The session “PowSyBl AC Optimal Power Flows” at LF Energy Summit 2024, presented by Nicolas Omont of Artelys, delved into the intricacies of nonlinear optimal power flows (OPF) and their application to power systems. Omont provided insights into the challenges and opportunities associated with solving AC (alternating current) OPF problems, contrasting them with simpler, linearized DC (direct current) OPF approaches. Key takeaways from the session follow (full video is at the end):
- What is OPF? OPF is a mathematical approach used to optimize the operating conditions of a power system, minimizing costs while adhering to system constraints such as power balance, voltage limits, and transmission capacities. By balancing these constraints, OPF ensures the system runs efficiently and safely.
- AC vs. DC Optimal Power Flows
- AC OPF retains the nonlinear nature of power systems, making it crucial for managing aspects like voltage control. It’s more accurate but computationally demanding.
- DC OPF, on the other hand, simplifies the problem by linearizing the power flow equations, making computations faster but at the cost of precision, particularly in managing voltage. Omont focused on the AC OPF during his talk, discussing how it better reflects real-world system behavior, particularly in voltage management.
- The PowSyBl Framework and OpenReac PowSyBl (Power System Blocks) is a modular framework developed to simulate and optimize power systems, and it includes tools for both AC and DC OPF. Omont introduced OpenReac, an open source AC OPF tool within the PowSyBl framework.The primary function of OpenReac is to initialize the voltage plan during grid simulations. When forecasting grid states, this tool optimizes settings like generator set points and reactive power sources to provide realistic operational data for further analysis. It also helps minimize system losses.
- Customization and Flexibility The OpenReac tool is highly customizable. Users can adjust objectives (e.g., minimizing losses, optimizing voltage set points) and add new constraints. This flexibility is enabled by a solid foundational model written in AMPL, which interacts with nonlinear solvers like Artelys Nitro. There are also ongoing efforts to integrate open source solvers such as those based on Julia and JuMP to enhance flexibility while ensuring the tool remains accessible.
- Performance and Real-World Application In terms of performance, Omont shared that OpenReac can optimize a French IGM (Interconnected Grid Model) with 7,500 nodes in just over 10 seconds. While performance isn’t the primary concern, this level of efficiency is suitable for tasks like voltage initialization, even in larger grids.
- Future Roadmap Looking ahead, the plan is to move away from proprietary software components like AMPL, focusing on open source alternatives such as Julia/JuMP. This will help keep the entire AC OPF process open source, enhancing transparency and collaboration within the power systems community. Omont also emphasized the potential for further integration of other nonlinear solvers to provide additional flexibility and performance options.
- Prototypes and Ongoing Work In addition to OpenReac, Omont highlighted prototypes such as a Divergence Analyzer and a State Estimator, both designed to enhance the accuracy of power networks by identifying inconsistencies and detecting errors in measurements and topology. While these tools are still in development, they demonstrate the potential for OPF to be used in more advanced grid analysis tasks.
