size and the soot and salt contained in the overall emitted particles.
Finally, some suggestions such as continuous measurements of background particle and CO2 levels are recommended. Additionally, further research ideas are also proposed.
Key words: Particulate matter emissions, cookstoves, gravimetric measurements, soot and salt particles, SMPS.
Table of Contents
Abstract ... - 2 -
1. Introduction ... - 4 -
2. Material and methods ... - 7 -
2.1 Laboratory setup and test procedure ... - 7 -
2.1.1 Setup description ... - 7 -
2.1.2 Test procedure ... - 8 -
2.1.3 Instrumentation ... - 11 -
2.2 Data analysis ... - 12 -
2.2.1 Mass loss ... - 12 -
2.2.2 Calculation of the dilution factor over the hood ... - 12 -
2.2.3 Offline particle emissions data ... - 14 -
2.2.4 Transient particle emissions data ... - 15 -
3 Results ... - 18 -
3.1 Results from the evaluation of the new laboratory setup ... - 18 -
3.2 Results from the comparison of three cookstove appliances ... - 22 -
3.2.1 Fuel consumption and boiling and simmering times ... - 22 -
3.2.2 Mass loss over time ... - 23 -
3.2.3 Water temperature and CO2 emissions ... - 24 -
3.2.4 Particle emissions based on offline measurements ... - 25 -
3.2.5 Dilution factor calculation ... - 26 -
3.2.6 Particle emissions based on transient measurements ... - 27 -
4. Discussion ... - 33 -
4.1 New laboratory setup and test procedure ... - 33 -
4.2 Comparison of the three cookstove appliances ... - 35 -
5. Conclusions ... - 39 -
6. References ... - 40 - Appendix 1 ... I
1. Introduction
Biomass combustion for cooking and heating purposes is an ancestral practice still in use by almost 40 % of the world’s population (1). The most widely used cooking technology in countries with developing economies is the 3-stone fire arrangement. Nevertheless, during the last decades, the development and implementation of improved cookstoves has been remarkable. Although improved cookstoves are relatively simple appliances, they can still improve the stove efficiency and reduce particle emissions considerably in comparison to the 3-stone arrangement (2). Recently, the usage of small domestic gasifier stoves operated with processed biomass, mostly in form of pellets, has been proposed as a way to further improve stove efficiency and reduce its emissions (3). In some countries with developing economies, such as Kenya, this technology is starting to be implemented and commercialized.
The use of each of the different available technologies has implicit health, social and environmental consequences. Emissions from traditional combustion of biomass have noteworthy health effects on the stove users, who are mostly women and children, which in turn leads to social inequalities. Moreover, the traditional usage of biomass has environmental impacts in form of deforestation, soil degradation and increased atmospheric particle and CO2 concentrations. CO2 emissions contribute to higher global concentrations, thus contributing to global warming, and particle emissions contribute to higher local, regional and global particle concentrations. Recently, the Global Alliance for Clean Cookstoves claimed that clean cooking can contribute to achieve 10 of the 17 UN Sustainable Development Goals (SDGs), that were adopted by 193 nations in 2015 (4). In the last decades, great initiatives have shed some light on the field of clean cookstoves, but nevertheless there are still some aspects to be studied more carefully, for instance particle characteristics and not least the toxicology of the particles emitted by different appliances.
Fine (sub-micron) particle emissions (<1 µm) can be classified into three categories, i.e.
inorganic salts, organic matter and soot particles. In biomass combustion, inorganic salt particles, which are small particles with diameters ranging between 20 and 70 nm (5,6), are mainly composed of potassium salts, e.g. KCl and K2SO4. Soot particles are typically larger particles, although sub-micron, composed of agglomerates of smaller solid spherical particles, often unburned carbonaceous particles, in the range of 100-500 nm. The group of particles that are comprehended in the soot particles usually have different particle characteristics, for
instance surface chemistry. Soot can be described by its optical properties as “black carbon”, which is considered to contribute to global warming through two main processes. First, particles that are suspended in the atmosphere absorb sun radiation, and second, particles that are deposited on top of snow and ice surfaces reduce the albedo of those surfaces which causes an increase in the local temperatures (7).
The need for standarized methods to test cookstoves made that already in the beginning of the 80s, the Volunteers in Technical Assistance (VITA) launched the first version of a standard procedure for testing cookstoves. Since then, the standarized Water Boiling Test (WBT) has been widely used. The WBT is intended to measure “how efficiently a stove uses fuel to heat water in a cooking pot” and “the quantity of emissions produced while cooking” (8). Despite the suitability of the WBT to test cookstove efficiency and to measure the quantity of total emissions, e.g. as total particulate matter (PMtot), the test is not developed and adjusted to enable detailed aerosol studies (e.g particle formation and detailed properties). This is mainly due to the high moisture content of the flue gases into the hood caused by the boiling water.
