GridShare: How AI software is transforming Virtual Power Plants
Climate change is a race against the clock. In order to prevent the worst effects of this climate emergency, we now face an urgent challenge to transition away from fossil fuels and develop new methods for more renewable energy to power our economies.
However, among the challenges of adding more clean energy to the grid is the unpredictable nature of renewable energy sources, such as wind or solar. These resources might also not always be located near where they are needed most, so we rely on transmission lines to move power from where it’s generated to other locations. Therefore, scaling renewables to the levels required to meet climate targets could result in substantial amounts of grid congestion.
To overcome these problems, finding ways to enhance flexibility on the grid is crucial to help balance supply and demand and keep the grid stable while also decarbonising our electricity supply and preventing transmission lines from becoming congested. An analysis led by the Carbon Trust and Imperial College London shows that a fully flexible energy system could cut the cost of reaching net zero by up to £16.7bn a year in 2050.
The importance of DERs and AI for grid services
As the residential energy storage system market is gaining momentum, flexibility markets that allow devices in homes to participate are going to become more established in the next few years. Consumers worldwide are taking centre stage in supporting the creation of a clean and flexible grid, and distributed energy resources (DERs), such as smart batteries and electric vehicles (EVs) aggregated in virtual power plants (VPPs), are becoming a critical flexibility source.
For grid operators to capitalise on the aggregated power of DERs and reach our global decarbonisation goals at the speed we need, it is clear that artificial intelligence and sophisticated software are a must for VPPs. Fast decision-making on a home by home basis is required on a vast scale to enable VPPs to succeed, which is where AI comes in.
At Moixa, we are helping accelerate the world’s transition to renewable energy by leveraging our GridShare software to deliver grid services with residential assets. In our journey to building and scaling GridShare, we have found that the architecture of our software is the key to its reliability and sophistication.
Below, we will take you through the evolution of our GridShare software and share some insights that have helped us tackle these complex and exciting challenges in distributed renewable energy.
GridShare’s architecture and data processing model
GridShare is our customisable cloud-based software platform that optimises residential assets to provide savings to customers and value to the grid. In increasingly complex markets, millions of distributed assets will need to be coordinated to create optimal value on a second-by-second basis to end customers, asset owners, and the grid. GridShare enables this, acting as an AI infrastructure layer for the future smart grid. Our software can optimise various device types, from energy storage systems to EV chargers, and will have additional use cases in the future.
GridShare software’s distinctive value comes from its unique architecture and data model that allows it to combine behind-the-meter optimisation and grid service revenue opportunities.
Currently, in the residential energy storage industry, standardisation of device communication is limited or in its infancy. Standards such as IEEE 2030.5 for batteries and OCPP for EV chargers are helping adoption but are not widespread. Data is often presented quite differently, generally based on what the manufacturer included in their own communication specifications. This means that data resolution, control options, and metering setups can all vary substantially from one asset to another. This aspect is particularly complex in the residential space due to the wide range of products available and the sheer number of assets you need to pull together to have significant volumes of energy available (in comparison with commercial and industrial based VPPs, which tend to have fewer assets of a higher power).
GridShare was built with a precise focus on the residential energy storage space; therefore, our software is well suited to deal with this complexity. We have created a robust model for mapping any home set-up, which has been validated across various markets: this is often referred to as a Home Energy Management System (HEMS).
This forms the basis for all the intelligence we build on top and makes GridShare extremely scalable. This scalability would not be possible if the platform had been built for fleet-level optimisation of only a few large scale assets.
GridShare manages to ingest data from a wide variety of sources, including meter readings, internal device data (such as temperature), device status alerts and market data (including the customer tariff). Data is stored in our data lake such that all raw information is available and accessible. Additionally, all data is regularised into a more structured store, allowing quick, intuitive access.
Thanks to its data model, data processing within GridShare is generalised and allows us to be device agnostic and scale quickly, without the need to develop use case-specific applications.
