Cover Image
close this bookBoiling Point No. 02 - Special Edition April 1991 (ITDG, 1991, 32 p.)
View the document(introduction...)
View the documentSmoke Pollution
View the documentDialectics of Improved Stoves by Kirk R Smith, East-West Centre, Hawaii, USA
View the documentWhite Rabbits!
View the documentThe Chimney Approach to Smoke Pollution
View the documentResearch Needs - Biofuel Stove Technology
View the documentWoodsmoke - who will put it out?
View the documentCookstove Smoke - The Other Side of the Coin
View the documentDomestic Air Pollution in Rural Kenya
View the documentA Chimney is Not Enough!
View the documentIndoor Air Pollution in Rural Malaysia
View the documentReferences:

Dialectics of Improved Stoves by Kirk R Smith, East-West Centre, Hawaii, USA

Reproduced from Economic & Political Weekly,
New Delhi, March 11th 1989

('Dialectics' - 'Method of logical investigation based on the statement and resolution of contradictory views').

The history of improved village stoves since mid-century has been characterized by three overlapping periods - the 'classic' period focused on reducing smoke exposures but generally did not apply scientific approaches to design, promotion and testing, and the 'energy' period stoves in the 1970s focused on improving fuel efficiency. All too often, however, these programmes also failed to apply scientific and critical methods. The programmes of the third or 'phoenix' period are now evolving, although programmes representing both older approaches are still active. A dialectic approach is offered in this paper to help make these new programmes successful.


Improvements in biofuel-fired stoves are nothing new. They have been part of human technological and entrepreneurial efforts since the discovery of fire. What seems to be new in this century has been the birth of what might be called 'self-conscious' stove improvement movements. These are motivated desires to stimulate social and economic progress in under- (or less) developed regions. The first group of these, here called the 'classic" phase, had as its principal members various Gandhian organizations and focused primarily on lowering smoke exposures. In spite of what now appear to be naive engineering and dissemination approaches, some of these programmes survive today, although most have been supplanted by the programmes of the second movement, here called the 'energy' phase. These programmes stand on the three legs created by global interest in appropriate technology, oil shortages and deforestation. Stove designers in this phase focused on fuel savings, with smoke exposure criteria being secondary or absent. The efforts of the energy phase have had mixed success to date, particularly with stoves designed to use unprocessed biofuels rather than upgraded forms such as charcoal.

The presently emerging 'phoenix' phase programmes are set off from those that preceded it by an ability to apply the lessons reamed from this recent history. Much of what can be gleaned from past experience, however, is not in the form of simple rules for the future, but rather a revealed set of dialectics on stove improvement. The programmes of the phoenix phase will be strengthened if they consider these dialectics although there is no universal set of answers, but rather a grouping dependent on local circumstances and priorities.

These dialectics can be organised under two broad categories: efficient versus smokeless performance and centralised versus decentralized dissemination. Here I will focus particularly on those issues related to smoke exposures and health.

Efficiency versus Smokelessness

Both improved fuel utilization and reduced smoke exposures need to be considered as primary goals for stove design and dissemination programmes. Indeed, as shown by the reports of this workshop, most post-dissemination surveys of improved stoves introduced to areas where traditional stoves cause large exposures have found that reduced smoke exposure is cited by users as often as improved efficiency to be the largest benefit.

One of the principal lessons coming from the energy period has been the importance of applying engineering techniques to stove design in order to reliably achieve higher fuel utilisation [Baldwin, 1987]. What has not yet seemed to be so well-recognised, however, is that the relationship between stove design, fuel use, and smoke exposure is also complex.