Thus, some modifications should be applied to the standard WBT in order to allow for more specific and advanced aerosol particle studies that can give further information about the gas- and particle emissions from cookstoves (9).
The WBT is typically conducted in the laboratory. However, particle emissions from cookstoves may be studied in the field as well. Both approaches have advantages and disadvantages. The challenge with a field study on particle emissions is that many parameters cannot be controlled as accurately as in the laboratory. Nevertheless, results obtained from a study carried out in the field, under real-operating conditions, are likely more representative than those obtained in the laboratory. In other words, the strengh of a laboratory study is that parameters can be better controlled and studied in designed experimental matricies, whereas its weakness is the fact that the simulated cooking event is not a real cooking event.
Furthermore, in the nature of a laboratory study, there is an intrinsic search of the ability to reproduce experiments. That is, the repeatability of the experiments. In contrast, field experiments are very difficult ro repeat under the same experimental conditions.
Due to the great contribution from many research groups to the field of cookstoves, focusing on efficiency and emissions, the knowledge in this field has increased considerably in the past couple of decades. In adittion, interdisciplinary studies involve people from different disciplines , applying advanced instrumentation to measure particles, have resulted in some more detailed studies on particle emissions from biomass cookstoves (5, 10,11,12). In general fine particles have been defined as PM2.5, i.e. particles with an aerodynamic diameter <2.5 µm, and ultrafine particles have been defined as PM0.1, i.e. particles with an aerodynamic diameter
<0.1 µm. However, in this work, fine particles (fine PM) are defined as PM1, which is in line with what has been discussed earlier for biomass combustion particle emissions (10). Insights on fine and ultrafine particles emission are of great interest from a human health perspective.
In contrast to bigger particles, which are deposited before they reach the lungs and that might be coughed out of the body when breathed (Fig. 1) , ultrafine particles might reach the alveolus, where they can affect to the actual lung cells or be taken up by the bloodstream (13).
Figure 1. Illustration of the respiratory system and relation to the deposition regions of different particle sizes (left).
Deposition fraction as function of particle size (right). Reprinted from (13,14).
A laboratory study was therefore carried out with the overall aim to gain novel knowledge of particle emissions from biomass cookstoves. The specific objectives of this study are to: I) evaluate a new laboratory setup for studies of efficiency and emissions from cookstoves, and II) test three different types of cookstove appliances, using wood fuels, with respect to fine and ultrafine particle emissions.
2. Material and methods
This study consists of an experimental part followed by a data analysis part. The experimental part comprises an evaluation of a new laboratory setup to study particle emissions from cookstoves as well as an experimental campaign in which three cookstove appliances were tested. The data analysis part aims to test a methodology to analyse data generated using the used laboratory setup.
2.1 Laboratory setup and test procedure
The laboratory setup is a combination of a fixed installation, including e.g. hood, flue gas channel and fan, as well as a number of instruments that can also be used in other experimental setups if needed.
2.1.1 Setup description
The setup is composed of a flue gas channel provided with a hood and a fan. The flow of flue gases at the flue gas channel is regulated by a frequency regulator. The system is provided with a scale to measure the mass loss of fuel during the combustion process.
Additionally, the temperature of the water, the temperature of the flue gases both under the pot and at the flue gas channel are also measured and recorded every five seconds. In the present study, only the water temperature is reported, while the other two temperatures were used to understand the collected data and to monitor the process while the experiments were performed.
Carbone dioxide (CO2) concentrations were measured with a non-dispersive infrared (NDIR) gas analyser (LI 840A, LI-COR Corp.) that sampled the exhaust gases from the flue gas channel.
Offline and transient particle emissions were measured using a total dust filter sampling line and a Scanning Mobility Particle Sizer (SMPS) system (TSI Corp.), respectively. The flue gases were diluted with an ejector dilutor (Dekati Ltd.) previous to be analysed by the SMPS (Fig. 2).
Figure 2. Schematic illustration of the laboratory setup used in this study.
Filter based offline measurements, e.g. of Total Suspended Particles (TSP), give information on the overall mass emissions of particulate matter and also allow to make further chemical analyses of the collected particles. In contrast, transient measurements are time resolved measurements that give information on the dynamic behaviour (changes) of the measured parameter. In this study, mass loss, water tempertaure, CO2 emissions and particle number size emissions (by SMPS) were measured.
