Summary
Achieving the clean, secure, and reliable energy future we all want requires innovation in power generation and transmission. Equally as important will be the orchestration of these new technology innovations into a flexible, highly automated energy system with minimal waste.
Software solutions that combine the power of artificial intelligence (AI)-driven forecasting and predictive asset models into an optimisation solver can fix this challenge. The result, an optimised power generation plan communicated to hybrid equipment controllers, will optimise utilisation of an energy system.
Why does that energy system need to be optimised? Often these energy systems consist of power generating equipment like solar / PV, wind and thermal power with new complementary technologies such as battery storage and electrolysers (referred to as a hybrid energy system).
Subscribe to PII and receive:
Our long-established bi-monthly digital magazine *Process Industry Informer*, featuring exclusive articles, news and updates across the Process Industries, plus commentary from our three industry experts:
Sean Moran
Gavin Smith
Dave Green
PLUS:
- Process Industry Update — our weekly e-newsletter
- PII Podcast
- Special messages from our advertising partners
Figure 1 depicts this type of mixed-energy landscape. Because there are so many different types of technologies that make up a hybrid energy system the decision for how and when to operate the individual assets is extremely complex.
This is where an energy management solution comes into play. With an energy management solution, asset dispatchers can optimise how they run these assets in a clean, secure and reliable way while maximising profit and minimising emissions.
The dispatcher’s dilemma
Every day, energy asset dispatchers have an increasingly complicated decision to make, how can I plan the dispatch of my assets to satisfy market demand, but in the most profitable way? Depending on the asset type the answer to this question can be quite different.
For Dispatchers with dispatchable thermal assets (think combined-cycle power plants) this question is challenging due to the introduction of non-dispatchable assets – solar and wind – that create wider fluctuations in demand for dispatchable energy from traditional power producers. When solar and wind are available, they are prioritised on the grid.
This situation leads to shorter market intervals for dispatchable thermal assets. These Dispatchers now must access and understand weather data to know when neither wind nor solar will be available, and they will need to fill the generation gap in a cost efficient way.
There are several markets around the world where we see these trends take hold such as the California ISO, ERCOT, Australia, and parts of Europe to name a few. These types of markets experience a phenomenon known as the “duck curve” (See Figure 2).
In duck curve markets, we observe a year-over-year increase in energy supplied by renewables in the midday, but require dispatchable energy sources during the highest demand times in the morning and evening. When exactly will energy demand shift? And what is the capability of dispatchable equipment to produce at these times?
These are the questions that must be answered so that the dispatcher or asset manager can create the most profitable generation plan for the day ahead. An energy management solution can help the dispatcher understand their options, create cost-efficient generation plans and deliver rapid response plans when market disruptions occur.
In real customer use cases involving combined cycle power plants, we have seen fuel consumption reduction in the range of 0.5% to 5.0%. This reduction can equate to millions of dollars in fuel cost savings per year and slashing of CO2 emissions of up to 7,000 tons annually – this is equivalent to removal of removing over 1,300 cars from the road for one year.
How to make the most profitable choice
For non-dispatchable assets the dispatch question is still challenging but in a different way. Often, the primary obligation here is to satisfy power delivery agreements. These agreements help keep the power grid reliable, but to ensure these agreements are routinely satisfied requires the asset power capacity rating (i.e., maximum power output) to far exceed the required output.
For example, a power delivery agreement for 10 megawatts (MW) delivery during the day might require a solar PV asset with 50 MW maximum capacity (it’s a similar case for wind power). This is because the solar PV asset’s output depends on the intensity of sunshine which in not consistent. So only 10 MWs can reliably be delivered. But what do you do when the sunshine intensity is high, and the solar PV asset is producing near its full capacity?
We have several options such as routing power to battery storage for later use, participating in ancillary services, or delivering to an electrolyser that can produce hydrogen fuel for sale to an off taker. Making the most profitable choice is a complex economics optimisation question that requires well planned and insightful timing down to increments of only a few minutes.
To ensure operators of non-dispatchable assets have a generation plan and ability to respond quickly when we expect cloudy days, energy management market forecasting models will provide this plan. Going even further, the full power of an energy management solution is unlocked when the software is integrated with a hybrid controller, as depicted in figure 3, to automate the rapid response plan.
Energy management optimisation
Now let’s look at optimisation technique innovations in energy management that address the challenges for dispatchers and asset managers just described. Optimising predictive asset models and AI forecasting models are key to creating a highly automated and agile energy system.