By comparison to studies of fuel utilisation, relatively little work has been done to determine how modifications in cookstove design affect smoke emissions. Much information can be gleaned, however, from the extensive research done with wood-fired metal heating stoves that have recently become popular again in many developed countries. [Smith, 1987b]. Indeed, some developed countries have found need to rapidly develop and promulgate rigid air pollution controls for household wood stoves because of the high emission levels characterising most traditional designs. In the US, for example, the Environmental Protection Agency, pushed by a lawsuit brought by the Natural Resources Defence Council, has recently announced woodstove emission standards to be enforced on new stoves next year. This has been justified because, by the mid-1980s, woodstoves had probably become the largest source of several important categories of air pollution in the country exceeding, for example, the carbon monoxide emissions of all US industry and matching the entire power sector in particulate emissions.

The concern in developed countries, of course, relates to outdoor air quality since metal heating stoves essentially all have flues or chimneys. Leakage from poor design, installation, or maintenance, however, can lead to significant indoor concentration [Traynor et al, 1987]. The village cookstove, on the other hand, typically does not have a flue and emits directly into the household environment. Many of the improved cookstove programmes around the world have promoted stoves with flues; sometimes called 'smokeless' although they are not designed to emit less smoke but to direct the smoke out of the house. Indeed, the most common designs probably actually increase total smoke output compared to the traditional open-combustion stove.

Fig 1 - The relationship of overall, combustion and heat transfer efficiencies to fuel moisture content (dry basis). Note that the three efficiencies do not peak at the same point. Since emissions are inversely proportional to combustion efficiency, therefore, the points of greatest overall efficiency and lowest emissions do not coincide. Since few such tests hove been done with biofuel cookstoves, this example is taken from a study of wood heating stoves done by J W Shelton at the Woodstove Research Institute, Santa FNew Mexico.

There are trade-offs between efficiency and emissions in many stove designs. Efficiency and low exposures may seem to be and indeed are, in general, compatible goals. After all, the source of most emissions from biofuel combustion is incomplete combustion and, thus, high combustion efficiency means low emission factors (emission per unit fuel). Unfortunately, however, some of the principal techniques used by stove designers to increase overall fuel utilization actually increase emission factors as well. This comes about because overall stove efficiency is a combination of two separate internal efficiencies, as illustrated in Fig 1. Enclosing the combustion chamber and reducing airflow - two common approaches for improving fuel utilization - may increase overall fuel utilisation by increasing heat transfer efficiency (shifting the heat transfer curve upwards in Fig 1). This may, however, actually decrease the combustion efficiency because of poorer turbulence and a lower airfuel ratio. The results therefore, can be increases in both overall thermal efficiency and emission factors [Ahuja et al, 1987]. Care must be taken, therefore, to improve or at least maintain combustion efficiency when seeking modifications to improve overall fuel utilization.

To predict the health impacts of changes in heat transfer, combustion, and overall efficiencies is not straightforward. This is because human exposure is not a direct function of emission factors, but is also affected by the emission rate, cooking time, room ventilation, proximity to stove, and other factors that may themselves be changed by modifications in stoves designed to improve fuel utilization. In some cases, for example, an increased emission factor may be more than compensated for by a decrease in total fuel usage and cooking time [Ahuja et al, 1987]. On the other hand, lower emission factors themselves do not guarantee decreases in exposures [Smith, 1987b].

Thus, as with fuel utilization, laboratory and simulated tests impart only limited ability to predict actual exposures. Field tests are necessary. In addition, even field verification of improved fuel utilisation is not sufficient by itself to conclude that exposures have lowered.

It might be thought that the above discussion refers only to stove improvements that do not incorporate flues. Unfortunately, this is not so. It is clear from studies in India, for example, that the existence of a flue is not always sufficient in itself to guarantee a significant reduction in human exposures under field conditions (Table 1). A number of factors seem to be involved but, in general, it is unfortunately true that stoves in the field are often not built, operated, or maintained in the ways intended by their designers. Such effects have also been found in tests done under simulated conditions, where it was found that under some circumstances improved unvented stoves release less smoke into the room than do flued models [Joshi et al, 1987]. In addition, users may frequently substitute fuels and pots in ways that lead to smoke releases. Thus, field tests are needed to verify the extent of exposure reductions.