2.1.2 Test procedure
In this study, a modification of the standard Water Boiling Test (WBT) was used to study particle emissions from three different cookstove appliances.
Given that some stoves have a big thermal mass, the WBT intends to assess whether the emissions from a stove are affected by the stove temperature. Thereby, the test is comprised of three phases, i.e. cold start, hot start and simmer phase. The test starts with a known amount of water in a standard sized pot which is heated by combusting pre-weighted fuel in the stove appliance to be tested. The first phase (cold start) starts with the stove and the water at room temperature and ends once the water reaches the boiling point. Before starting the second phase, the water in the pot is replaced by new water at room temperature, whereas the stove remains hot from the previous phase. The third phase (simmer phase) starts once the water used in the second phase (hot start) starts to boil and lasts for 45 minutes.
As previously described, the aims of this study were to gain novel understanding of the particle emissions from different cookstove appliances and to test a new laboratory set up. Given that the water boiling test WBT is a test designed to, mainly, test cookstove efficiency (8), the test was modified in order to be able to also gain insights into particle emissions.
In indoor cooking, the emitted gases and particles from the cookstove are progresively diluted as the flue gases move away from the fireplace, whereas in the herein presented setup, the emitted gases and particles are gathered by the hood and conducted into the flue gas channel (100 mm in diameter). When pre-testing the setup, it was seen that when the water started evaporating, the increment in water content of the flue gases affected the results from the particle emissions measurements.
One of the most important modifications of the WBT implemented in this study was therefore to use a tight lid to cover the pot. Water vapour was then conducted out of the extraction hood through a teflon tubing connected to the lid. Adittionally, a thermocouple was inserted on the pot through the lid. The pot used to boil the water is a ‘Sufuria’- the most common cooking pot used in Sub Saharan Africa- here acquired in a local market in Kenya.
Another substantial difference between the standard WBT and the test procedure used in this study was related to the test phases. The WBT divides the test in three phases, i.e. cold start, hot start and simmering phases with the aim of identifying effects of the stove thermal mass.
In contrast, and given that the stoves tested in this study are low mass stoves, the test used in this study was divided into two phases, namely boiling phase and simmering phase (Fig. 4).
Thus, the water that was already heated during the boiling phase remained in the pot until the test was completed.
Figure 4. Illustration of the water temperature profile over time in the WBT (left) reprinted from (8), illustration of the water temperature profile over time for the runs carried for this study (right).
Three cookstove appliances were tested: a 3-stone fire, a rocket stove and a gasifier (Fig. 5).
Figure 5. The tested cookstove appliances. From left to right: the 3-stone fire arrangement, the rocket stove (Zama Zama) and the gasifier stove (ACE 1).
The 3-stone fire is the most traditional and extended way to arrange a fireplace for cooking purposes. The arrangement used in this study consisted on 3 stones (10x10x4 cm) evenly distributed around a metal ring of 250 mm in diameter. Once the stones were placed, the ring was removed and the wood logs were placed.
The technology known as rocket stove is based on a combustion chamber with an air inlet and a small “chimney”. There are many different designs but the principle is that the flames are gathered in the chimney and conducted to the pot. The fact that all flames are conducted to the pot and that the combustion gases have more time to be combusted than in a 3-stone fire arrangement, makes the rocket stove design a more effective stove than the former. The rocket stove used in this study was a commercially available cookstove known as Zama Zama.
The fuel supply for the 3-stone and the rocket stove appliances is controlled by the user, which means that the fuel sticks are added progressively as they are consumed, i.e. a batch-wise manner. The sticks were obtained from manually chopped wood logs, and the composition of the used birch fuel is given in Table 1 in Appendix 1.
The third appliance tested in this study was a biomass gasifier known as ACE 1. In this case, standard softwood pellets, commercially available on the Swedish market, were used as fuel, and the composition of the used pellets is given in Table 2 in Appendix 1.
Cookstoves dimensions, fuel type and dimensions and pot characteristics used with the different appliances are compiled in Table 1.
Table 1. Summary of some of the key components of the laboratory setup.
Cookstove Cookstove dimensions [mm]
Fuel Fuel dimensions [cm]
Pot dimensions [mm]
Pot capacity [L]
Water volume [L]
3-stone fire ø 250 height 100
Birch sticks
1.5*1.5 to 3*3 by 30 length
ø 280 height 180
7 liters
5 liters Rocket stove ø 240
height 270
1.5*1.5 to 3*3 by 20 length
Gasifier stove ø 220 height 240
Spruce/pine pellets
0.8 ø
2.1.3 Instrumentation
The scale used in this study was a KERN DS 60K with a precision of ±0.2 g. Data was recorded every five seconds. The mass loss data was then used to further calculate mass loss rate (g/s).
