Effect of wood moisture content on the performance of wood burning cook stoves

ABSTRACT Rural Ethiopian communities rely mostly on biomass fuels for subsistence and productive purposes. With an increasing population and dwindling forest resources, wood fuel consumption has exceeded its supply. Besides, people do not know ways of maximising biomass efficiency. The study involves the burning efficiencies of woods with varied moisture contents. Eucalyptus wood of five different moisture contents was used in five identical wood-burning stoves purposefully constructed for this purpose. Tests were conducted over 5 days, thereby giving five replications. In the tests, food temperature, stove body temperature, and stove smoke outlet temperature measurements were conducted along with ambient air temperatures. Wood with 10% moisture content performs better during combustion with respect to low and high moisture content wood. High moisture content delayed the cooking onset and low moisture decreased the duration of effective cooking. The 50% moisture content or greater with respect to oven-dried wood failed to cook. Moisture content of around 30% delayed the time to reach cooking temperature by two and hence elongated the cooking time. There is also the extended smoke time as observed from the smoke outlet temperature.


Introduction
About 16% of global energy consumption is coming from renewable energy sources, out of which biomass has the largest share of 10%.Fuel wood is used for cooking and heating, especially in developing countries of Africa, Asia and Latin America (Kumar, Prasad, and Mishra 2013;Lu et al. 2008).On a global scale, nearly three billion of the world's population has no access to modern energy alternative and still depending on biomass.In Ethiopia, 29.6% of the population live below the poverty line (Iscaro 2014) and most of the rural population entirely depends on biomass for its energy needs (Amare, Endeblhatu, and Muhabaw 2015).The categories of the biomass energy sources are as follows (L'Orange, DeFoort, and Willson 2012): woody biomass (78%), dung (8%), crop residue (7%) and the remaining is from petroleum (5%) (Eshete, Sonder, and Ter Heegde 2006).Wood as a fuel has several environmental compensations compared with fossil fuels (Kumar et al. 2010) and which repeatedly replenished and can be considered renewable.Hence, if managed properly, wood is a sustainable and dependable supply.This requires proper forest administration practice to ensure that growing conditions are not degraded throughout biomass production.However, burning wood has environmental impact since it brings greenhouse gases by producing gases such as carbon dioxide (CO 2 ) and oxides of nitrogen.In order to alleviate this problem, the wood that is used as a fuel must be minimised without compromising the energy needs of the people.
Energy generated from wood fuel can be increased by improving the wood quality.Moisture content in the wood is one of the most important parameters influencing the energy production from wood biomass (TUCHO 2019;Routa, Kellomäki, and Peltola 2012;Routa et al. 2015;Korenaga, Liu, and Huang 2001).Energy is required to dry wood to an acceptable level.This means, from the energy produced by wood, some is lost in drying the wood, particularly if the wood has a high moisture content (Lukáč et al. 2014).The increased moisture content raises the demand on the energy consumption, which is desirable to produce the heat required to remove water from the wood in the form of evaporation.This adversely influences the wood quality and input amount of wood.To put it differently, high wood moisture content increases wood consumption.Wood moisture content determines the net calorific value available in wood.The two are inversely related.When there is high moisture content in wood, the energy produced from the wood is required to do two things.On the one hand, the energy is required to heat and vaporise the free water in the wood.The heat energy that is used to remove the moisture content dissipates into the surrounding atmosphere (Jekayinfa and Scholz 2009).The wood must also reach the combustion temperature needed for combustion to occur (Miljković 2015).External factors such as ambient temperature and relative humidity can also influence the temperature a sample of wood will eventually reach.
In Ethiopia, rural communities have less access to electrical energy than the urban population enjoys (Sundblad 2014).
This community must rely almost entirely on biomass fuels (Westerlund, Hermansson, and Fagerström 2012;Anderson and Westerlund 2014).In short term, the use of woods as a source of energy for cooking and heating is going to continue until alternative energy sources are obtained.People use woods of different moisture contents for their energy needs without taking into consideration which moisture content provides high burning efficiencies.Finding the right moisture content of wood is one-way to partially reduce the quantity of wood used thereby curbing the demand.Determining the optimum moisture content of wood is essential to maximise calorific value and to reduce the quantity of firewood required for a specific cooking.Of course, the effect of wood moisture on the performance of cookstoves is already known.However, we have significantly obtained the optimum moisture content of wood that results in maximum efficiency such that it ultimately economises the quantity of wood used.By doing so, users benefit economically, slightly less trees are used and the environment benefits.The effect of wood moisture on the performance of cookstoves was done on different stoves, some time on moisture of different wood type and only studied on gas energy emission reported by different groups, but in this work, the burning efficiency of five different moisture contents of the Eucalyptus wood with constructive stove and one conventional stove (local stove) were compared.The issue of this study is to assess the burning efficiencies of woods of different moisture contents.Thus, we expect the test efficiency of wood burning in relation to its moisture content (time and temperature).

Materials
The materials used in this study were as follows: metal sheets, metal tubes for the air outlet, iron bars to construct the grates and the stove handles, welding machine, welding rods, a digital kitchen thermometer (AMB.TEMP, No: 10L03250, Japan) laboratory oven (SONYO OMT Pf120, 200, UK) were purchased, beam balance, mercury thermometers, metre stick, mobile clock, matching pots of diameter 20 cm and height 20 cm, wood samples of Eucalyptus with the dimension of 15 cm × 3 cm × 4.5 cm of different moisture contents (0, 5.2, 10, 14, 34, 54%), axe, a lighter, rice grains and water.

Construction of the stoves
In this study, five identical wood burning stoves were designed and constructed to compare the treatments.The stove bodies were made from sheets of iron and grates, from 10 mm iron bars.The scheme of the stove is made from locally available material as shown in Figure 1(a) , while stoves used in the experiment including conventional stoves are shown in Figure 1(b).

Preparation of wood samples
The Eucalyptus woods for all samples were cut into 15 cm × 3 cm × 4.5 cm dimensions by using axes.Samples of each treatment were oven-dried at 105°C for the duration specified, and they were identified from treatment one to five treatments (S 1 -S 5 ) as indicated in Table 1.The remaining pairs for water content determination were for all oven-dried for 24 h.

Experimental procedures
A treatment number was identified for each stove and the corresponding treatment (wood sample) was arranged horizontally on the wood rack of each stove.Mercury thermometers were inserted into the slots prepared to hold the thermometers on the body and smoke outlets on each stove except the conventional stove.A cup of rice (242 g) was added to each pot along with one litre of water.The initial water and rice mix temperature was recorded using a kitchen thermometer inserted through a hole prepared on the lid of each pot.The experiments were carried out outdoor with all the air inlets of the stoves facing the same direction.The direction selected was the direction of the prevailing wind.The arrangement of the stoves was randomly changed after every test.Another mercury thermometer was placed in open air away from the stoves to record the initial and final ambient air temperatures during the cooking.A small amount of kerosene was used to start the fire in each stove nearly at the same time.Each stove was ignited and after making sure the fire has started fairly well, the respective pots were placed on the stoves and the time was recorded at the same time.Thereafter, the three (rice, stove body and smoke outlet) temperatures and the time were recorded every 7 min for the same treatment.The kitchen thermometer was used turn by turn to measure the temperature of rice in the pot.This was necessary since there was only one kitchen thermometer.The measurement continued for 85-90 min (until the temperature of the food in the pot reduced roughly close to ambient temperature).The ambient temperature measurements were taken twice, the first at the beginning and the other at the end of the cooking process.The experiment was performed for 5 days, thereby giving five replications.
The moisture contents of the firewood were determined using Equation (1) below, where MC db is the moisture content on dry basis, M wet is the mass of wet wood, M d the mass of the dry wood and M w is the mass of water.Operationally, the moisture content of a given piece of wood can be calculated by using Equation 1 (Zheng et al. 2013).
Comparisons among treatments were made instead of food, stove body and smoke outlet temperatures.First plots of the food temperatures were made for all the treatments together using Origin software.From the plots, the three parameters, namely, the time of onset of cooking (t i = time at which 75°C was reached), maximum temperature (T mx ) reached and effective cooking time (Δt = duration for which the temperature stayed equal to or above 75°C) were determined.The three parameters were then tested using one-way ANOVA and whenever the ANOVA showed significant differences, pair comparisons were made.Plots of stove body temperatures and smoke outlet temperatures were also done for all the treatments together in the same manner as was done for food temperature.From the plots, the treatment that registered the longest time during which 75°C was recorded was selected and this time was taken as the upper limit of integration.Using zero as the lower limit of integration, areas below each curve were determined for each treatment using Matlab.The relative areas were then compared to estimate relative heat losses through the body and the smoke outlets of the stoves.

