Enhancing Off-Grid Appliance Maintenance with Remote Monitoring: Insights from Innovex

The Efficiency for Access Research and Development Fund has provided funding to Innovex, a pioneering technology company based in Uganda. Innovex used the funding to develop a remote monitoring hardware, and a cloud-based platform called REMOT, which are used to support the operation and maintenance of off-grid appliances. In this interview, Hannah Mottram, Senior Research Analyst, and Mahta Ameli, Project Co-ordinator, Energy Saving Trust, co-Secretariat of Efficiency for Access, speak with Ben Wycliff Mugalu and Benjamin Bwowe from Innovex about their research into how remote monitoring can be used to enhance the operability and repairability of solar-powered refrigerators and solar water pumps.

What challenges do people have with repairing and maintaining off-grid appliances?

The main issue is high maintenance costs for components, especially for rotors or inverters.  Not only is there a high upfront cost, but it also affects users’ income, as they can’t earn money when their appliance isn’t working.

It is also hard to ensure that appliances are maintained consistently due to farmers’ day to day schedules. They are so busy that they often forget to follow the maintenance schedule provided by the distributors, meaning that their appliances face issues frequently. Another challenge is that new users may not be trained to maintain appliances.

How has the funding from the Efficiency for Access Research and Development Fund helped you solve these challenges?

The R&D Fund helped us to move from our assumptions to actually tackle the problem and get insights into what’s possible.

The research that we’ve done has given us a clear picture of the frequently occurring user-related faults in the field that could be predicted by monitoring systems. We also set up controlled experiments and used the data to build AI models that can detect how and when faults take place in real time. These include issues such as clogged impellers for the pumps, thermostat failures for the freezers, and blocked vents. There are also user-related faults – particularly when it comes to freezers – as people sometimes put hot food into their freezers, which can cause long term damage.

What were the limitations of your research project or the challenges you encountered when you were doing it?

Most of the challenges we encountered related to data.

We had challenges collecting data from solar mills, as there was inconsistency with the data sheets and the actual consumption, which meant that the sheets damaged our equipment. Some faults were difficult to measure because they were dangerous – for example refrigerant leakage and loose belts in milling machines.

We would like to build a digital twin. To do this, you need a diverse data set, but our sample space was quite small. We’d like to expand the types of appliances and collect the data in real time to understand the gradual manifestation of faults.

How will this project help end users on the ground?

Our objective is to prevent faults, not just fix them. We’ve implemented real-time alert systems using deployed models.

As data is coming through the remote platform, it’s constantly being scanned for potential faults.  If a fault is detected, the user receives a text message alerting them to the issue and providing a simple recommendation to fix it.

For example:

• ‘It appears the freezer is overloaded. Check if you’ve placed hot items inside.’

• ‘The flow rate of your water pump is low. Check for any blockages.’

This system allows users to identify and address faults before they cause appliance breakdowns, enabling timely fixes or technician calls.

What’s next for your work on repairability and predictive maintenance?

We plan to install these systems with some of our existing clients who have a large pool of devices, like Koolboks for freezers and ennos for water pumps, and then try to evaluate how our models can adapt to different appliance types.  Through market research, we found that our users don’t like complicated dashboards and prefer simplified text alerts.  We’re going to explore integration of large language models into our work. Instead of receiving a notification via email or text message, there is an intelligent agent that delivers that message directly to you, and you can ask it further questions to learn more.

Dr. Hannah Mottram delves further into Innovex’s REMOT platform in our new publication ‘Smart Off-Grid Solar Appliances’. Please contact us at eforagrants@est.org.uk if you’re interested in discussing this exciting and innovative topic.