Water temperature was measured with a thermocouple (type-K) located in the centre of the pot and at 50 mm from its bottom as defined in the WBT. Data was recorded every five seconds.
CO2 emissions were measured with the LI-840A instrument and recorded every second.
In order to quantify the total mass of the particles emitted during a whole run, a standard total dust filter setup was used. The total dust setup was provided with two fiber glass filters of 90 mm in diameter (manufactured by Munktel). The filters were weighed before and after the test with a laboratory scale with a precision of ± 0.01 mg. The total dust setup was provided with a pump that sucked the flue gases through the fiber glass filters.
Transient particle emissions measurements were performed using a Scanning Mobility Particle Sizer (SMPS) (Fig. 6).
Figure 6. The used SMPS instrument (left) and a typical screenshot of the SMPS under operation (right).
2.2 Data analysis
2.2.1 Mass loss
The mass loss data obtained from both the 3-stone and the rocket stove appliances had to be processed due to the fact that the fuel additions led to large steps in the weight data and also since the rearrangement of wood sticks by the user led to some deviations.
Thus, the raw data was first manually processed in Excel to remove the steps corresponding to fuel additions. The obtained data series were then smoothed in MATLAB using the smoothing function “smooth” five consecutive times in order to get rid of the smaller deviations corresponding to wood rearrangement and thus obtain a derivative curve with less noise. Finally, the derivative curve, illustrating the mass loss rate, of the smoothed curve was calculated.
For the gasifier stoves, the obtained data from small scale the runs performed did not need to be processed in the same manner given that the mass loss data was already smooth.
2.2.2 Calculation of the dilution factor over the hood
Other laboratory setups designed to measure particle emissions from small scale biomass combustion, such as woodstoves or pellet burners, have in general some differences with the herein presented setup. One peculiarity of this setup is that the hood is open to the laboratory and therefore the sample becomes diluted before it enters into the flue gas channel (Fig. 7).
Thus, the dilution occurring over the hood needs to be considered.
Figure 7. Schematic illustration of the laboratory setup with the most relevant flows along the system.
The amount of produced combustion gases was calculated based on the mass loss of fuel per unit time and the fuel composition (C/H/O, moisture and ash), assuming stochiometric combustion. From this calculation, the volume of the total produced gases as well as the volume of produced CO2 per gram of combusted fuel were obtained.
Thereafter, the total produced gases and the CO2 flow were calculated according to equations (1) and (2).
Flow = Tpg × − ∆
∆ (1)
Where:
= Flow of total produced gasescalc, in Tpgcalc = Total produced gasescalc, in
∆
∆ = Mass increment per increment of time, in
Flow CO = CO × − ∆
∆
(2)
Where:
Flow CO2 calc = Flow of produced CO2 calc, in Prod CO2 calc = Produced CO2 calc, in
∆
∆ = Mass increment per increment of time, in
Given that the flow of flue gases is constant and that the flow of the total produced gases varies over time, the dilution factor over the hood will change over time as well which was calculated according to equation (3).
= (3)
Where:
Dil factHood = Dilution factor at the hood Flowfg = Flow at the flue gas channel in
= Flow of total produced gasescalc, in
Moreover, the flow of CO2 in the flue gas channel was calculated by the addition of the calculated flow of emitted CO2 plus the atmospheric concentration of CO2 times the flow of dilution air, according to equation (4).
= Flow CO + 0.0004 × − (4)
Where:
c = Calculated flow of CO2 at the flue gas channel Flow CO2 calc = Flow of produced CO2 calc, in
0.0004 = [ ]
FlowTfg calc = Flow of total produced gasescalc, in
Flowfg = Flow at the flue gas channel in
Finally, the flow of CO2 at the flue gas channel was converted to parts per million (ppm) according to equation (5).
. = × 106 (5)
Where:
CO2 conc.calc fgc = CO2 concentration at the flue gas channel [ppm]
c = Calculated flow of CO2 at the flue gas channel Flowfg = Flow at the flue gas channel in
The calculated time resolved CO2 concentration in the flue gas channel were then compared with the measured CO2 concentration in order to evaluate the predictive power of these calculations.
Furthermore, and given that there are clear devolatilization and char combustion phases during the batch-wise combustion of pellets in the gasifier stoves (15) , two different fuel compositions
were assumed when calculating the flow of total produced gases and the produced CO2.
The assumed fuel compositions for the devolatilization and char combustion phases, respectively, are given in Table 3 in Appendix 1.
