Model predictive control for wind power gradients

Tobias Gybel Hovgaard, Stephen Boyd, John Bagterp JørgensenWind 2014 · DOI

TL;DR

This paper presents a novel convex formulation for optimizing wind turbine power output to meet strict grid code requirements for power gradient limits. By cleverly reformulating the problem to focus on power flows and utilizing turbine inertia as an energy storage mechanism, they enable efficient, global optimization and demonstrate significant improvements in power smoothing using real-world wind data.

What Problem Does This Paper Address?

The increasing penetration of wind power into electricity grids presents a challenge due to its inherent variability. Grid operators are imposing stricter "grid codes" that limit how quickly wind power output can change (the power gradient). This is crucial for grid stability, preventing cascading failures, and allowing grid operators to manage supply and demand effectively. Traditional wind turbine control strategies often struggle to meet these stringent, rapidly evolving requirements, especially when dealing with fast fluctuations in wind speed. The paper addresses the need for a control strategy that can actively manage wind turbine power output to smooth these fluctuations and adhere to power gradient constraints, while minimizing the loss of energy production.

Methods

The core innovation lies in transforming a complex, non-linear optimal control problem into a convex one. The authors achieve this by:

  • Novel Change of Variables: Instead of directly controlling physical parameters like pitch and torque, they focus on power flows (generator power, grid power, charge/discharge rates).
  • Utilizing Turbine Inertia: They treat the wind turbine's rotational inertia as a form of distributed energy storage, allowing it to absorb or release energy by slightly varying its rotor speed. This complements a central battery or other storage device.
  • Convex Formulation: Through these changes, the system dynamics become linear, and the constraints and objective functions become convex (or concave for maximization). This is a significant breakthrough because convex optimization problems can be solved globally and efficiently using specialized solvers.
  • Model Predictive Control (MPC): The optimized control strategy is implemented in a receding horizon fashion. At each time step, the controller predicts future wind speeds and solves the convex optimal control problem to determine the best control actions for the immediate future, then implements only the first step and repeats the process.

Datasets

The paper utilizes real wind speed data measured at the Høvsøre test site in Denmark in 2004. The wind speed predictions used in the MPC controller were generated using modern continuous-time formulations and spline basis expansions, drawing on upstream wind speed information.

Key Insights and Ideas

  • Turbine Inertia as a Free Resource: The paper highlights that the inherent inertia of a wind turbine can be actively leveraged as a form of energy storage, reducing the need for expensive external storage solutions.
  • Power Flow Focus Simplifies Control: Shifting the control focus from physical actuators to power flows simplifies the problem and makes it amenable to convex optimization techniques.
  • Convexity is Key to Global Optimality: By transforming the problem into a convex one, the authors ensure that their solution finds the true global optimum, rather than getting stuck in local optima, which is a common issue with non-convex problems.
  • MPC for Grid Code Compliance: The MPC framework, combined with the convex formulation, provides a robust way to dynamically adjust turbine operation to meet strict, time-varying grid code requirements.

Key Results

  • Effective Power Gradient Control: Simulations using real wind data demonstrate that the MPC controller can effectively reject disturbances from fast wind speed changes, ensuring power gradients are limited to ±3% of rated power per minute.
  • Minimal Energy Loss: The proposed method achieves this power smoothing with an insignificant loss in total energy production (e.g., 0.57% and 0.42% reduction in two scenarios).
  • Reduced Storage Requirements: Utilizing turbine inertia can reduce the required capacity of external energy storage by up to 30% without sacrificing power output.
  • Improved Power Output in Overspeed: In scenarios where wind speed exceeds the rated speed, the MPC controller can even increase power output by up to 2% by leveraging kinetic energy storage.

Limitations and Future Work

  • Approximation of Power Function: The paper acknowledges that the available power function is nearly concave and uses piecewise linear approximations. While shown to be accurate, this is still an approximation.
  • Focus on Single Turbine: The current work focuses on a single wind turbine. Future work could extend this to coordinated control of wind farms.
  • Uncertainty in Wind Forecasts: While the MPC uses wind forecasts, the paper doesn't deeply explore the impact of forecast uncertainty on the controller's performance. Robust MPC or stochastic MPC approaches could be investigated.
  • Economic Optimization: The paper mentions economic MPC and maximizing profit as a goal. Further work could delve deeper into the economic implications and trade-offs in more complex market scenarios.

Summary updated

Abstract

We consider the operation of a wind turbine and a connected local battery or other electrical storage device, taking into account varying wind speed, with the goal of maximizing the total energy generated while respecting limits on the time derivative (gradient) of power delivered to the grid. We use the turbine inertia as an additional energy storage device, by varying its speed over time, and coordinate the flows of energy to achieve the goal. The control variables are turbine pitch, generator torque and charge/discharge rates for the storage device, each of which can be varied over given ranges. The system dynamics are quite non-linear, and the constraints and objectives are not convex functions of the control inputs, so the resulting optimal control problem is difficult to solve globally. In this paper, we show that by a novel change of variables, which focuses on power flows, we can transform the problem to one with linear dynamics and convex constraints. Thus, the problem can be globally solved, using robust, fast solvers tailored for embedded control applications. We implement the optimal control problem in a receding horizon manner and provide extensive closed-loop tests with real wind data and modern wind forecasting methods. The simulation results using real wind data demonstrate the ability to reject the disturbances from fast changes in wind speed, ensuring certain power gradients, with an insignificant loss in energy production.
BibTeX
@inproceedings{hovgaard2014model,
  author = {Tobias Gybel Hovgaard and Stephen Boyd and John Bagterp Jørgensen},
  title = {{Model predictive control for wind power gradients}},
  booktitle = {Wind},
  year = {2014},
  doi = {10.1002/we.1742},
}
Figure