Unfortunately, such field tests are difficult to conduct and quite labour intensive. This is because of the necessity of utilising personal rather than paint or area monitoring techniques. Since smoke levels are extremely inhomogenous in time and space within typical village houses, measured concentrations are sensitive to the location of sampling instruments [Smith, 1987b: Mumford et al, 1987]. But because of the many differences in kitchen design, stove location and personal behaviour among households, it is difficult to standardise the location of stationary monitors or conduct meaningful exposure tests in a simulated situation. The only viable alternative is the use of monitoring devices actually worn by family members during normal daily routines, which requires substantial supervision by the investigators and can introduce other forms of bias.

Although there have been only relatively few studies to date [Table 25.3 in Brewster et al, 1987], it seems possible to extract some tentative lessons for future studies. In combustion labs, for example, thermal transfer and combustion efficiencies should be measured separately along with critical pollutants. Without sophisticated apparatus, it is possible to obtain a reasonably complete understanding of the smokiness of a stove in its field setting by conducting three types of tests.

Emissions: Using the chamber technique [Ahuja et al, 1987], emission factors and rates can be monitored during performance of standard fuel utilization tests. This can actually be done in a village house using portable air pollution monitoring equipment and a portable fan.

Concentrations: Using stationary monitors, 24-hour and weekly concentrations can be determined. Placement might be standardised at the place where young children sleep, since exacerbation of childhood respiratory infections is thought to be one of the chief effects of smoke exposure.

Exposures: Personal monitoring can be conducted during cooking or other high-exposure periods. Depending on local customs, the monitoring device may be worn by the cook as well as other family members.

Table 1: Smoke Exposures and Concentrations Due to Traditional and Improved Cookstoves with Flues in South Asia. Cross-Sectional Comparisons of Matched Sets of Neighbouring households

Large natural variations in smoke levels exist in these relatively well-ventilated houses (poor mixing), when high-volatile solid fuels are used (great temporal differences in emission rates) in unvented or partially vented stoves (heat pumps). As a result, intrahousehold (eg, between different days) statistical variation is usually greater than interhousehold variation [Bolej et al, 1987]. This means that cross-sectional experiments to detect the effect of improvements in fuel, stove, ventilation, or behaviour will require large sample sizes and careful stratification to achieve statistical significance. Prospective (before and after) studies would be likely to produce more reliable results.

Another factor that tends to limit the exposure reduction of flued stoves is the entry of smoke from outside the house. Since smoke is still produced (even, in some cases, in greater amounts) by flued stoves, the outside air can become heavily polluted in some conditions. When houses are close together, stoves are used at the same time of day, and outdoor ventilation is low (as in the dry winter season characterizing many continental areas), for example, local ambient air pollution can reach high levels. In these cases, the relatively high ventilation rates of village housing can lead to a significant indoor concentration even when the flued stoves are working well [Smith and Durgraprasad, 1987]. In such conditions, even homes using biogas for cooking can experience nearly as high concentrations as nearby homes using traditional fuels even though biogas combustion itself contributes little. A study of improved flued stoves in Nepal, on the other hand, where houses were widely spaced horizontally and vertically, found significantly lower exposures among women cooking on smokeless stoves [Reid et al, 1986]. To be truly smokeless, stoves need to incorporate features such as secondary combustion chambers that directly decrease emissions. Unfortunately, it has fumed out to be difficult to design such devices to operate reliably. This is true even for metal heating stoves in developed countries, which cost many hundreds of dollars. In what might be called the 'woodstove dilemma', the rate of energy (power) needed by typical houses (0.5-2.0 kg wood/in) occurs just at the lower limit of wood bum rates at which high combustion efficiency and low emissions can be maintained (see Figure 2). Unfortunately, the typical power needs for household cooking are within the same range.

