Predicting failures for zero handpump downtime

Patrick Thomson received an EPSRC Impact Accelerator Account grant for “Predicting failures for zero handpump downtime”. He will be working with FundiFix and other OWN members Heloise Greeff and Cliff Nyaga to implement and test failure prediction algorithms into operational handpumps.

Embedding this machine learning into the Smart Handpumps will enable FundiFix to repair handpumps before they break. FundiFix has already reduced pumps’ downtimes from weeks to days. Reducing them to zero is expected to improve public health; even a few days without a quality water supply is associated with an increase in diarrhoea morbidity.

This research builds on earlier aspects of the Smart Handpumps project and is complementary to the work REACH and Uptime are doing with FundiFix in Kenya. After completed the technical work and initial testing in 2021, Thomson and others will be running a formal trial through the REACH programme in Kitui in 2022 and incorporating this new data stream into FundiFix’s operations.

Photo credit: Rob Hope