Determination of wood moisture contents
It is noted that the moisture content of wood samples that were dried in the oven for different times (1, 2, 4, 8 and 24 h) were 54%, 34%, 10%, 5.2% and 0%, respectively (see in Table 2).The percentage (relative moisture content) was taken with reference to the wood that was oven-dried for 24 h.The other sample was air-dried and had a moisture content of 14%.

Food temperature comparisons among treatments
As far as food temperature is concerned, three things were considered.These are as follows: 1) the onset of effective cooking (the time at which the food temperature reached 75°C), it is denoted as t i and indicates how fast (early) the stove starts to cook.The less value of t i is always preferable. 2) maximum food temperature (T max ) reached.It is the indicative of the intensity of cooking and it must be greater than the critical temperature (75°C) for effective cooking.3) effective cooking time (represented as Δt), which is the time interval between the onset of cooking and the time at which the food temperature drops back to 75°C (348 K) mark, on its way to cooling.It is the difference t i and final time (t f ) and Δt indicates the time during which the food is actually cooking.Three of the parameter (t i, T max and Δt) each were compared for all the treatments using graphical and statistical methods.The graphical method was done by plotting food temperature versus cooking time of all the treatments together.This was done after the data were temperature corrected and time synchronised, as explained in section 2.4.Plot of treatment-averaged temperature of cooked rice versus cooking time is shown in Figure 2.
The left-hand side before the inflection point marks, the heat generation phase.During this phase, the temperature goes on increasing until a certain maximum temperature is reached: the time up to the maximum temperature and the time range when the fuel is actually generating hot gases that are responsible for the cooking process.The maximum emission of these gases is during the maximum temperature.After the maximum temperature, there is a cooling or heat dissipation or heat extraction phase during which the actual heat supply is reduced, and the temperature decreases as more heat escapes into the surrounding.On the temperature axis (see in Figure 2), the 75°C (dashed horizontal) line is selected as the basis to estimate the three parameters.The first intersection points before reaching the maximum temperature line indicate the t i values of the treatments and the maximum temperature (T mx ) is at the inflection points of all the curves.The second intersection points of the curves with the 75°C line (after the inflection) mark the final time of effective cooking represented as t f , which makes effective cooking time, Δt = t f -t i .
As revealed from the above Table 3, cooking with wood of moisture content greater than 50% could not yield an effective cooking temperature of 75°C in this test.There was no sufficient heat to dry the wood and to heat the food to cooking temperature at the same time.This treatment was not considered for comparison since the three comparison parameters are absent in this treatment.
As observed from the remaining treatments, the increase in moisture delayed the ignition time (for instance, increased t i value to 18 min for S 2 or 34% moisture content).As wood moisture increased, the maximum temperatures attained were also reduced.Because the cooling effect reduces the combustion and cools down the gases produced, high moisture reduces the cooking performance and extended good secondary combustion.High levels of moisture cause lower heating value of biofuels and incomplete combustion, which is responsible for emissions of gases such as carbon monoxide, carbonaceous particulate and unburned hydrocarbons (Possell and Bell 2012) thereby affecting the environment.Incomplete combustion also decreases carbon dioxide production and combustion efficiency.According to Palacka et al. (2017), high moisture content results in lower combustion efficiency (results a higher production of CO).Accordingly, the increase in the moisture and the presence of CO produce H 2 which produces more CH 4 by direct hydrogenation.
Completely dry wood seems to burn faster and decreases the effective cooking time (S 5 or 0% moisture content exhibited the lowest effective cooking time of 36 min0 (as seen in Table 3).The result is in agreement with what is observed by other researchers as seen in Grainger (1996) where the effect of moisture on combustion shows a fall in efficiency with increase in moisture content of wood.However, complete dryness does not seem to affect the maximum temperature reached.Slight moisture in wood seems to elongate the cooking time since the time to reach cooking temperature is initially delayed.Even though the highest efficiency was achieved at a moisture content of about 5% in the previous work (Deforestation 2010), the optimum relative wood moisture content in our experiment was achieved around 10% which revealed the benefit of relatively fast onset of cooking and long effective cooking time of 8 and 49 min, respectively (as seen in Table 4), and is in agreement with work of Deng (2006) where for moisture content of 10%−20%, the fuel combustion reaction was found to be intense, full and complete.As observed by Zheng et al. (2013), with slight moisture in the wood combustion reaction becomes complete and the combustion emissions of CO 2 increase significantly.Febriansyah et al. (2014) also discovered high combustion temperature when the fuel has a relatively low moisture content.It is important to see how t i and Δt are influenced by wood moisture content as seen in Figure 3.
The full data plot of Figure 3 has some distortion after moisture content of 15% because of the nature of the curve fitting.Therefore, it is fair to take 10% as the optimum moisture per cent.At this moisture content, t i is fairly low, but Δt is maximum.Since these are the required combinations, the 10% moisture can be considered as optimum moisture content.In the Statistical method of t i comparisons among the five treatments, the three parameters were determined from the plots of food temperature versus cooking time of each replication.The data used in the plot were temperature corrected, but there was no need for time synchronisation.The three parameters were determined from each plot and tabulated for all the treatments.Such data can be used for statistical comparison.
The ANOVA table indicates significant differences among the treatments at α or p = 0.05 (see in Table 4).However, in order to know which pairs are significantly different, it is  The pair comparison shows wood with moisture content of 34% has a delayed time on the onset of cooking, and it is significantly different from all the rest (see Table 5).The initiation of onset of cooking is delayed since the wood has to first be dried to burning temperature.Once the burning starts, the wood, however, continues to burn for a longer time.
For all the others, the time it takes to reach 75°C (onset of cooking) is not significantly different at p = 0.05.It means as far as the time of initiation of cooking is concerned there are no differences among woods with moisture contents from 0% to 14%.All of them exhibited initiation times of less than 9 min, which is half the time recorded for the wood with 34% moisture content (seen in Table 6).
The maximum temperature is the maximum temperature reached through the cooking process regardless of the occurred one-way ANOVA of the five treatments (wood moisture contents, see shown in Table 7).
As seen from Table 7 above, no significant differences were observed among all the treatments considered.The stove using the five the wood samples (0-34% moisture content) reached the averaged maximum temperatures, which were between 94.2°C and 98.6°C regardless of their moisture contents.Variations among treatments were similar to variations within treatments; therefore, one can conclude that moisture content of woods does not affect the maximum temperature of the food cooked at least for the moisture content between 0% and 34%.Thus, the mean values of the treatments (seen from Table 8) also reveal the same fact.
In this case, the cooking time considered is the time difference (Δt) between the initial time 75°C is reached and the final time it dropped below 75°C and there are significant differences among the different moisture contents as far as effective cooking times are concerned (see in Table 9).
Pair comparisons of Δt between the different treatments are different, so it is necessary to make pair comparisons (see in S 1 ).From the pair comparisons, wood with moisture content of 34% (Δt avg = 49.6 min) showed significant cooking time difference from those between 0% and 5.2% moisture contents with Δt avg = 40 and 32.6 min, respectively, but not with that of 10% moisture content (Δt avg = 44.2min).Hence, as far as effective cooking time is concerned the 10% and the 34% moisture contents are not significantly different at p = 0.05 level (see in Table 10) In general, more moisture delays the onset of effective cooking but prolongs effective cooking time.Since cooks are interested both in fast initiation of onset of cooking and long effective cooking time, the 10% wood moisture gives a better option in both cases and it enables to start cooking earlier and cooks for a long time as well.The statistical result is in good agreement with the graphical result in terms of t i (roughly 8 min in both cases) but not so with Δt, which showed differences of about 5 min.

