If you are familiar with the energy markets, then you know the importance of balancing on the gas and power grids. We’ll focus shortly on gas below and then move on to power. This blog will mainly focus on the benefits of forecasting for the power market, since the biggest gains can be made there.
For gas, system operators have a responsibility to keep the pressure on their network of pipelines and storage sites within safe and acceptable limits. Too much incoming gas will increase the pressure, too much outgoing gas will decrease the pressure. For this reason, shippers & market participants who are long or short on the network will have to pay the system operator a fee in most countries, the ‘imbalance cost’.
Compared to electricity, gas moves more slowly through the system. This has drawbacks but also benefits. The TSO (Transmission System Operator) has more time to act and prevent imbalances at certain points of the grid. Another difference is the fact that large quantities of (liquefied) natural gas can be stored much easier than power. Finally the production and supply of natural gas is easier to regulate and predict than for power, especially when taking into account renewables.
A power TSO is responsible for managing the security of the power system in real time and co-ordinate the supply of and demand for electricity, in a manner that avoids fluctuations in frequency or interruptions of supply. They are the end-responsible for the whole grid.
When a market participant is long or short on the grid during a certain time, they have to pay imbalance costs. You can see the imbalance prices on the Belgian Power grid for the 15-min slots in real-time here: https://www.elia.be/en/grid-data/balancing/imbalance-prices-15-min
For Power and Gas it is important to have accurate forecasts. There are many models and solutions out there that provide forecasts. There are also many applications of forecasting solutions and the following examples are the ones we see most often in the industry:
- Consumption forecasts: Demand or consumption forecasting is one of the most essential tasks in the gas & power markets.
- Production forecasts: Using weather data and other variables to predict solar production, wind production, gas production rates, etc.
- Price forecasting: Using various methods to predict the price level of a certain commodity for various timeframes.
- Load forecasting: Load forecasting remains a challenging problem in power system operation due to growth in low carbon technologies and distributed small scale renewable generation.
- Anomaly detection: especially for production assets this is important
- Maintenance prediction for physical assets such as wind turbines
How do we incorporate forecasts in egssPort?
We allow our clients to import their forecasts straight into EgssPort Gas & Power, our SaaS platform for scheduling, balancing, nominations, shipping, and managing all operational aspects.
There are several ways we allow this:
- Our clients use their own forecasting models & tools and export the data in a supported format – then they import these files into EgssPort on the related screens
- The second option is that they use one of our preferred partner solutions which is integrated via API with EgssPort. This allows clients to automatically import and run the forecasts for the desired variables.
One of our preferred partners is Tangent Works, with their Tangent Information Modeler (TIM).
The benefits of our integration with TIM is that our clients can interact with TIM from within the EgssPort platform! They don’t need to log-in to another platform to create forecasts. Even (re)training the model happens within EgssPort. Below you can see a screenshot:
We also have APIs set-up with weather data providers, to import & visualize a range of data such as temperature, cloud coverage, wind speed, and global radiation. As you can see below:
We currently integrate with several weather data providers such as Meteomatics, MeteoGroup, etc.
What makes TIM different from other forecasting solutions?
In a nutshell: TIM, Tangent Information Modeller, is a tool that allows one person to do the work of multiple Data Scientists. Thanks to instant machine learning and information geometry, it can come up with forecasting models automatically and in seconds instead of days.
Tangent Works has developed the next step in Machine Learning (ML), improving greatly upon hand-crafted models and Automatic Machine Learning (AutoML). TIM actually moves into Instant Machine Learning (InstantML) territory, and even Real-Time Instant Machine Learning (RT InstantML).
The main benefits of this approach:
- no data engineers & data scientists required: these profiles are in high-demand, and even if you find them they can have a large impact on the payroll of a company. However, the experience from these profiles and the technology of TIM can supplement each other and create valuable insights in your data.
- no ‘heavy IT’ infrastructure required: TIM does not require high-end hardware and runs well on a laptop or even on a mobile device. Nevertheless, the cloud architecture of TIM is designed to scale whilst using computational resources in a cost-efficient way.
- many possible use cases: TIM offers a solution for all your forecasting needs, we will explain this last point in more detail below
The core of TIM is designed for generic time-series forecasting. In the energy industry, TIM covers most sub-domains, i.e. electricity load, gas consumption, district heating and cooling, solar production, wind production, price forecasting, and others using the default pre-built dictionaries.
Traditionally, specialized solutions and models are developed to address these diverse forecasting challenges. It is often the case that within a company there are several solutions from different vendors in use to cover e.g. electricity load forecasting, solar and wind, and price forecasting. TIM unifies this traditionally siloed view – and offers a one-stop-shop for all your forecasting needs!
Find out more about TIM here: https://www.tangent.works/how-to-reduce-imbalance-costs-with-automatic-forecasting-2/