Self-organising Thermal Operational Resource Management

Informacje na temat projektu

STORM

Identyfikator umowy o grant: 649743

DOI

10.3030/649743

Projekt został zamknięty

Data podpisania przez KE 5 Lutego 2015

Data rozpoczęcia 1 Marca 2015

Data zakończenia 31 Marca 2019

Finansowanie w ramach

• SOCIETAL CHALLENGES - Secure, clean and efficient energy

Koszt całkowity

€ 1 972 125,94

Wkład UE

€ 1 972 125,94

Koordynowany przez

VLAAMSE INSTELLING VOOR TECHNOLOGISCH ONDERZOEK N.V.
Belgium

Periodic Reporting for period 3 - STORM (Self-organising Thermal Operational Resource Management)

Okres sprawozdawczy: 2017-10-01 do 2019-03-31

Podsumowanie kontekstu i ogólnych celów projektu

The STORM project tackles energy efficiency at district level by developing an innovative district heating & cooling (DHC) network controller. The project partners have developed the STORM controller based on self-learning algorithms and is demonstrated, implemented and evaluated in 2 demo sites, Mijnwater BV in Heerlen (NL) and Växjö Energi in Rottne (SE).

The technology consists of different control strategies: peak shaving, market interaction and cell balancing. Peak heat reduction tests led to a long-term peak heat reduction of 12.75% on average compared to the reference scenario without the STORM controller. The Market Interaction strategy is a strategy that uses both charging and discharging capabilities to adapt to a set of electricity spot prices. Based on these prices, the STORM controller moves heat demand to match spot prices, thereby ensuring heat delivery and comfort. This strategy resulted in a 15% reduction on the electricity purchase price and an overall electricity procurement costs reduction by 6%. This option of the controller is beneficial for electric systems such as heat pumps and cogeneration units, especially when sufficient thermal buffering is provided in the system, making it possible to charge energy independently of the energy demand at times when the electricity price is most favourable. For the cell balancing strategy in the Mijnwater system the controller was able to reduce the flow over the entire test period without jeopardising the energy delivery to customers. A peak shaving potential of 17.3% could be determined here. Furthermore, an improved capacity could be derived ranging from 37% up to 49% (median value 42.1%) which corresponds to a total of 48,200 normative Home Equivalents (nHE) that can be additionally connected to the existing system.

The project consortium succeeded in an implementation of the STORM controller technology in the district heating network of Ennatuurlijk, Eindhoven, Furthermore, within the framework of the Interreg project Heatstore, the STORM controller technology is implemented in the deep geothermal project in Mol, Belgium. Commercial deployments are explored with Veolia and Dalkia as large district heating companies. So practical applications of the STORM controller technology are realized and the involvement of industry engaging them into the STORM controller technology was addressed. All demonstrations have already now delivered highly relevant lessons for further market roll out of the technology. The STORM project, associated with this dedicated hashtag, was hence successfully positioned as one of the projects in the forefront of digitalisation in heating and cooling and contributed to the impact.

Prace wykonane od początku projektu do końca okresu sprawozdawczego oraz najważniejsze dotychczasowe rezultaty

The main achievements for the project are:
The algorithms for the STORM controller were developed, demonstrated and evaluated and implemented on the 2 demo sites.
The features peak reduction, cell balancing and market interaction were developed and implemented on the NODA smart heat platform.
The algorithms were implemented on the commercial NODA platform and tested in the demonstration sites.
Reference data was analysed from VEAB and NODA as well as from the Mijnwater PRIVA operating system.
The performance of the STORM controller was determined for the 2 demo sites and for the 3 features.
Commercial deployments are explored with Veolia and Dalkia as large district heating companies.
So practical applications of the STORM controller technology are realized and the involvement of industry engaging them into the STORM controller technology was addressed.
All demonstrations have already now delivered highly relevant lessons for further market roll out of the technologies as part of the replication plan that was made.
VITO and NODA are continuously looking at the market potential of the controller and how the controller can be exploited on a wider scale.
Activities as licensing, setting up of a spin off company, third party licensing, etc. are investigated.
Many commercial interest were identified in relation to the STORM technology and the digitalisation of the DHC system.
A joint ownership agreement was signed in 2016 by the partners NODA and VITO ensuring their long term cooperation.
Dissemination activities were successfully continued and most KPIs were achieved.
The project results were presented at 20 national and international conferences and events reaching more or less 100,00 users.
The objectives of the STORM controller is still relevant for the district heating and cooling community.
A new H2020 project was granted and continues on the STORM work and focuses on low temperature district heating network and the faults that occur in these networks and substations.
This ensures a continuation of the STORM technology for the following years.

Innowacyjność oraz oczekiwany potencjalny wpływ (w tym dotychczasowe znaczenie społeczno-gospodarcze i szersze implikacje społeczne projektu)

The work done in the STORM project is related to the development and implementation of a district heating and cooling controller based on self learning techniques. Heat demand forecasting is in one form or another an integrated part of most optimisation solutions for district heating and cooling (DHC). Since DHC systems are demand driven, the ability to forecast this behaviour becomes an important part of most overall energy efficiency efforts. The prior work of the STORM partners, based mainly on certain decision tree based regression algorithms, is expanded to include other forms of decision tree solutions as well as neural network based approaches. These algorithms were analysed both individually and combined in an ensemble solution.

The primary purpose of this project is to implement online machine learning forecasting algorithms in an industrial, fully operational real-time environment using actual weather forecasts as input for the system. Basically to run the system as it would have been used in an operational environment. A selection of three algorithms was used to perform the tests, one of which is similar in construction to the algorithm presented in the previous work. Based on the results from the previous study only aggregated data was used to construct models (second approach). This also means that data such as meter data from production facilities can be used as input for the training. The first algorithm used is the Extra-Trees Regressor (ETR), which uses randomized decision trees in relation to sub-sets of the dataset in combination with averaging to increase predictive accuracy and to minimize over-fitting. The ETR algorithm is the one similar to algorithms used previous work. The second algorithm is Extreme Learning Machines (ELM) which is a feed-forward neural network used for regression analysis. An ELM uses a single layer of hidden nodes with randomized weights assigned to the input to the hidden layer. The third algorithm is an expansion of the ELM algorithm, in which a regularisation factor was added to prevent over-fitting of the training data. These algorithms were implemented in the demo sites.

It is shown that the results are in line with expectations based on prior work, and that the demand predictions have a robust behaviour within acceptable error margins. Applications of such predictions in relation to intelligent network controllers for district heating are explored and the initial results of such systems are discussed.

Final results of the project

Results of the STORM self learning algoritms in Rottne