To some extent, this dilemma can be overcome by redesign of combustion chambers (eg, to concurrent air and fuel flows) and more meticulous tending [Smith, 1987b]. Wide adoption of the practice of lighting the stove outside the house and moving it inside only after the most smoky phases of the burn are complete can also reduce exposures at some cost in efficiency, convenience, and safety.

To solve this dilemma in developed countries, many stove manufacturers have fumed to catalytic converters. The catalytic conveners increase the efficiency of combustion such that, in typical developed-country conditions, they are usually cost effective for the users.

Fig. 2: The effect of burnrate on carbon monoxide emission factor for a wood-fired heating stove.

The cost of woodstove catalytic converters has decreased dramatically since the early 1980's, being now something less than $40. This is still too high for consideration by many developing country users but, just as has been shown with photovoltaics, there may well be appropriate niches in developing countries for high-technology devices that are user-friendly (particularly if they can be made in-country). Application to cooking stoves, however, may not necessarily be easy.

A more modest approach to accomplishing the sometimes conflicting goals of low exposure and high efficiency is to optimise stove design for efficiency without a flue and to use the stove on a fireplace-like hearth under a chimney. Such arrangements have been found to be quite effective in field studies in India, for example [see Table, note d; Ramakrishna, 1987]. In addition, the chimney arrangement can often be made of the same kind of materials used for the walls of the house itself.

The basic dialectic to be considered by stove designers, therefore, involves the potential compromise between efficiency and smokiness. Stoves designed to accomplish one goal may not achieve the other. There are several related dialectics that are also worth mentioning.

Efficiency Fuel Economy: Stove researchers have re-discovered what has been known for many years by investigators of other household appliances: efficiency is a strong function of use cycle. Put another way; reliable comparisons of devices that are controlled by individuals can often only be achieved after careful definition of a use cycle. The 'efficiency' of automobiles, for example, is recognised to be meaningless for most practical applications. Measured instead is fuel consumption during some standardized driving cycle.

The word efficiency is thus to be shunned because it implies a degree of universality that is rarely achievable with devices that are operated in such a variety of ways. An additional implication is that laboratory measurements based on standardized use cycles should not be expected to represent actual fuel economy in practice. One can hope to design a laboratory test such that the relative rankings of different stoves under field conditions can be determined but it will be some time, if ever, before absolute fuel economy can be predicted. In spite of this now well-known phenomenon, however, we still too often hear both the technical and policy communities use the results of laboratory tests to predict actual fuel use of savings, sometimes, as a result, becoming frustrated and discouraged when actual field measurements do not correspond [Gill, 1987].

The dialectic that results might be called the use-cycle uncertainty (Heisenberg) principle. The more closely that the use-cycle mimics the true use of the stove, the less it can be used to compare different stoves or the same stove in different situations. The designer and disseminator must compromise between accuracy and comparability when choosing a metric for evaluation.

Fuel Economy versus Deforestation: Even less well established is the connection between fuel shortages and deforestation. Indeed it now appears that, except in unusual circumstances (principally in Africa), it can be difficult to determine even which way the causation runs. Indeed, other sources of biofuel may make the connection in either direction be most tenuous [Leach, 1987]. Too often, however, we hear calculations of forest area preserved by virtue of a fuel saving that itself may be determined on the basis of dubious assumptions about the relationship between fuel use and laboratory efficiency.

This is not the place to go into the many reasons why such calculations are usually too simplistic [Goodman, 1987]. Given the initial impetus behind the 'energy' period stove programmes, the felt need to tie stoves to forests is understandable. The very complexity that is ignored by making such a direct connection, however, may well be the source of approaches to the actual problems that underlie the biomass/biofuel/food crisis. The administrative advantage of assuming such a simple direct connection, therefore, must be balanced against the loss of information and of the potential to perceive long-term solutions that could result. This is the dialectic.

