Discovering the data landscape of electricity balancing

Abstract (3:50 mins)

1. Introduction

Electricity systems are increasingly becoming decarbonised and decentralised. Data and data-driven solutions will play an important role in catalysing the transition. This article aims to discover the data that matter in the context of electricity balancing. A follow-up article will explore the relevant data-driven solutions.

Electricity balancing: The tradition of matching electricity generation and demand based on plans made months-ahead and tweaked until the very moment. GIF courtesy (source).

This investigation is based on the premise that, in a highly decarbonised and decentralised future, electricity balancing will be predominated by the following participants, who own one or more distributed energy assets that can generate and/or consume energy:

  • Renewable generators
  • Energy storage systems
  • Electric vehicles
  • Large prosumers
  • Small prosumers

Due to their interplay and the many uncertainties associated, electricity balancing is expected to become an incredibly complex process (not that it is any less at the moment). Understanding the relevant data and the aspects related to their availability and accessibility will help develop data-driven solutions that benefit these participants as well as improve the electricity grid's reliability. Some of the questions that motivated this investigation are:

  • What distributed energy assets are owned by the above mentioned participants?
  • How do they currently participate in electricity balancing?
  • What factors influence the operations of these assets?
  • How are the operations of the assets recorded? Who owns these data?
  • What data are relevant, but not being recorded yet?

Let us try finding these answers for each of the outlined participants.

2. Distributed energy assets and the data that matter for electricity balancing

2.1. Renewable generators

This has been used as a generic term to refer to any large-scale renewable energy based electricity generation. Since wind and solar generation technologies take the major share in global renewable energy capacity (excluding hydro) [IRENA], the diagrammatic representation of a renewable energy farm given below is understandably biased.

Electricity generation from wind turbines primarily depends on wind speeds [DWIA]. That from solar panels depends on solar radiation, influenced by factors such as temperature and wind speeds [Malvoni et al]. By and large, weather is a major influencing factor that dictates the electricity generation in these cases. Consequently, electricity planning into the future should account for the forecasted weather conditions. Since the wind turbines and solar panels generate electricity and deliver this to the grid, it is also important to collect the operations data such as the energy generated, health of the asset, among others.

Here, the energy assets (wind turbines, solar panels) are shown in blue and the associated data (weather, operations) are shown in red. The 'candy' like symbols represent the devices that monitor and record the operations of the energy assets. When connected to the internet, these devices would become part of the Internet of Things (IoT).

While weather data can be collected from the nearby meteorological stations or from the on-site weather monitoring systems (if any), operations data are usually recorded by the renewable generator owner/operator.

2.2. Energy storage systems

Over the past few years, many large-scale energy storage technologies have proven themselves. Some of these even make the best use of gravity for energy storage. They are capable of addressing the electricity imbalances in the dual role of a generator (export energy) as well as a load (import energy). Such systems can be directly connected to the grid as standalone units or co-located with renewable energy farms / buildings. Among the proven large-scale (kilowatts and above) energy storage technologies, Lithium-ion (Li-ion) batteries have high energy densities and can respond to imbalances in sub-seconds. The fast response capability makes them lucrative for delivering frequency regulation/response. The plummeting cost of Li-ion technology is encouraging many developers to invest in it. As a result, Li-ion batteries are leading the pack of large-scale energy storage systems with over 90% market share [EESI].

For representation, the sketch above shows a containerised battery energy storage system. The energy asset here is the entire system itself that comprises of many tiny battery modules. Weather factors such as temperature influence the battery's performance [Ma et al], that in turn affect its energy delivery or consumption. The battery energy storage come with battery management systems (BMS) that keep track of their operations and collect data such as the power exported/imported, state-of-charge, voltage levels, current levels, cell among others. Hence additional devices are not necessary for monitoring and recording their operations. These data are essential to drive decisions that ensure reliable participation of battery energy storage systems in electricity balancing.

2.3. Electric vehicles

The future of mobility is expected to be electricity based. Electric vehicles (EV) largely come with small-scale (<= kilowatts) Li-ion batteries. In aggregated forms, they can represent a large-scale energy storage system - imagine 1000 Nissan Leaf cars charging @6.6 kW simultaneously, adding an abrupt load of 6.6 MW to the grid. While charging is the most common behaviour, trials on vehicle-to-grid (V2G) discharging are proceeding successfully. If EV charging/discharging can be aggregated and intelligently managed, we could achieve a well balanced and truly decarbonised electricity grid. Some examples would be: charging during a period when renewable generation is in excess and discharging when a thermal power generator fails.

