Week 5 at the Centre for AI & Climate was a relatively quiet one. Steve is off on annual leave and I took a few days off myself as well.
That being said, I attended a fantastic webinar by Beringa on the potential power of DNO smart meter data, and I thought it would be useful to share my key takeaways and thoughts.
The webinar was part of a study that Beringa are carrying out, looking into the publicly available smart meter data, how it’s being used, and how it needs to be improved. Smart metre data being published at this granularity is a huge step forward, but as anyone who has tried to work with the data knows, it’s pretty difficult to extract value in its current form.
Here are some of my key takeaways:
- The data as it currently stands, even with all of its shortcomings, is far better than what was previously available. The alternative is national level demand profiles, which are far too abstracted, or small samples of smart metre data from over a decade a go. So all in all, this is progress.
- A geospatial element needs to be added. This is something we thought of immediately after working with the data for a few days and seems to be a popular suggestion.?
- Total counts of metres that the data represents is essential. Not all smart meter data is represented by each DNO, and not all meters are smart. Without this, it’s really hard to know what you’re actually looking at.
- More granular data. At the moment, smart meter consumption is aggregated at low voltage feeder level. It’s been done for data privacy reasons, but without the meter level consumption profile, the important details are getting lost in the noise.
- At the moment the smart meter data only covers domestic properties, but commercial is being considered.
- Current uses were mostly associated with energy demand modelling, local area energy planning and provision of new products and services.
- Contextualising the data with complimentary datasets was discussed. In particular the inclusion of low carbon technologies.
- The geospatial element seems like a very sensible next improvement to me. All DNOs would need to use the same structure and hierarchy as the value comes from combining the datasets to get a national picture.
- The meter level (non-aggregated) data is a really interesting one, because even though I think it would be unbelievably valuable, I very much understand the data privacy issues. What I would be interested to see is an anonymised meter level data set that is separate from the geospatial-based dataset, but can provide insights into the variation in demand profiles.
- Commercial property needs to be included. If the main use of this data is going to be demand modelling and energy planning, we need to make sure all demand profiles are represented. Commercial is where most of the variation is found.
- Total count of meters and smart / dumb split has to be included. It’s a bit of a shocker that this was missed. That data is essentially unusable until this is included.
- In terms of contextualising the datasets, we might be thinking about this too narrowly. I don’t think the aim should be to create a behemoth dataset that includes every field and feature. Instead, we should create multiple datasets that can be linked in analysis. So the key here for me is to make sure the smart meter data can be easily referenced and queried when using other datasets.
- Something that wasn’t discussed in much detail and I really think is the elephant in the room, was the sheer size of the datasets and how difficult they are to work with. We’re going to follow up with a full blog on this topic as we believe this is what it really comes down to.
