IEA Wind Task 52 LAC Summer Games 2025

Authors: David Schlipf, Feng Guo, Maarten van den Broek, Simon Weich Content Introduction ............................................................................................................................ 2 Background ............................................................................................................................. 2 Our Categories ........................................................................................................................ 2 Evaluation and awards............................................................................................................. 3 Our Disciplines ....................................................................................................................... 3 The 30 s Sprint ..................................................................................................................... 3 The 18 m/s Hurdles .............................................................................................................. 7 Our timeline ............................................................................................................................ 9 How to get started ................................................................................................................... 9 Support ................................................................................................................................... 9 References ............................................................................................................................ 10 Task 52 LAC Summer Games 2025 2 Introduction Following the success of the Lidar-Assisted-Control (LAC) Summer Games in 2024, this new edition builds on the lessons learned that were presented at the Wind Energy Science Conference in June 2025 [1]. From an educational perspective, the Summer Games 2024 proved to be a great opportunity for problem-based learning and encouraging collaboration between universities and professionals from the industry. In total, 15 contributions from students and professionals across 6 countries have been submitted to three different disciplines. At the same time, the experience showed how challenging it is to design LAC exercises that are realistic enough to be meaningful, yet manageable within a limited timeframe – especially for students at universities. For the Summer Games 2025, we want to continue this collaborative format with updated scenarios and more realistic wind conditions. Background Wind characteristics are inherently dynamic, subject to sudden changes in speed, direction, and turbulence. Consequently, wind turbines are highly dynamic systems, excited by such stochastic influences and must be designed to handle these variations and disturbances. Traditional controllers react only after wind changes have impacted the wind turbine, resulting in an offset operation until the controller handles the disturbance. Contrary, lidar-assisted control concepts use preview information of the wind to anticipate changes, hence reacting before the disturbance approaches the rotor and therefore improving the operation of the turbine. The mission of our IEA Wind Task 52 working group is to push forward the lidar-assisted control technology and simplify its application by recommended practices, open-source tools, and joined exercises. In line with our mission, we are very glad to launch our Summer Games 2025 after the successful Summer Games 2024 to continue to encourage students and professionals in the art of designing and deploying lidar data processing algorithms and lidar-assisted controllers. By participating in the games, you can increase your knowledge in LAC and enjoy the design process by competing with others. Moreover, we aim to trigger your creativity in three different disciplines and with it motivate the development of new concepts. During the competition, it is intended that the participants learn the existing open-source tools and their application will result eventually in the further development of the tools. Additionally, during the games you will be able to exchange knowledge with different parties of the wind energy community. We are looking forward to strengthening your passion for advancing the vision of modern wind turbines and getting the best LAC out of you! The tools are based on previous publications [2], [3] and built up on the IEA 15 MW reference wind turbine [4] and the open-source controller ROSCO [5]. We intend to write a publication on the results similar to [1] and integrate the learnings in our planned IEA Recommended Practices on Lidar-Assisted Control. Our Categories For the Summer Games we have the following categories: 1. Students (including PhD students) 2. