[Stoves] Tom Miles' Highlights from ETHOS 2007

[Crispin Pemberton-Pigott]

Dear Kevin

You wrote:

"I would like to see the laboratory techniques for measuring performance and
emissions become closer to what is experienced in the field so that the
transition from lab to field is smoother. As Dean and others have emphasized
the lab tests are nothing more than development tools, but we need to make
them as close to the actual use as it is practical. Step by step maybe we
can reduce the differences between them."

On the serious side, I see that there is a problem aligning this 'it's only 
for the lab' idea with what happens in practise.  People who know little 
about stoves use the lab-based 'performance' tests for making decisions 
about what stove to build.  It is as simple as that.  While it is obvious 
that one should use a test of cooking in the homes using normal fuel (etc) 
when making such selections, these are 'early days' and there is precious 
little information available on stove performance anywhere.

On the very serious side, a stove test that does not represent something 
close to reality is of questionable value when developing a product to be 
used in a real place.  I give as a single example the surface area to volume 
ratio (A/V) of the fuel.  The gassing rate of the fuel is a very important 
factor during the burning as it determines the need for primary and 
secondary air.  Preheating air that is headed toward high A/V fuel requires 
that the air supply be choked, or chokeable, and secondary air provided as 
appropriate.  A good example is grass.  Burning grass 'openly' would result 
in a huge, brief plume of white smoke in most stoves because the air supply 
is completely wrong.  Conversely, putting one fat piece of hardwood in a 
Rocket Stove with yield nothing because it will go out.

On the really serious side, a stove test that purports to aid developers 
cannot report incorrectly or it will mislead the developer to think that an 
improvement is making things worse, and vise versa.  For example, it you are 
testing two stoves which are very similar in all respects save for a 
slightly different pot-skirt gap, you would expect a standard stove test to 
show whether or not the gap change produced a better or worse performance. 
Let us take the boiling phase of the test.  This test can be reported in 
tems of Percentage Heat Utilized (PHU) though now it is popular to use 
Specific Fuel Consumption (SFC).

A PHU test reports how much water was boiled out during the boiling phase, 
the time, and the fuel required.  To report this a combination of the 
initial water mass, final mass, and fuel burned is used with calculations 
for heat absorbed in the water, evaporation of the missing water and fuel 
burned, adjusted for moisture in the fuel (the heat required to heat and 
evaporate that moisture).  Given identical initial water and fuel masses, 
the principle numbers are the final mass and the fuel burned and the portion 
of heat that got into the pot.

An SFC test reports on how much wood was burned to boil a unit amount of 
water (i.e. per litre).  If the initial water mass is taken, the result is, 
say, 500 grams of wood burned divided by, say, 5 litres of water to give a 
rating of 100gm/litre.  If the final mass is 4.92 litres and is used for the 
calculation, it is given as x-many grams of wood burned divided by 4.92 
(litres) to give an SFC of 101.62.  Thus the same test, reporting on the 
same basis, gives different results depending on whether you use the initial 
or final mass of water in the pot.

It would not matter how this is reported unless you are trying to improve 
the stove.  Suppose a second stove is 1% more thermally efficient than the 
first.  The fires are identical, the wood and water masses too.  The result 
will be that the second stove evaporates more water than the first for the 
same amount of wood burned during the boiling phase.  If you are reporting 
on an SFC basis, they have the same result when taking the initial water 
mass, showing no performance increase: 100 gm/litre so a comparison of SFC 
based on initial mass is literally of no value to a designer.

Suppose the first stove evaporated nothing during the boil, and the second 
boiled 80 cc leaving 4.92 litres.  If you use the final water mass for an 
SFC calculation, the stove with the better performance shows a _higher_ fuel 
consumption, viz. 101.62 v.s. 100 gm/litre.  This gives the stove with the 
better thermal efficiency a worse performance.  Obviously this is going to 
mislead the designer.  The effect of calculating stove performance in this 
manner was shown in Bailis' talk at ETHOS.  He got a negative correlation 
between lab and field results, which is what one would expect if the manner 
of calculation were to rate the more efficient stove as having a higher 
specific fuel consumption.

Thus it is shown that the SFC method is not useful for determining 
improvements to stove performance or comparing the thermal performance of 
different stoves during the boiling phase.  If the initial water mass is 
used, there may be no difference shown in the test result unless one fire is 
smaller than the other and still manages to boil the water in exactly the 
same time - something very difficult to achieve in practise.  During the 
boiling phase, only a PHU test can be guaranteed to show an increase or 
decrease in performance.

However this is not the case with the simmering phase because the simmer has 
a clear definition: keep the water in a narrow range of temperature between 
the local boiling point and three degrees below that point.  The fuel used 
to do this might be given in SFC terms or PHU terms, and the initial or 
final water mass could be chosen, with quite different results.  This is not 
a calculation of work done, it is a calculation of how much fuel was used to 
accomplish the defined task so the SFC is attractive.

Given two stoves of slightly different heat transfer efficiencies, what 
information do these approaches provide to the stove developer?

SFC based on the final mass will tell the developer that the more efficient 
stove used a higher mass of fuel per litre to simmer the water, giving it a 
lower performance rating because more will have boiled away during simmering 
(presuming a similar fire size).

Using a PHU calculation taking into account the work done boiling off excess 
amounts of water, will rate the more efficient stove higher as it did more 
work, but does this also presume the stoves used exactly the same amount of 
wood?  If they used different amounts of wood and boiled different amounts 
of water, it is not actually interesting to know how much work was done 
because the work in question is to maintain a temperature slightly below 
boiling.  The amount of water boiled off is of no consequence.  That being 
the case, only the initial amount of water can be used, together with the 
wood consumed to give a useful figure for simmering, and it will be an SFC 
rating.  Only reducing the amount of wood burned can give a figure of value 
to a designer, and it should be expressed in gm/litre simmered.  Using the 
final water mass gives an poorer rating to the better performing stove.

In conclusion, it can be argued that if the purpose of a lab test is to 
assist the stove developer, then the test should give results that indicate 
an improvement when there is one, either in the boiling or simmering phase. 
The proposed Modified Water Boiling Test does not accomplish this, as ably 
demonstrated by Bailis for which I thank him.  He has not proposed what I 
have written above, but he did show in carefully run experiments that there 
is a negative correlation between lab tests done according to the modified 
WBT and field performance which can be explained by the analysis above.

Best regards
Crispin