Energy distribution infrastructure operator Jemena said it has found a way to “really mine” the tens of billions of data points collected by electricity smart meters in Victoria for the first time.
The company late last year said it had built a platform called JAWS - Jemena AWS - with the assistance of AWS and Deloitte, in the hope of influencing customer behaviour but also its own infrastructure investment decisions, a key input to energy pricing.
JAWS is now about two years old and was used during the pandemic to identify vulnerable end users - those whose usage had suddenly dropped - and advise them of support options, and to identify at-risk users whose energy needs had suddenly increased.
Jemena strategy and commercial general manager Andrew Davis told an AWS webinar late last year that despite electricity smart meters in Victoria collecting vast amounts of data, the sum of that data hadn’t been effectively harnessed.
“The numbers are quite staggering: [there’s] 500 billion network data points. That's for a base of 360,000 residential and small business customers,” Davis said.
“When you look at the customer data points themselves, [there’s] 50 billion.”
In a case study, these numbers are explained further: the 50 billion data points are on energy consumption, while the 500 billion data points cover “induction, voltage, and power factors”.
“What we had struggled in the past with was how do we actually derive value from [all this data],” Davis told the webinar.
“While we had an on-premises data warehouse collecting the data, we actually lacked the nous and also the ability to leverage it effectively.”
Out of this came JAWS, though not as a specific project aimed at standing up a cloud-based data lake.
Rather, Jemena entered a “timeboxed” 16-week “sprint” with Deloitte and AWS.
“We were able to create a small datathon rather than a data lake to bring in the data that we needed to run unsupervised machine learning to understand if we actually could see that there was insight out of how our consumers use electricity,” customer innovation lead Jesse Mills said.
“By taking this approach, we wanted to make sure that we were actually investing in something that was going to give us insights, and if we weren't going to get the insights, we actually hadn't over committed ourselves.
“Within that 16 weeks, we partnered with Deloitte and AWS, we set JAWS up, and then we were able to create six distinct unsupervised customer segmentation models.
“We were able to really dive in and find statistically significant and distinct models based on how customers use electricity during the day, and how they responded to temperature changes.”
Davis said the sprint showed Jemena that JAWS had legs.
“For the first time we've been able to really mine the data that's just richly sitting there and start to move it towards how we can support our customers more effectively and actually leverage value, not only for us but most importantly for them,” he said.
“More than anything else in current times, what JAWS has allowed us to do is implement something that's really cost effective.
“From the outset, we weren't dealing with big dollars. It's safe to say that during this process we've done it really quite quickly and JAWS has moved from strength to strength.”
Exploring data use cases
With Jemena on the distribution side, the company needed the involvement and assistance of energy retailers, both to understand what customers wanted to know from their usage data, and to provide those insights back.
“What we were hearing from our customers through our customer research, or consumer advocacy groups through partnerships, through the Energy Charter and discussions with retailers, was that customers expected us to be able to take the data on how they were using the electricity network, and actually turn that into both better business decisions on where we would invest funding for things such as asset replacement or expansion of the network, but also to provide them with a targeted, more personalised service and support,” Mills said.
Some of the personalised service use cases - at least initially - were Covid specific, recognising immediate or future hardship caused by either cessation of service or a spike in energy demand, which could lead to substantially higher bills.
“The results for utilising JAWS have really been more than we'd hoped for, and what it's allowed us to do during periods like Covid-19 is actually tailor our messaging to [end users],” Davis said.
“We've had some really great work with a community reference group, comprising customer advocates, where we've been able to identify through JAWS really quite quickly customers that a) could have been vulnerable, and b) have had over-usage.
“What we're trying to do in both instances is ensure the message is getting to them that they can change their behaviour.
“By working with the community reference group, we've also been able to tailor our SMS messaging as well, so that the messaging isn't intrusive, and it doesn't look like ‘big brother’ is looking at you.”
Davis said that vulnerable customers could be advised of support options available to them.
This is explained more in the written case study: “Jemena sent communications to its electricity customers who had demonstrated a loss in consumption to ensure they were aware of the support programs available, such as their energy education partnerships with Australian Energy Foundation (AEF) and Uniting.”
Jemena said it also intended to use JAWS to trial proactive demand response management on particularly hot days over summer.
“We're going to reach out to customers that, during peak events, use more than 30 percent [more] power than they traditionally would,” Davis said.
“The real beauty of it is rather than just looking at days, we're actually taking a number of variables like the weather, the day of the week, and other other key variables [into account to] give us really rich insight.
“We trialled [this] on a small scale with [in 2019-20] and we had really strong results around people proactively wanting to do the right thing by the community, and by the grid as well.”
Jemena also intends to use JAWS to glean insights that could help it make future investment decisions with regards to its infrastructure.
This may help the company keep a lid on its charges and also respond to changes in the distribution market, as more households run their own generation assets like solar arrays.
“What we're also looking at - and we've successfully implemented - is trying to figure out what the future network will look like and how customers' consumption and purchasing behaviour is changing,” Davis said.
“We thought we had a very good handle on the amount of batteries, solar and EVs [electric vehicles] in our network, but what JAWS has actually allowed us to do is drill down and figure out which households actually do have solar, and that's starting to help shape our thinking around where we need to invest for a DER [distributed energy resources]-rich environment, which as everybody knows isn't too far away.”
Mills said the customer segmentation models run over data in JAWS helped Jemena understand these changing patterns and trends.
“We believed that if we didn't actually really understand how our customers were using our network today, how they responded to different times of day or to different temperature changes, we wouldn't be able to prepare for the future, prepare for more solar on rooftops, prepare for electric vehicles/ batteries, and a number of other different technologies that probably haven't been invented yet, but will be coming onto our network in the future,” she said.
“If that’s the future, we need to start to understand our customers now, so we can work with them to respond to that.
“That's where the customer segmentation modelling came in: really looking at actually understanding the customer's behaviors, because that's what we had to be able to inform what we knew about the customers.”
Davis said that insight would ensure Jemena remained both “relevant” and cost-efficient.
“It's starting to inform our future investment decisions and that's only an upside for two reasons,” he said.
“One is obviously we will build a network that people want and we remain relevant.
“But most importantly, it helps ensure that we keep our prices that we're charging the customers as low as reasonably possible, and that's only a good thing.
“So we're starting to make really informed decisions.”