Smokeless Stoves versus Smokeless Kitchens: As mentioned above, what have been called smokeless stoves are often actually quite smoky but flued stoves that may, under proper conditions, maintain low smoke kitchens. Unfortunately, flues by themselves do not guarantee low exposures and a true small-scale low-cost smokeless stove using unprocessed biofuels has yet to be developed.

Emissions versus Exposures: While seemingly obvious, it is worth pointing out that pollutant emissions by themselves are not a health problem unless they are retained in a volume of air (concentrations) and breathed by people for a significant time (exposures). Thus, the health effects resulting from cookstoves are influenced by several factors including room volume, room ventilation, wind, cooking time, cooking behaviour, which themselves are not completely independent. The interaction of these factors can be modeled but only in a rough manner [Smith, 1987b]. Thus, the relationship of laboratory emissions measurements to true exposures may not be straightforward. Actual field measurements may be needed.

Flued versus Flueless Stoves: it may seem like a paradox, but in some cases improved unflued stoves may actually create lower exposures than flued ones. This is because some unflued designs improve combustion efficiency as well as heat transfer efficiency while flued designs tend to focus on the latter. All too often in the field, flues do not work well and may not only fail to remove most smoke from the room, but also result in lower rather than higher overall fuel utilization because of user-reluctance to use dampers.

Measured versus Perceived Improvements: Although simple in concept, rural energy specialists will agree that it is quite difficult in practice to determine fuel usage accurately across different households and seasons. Indeed, the gap between concept and practice can be most frustrating and is evidence of the classic uncertainty principle of the social sciences: the more effort put into monitoring at the household level, the more change is introduced by the investigation itself. While involving more sophisticated monitoring equipment, the same principle operates for smoke monitoring: the measurer disrupts what is measured.

It is often important to determine perceived fuel use and smoke exposures such that stove dissemination can be successful. Unfortunately, however, reported perceptions too can be greatly influenced by the investigation itself. In addition, particularly with smoke, perception may not entirely mirror reality since some of the most important pollutants are wt. sensed accurately by humans.

Balance points are needed, therefore, between efforts to measure physical and perceived improvements and between the needs for accuracy and for minimizing uncertainty introduced by the measurement process.

Centralised versus Decentralised Dissemination: The most persistent dialectic in development is that between centralised and decentralized approaches (many other terms are also used including top-down and bottom-up). Indeed, it is possible to trace the pendulum swings between the extremes of this dialectic far back into history [Stohr and Taylor, 1982]. Perhaps the most fundamental differences among stove dissemination programmes is described by this dialectic [eg, Sarin, 1986 and Upadhyray, 1987].

India, for example, provides examples at almost any point along the spectrum from highly centralised nationally-managed dissemination programmes to highly participatory locally-managed ones. At the risk of simplification, it is probably reasonable to say this choice is also one between quantity and quality, alacrity and thoroughness and single and multiple objectives. Centralised programmes tend to result in more stoves being built per unit of time and money, but also suffer from high rejection rates and lack of integration into local social and economic development programmes.

As with every dialectic, the answer lies not at the extremes but in a balance. Social niches exist for both locally-made stoves utilizing mostly local materials and labour as well as centrally-made devices of metal or ceramic in which stricter quality control and economies of scale can be applied. Indeed, in many ways the two approaches are complementary, each having strengths where the other has weaknesses. A conscious centralised programme, for example, cannot long proceed without arranging to continually learn from the results of decentralized programmes with their careful attention to local feedback. On the other hand, national attention and commitment may be needed in many countries in order for a significant number of people to be affected.

Commercial versus Welfare Approaches: Should stoves be given away and otherwise highly subsidised or should they be priced such that the users bear a significant fraction of the cost? There are good arguments on each side and no easy answer. For example, large segments of many rural populations are not able to afford even minimal costs. In addition, many of the benefits of improved stoves may accrue to other groups or to the public 'commons'. These are the arguments for the welfare approach (cross subsidization). On the other hand, there is evidence that people often do not value or take seriously what comes to them at no cost. Furthermore, market approaches offer the workings of the 'invisible hand', which can operate to continually press for improvement through competition.

