Cover Image
close this bookBoiling Point No. 21 - April 1990 (ITDG - ITDG, 1990, 44 p.)
View the document(introduction...)
View the documentCoal Briquette Technology & Policy for Development
View the documentCookstove Smoke & Health
View the documentFuel Consumption Per Head a Misleading Concept ?
View the documentWhere Expertise Counts ?
View the documentMonitoring & Evaluation of Stove Programmes
View the documentFocus on the Shortage of Metal in the Sudan
View the documentPublicity for Stoves Programmes in Fiji
View the documentAn Aspect of Women & Stove Production in Tanzania
View the documentBetter Bread Ovens
View the documentThe Silkalon Stove
View the documentClean Combustion of Wood
View the documentAlternative Rural Energy Strategies in Zimbabwe
View the documentThe Stove - A Target Orientation Programme
View the documentA View of Improved Stoves Prospects in Sudan
View the documentRWEPA NEWS
View the documentStove Journal Profiles
View the documentLetter to the Editor
View the documentNews

Fuel Consumption Per Head a Misleading Concept ?

by Brian MacGarry, Silveria House Training Centre, P O Box 545, Harare, Zimbabwe.

It is perhaps a built-in bias of our age and of the training of many of us who work with woodstoves, to attach an almost magical value to any quantity which we can define clearly and attach a numerical value to, while forgetting that what is finally defined may still prove very slippery to use.

Fuel survey literature abounds in woodfuel consumption figures expressed in kg/person/meal (eg Hancock & Kaeser p. 107), kg/person / yr (ibid p.109) or m3/person/year (FAO, 1981); but can we expect to go reliably from the measured consumption per household to a reliable 'per head consumption' and from there, to make comparisons between different areas and environments or useful projections into the future?

The first question to be asked here is: "which is the more constant value within one situation: consumption per head, or per household?" The figures which follow, taken from two small surveys, one urban and one rural, in Zimbabwe (MacGarry 1987), give indicative answers.

The purpose of the first study was to determine whether urban low-income households could benefit economically from changing to cooking on a charcoal stove of a type being newly manufactured in Zimbabwe, where charcoal, although available as a by-product from extensive wattle plantations, is not a commonly used fuel.

Households which agreed to take part in the survey received a cooker on loan and a 5kg bag of charcoal which they would use for all their cooking as long as the charcoal lasted. The monthly cost of cooking with charcoal was then calculated and compared with the cost of using their usual fuel. Households studied included users of wood (open fire), kerosene and electricity. Varying tariff rates made it difficult to translate cost figures for electricity into an energy consumption figure, but fuel consumption figures could be fairly easily derived for wood-using and kerosene - using households (see tables 1 & 2).

Table 1: Total paraffin consumed per urban household (abbreviated)

23 households surveyed with 1-8 occupants

paraffin consumption

litres / month

litres / month / head




Standard deviation in litres/month



Standard deviation as % of mean:



Table 2 - Total urban household fuelwood consumption

7 households 5-8 occupants

wood used


per h'hold

per head




standard deviation (kg/montht)



standard deviation as % of mean.



The second study measured family fuelwood consumption of a number of families in one rural area over several months. Some results are shown in Table 3.

Table 3 - Total rural household fuelwood consumption March-Oct 1986 12 households 3-10 occupants

total wood consumption

per household

per head




standard deviation kg:



standard deviation as % of mean




From tables 1 & 3, the consumption per household is clearly a more nearly constant quantity than the consumption per head. From table 2, there is little to choose between the two; this table, containing the smallest number of cases, and the smallest range of household sizes will give results of lower statistical significance than the other two.

Ale attempts which have been made to 'correct' per capita consumption figures seem unlikely to succeed in their attempt to produce a more usable consumption per head assuming a 'standard adult'. The most commonly proposed coefficients rely on the dubious assumption that fuel consumption per person is proportional to the calorific intake of that person as food. Even so, as comparisons by IDRC from figures obtained by Lusk, League of Nations and FAD/WHO show, different sources give very different values for calorific intake variations with sex and age.

The differences for children especially, who make up the majority of nearly all households of more than 4 members, make it particularly difficult to place any reliance on any of these coefficients. The strange thing about attempts to use any such coefficients is that two important points are not mentioned at all:

· whatever numerical value is attached to the fuel consumption of one individual, the consumption (or, ea. six such individuals living and cooking together is very much less than six times that of one individual living as a one-person household.

· if any value can be attached to coefficients for people of different ages, an infant will usually require more fuel than an adult, for separate corking, extra washing etc. although all the coefficients proposed give lower values for a child than for fan adult.


The range of variation to be expected between households of different sizes and among households of the same size in the same society are of the same order, and define the limitations on the accuracy with which we can expect to measure the overall fuel consumption of larger groups.

When making comparisons between different groups (one rural area and another, or between rural and urban areas) or for the same group at different times (allowing for projected population changes), the total fuel consumption will be more reliably obtained as = (consumption per household) x (total number of households) than as = (consumption per head) x (total number of individuals).

Projected future consumption should be calculated using consumption per household, not per head. If average household composition does not change over the period of the projection, both will, of course, give the same result, but in many cases this condition will not apply.


D Hancock & G Kaescr-Hancock: Cooking patterns and domestic fuel use in Marvingo Province, GART, Masvingo, Zimbabwe/GATE Eschborn, W Germany 1985

M Howes: Rural energy surveys in the third World - a critical review of issues and methods IDRC Othawa' 1985

B Mac Garry: Measuring family fuelwood consunption in Zimbabwe Commonwealth Social Science Council London 1987. (CSC technical publication series No 217)