[Stoves] (stoves) Haitian cooking

[Crispin Pemberton-Pigott]

Dear Tom

You ask a question that is not easily answered.

>So what is the range of values that we should be considering for wood vs
charcoal in improved and unimproved stoves? What are reasonable reference
values for a given meal size?

The improvement has to be stated along with a method of evaluation. This is
a can of worms because as everyone knows, the WBT3, widely used to evaluate
stove performance, does not predict field performance. This has apparently
not bothered programme implementers at all and the WBT results are often
used, trumpeted even, as justification for selection of stoves to spread
abroad. The situation might be a bit different if the intrinsic errors of
the WBT were declared (i.e. 'error bars') so that people knew in advance
what range of performance they could expect, not because of stove
variability, but because the test itself gives a performance that is
actually a range, not a firm number. What I am getting at is akin to the
difference between accuracy and precision. Performing a WBT can increase the
precision of the performance rating but not the accuracy.

It is very interesting that Jim Jetter (PCIA) has been performing something
closer to a heterogeneous test (like the SeTAR HTP) in order to get two
performance points (different moisture levels, but still only one pot size).
Significantly, one stove was not able to work at all with wood that was
moist. Suppose it have been chosen for promotion on the basis that with dry
fuel it did really well? Clearly a heterogeneous test is needed to provide
performance curves.

With that in mind, and close to answering your question, the testing in
Maputo of the traditional and POCA stove provided the following formulas:

Traditional stove, charcoal burned (uncontrolled cooking test, UCT)
y = 7.002x + 418.36 where x is the meal size in grams.

Improved POCA stove, charcoal burned (uncontrolled cooking test, UCT)
y = 1.439x + 373.8 where x is the meal size in grams

If you plot the curves you will see they meet at small meals (about 1.5 kg
total) and diverge significantly as the meal gets large (8 kg). For an
average 4.5 kg meal the saving is approximately 40%. This rises to 50% at 9
kg. The analysis includes 37 tests which cooked a total of 266 kg of food.
Many of the individual tests had separable cooking events because they used
two stoves at the same time. All this was considered.

So, the savings are real, across the board for all meal sizes, but highly
variable, depending especially on the meal size and operator technique. Some
people are hopeless at operating a stove. We did not weed those tests out as
that is their reality.

Regards
Crispin