In either scenarios, battery operations data recorded by the battery management systems are critical. Another unique factor here is the behaviour of the car user who decides when to charge (or to discharge). While it is not easy to determine what triggers these behaviours (personal calendar, battery state-of-charge, availability of chargers?), it is possible to keep logs of the charging/discharging operations using intelligent devices attached to (or embedded into) the electric vehicle charging points. These data are indeed valuable for electricity balancing decisions.

2.4. Large prosumers (consumers who also produce)

The commercial and industrial buildings with distributed energy assets such as on-site generators, battery energy storage systems and/or flexible loads, with the capability to 'produce' and not just 'consume' are referred to as large prosumers here.

Large prosumers may have on-site renewable generators such as small wind turbines and/or solar panels (encouraged by their declining costs). Combined heat and power (CHP) generators are also being increasingly adopted since they generate both heat and electricity. The excess electricity generated can be stored in batteries or sold into the grid as distributed generation. Battery storage systems can backup the building load and hence take that load off the grid, while getting incentivised. Flexible loads such as heating, cooling or lighting units can also help earn additional revenue for the building owners. The use of such distributed energy assets for electricity balancing is already being facilitated and rewarded through demand response programs in markets such as the United States, the Great Britain, Continental Europe and Oceania [Paterakis et al].

As discussed earlier, weather factors influence the wind and solar electricity generation. Operations data from the CHP units may need to be recorded using additional devices. Battery energy storage systems are usually equipped with BMS that collect the operations data.

Weather factors such as temperature, humidity, solar radiation and wind speeds influence the loads such as heating, ventilation, air-conditioning and cooling (HVAC). In addition to weather, behaviour of the building occupants also impact building energy consumption. For instance: a) a team meeting called inside an office building may prompt the people to switch on additional lights, b) an upcoming festival season may encourage a factory owner to ramp up production processes to meet the expected demand, c) decline in demand of a certain perishable product may force a cold storage building to store that for longer periods, hence running the cooling loads at their peak. These factors add a lot of unpredictability to building energy consumption.

While it is hard to monitor human behaviour, sensors for monitoring head count and carbon dioxide levels may provide useful proxy data. The human actions also reflect in the energy consumption patterns recorded using smart meters. An even better solution would be to collect sub-metering (load specific) data. Unfortunately not all buildings are equipped with such smart monitoring technologies yet. This is a major gap that could hinder the development of data-driven models for enhancing the large prosumer participation in electricity balancing.

2.5. Residential prosumers

Similar to large prosumers, the residential prosumers are also expected to have energy assets such as small wind turbines, solar panels and battery energy storage systems (small-scale). The surplus energy can be delivered into the grid and in aggregated scales utilised for electricity balancing. The relevant datasets include weather data, energy asset operations data and user behaviour data.

A thought when it comes to residential prosumers: if all the required energy is generated, stored and consumed on-site, and the incentives from grid balancing are not substantial enough, will they go completely off-grid? In that scenario, would it be worth tapping into their data for grid balancing?

3. Symphony of the distributed energy assets

When the distributed energy assets are physically connected to the grid, the whole act of electricity balancing appears to be a coordinated symphony of generation and consumption actions. As represented in the sketch below, the electricity market plays the role of a symphony maestro who directs the simultaneous operation of the energy assets to ensure that the grid is stable. A lot of planning goes into deciding which asset should do what. If any one of the assets fail, backup measures are taken. Looking back, the assets are rewarded or penalised based on their commitments and delivery. All the data related to these actions are categorised as the electricity grid/market data here. In most of the electricity markets, these data are recorded by specialised parties and managed by the system operator. While many of the datasets are open for public access, a lot remain hidden in deep vaults.

So, what does the data landscape of electricity balancing look like? In the sketch below, let's say the red circles around the participants represent the data that are relevant to their involvement in electricity balancing. As already investigated, these are largely comprised of the weather data, the energy asset operations data and the user behaviour data in some cases. The red circle in the centre, around the 'physical' electricity grid and the 'virtual' electricity market, refers to the grid/market data.

4. Concluding thoughts

The article presented a route to discovering the data landscape of electricity balancing. The weather, energy asset operations, user behaviour and grid/market data are identified as the main components of this landscape. This is indeed a very broad categorisation and would need further refining. For the participants to gain the most out of their distributed energy assets in electricity balancing, it is necessary to tap into these data sources first.

Going further, they need access to a layer of applied intelligence that helps derive insights out of these data. These insights may also need to be operationalised such that business continuity is maintained. This is one of the value propositions that Dtime aims to deliver and will be discussed in a follow-up article.

In the meanwhile, please do share your thoughts about the data landscape of electricity balancing from your viewpoint in the LinkedIn comments section.

Author: Dr. Gautham Krishnadas

Share this page