Wood moisture content and stove body temperature
A substantial quantity of the heat produced in a stove is partially lost through the body of the stove or through the cracks and openings in the body of the stove.Because it is imperative to measure the body temperature of each stove that uses wood samples of different moisture contents.The stove body temperature takes place during the cooking process for the stove using wood samples of different moisture contents (see in Figure 4).
One of the ways to compare heat energy emissions from the stoves is by looking at the peak temperatures reached, but the more appropriate way is by calculating the areas under the  curves of Figure 4.As far as the limits of integrations are concerned, the lower limit is taken as zero (assuming all of the stoves were initially at ambient temperature, which is close to 20°C most of the time).The upper limit is estimated from the curve that exhibits the maximum time of intersection of the 75°C line and to determine this time, it is necessary to plot t f , T b and T s of the second treatment against time and where the horizontal line of the 75°C cuts the T b versus time plot.The horizontal axis that cuts through the 75°C temperature cuts the Tbf versus cooking time graph at 53 min (shown by the dotted vertical line, see in Figure 4).Based on this estimation, the higher limit of integration is 53 min.For area determination, it suffices to take the upper limit of 53 min since above this time; the actual effective cooking is not going to take place.That means anything in excess of this time limit is not going to contribute to the cooking process.Using the equations and the limits of integration considered, the value of the integration obtained by Matlab for each treatment is summarised in Table 11.
The maximum area was exhibited by treatment S5 (0% wood moisture content).In other words, this treatment has the highest relative stove body heat loss.Also, S5 has the highest body temperature since the wood did not have moisture that has taken some of the heat energy to vaporise the water from the wood (see in Table 11 and Figure 5).In the absence of latent heat energy, all of the heat energy has gone to the food, body of the stove and to the smoke outlet.Similarly, all the S parameters are compared with this treatment, and similar arguments can be made for S4, S3 and S2.The conventional stove could not be considered with the other stoves since heat energy from this stove could easily escape to the ambient air.Thus, the reduction in relative heat loss from the body of the stove is correlated to wood moisture content as shown in Figure 6.
With higher wood moisture, there is a lower amount of heat energy that escapes through the body of the stove (see in Figure 6).The 15% (shown by the middle dotted vertical line) separates the curve into two parts: with upward curvature below 15% and downward curvature above the 15%.The relative heat loss below 15% is less than 7% while from 15% to 50% the drop could reach as high as 50%.Such a drop is not due to the efficiency of the stove (since all the stoves are identical) but it is due to the inefficiency of the burning wood.This implies that much of the heat escapes the stove in the form of latent heat (vaporises the moisture in the wood) rather than translating into sensible heat.

