Practical Deterministic Optimization: Profit-Optimized Battery Scheduling

Summary

This experiential workshop will focus on the critical link between Data Science (DS) and deterministic optimization. We will explore how models are developed and specifically how to translate complex, real-world operational requirements into solvable optimization models. 

The core interactive problem will involve creating an optimal scheduling model for battery management to maximize profit based on day-ahead pricing. This approach is essential for applications like PV forecasting and grid flexibility management. The target audience includes individuals— of varied prior knowledge—who seek to master the skill of modeling complex conditions using linear optimization.

Syllabus

The workshop will cover:

1. The importance of optimization for Data Science, defining data, and model development.

2. Introduction to optimization theory, what methods exist, and what Abelium uses (including pros/cons).

3. Presentation of a simple linear model for battery scheduling.

4. Interactive application where participants add complex constraints (e.g., specific charge/discharge windows) to the simple model using techniques like Big M or new variables.

5. Group review of outputs and a debrief explaining the theory behind modeling complicated conditions into simple linear formulations

Instructor

Nevena Pivač holds a PhD in Mathematics with a specialization in algorithms, graph classes, and combinatorial optimization. She is currently employed as a Researcher and Developer at Abelium d.o.o. Her expertise includes research into graph theory, minimal separators, fair allocation problems, and the complexity of methods and problems. She is also involved in the European Space Agency (ESA) funded project EO x Grid, focusing on AI-driven PV forecasting for grid optimization.

Attendee equipment prerequisites

Participants should bring a laptop to fully engage in the interactive modeling exercises.