Figure 1: Illustrative image of GridShare’s Site and Data Model.
Behind-the-meter optimisation of sites with GridShare
GridShare deploys its AI at a site level in order to optimise for “local” or behind the meter factors, such as minimising cost or carbon footprint.
GridShare identifies a household’s solar generation and consumption patterns alongside weather forecast data and creates generation and consumption predictions based on these inputs. The AI software generates a personalised daily plan for the on-site devices for each household based on these predictions, and the import and export tariffs of each household, or the grid carbon intensity, to ensure the best performance for cost or carbon.
For battery owners on a time-of-use tariff, GridShare will ensure their battery and solar cover their household’s electricity needs, charging from low-cost grid energy if solar generation is predicted to be insufficient. This enables customers to get maximum benefits from their self-generated power from both an environmental and a financial standpoint.
GridShare will also take into account any user-set preferences, such as backup reserve, and any local grid constraints.
Figure 2: Illustrative image of GridShare’s Site Optimisation.
Co-optimising behind-the-meter and grid services revenue opportunities with a scalable and flexible approach
On top of behind-the-meter optimisation, GridShare also optimises for revenues from grid services, allowing us to aggregate individual assets into VPPs to help keep the power grid stable.
Getting a fleet of devices “VPP ready” is a challenge and requires planning and support in various activities, such as IoT data connectivity, cloud data management and optimisation strategy. All these components are easily manageable through GridShare.
GridShare predicts the fleet’s aggregate power behaviour over a given time period, which can be used as a baseline for the grid service (if required). When GridShare receives a command to dispatch, it uses AI to provide flexibility dispatches most efficiently in terms of monetary savings or carbon emissions for the fleet as a whole.
GridShare’s main point of differentiation is that, instead of optimising at the fleet level only, the software can also take into account the behind-the-meter value delivery to the end customer. In fact, GridShare’s AI optimises the commands sent to fleet participants to ensure that flexibility is provided with minimal impact on individual assets and the customer’s bill.
As more devices join the aggregation, GridShare is still able to optimise per site while having the capability to aggregate and directly control thousands of batteries in people’s homes for energy market participation.
Flexibility markets vary considerably internationally, with variations in how markets are structured, the role of the different market players and the types of devices which can participate. One thing that doesn’t vary is the physical structure of the grids themselves, and by building GridShare to be the optimisation layer for the smallest component of this system, the behind-the-meter site, we can develop confidently on top of this. Thanks to its architecture, GridShare is able to be both scalable and flexible to different use cases and device types without having to redevelop core algorithms.
Figure 3: Illustrative image of GridShare’s Fleet Optimisation.
Our work in Europe and Japan
Grid services with residential participation are already a reality in Europe, and this market segment is also increasing its share in the global VPP market. Moixa is thrilled to be at the forefront of this innovation and be able to provide flexibility services on a worldwide scale.
In the UK, we are involved in several contracts to support the grid in collaboration with Distribution Networks Operator UK Power Networks. Our GridShare software provided grid services in a few different constraint areas across the country, such as in the Worthing and Littlehampton areas. Read our latest blog post to find out more.
In Japan, since 2018, we have partnered with ITOCHU, a leading Japanese trading house, to deploy their Smart Star ESS to customers in tandem with Moixa’s GridShare software. By leveraging our technology, ITOCHU has been able to deploy arguably the largest connected battery fleet globally, reaching the milestone of over 32,000 residential batteries connected to Moixa’s GridShare platform. In Japan, residential participation in grid services is expected to begin in 2024, with some trials due to be rolled out soon. We are ready to generate more value for the grid and for consumers when this reality arrives.
As more markets worldwide are opening up to flexibility services, we are thrilled to have the opportunity to deliver on a global scale and accelerate the transition to a low-carbon future by leveraging GridShare and adding significant value for customers, fleet owners, and grid operators.