Predictive Testing of Cars and Wood Stoves

Written decades ago, the lab based tests for both biomass heating and cooking stoves were designed to achieve statistical validity by controlling variables. Because many real world variables were removed from the heating and cooking stove protocols, the results were known not to predict real world performance.

Automobiles are currently tested on a dynamometer instead of being driven around town. EPA estimates are based on dyno tests designed to reflect “typical” driving conditions and driver behavior. Even so, The EPA warns customers that actual mileage will probably be significantly different.

To predict real world performance, each car could be driven around town by enough people until a meaningful average was mathematically determined. Cook stove tests could also rely on field tests generating complicated data resulting in accurate predictions. However, doing real life testing for every manufactured car or stove has been thought of as rather cumbersome.

Another approach might be to have regional survey data inform a predictive model. The model teaches the dynamometer how to test the car. Stove use could be modeled in the same way, so lab tests get closer to predicting what actually happens when cooks use wood to make meals.

Making a model probably didn’t seem to be worth the trouble in the past. However, things have changed. The harmful emissions from cars and biomass stoves damage health and contribute to climate change. Actually knowing what a new car or stove will do when used should help to create better technologies and reduce pollution. Any kind of predictive testing seems like a great idea!