2.2.3 Offline particle emissions data
The influence of the background concentration of particles by the dilution air was calculated and thereafter subtracted from the total sampled mass on the filters by applying the dilution
factor for the hood to the total amount of flue gases and assuming a typical indoor background particle concentration of 10 000 part/cm3 and a particle density of 1 g/cm3.
Firstly, the data obtained from the gravimetric analyses performed with the total dust filter sampler is presented as milligrams of emitted particles per completed test (mg/test). The emissions from each run were divided by the number of minutes that it took for each run to be completed, thus obtaining emissions on time basis (mg/min). Additionally, the emissions were divided by the mass of fuel used to complete the run, which allowed to present the data as emissions on fuel basis (mg /gfuel). Furthermore, emissions on an energy basis (mg/MJ fuel) were obtained by dividing the emissions on fuel basis by the heating value of the used fuel.
Finally, the emitted particle mass per delivered energy (mg/MJdel) was obtained by applying the thermal efficiency factors (16) compiled in table 2.
Table 2. Fuel and gross heating values as well as thermal efficiency for the three tested cookstove appliances.
Appliance Fuel Gross heating value [MJ/kg], dry fuel
Thermal efficiency [%]
3-stone fire
Birch sticks 18.7
18
Rocket stove 30
Gasifier Pellets 20.3 35
2.2.4 Transient particle emissions data
The data obtained from the measurements carried with the SMPS is reported in number of particles per cubic centimetre. However, as showed in Figure 8, the SMPS instrument sampled after the ejector [Part]Ae, i.e. giving the particle concentration after the ejector. Thus, in order to obtain the particle concentration of the produced gases, both the ejector and the hood dilution factors needed to be considered. The ejector had a dilution factor of 1:18. The particle concentration before the ejector is presented as [Part]Fg, i.e. particle concentration at the flue gas channel. Moreover, the fact that the dilution air had a background level of particles, [Part]Dil Air was also taken into account in the calculations, i.e. the particle concentration of the ambient air. Based on the calculated time resolved dilution factor and the assumed particle concentration in the dilution air, the estimated particle contribution from the background levels to the actual readings was subtracted and therefore the presented results are on a free background contribution basis.
Figure 8. Schematic illustration of the laboratory setup with the most relevant flows and particle measures along the system as well as the relation from which the particle concentration in the produced gases was calculated.
The relationship between the flows and the particle concentrations along the different points of the setup is represented in equation (6):
[ ] × = ([ ] × ) + ([ ] × ) (6)
Where:
[Part]fg = Particle concentration at the flue gas channel, in Flowfg = Flow at the flue gas channel, in
[Part]Tpg calc = Particle concentration at the calculated produced gases, in = Flow of total produced gasescalc, in
[Part]da = Particle concentration at the dilution air, in Flowda = Flow of dilution air, in
Based on equation (6), the particle concentration in the produced gases can be calculated according to equation (7):
[Part]Tpg calc = [ ] × [ ] × (7)
Where:
[Part]Tpg calc = Particle concentration at the calculated produced gases, in [Part]fg = Particle concentration at the flue gas channel, in
Flowfg = Flow at the flue gas channel, in
[Part]da = Particle concentration at the dilution air, in Flowda = Flow of dilution air, in
= Flow of total produced gasescalc, in
In this study, and in contrast to other studies in which particle emission data normally has been presented as typical overall size distributions of the emitted particle, it was here considered of vital importance to present the change on particle emissions over time.
Additionally, a “Salt versus Soot mode analysis” was carried with the aim to gain some understanding on the relationship between the number of different kinds of emitted particles and the size of those particles and how this relationship changed over time.
Particles with a mobility diameter between 15 and 70 nm were therefore considered to be alkali salt particles and particles between 71 and 553 nm were considered to be soot particles.
The number of emitted particles per volume were then analysed in relation to the particle emissions in terms of mass per volume. Salt particles were assumed to have a density of 2.5 g/cm3 whereas soot particles were assumed to have a density of 0.8 g/cm3.
Finally, the change in the relation between particle number and particle mass was determined by comparing the sum of normalized particle number concentration and the sum of the particle mass concentration by using a “bubble plot”.
3 Results
3.1 Results from the evaluation of the new laboratory setup
As described in the material and methods section, the mass loss data was treated with the aim to smooth the abrupt changes in mass caused by wood sticks being introduced or rearranged in the fire-place. Figure 9a shows the raw data obtained from an experiment carried with the 3-stone fire arrangement, while Figure 9b shows the mass loss treated curve after manually removing the original “jumps”. In Figure 9c, the derivative of the mass loss treated curve is also included in the graph. Finally, the resulting curves from smoothing the mass loss treated curve with the smooth function in Matlab 5 consecutive times, and the derivative of this smoothed curve, are shown in Figure 9d.