Smoke outlet temperature and wood moisture content
The tube that is attached to the stove for smoke removal does not help the environment, but at least, it partially protects the cook from direct contact with the smoke (the pollutant gases).This is because there can be volatile and harmful compounds in the smoke that could have deleterious effect on the cook.Since the smoke outlet is an additional surface area exposed to  the surrounding air, it would have its own contribution to transmit thermal energy to the surrounding particularly if it is made from a heat conducting material (metal).
As far as the limits of integrations are concerned, the initial limit is taken to be zero.Similar to the body temperature, the time after which the food temperature has returned to 75°C is not of interest as far as cooking is concerned and therefore the time limit that corresponds to the curve that depicts the highest value of time at the intersection of the curve and the 75°C was considered as the final time.Based on this estimation, the upper limit of integration is 45 min (the dashed vertical line in Figure 4 and dotted vertical line in Figure 7).Using the equations and the limits of integration considered, the values of the integration obtained by Matlab for each treatment are summarised in Table 5.
The highest calculated area from the smoke outlet temperature versus time curve was exhibited by S3 (10% moisture, see Table 12).At this moisture, there is intense heat produced as well as increased products of combustion such as carbon dioxide and oxides of nitrogen as also found by Zheng et al. (2013).Per cent change in relative area and per cent area reduction calculations were made by taking the S3 area as a reference.S1 showed the highest reduction but that does not matter since this stove did not perform its duty (was unable to reach cooking temperature because of the high moisture content of the wood).The next highest reduction was by S 2 (34% moisture).Because of the relatively high moisture content in the wood, this stove showed both high t i and high t f .In other words, unlike the other treatments, there is a substantial area of this treatment that has not been accounted for after t f .
Both S 4 (5% moisture) and S 5 (0% moisture) showed slightly low moisture per cent reduction.With less moisture content, there is less heat emission through the smoke outlet since wood with low moisture has less smoke.The low heat emission through the smoke outlet at both high and very low moisture contents perhaps indicates a low intensity of heat that results in lower gas emissions.At moderate moisture (such as 10%) there is intense heat and CO 2 production also increases.With higher moisture, heat energy produced is partially spent on vaporising water from the wood, which means, the intensity of combustion reduces.In the case of low moisture and absolutely dry wood, duration of fuel combustion and emission of various gases are the lowest (Zheng et al. 2013).Since heat energy is carried to the smoke outlet by gases such as carbon dioxide, water vapour, carbon monoxide, and so on, low quantities of these gases mean less heat energy release through the smoke outlet.
The maximum smoke outlet versus cooking time area could have occurred around 20% moisture of instead of 10% (Figure 8).If this moisture content were in the treatment, the loss could have been even higher.This does not seem to reflect intense heat but rather the escape of large quantity of gases through the smoke outlet at this moisture content.
Woods with both high and low moisture contents did not do well in the combustion process compared to the wood with   moderate heat content (around 10%).Cooking with wood with a moisture content greater than 50% could not yield cooking temperature (T = 75°C) in this particular case.Slight moisture in the wood seems to elongate the cooking time since the time to reach cooking temperature initially is delayed.High moisture content delayed the onset of cooking and low moisture content decreased the overall duration of effective cooking.
With wood of more moisture, the maximum temperature reached is reduced.This is because the cooling effect slows and cools down combustion of the gases produced and similarly high moisture reduces efficiency and sustains good secondary combustion.High levels of moisture cause lower heating values of biofuels and incomplete combustion, which in turn affects the environment.Wood with moisture content of 5.2% resulted in more heat loss through the stove body, while the one with 20% moisture (from Figure 9) exhibited the highest heat loss through the smoke outlet.In general, the onset of cooking does not seem to change much up to moisture content of 20%.Completely dry wood seems to burn earlier and decreases the cooking time range.The optimum moisture in this experiment is around 10% since it has the benefit of fast onset of cooking and long effective cooking time.

