This is my final master thesis project that I worked on alone for my Master of Science (Msc) in Computer Science and Informatics at Roskilde University. The subject area for the thesis was Computer Science focusing on exploring new innovative solutions for green it and sustainable energy. The project duration was 6 months starting in June 2021 and ending in January 2022 with a work input of 480 hours worked.

My master thesis was quite an adventure where I went really deep on a single technical topic in the energy sector, that I called digital twin data modeling with internet of things streaming telemetry for peak-shaving in virtual power plants.

Research problem

My research hypothesis was to test if battery-storage devices can be used to level out peak-load periods in the energy sector. Peak-load periods are short time-periods where electricity is in high demand for example in the evenings when most people are in their home making dinner, washing clothes etc. Peak-load periods are currently very expensive due to challenges with meeting the peak demand for electricity. New innovative research in software and hardware solutions for sustainable energy are needed to solve and mitigate this problem.

Potential solution

My job was to explore new innovative solutions for this problem. Battery storage devices combined with sustainable energy (wind and solar) is a solution that can potentially help offset peak-load periods by discharging electricity from the battery when electricity is needed most, which results in reduced cost for both producers and consumers of electricity. The battery can recharge when electricity production is high for example when the wind is blowing or the sun is shining.

Methodology

My methodology for the project was an experimental computer science approach that I used to explore new innovative solutions through the construction and evaluation of a virtual power plant prototype system. A virtual power plant is a virtual representation of a power plant, that can aggregate energy from a large set of smaller heterogeneous energy devices such as residential and commercial battery-storage devices, electric cars, smart thermostats etc, to provide peak-shaving solutions by using the aggregated energy at peak-load periods or trading it on the power markets if needed.

I used computer simulation as a method to simulate battery-storage devices, which I used to evaluate my prototype system by experimenting with large quantities of simulated streaming telemetry data.

Results

The virtual power plant prototype consisted of a microservice based data platform built with state of the art technologies such as Apache Kafka, to provide high throughput, scalability, permanent storage and high availability. I combined Apache Kafka with actor model programming in Scala and Akka to leverage actors, advanced streaming capabilities and following the reactive principles to design a digital twin prototype to provide real-time insight into battery-storage devices for peak-shaving use-cases.

I ended up collecting and aggregating weather data and power market data that I used to train an artificial neural network for electricity price forecasting to find out when future peak-load periods would take place. I then developed algorithms for scheduling battery-storage devices to reduce peak-load periods based on forecast.

My results showed that a virtual power plant system can be built with state of the art technologies such as Apache Kafka and actor model programming, and it is feasible to use electricity price forecasting and battery-storage scheduling algorithms to level out peaks.

You can check out combotto.io if you are interested in learning more about this project.