As discussed before, the mass loss data obtained from the gasifier stove did not need to be processed in this manner.
Figure 9. Illustration of the different steps used to treat the mass loss data obtained from tests performed with 3-stone and rocket stoves appliances.
In order to get a better understanding of the system, the time resolved dilution factor calculated with help of equation (3) was further compared with the calculated flow of total produced gases, as illustrated in Figure 10.
0 300 600
0 20 40 60 80 100
Mass loss [g]
Time [min]
3 s tones
Mass loss raw
a)
0 1000 2000
0 50 100
Mass loss [g]
Time[min]
3 s tones
Mass loss treated
b)
-1 -0.5 0 0.5 1
0 1000 2000
0 50 100
Mass loss rate [∆g/∆s]
Mass loss [g]
Time [min]
3 s tones
Mass loss treated Mass loss rate treated
c)
-1 0 1
0 1000 2000
0 50 100
Mass loss rate [∆g/∆s]
Mass loss [g]
Time [min]
3Stones
Mass loss treated Mass loss Mass loss rate
d)
Figure 10. Typical trend of total produced gasescalc and dilution factor over a run for data obtained from a test carried out with the gasifier stove.
Further, the calculated CO2 emissions were compared to the measured CO2 emissions to see whether the calculated emissions were consistent with the measured, as illustrated in Figure 11.
In the cases of the 3-stone fire and the rocket stove, the fuel is constantly being added as it is being consumed, therefore the fuel composition was assumed to be constant over the whole test.
Figure 11. Calculated CO2 emissions versus measured CO2 emissions for data from the 3-stone fire and the rocket stove appliances.
In contrast to the other tested appliances, the gasifier stove is operated as a batch-wise process with a batch of pellets that lasts for the whole test. This implies that the devolatilization and char combustion phases are much more defined during the conversion process, than for the other studied appliances (3-stone fire and rocket stove). Nevertheless, since the batch, i.e. 500 grams, of pellets is in the cylindrical stove compartment, there is a first phase in which the flames are not in contact with the bed of pellets due to the devolatilization process. And a second phase in which a glowing front moves down as the char is being combusted.
0 100 200 300 400
0 0.0005 0.001 0.0015
0 10 20 30 40 50 60
Dil factHood
Tot prod gases [Nm3/s]
Time [min] Tot al produced gases calc Dilution fact or
0 5 000 10 000
0 20 40 60
CO2[ppm]
Time [min]
3-stone fire
Cal cul ated emissi ons Measured emissions
0 5 000 10 000
0 20 40 60
CO2[ppm]
Time [min]
Rocket stove
Cal cul ated emissi ons Measured emissions
In Figure 12, it can be seen how the CO2 emissions are influenced by the change in fuel composition during the combustion process, i.e. going from devolatilization to char combustion. In this study, these two different fuel compositions were used with aim to obtain more accurate CO2 emissions, based on an assumed fuel composition during the respective phase.
Figure 12. Comparison between the measured CO2 concentrations and the calculated using the assumed fuel compositions (devolatilization and the char combustion) for the whole combustion cycle (left), and a comparison between measured CO2
composition and combined calculated CO2 concentrations using the adjusted fuel compositions for the respective phases.
By knowing the dilution factor over the hood and the dilution factor of the ejector, measured particle concentration after the ejector [Part]Ae was converted to the particle concentration at the flue gas channel [Part]Fg and finally to the particle concentration at the produced gases [Part]Pg calc (fig 13).
0 5 000 10 000 15 000
0 20 40 60
CO2[ppm]
Time [min]
Gasifier stove
Measured Char combustion composition
Devolatilizacion compos ition
0 4 000 8 000
0 20 40 60
CO2[ppm]
Time [min]
Gasifier stove
Combined calc ulated Measured
Devolatilization composition
Char combustion composition
Figure 13. Typical particle number concentrations (by the SMPS) for one test with the gasifier stove, illustrating both the measured concentrations in the diluted gases and the calculated concentrations in the produced (primary) combustion gases.
Once the emissions were converted to particle concentration at the emitted gases, the evaluation of the particle emissions were divided into salt and soot fractions. Additionally, the relation between particle number and particle mass was estimated by calculating the cumulative particle mass over time (fig 14).
Figure 14. Typical time-resolved particle number concentration for the estimated salt and soot mode (left), and time- resolved total mass and number concentrations results (right).
0 60 000 120 000
0 20 40 60
Part conc [Part/Ncm³]
Time [min]
After ejector
0 1 000 000 2 000 000
0 20 40 60
Part conc [Part/Ncm³]
Time [min]
At flue gas channel
0 70 000 000 140 000 000
0 20 40 60
Part conc [Part/Ncm³]
Time [min]
At calc produced gases
0 60 000 000 120 000 000
0 20 40 60
Part num conc [Part/Ncm3]
Time [min]
[Part]Pg calc
SALT SOOT
0 0.2 0.4
0 60 000 000 120 000 000
0 20 40 60
Part mass conc [µg/Ncm3]
Part num conc [Part/Ncm3]
Time [min]
[Part]Pg calc
[N/cm3 ] [µg/cm3 ]
Finally, the time resolved relation between the particle number and particle mass was plotted.
A typical particle mass to number concentration plot corresponding to a test performed with the gasifier stove is shown in Figure 15.
Figure 15. Illustration of the time resolved relation between particle number concentration and particle size in a typical combustion test.
3.2 Results from the comparison of three cookstove appliances
Once the evaluation procedure of the set-up and measurement systems was completed, a comparison of three cookstove appliances was performed. As mentioned, the used appliances were; a 3-stone fire arrangement, a rocket stove and a gasifier stove.
3.2.1 Fuel consumption and boiling and simmering times
Fuel consumption and combustion times for the boiling and simmering phases were recorded for all nine runs, including three replicates for each of the three tested appliances.
The highest fuel consumption for both the boiling and simmering phases was registered for the 3-stone fire, followed by the rocket stove and last the gasifier stove. The largest variation in fuel consumption between the experiments was registered for the 3-stone fire as well, closely followed by the one for the rocket stove, while the gasifier stove had the lowest variation in fuel consumption (Fig. 16).
The “fastest” appliance to complete the WBT, i.e. the cookstove that boiled 5 litres of water in shortest time was the gasifier stove, followed by the 3-stone fire and finally the rocket stove.
The results of fuel consumption and the times to complete the boiling and the simmering phases are shown in Figure 16, and the full data given in Table 4 and 5 in Appendix 1.
3.2.2 Mass loss over time
Mass loss and mass loss rate were smoother for the gasifier stove than for the other two tested appliances, as seen in Figure 17.
The observed pattern in mass loss was also very different in the case of the gasifier stove compared to the other two tested appliances. A more drastic mass loss was observed in the first part of the test for the gasifier stove which illustrates a clear devolatilization phase of the biomass in the first part of the conversion process. In contrast, in the case of the 3-stone fire and the rocket stove, the devolatilization and char combustion phases took place simultaneously, which gives a more constant mass loss rate.
-1 -0.5 0 0.5 1
0 400 800 1200 1600
0 20 40 60 80
[g/s]
Mass [g]
Time [min]
3-stone fire
Mass loss Mass loss rate
-1 -0.5 0 0.5 1
0 250 500 750 1000
0 20 40 60 80
[g/s]
Mass [g]
Time [min]
Rocket stove
Mass loss Mass loss rate
Figure 16. Average ±SD (n=3) fuel consumption for the three tested appliances for boiling and simmering phases (left).
Average ±SD (n=3) time to complete boiling and simmering phases for the three tested appliances (right).
0 500 1000 1500
3-stone fire n=3
Rocket stove n=3
Gasifier stove n=3
Weight [g]
Fuel consumption
Fuel til l boiling Fuel simmering
0 20 40 60 80 100
3-stone fire n=3
Rocket stove n=3
Gasifier stove n=3
Time [min]
Time Time to boil Simmering time
Figure 17. Typical mass loss and mass loss rate for the three tested appliances.
3.2.3 Water temperature and CO2 emissions
The overall CO2 concentrations were highest for the 3-stone fire, followed by the rocket stove and the gasifier stove, which had the lowest emissions of all (Fig. 18). In the case of the 3- stone fire and the rocket stove, it was observed that the CO2 emissions decreased when the simmering phase started. In contrast, the CO2 emissions for the gasifier stove did not decrease after the boiling phase ended.
The pattern of CO2 emissions over time was smoother for the gasifier stove than for the 3- stone fire and the rocket stove. In addition, in the case of the gasifier stove there was a clear peak in CO2 emissions, which took place approximately in the middle of the boiling phase.
-1 -0.5 0 0.5 1
0 150 300 450
0 20 40 60
[g/s]
Mass [g]
Time [min]
Gasifier stove
Mass loss Mass loss r ate
0 20 40 60 80 100
0 4000 8000 12000
0 20 40 60
TEMP water [°C]
CO2 [ppm]
Time [min]
3-stone fire
0 20 40 60 80 100
0 4000 8000 12000
0 20 40 60 80
TEMP water [°C]
CO2[ppm]
Time [min]
Rocket stove
Figure 18. Typical CO2 emissions in the flue gas channel versus water temperature of the three tested appliances.