Mathematical optimisation techniques and a mathematical solver are used to calculate economically optimised asset utilisation schedules and – if applicable for the use case – economically optimised market bids and (corresponding) power dispatch schedules. These calculations are based on the forecasts of availability of energy assets and unique business objective functions.
Optimisation techniques include:
- Mixed-Integer Linear Programming (MILP): Some decisions that need to be taken when optimising the use of assets in an energy system are of a discrete nature, such as the question whether to use a storage in its charging or discharging mode, or to switch an asset on or off. To this end binary (or integer) decision variables are introduced in the model. MILP theory and solver implementations extend their linear counterparts with cut and branching mechanisms to cope with discrete variables in otherwise linear models.
- Model Predictive Control (MPC): The optimal bidding decisions at electricity markets and with that the decision to turn an asset on or off, or to switch from charging to discharging of a storage asset, rely on future states of the system.
To incorporate the future behaviour into the MILP model, the time interval which covers the relevant future is divided into convenient smaller time steps. On each subinterval a structurally similar model is set up, with parameters related to the specific situation in the underlying subinterval.
These individual models are then linked together with state update equations or ramping restrictions, etc. This approach is called Model Predictive Control (MPC). The set points in its solution indicate how to control the system in the current and considered future time intervals.
Usually only a subset of these set points must be communicated to the components or markets, since set points further in the future might become irrelevant due to changing forecasts or unpredicted state changes. By using repeated solutions with a rolling horizon, optimisation can adapt to such changes and improve the solution of earlier calculations concerning later decisions. - Stochastic Programming: The volatility of renewable resources and the uncertainty of market price forecasts must be addressed. This can be accomplished by applying a scenario tree approach and a corresponding optimisation time structure.
This amounts to a flexible multi-stage stochastic optimisation model: each new branch introduces a new stage, in which several possibilities are elaborated and considered, e.g., after a market closure time it is reasonable to distinguish the availability of renewable energy sources between a high, medium and low scenario. In this way, the bids can be appropriate for all three of these situations. This fits well together with the capability of the forecasts to produce several possible realisations.
Together these optimisation techniques allow for rapid multi-dimensional or multi-branch scenario simulations and when applied to energy systems, result in day-ahead and short-term future generations plans that are economically optimised to the plants circumstances.
Once the generation plan is in place, the communication with hybrid controller(s) can take stage with setpoints for only the appropriate time frames being executed and updated as circumstances change.
A hugely important role in the future of energy
Regardless of the energy asset being operated, we hope this article makes clear why energy management will have a hugely important role in the future of energy. The solution that we have developed is Omnivise Energy Management and has three parts: First, the AI-driven forecast models, second, the high-fidelity predictive asset models and third, sophisticated optimisation techniques and automation.
These techniques result in an optimised dispatch plan that will have serval benefits depending on the type of energy system. Such as minimising fuel costs and emissions, maximising the economic utilisation of system assets, automating the bidding strategy with optimally prices energy bids, and automating the dispatch with hybrid controller integration.
Omnivise Energy Management will help dispatchers and asset managers orchestrate their energy systems. Thanks to that orchestration they can now operate more profitably, minimise waste, and use carbon-free assets in a way that opens new revenue streams.
These are the reasons why we see Omnivise Energy Management as important as any other innovation in the energy sector. It helps make use of all the equipment innovations and turn their outputs into a symphony of energy assets working together to create the clean, secure, and reliable energy future our world is working so hard to achieve.
Darren McGuirre, Digital Business Development Manager, Siemens Energy Orlando
Darren is a member of the Siemens Energy Omnivise Performance team and leads business development activities for Omnivise Energy Management… He has been with Siemens Energy for 15 years with special focus on power generation and digital solutions. He also holds a masters degrees in both engineering and business administration from the University of Florida.
Katie Hanley, Global Head of Omnivise Performance, Siemens Energy Washington DC
Katie leads the Siemens Energy Omnivise Performance team where she and her team create digital products that improve energy management and asset performance, resulting in optimised energy systems, reduced emissions, and increased generation. Katie has over 15 years of experience building and deploying data science and digital solutions for Fortune 50 companies and the US Federal Government. She graduated from Purdue University (Bachelor’s degree) and University of Missouri in St. Louis (Master’s degree).
