Shining a light into the black box: Disaggregation technology

Many use the metaphor of a seed sprouting and growing to describe technology development: there first needs to be an idea (seed), but it requires the right conditions (soil, water, temperature, light) to start to grow. For disaggregation technology, the market is not quite there to create the conditions which allow this idea to grow.

Foto von Daniel Norris auf Unsplash

The advantages of disaggregation appear clear: It shines a light into the blackbox of a smart meter reading. Disaggregating allows us to see all the consumption “tributaries”, like a boiler, fridge and TV, which make up the entire household meter reading “river”. In knowing how each tributary is consuming, in volume and time, we can optimize its efficiency (e.g. using less or at less high-demand times of the day, changing standard settings, or replacing with more efficient versions). More information = more knowledge and ability to make impactful change.

However, it is less clear who will pay for this technology and what is a business model that allows further innovation development. Technically quite complex, disaggregation is not a standard technology and is still developing to handle some major challenges: each house is uniquely made up of different appliances, different consumers with different daily patterns. The “fingerprint” of a single appliance cannot be seen in the consumption data, just like how we cannot know where each drop of water in a river originated from.

Rather, a combination of indicators allows developers to start to isolate unique events in the consumption data patterns and associate them to specific appliances. While for some appliances this is easier, like for a fridge or heatpump, which have a recognizable demand pattern over a day, for other appliances the randomness of use and demand is harder to separate out.

Another difficulty is to separate similar appliances by their actual electrical fingerprints. Motors used in heatpumps and fridges have different electrical characteristics than non-rotating machines, but within each group it is difficult to separate them from each other. An internet router has a very similar electrical fingerprint as a LED lightbulb. And when a large appliance is running, like a heat pump, this masks the fingerprints of any smaller appliances being used at the same time, like TV’s, computers or light bulbs.

Thus, in order to maintain a reasonable accuracy, as incorrect attribution may cause changes that could actually increase consumption, disaggregation was limited to heating, non-heating and standby within Social Power Plus.

See more about the disaggregation technology, its potential and challenges, developed for reading heat pump consumption within the project here from CLEMAP.

Within the Social Power Plus project, with a reduced, but therefore more accurate, disaggregation of heating and non-heating energy use, several participants were able to optimize poorly-operating heat pumps, replace their boilers for smaller and more efficient versions, identify the savings from replacing halogen lightbulbs with LEDs, and adjusted automatic heating setting in bathrooms and laundry rooms.

You may ask- but if this is just a question of innovating until someone figures out how to read all the electricity consumptions, why isn’t research or the market there yet? Good question, and the answer is because the market conditions are not favourable to continue to put resources (time, money, and brain power) in this direction.

While the technology may be interesting for households who want to save on their electricity bills or reduce their environmental impact, energy utilities are not yet working with this type of data to innovate their own business models (e.g. to manage grid loads through real-time feedback to customers) and thus are not yet developing the technology themselves. Government research funding often steps in to fill this gap. With the support of the local stakeholders, like the energy utilities, more information on infrastructure and installations (e.g. where a PV panel has been installed or an EV charging station) can be known which can support a more goal-oriented and high accuracy disaggregation.

The potential of the technology is also not yet known (a classic chicken and egg problem). Projects like Social Power Plus, which involved some forward-thinking energy utilities willing to take a risk, will allow us to estimate what the actual savings potential may be when disaggregated information is provided. The results of the project will be read by the Swiss Federal Office of Energy, who funded the project, and hopefully they will see the potential and continue to finance further efforts to develop disaggregation technology.

With support from the private sector and government, the right conditions can be created to make this technology grow.