3.2.4 Particle emissions based on offline measurements
The results from the total suspended particulate matter (TSP) collected with the total dust filter setup are presented based on five different parameters as illustrated in Figure 19.
Regardless of the unit in which the emitted particles are presented, there was a clear trend in the amount of particle emissions from the three tested appliances. The 3-stone fire was the appliance with the highest emissions, followed by the rocket stove and the gasifier stove, respectively.
The standard deviation of the emission values registered for the tested appliances was largest for the 3-stone fire, followed by those for the rocket stove and the gasifier stove.
The raw values used in the graphs presented in Figure 19 are compiled in Table 6 in Appendix 1.
0 20 40 60 80 100
0 4000 8000 12000
0 20 40 60
TEMP Water [°C]
CO2[ppm]
Time [min]
Gasifier stove
0 5000 10000 15000
3-st one fire n=3
Rocket st ove n=3
Gasifier stove
mg/completed test emitted n=3
Emission per completed test
0 40 80 120 160 200
3-st one fire n=3
Rocket stove n=3
Gasifier stove n=3
mgemitted/min
Emissions on time basis
Figure 19. Particle emissions (PMtot)for the three tested cookstove appliances given as different emission factors on a test-, time-, fuel- and energy basis.
3.2.5 Dilution factor calculation
The calculated vs. measured CO2 concentrations as well as the calculated dilution factor over the hood for the three appliances are shown in Figure 20.
As mentioned earlier, in the case of the gasifier stove both the calculated CO2 emissions and the dilution factor are based on two different assumed fuel compositions.
The best match between calculated and measured CO2 emissions was for the gasifier stove.
Nevertheless, the calculated CO2 emissions clearly followed the same trend as the measured CO2 emissions also for the 3-stone fire and the rocket stove.
It was obvious that the dilution factor was much lower for the 3-stone fire and for the rocket stove than for the gasifier stove. In addition, the dilution factor, for the gasifier stove, increased considerably in the second half of the test.
0 2 4 6 8 10
3-st one fire n=3
Rocket stove n=3
Gasifier stove n=3
mgemitted/gfuel
Emissions on fuel basis
0 200 400
3-st one fire n=3
Rocket st ove n=3
Gasifier stove n=3
mgemitted/MJfuel
Emissions on used energy basis
0 1000 2000 3000
3-st one fire n=3
Rocket st ove n=3
Gasifier stove n=3
mgemitted/MJdel
Emission based on delivered energy
Figure 20. Typical measured CO2 emissions versus calculated CO2 emissions and dilution factors over the hood for the three tested appliances. In the case of the gasifier stove, two fuel compositions were assumed for the calculations of the CO2
emissions. The vertical purple dashed line indicates the point at which the fuel composition was assumed to change.
3.2.6 Particle emissions based on transient measurements
The particle number emissions (normalized) in the produced gases was found to be much higher for the 3-stone fire and the rocket stove than for the gasifier stove, as seen in Figure 21.
0 200 400
0 5000 10000
0 20 40 60
Dil. fact.
CO2[ppm]
Time [min]
3-stone fire
Cal cul ated Measured Dilution fact or
0 200 400
0 5000 10000
0 20 40 60
Dil. Fact.
CO2[ppm]
Time [min]
Rocket stove
Calculated Measured Dilution fact or
0 200 400
0 5000 10000
0 20 40 60
Dil. fact.
CO2[ppm]
Time [min]
Gasifier stove
Calcul ated Measured Dilution fact or
0 400 000 000 800 000 000
0 50 100
Part conc [Part/Ncm³]
Time [min]
[Part]Pg calc
3-stone fire
0 400 000 000 800 000 000
0 50 100
Part conc [Part/Ncm³]
Time [min]
[Part]Pg calc
Rocket stove
Figure 21. Typical normalized particle concentration at the produced gases for the three studied appliances.
In order to get an even better understanding on the particle emission properties of the tested cookstoves and fuels, further analyses of the SMPS data were carried out. In the following, an assessment of different particle parameters is therefore performed, i.e.; salt and soot mode particles; particle number compared to particle mass; particle mass as function of CO2 concentrations and finally particle number in relation to particle size over time.
0 400 000 000 800 000 000
0 20 40 60
Part conc [Part/Ncm³]
Time [min]
[Part]Pg calc
Gasifier stove