Breeding a Better Stove

1
Breeding a Better Stove

• the use of Genetic Algorithms and Computational
  Fluid Dynamics to Improve Stove Design
• H Burnham-Slipper MJ CliffordSJ Pickering

2
IntroductionAppropriate and Inappropriate
StovesExperimental WorkComputer
Modelling CFD Genetic AlgorithmResultsConclusi
ons
Outline
3
Global problem half the world cooks on wood
burning stovesIndigenous stoves can be
inefficient, dangerous, smoky, hazardous to
healthIntroduced stoves can be unpopularOur
approach is to combine local indigenous knowledge
and preferences with advanced computer modelling
techniques to develop an improved stove for use
in Eritrea
Introduction / Motivation
4
Appropriate and Inappropriate Stoves
Classic Eritrean mogogo smoky, inefficient, but
free
5
Eritrea Research and Training Center Mogogo - 40
Appropriate and Inappropriate Stoves
6
MIRT improved efficiency, but developed in
Ethiopia. The stove of our enemies
Appropriate and Inappropriate Stoves
7
Aprovecho design improved efficiency, but heavy
use of material and poor thermal distribution
Appropriate and Inappropriate Stoves
8
CleanCook alcohol stove unfamiliar technology
and materials. Unsuitable for cooking injera
Appropriate and Inappropriate Stoves
9
Experimental Set-up

• Experimental aim
• mass-rate data
• temperature data
• Apparatus
• regular wood cribs
• mass balance
• K-type thermocouples
• extractor hood
• a tiny bit of fire-lighter

10
Experimental Results
11
Numerical Model Formulation

• Assume
• char combustion limited by diffusion of oxygen
  through species boundary layer
• volatile release limited by conduction of heat
  through char layer
• volatiles burn in air, limited by turbulent
  mixing

12
Numerical Model Formulation

• Fluent 6.2 CFD code
• buoyancy-driven flow
• k-e turbulence model
• species transport
• DO radiation model
• UDF fuel model
• lumpiness function

13
Numerical Model Results

• Fluent 6.2 CFD code
• burn-rate agrees with experimental
• temperature velocity fields agree with
  experiment literature

14
Stove Modelling
15
Stove Characterisation
Aprovecho rocket, HBS rocket, mogogo, 3-stone
firewith without grate
16
Stoves have evolved over hundreds (maybe
thousands) of yearsThere may be good reasons
why stoves are the way they areA genetic
algorithm can speed up the natural evolution of
stove design
Genetic Algorithm
17
Genetic Algorithm
Take two stoves Allow the stoves to mate Define
ten children (new stoves) using randomly selected
genes from parent stoves Test the efficiency of
the new stoves using CFD Discard all but the best
two stoves Repeat (The method can be adapted to
include genetic abnormalities / random mutations)
18
Genetic Algorithm
19
Genetic Algorithm Progress
20
Genetic Algorithm Result
21
Conclusions and Further Work
Engineers need to take many factors into account
when designing stoves Respecting local stoves and
building on indigenous knowledge is vital if a
new design is to be successful Combining genetic
algorithms and CFD represents a novel approach to
stove design, mimicking the natural evolution of
stoves It remains to be seen if the new design
can be manufactured and tested, we also have a
lot of field work to do