Data science helps to drive energy storage optimisation
Energy Mutual works with distributed renewable energy sources, such as small to medium scale hydro projects, that are now proliferating across the distribution network.
Energy Mutual
Lochaber-based Energy Mutual has created an online asset management software system to support the owner/operators of small to medium scale renewable energy projects such as hydro, solar PV, wind and biomass. In addition to helping monitor and maintain asset performance, in 2021 Energy Mutual began work on a framework that will help a range of asset operators to optimise energy storage.
Finding a path to The Data Lab
For Energy Mutual, the journey to The Data Lab’s MSc Placement Programme began when managing director, Kyle Smith, joined the Pathfinder Accelerator programme hosted by the Northern Innovation Hub and HIE in September 2020. Pathfinder supports entrepreneurs in the Highland region with a fully funded opportunity to accelerate their new product or business in only six months.
“I joined Pathfinder during the pandemic,” recalls Kyle, “It was a time when it was good to be able to establish a new connection with other business owners in the Highland area, and to have dedicated time to do some thinking for the long-term strategy for the business.
“Through Pathfinder, we were signposted to the EU-funded Co-Innovate programme, where we were successful in getting funding that enabled us to appoint a new Chief Technology Officer to drive our software development, and we were also directed to The Data Lab’s MSc Placement Programme.”
A spotlight on energy storage optimisation
Energy Mutual works with distributed renewable energy sources, such as small to medium scale hydro projects, that are now proliferating across the distribution network. These tend to be owned by non-traditional energy companies – typically farmers, estates, community groups and councils – for whom monitoring and maintaining performance can be a time-intensive process. Energy Mutual’s asset management software collects performance data from these assets to enable asset owners to establish if they are working correctly or optimally.
“What we are seeing across our clients,” explains Kyle, “is that where these different assets have a storage element, whether that’s a little bit of hydro storage, battery storage onsite or individual EV chargers, there is a need to optimise how and when that stored energy is being used.
“So with a hydropower system, for example, you can produce power at certain times of day when it makes most sense to generate, and the same is true for battery storage onsite or EV chargers, when it might make sense to charge a vehicle at different times of the day. So our aim was to create an energy storage framework or optimisation framework that could be applied in these three use cases initially.”
The challenge: effective use of existing data
For Energy Mutual, the brief for the MSc placement project was designed to start to investigate how an energy optimisation framework might work and what it might look like. With The Data Lab’s support, Kyle was quickly introduced to Farheen Tamkanath, a student on the MSc Data Science postgraduate programme at Robert Gordon University.
“We decided to use a bit of the extensive hydro power data we gather to investigate energy storage, and to begin to build a baseline model that would allow us to then subsequently look at how energy storage can be optimised,” explains Kyle.
“We had three years of data collected from this particular hydro scheme, including data on spillage over the dam, and minute-by-minute performance stats. So a big challenge for Farheen at the outset was to clean up that data set, and to get it into a useable format so that she could then actually start investigating different models. That in itself was a big help, and it has saved us a lot of time to have that data in a useable format, but the overall objective was to create a model that would help us to minimise water spillage – and optimise water usage for the hydropower system.”
Overall, it’s been a very positive experience for us. As a relatively small company, it was great to have an extra team member, as an extra pair of hands can make a big difference, even in a short period of time.Kyle Smith, Energy Mutual
The solution: establish a good baseline model
To deliver on this objective, having cleaned the huge data set, Farheen proceeded to complete a detailed analysis of the date using Python coding, before she applied different machine learning techniques to establish the best baseline model.
“As it turned out I explored three different machine learning techniques in total,” explains Farheen. “The first was decision trees, then random forest, and finally super vector machine – which proved not to be very accurate. Among the three, random forest machine learning produced the best model with excellent accuracy of over 99%.
“While the overarching goal for Energy Mutual is to optimise energy storage, we first had to make a baseline model, and I’m delighted it was possible to achieve that in the three-month timeline. Now the next stage for the team at Energy Mutual will be to use that model to develop a framework to optimise energy storage.”
Focused learning and commercial progress
Farheen was relatively new to data science, as she originally comes from an electronics background and holds a degree in Electronics and Communication. However, following a career break, she opted to pursue a change of direction, and was able to secure funding from The Data Lab to complete her MSc in Data Science at RGU.
“Because I’m not from an IT background, the data science course was very challenging at first, but it gave me excellent exposure to everything from data warehousing to advanced data management over the two semesters. Beyond that, the placement project with Energy Mutual, definitely helped me to gain a deeper, more practical understanding of the subject,” said Farheen.
As Farheen was working towards her MSc, the project needed to have a research component, and so The Data Lab worked with both parties to ensure the agreed scope of work matched their respective needs – and that realistic expectations were set regarding what could be achieved in three months.
“Farheen was able to look at different types of optimisation algorithms and to recommend different solutions. And now, on the back of her work, we are in a position to take forward the optimisation work internally with the goal of delivering a commercial solution later in 2022.”
A very positive outcome for both parties
With Energy Mutual now working on several projects that will involve the implementation of an energy optimisation solution, Farheen is also taking the time to reflect on her experience, and to look ahead to a new challenge.
“For me the MSc placement was excellent” she confirms. “Although the work was all virtual, Kyle and the Energy Mutual team were very helpful, taking the time to ensure I understood the domain and how it functioned, and guiding me on a weekly basis. Overall, I had a really good experience.
“I’m thankful to The Data Lab for giving me the opportunity in the first instance to study data science, and for then subsequently arranging the MSc placement with Energy Mutual. Having completed my MSc, I’ve now secured a data analyst’s job with NCR, and for that I need to give full credit to The Data Lab and the MSc Placement Programme.”
Find out more
Find out more about the Data Lab, Pathfinder and our other support programmes.
Browse all support services
We support businesses, communities and social enterprises through a range of support services, including advice, funding, events and property. Find out how we can help you.