Quantifying the impact of improved stoves and fuels designed to combat the health and environmental burdens of traditional cooking is necessary to ensure sustainable outcomes but remains challenging for practitioners. The current standard method to determine household fuel consumption, the Kitchen Performance Test, is costly, time intensive, and subject to error. To address these challenges, the Fuel Use Electronic Logger (FUEL), a sensor-based system that monitors fuel consumption in households was developed. In this study, the accuracy, granularity, and cost of FUEL were compared to that of the standard Kitchen Performance Test through simultaneous testing. Monitoring was conducted over four and five consecutive days in 10 households in Burkina Faso that were each stacking LPG, charcoal, and wood stoves; and in 20 households in Uganda stacking multiple wood stoves, respectively. Results show good agreement between the two methods on an aggregate level, with an overall R2 value of 0.81, and more varied agreement when comparing fuel consumption on a day-to-day basis. The sample variation was found to generally decrease with increasing monitoring length, pointing to value in monitoring over longer durations afforded by the FUEL. There was no systematic over- or under-prediction of fuel consumption between FUEL and the KPT, suggesting that the FUEL method does not have significant bias relative to the KPT, but the accuracy of the methods relative to the true, "ground truth" household fuel consumption value was not known. There was no agreement between either method with self-reported survey data, further illustrating the unreliability of quantitative survey data. Moisture content and Standard Adult Equivalence measurements were found to be similar whether measurements were taken only on the first and last days of the study period as compared to each day, although this should be evaluated over a longer time period for future studies. Potential errors in each method are discussed and resulting suggestions for developing an effective study with the FUEL system are presented. An economic analysis shows that the FUEL system becomes increasingly economical as monitoring duration increases or new studies are conducted, with a breakeven point at 40 days in this case. Overall, these results point to the viability of the FUEL system to quantify long-term, in-situ fuel consumption with similar accuracy to current methods and the capability for more granular data over longer time periods with less intrusion into households.