Conclusions
The efficiency of wood burning in relation to its moisture content of Eucalyptus wood of five different moisture contents was used in five identical wood burning stoves.Tests were conducted over 5 days, thereby giving five replications.Woods with both high and low moisture contents did not do well in the combustion process compared to the wood with moderate water content of 10%.High moisture content delayed the onset of cooking and low moisture content decreased the overall duration of cooking.Wood with moisture content of 5.2% resulted in more heat loss through the stove body, while the one with 10% moisture exhibited the highest heat loss through the smoke outlet.
Wood with moisture content of 50% or greater with respect to oven-dried wood failed to cook (reach cooking temperature).No attempt is made to cook food if it is known beforehand that the wood moisture is in excess of this value.Increased moisture content in wood (around 30%) delays the time to reach cooking temperature by about two times and hence elongates the cooking time and also extended smoke time as observed from smoke outlet temperature.As far as maximum cooking temperature is concerned, statistically there is no significant difference between woods with moisture contents 0% and 34%.
There is no need to worry about the maximum temperature reached in so long as the moisture is within this range.
It is sufficient to air-dry any freshly cut wood for 3 days when the ambient temperature is around 18°C or more.However, since excess loss of moisture has a counter effect on the burning efficiency of wood, it is advisable to keep wood which is already air-dried under shade to prevent excessive drying.For wood that is to be used after a long time it is recommended not to split it to reduce moisture loss.For already excessively dried wood, it is better to test if socking the wood to bring to the required moisture content could bring a satisfactory result.

Figure 1 .
Figure 1.(a) Schematic diagram of proposed stove (b) Represented six different stoves used in the experiment.

Figure 2 .
Figure 2. Effect of time on the food temperature for all samples.

Figure 4 .
Figure 4.The effect of cooking time on stove body temperature of all samples..

Figure 5 .
Figure 5.Effect of cooking time on food, stove body and smoke outlet temperatures for sample S 2 .Figure6.Role of percent wood moisture on reduction in relative heat loss.

Figure 7 .
Figure 7. Effect of cooking time on stove smoke outlet temperature for different samples.

Figure 8 .
Figure 8.Effect of percent moisture on relative smoke outlet heat loss.

Table 1 .
The specifications of the samples of wood used in the experiment.

Table 2 .
Moisture contents of air-dried wood and woods dried in oven for different hours.

Table 3 .
Cooking times and maximum temperatures reached by each treatment.
important to see the pair comparison and Table5pair comparison.

Table 4 .
One-way ANOVA significance test of t i among treatments (Comparisons t i among the woods of moisture contents of 0-34%).
Figure 3. Role of wood moisture content on the time of onset cooking and effective cooking.

Table 5 .
Pair comparisons of t i between the different treatments.

Table 6 .
Mean and standard deviations of time of onset of cooking (t i ).

Table 7 .
One-way ANOVA table to compare T mx of different treatments.

Table 8 .
Averages and standard deviations of maximum temperatures (T mx ) reached.

Table 9 .
One-way ANOVA to test δt differences among the treatments.

Table 10 .
Averages and standard deviations of effective cooking times (δt).

Table 11 .
Areas calculated from the curves to compare relative heat losses through stove bodies.

Table 12 .
Areas calculated from the curves to compare stove smoke outlet temperatures.
Area* = the area is given in arbitrary units; Δ W.r.t A mx **= difference in relative area with respect to the maximum area calculated (S 3 ); % reduction***= percentage of area reduction compared to the maximum area..