Again, neither approach will be appropriate in all circumstances [Evans, 1987]. Improvements in welfare and market dissemination methods must be pursued for use in different places, different times, and different groups.

Few versus Many Models: One of the truisms of rural development is that every village is different. This concept implies to some that a vast multiplicity of stove designs should be developed so each fits as closely as possible into a particular local living style. The other extreme is the 'any colour as long as it's black' approach where the dictates of mass production and quality control are allowed to prevail.

Simple versus Sophisticated Operation: It is difficult to design stoves that have high fuel utilization, low smoke emissions, and significant fuel flexibility without incorporating some user-operated tuning apparatus such as baffles and variable combustion chamber volume. Such apparatus, however, greatly increases the vulnerability of the stove to poor operator training and motivation. Another dialectic.

Design Evaluation: Stove programmes should not expect to be able to optimise one or two aspects of traditional stoves while maintaining all their other characteristics such as portability and fuel flexibility and continuing to give side benefits such as insect fumigation and room lighting (Figure 3). This is not to argue that such functions are unimportant but that they will need to be addressed by other means. Economic and technological development have nearly always been accompanied by specialization and there is no obvious reason that the evolution of cookstoves will be different [Smith, 1987a].

Conclusion: The changes in relative fuel costs and availabilities characterizing the 1970s led to changes in the perceptions of the likely evolution of domestic energy use. In contrast to the once inevitable conversion to petroleum-based fuels and electricity, it is now thought that biofuels may have a long future in a large percentage of the world's households. There are a number of implications of this view. The most obvious is that in most areas the biofuel cycle will have to change. Managed production must replace the unmanaged 'hunt-and-gather' techniques relied upon for harvesting most of today's household fuels. In addition, to serve development as well as survival needs, there will be a need for a greater degree of upgrading to higher quality solid, liquid and gaseous fuels. Finally the fuel cycle must end with both to be harvested on a sustainable basis and to continue to meet household fuel demand [Smith, 1987a].

Fig. 3

There are also implications for health and safety. No more can it be expected that existing problems will go away by themselves. They must be directly addressed at each step of the fuel cycle. Because there may well be difficult trade-offs among the desires for economy, efficiency, cleanliness, and other characteristics, increased quantification of the impacts on health will be required to make rational choices.

Such factors as economy, efficiency, and, to some extent, safety are fairly easily perceived by the users themselves. It can thus be argued that, given the opportunity, they are best qualified to choose among alternatives a way that best serves their own interests. Environmental contaminants, however, present a more difficult problem. Their impacts are often delayed and otherwise difficult to link directly to exposures. The health effects are not easily distinguishable from those with other causes. Thus, to pin down effects, it is necessary to rely on instrumentation, statistical judgments, and expert opinion. This is sometimes even true when the effects are great, as they are, for example, with tobacco smoking. This places responsibility on experts to communicate their findings in a way that is useful for people in their own thinking about relative risks.

The history of the world has shown that at every occasion where alternatives have been available and affordable, people eagerly turn away from unprocessed solid fuels for cooking, perhaps, I do not believe that the 'classic' and 'energy' periods of improved biofuel-fired stoves have provided convincing evidence that this trend will change in the future. Such fuels are always fated to be simply too inconvenient, dirty, bulky, hard to control, inefficient, and otherwise unsuited to cooking. The improved programmes of the 'phoenix' period promise to mitigate the impact of some of these characteristics and thus help make more comfortable and sustainable the unavoidable reliance on such fuels by large populations. It is well worth considering, however, at what level of effort we may actually start to engage in sub-optimising and counter-productive activities by pursuing further improvements in cookstoves burning unprocessed biofuel rather than the means to accelerate the natural trends leading away from them. Establishing broader programme goals such as improvements in cooking practices and housing may be more rewarding.