Optimization of a Li-Ion Storage System for the Integration of Renewable Energies in Industrial Processes
|Aushang||DW Aushang MA LiIon Optimization.pdf|
|Betreuer||Wenden Sie sich bei Interesse oder Fragen bitte an: |
Global warming is one of the big challenges in this century. To minimize the rise of global temperatures and the potentially devastating consequences, CO2-Emissions must be reduced. As the largest consumer of electrical energy in Germany, the industrial sector offers huge potential to utilize green, sustainable energy. However, the supply from renewable energy sources is fluctuating, because of strong dependencies on the weather. Electrical Energy Storage Systems (EESS), such as batteries, store energy to stabilize the supply from renewable energy sources. The dimensioning and the operation of EESS is focus of current research.
During the process of generation, of transmission, of storage and of consumption of electrical energy, a huge amount of data is collected. This data is usually time series data from measurement devices such as smart meters. With the evolution of smart grids, the expected amount of available data increases. Novel and efficient data analytics methods provide important information on the electrical systems, which engineers can then use.
In Model-Driven Engineering (MDE), models of a software system are applied during the development process. The objective is to predict how the design influences different quality dimensions of the system. For example, a large-scale data analytics application might need much less resources when the data is reduced from measurements per second to aggregated fifteen minute intervals, while still providing a satisfactory result for the application at hand.
The focus of this thesis is to model the effects of data granularity on the results of an optimization problem for Electrical Energy Storage Systems in an industrial environment and to build a predictive model of this effect that can be used in MDE.
The work will be done in close cooperation with the Institute of Data Processing and Electronics (IPE). The IPE operates a factory at the Campus North for specialized electronics and sensors. The factory is equipped with Smart Meters to monitor the energy consumption on the machine level.
In particular, the following aspects should be investigated:
- How do different data aggregation levels affect the optimization algorithm (quality vs. runtime)?
- Can the tradeoff between data quality and system performance be expressed in a predictive model?
For this thesis, you will:
- Extend an existing optimization problem to find the optimal charge/discharge profile of battery with multiple constraints (battery capacity, maximum charge/discharge power, …)
- Run experiments on aggregated data while monitoring the utilization of system resources and evaluate the trade-off between system utilization and optimization results.
Our technology stack builds upon modern data procession frameworks such as Apache Cassandra and Apache Spark. Experimental evaluation can be run on a cluster with 512 GB RAM and 48 Cores. In this thesis, you gain deep insight and knowledge on large scale data analytics.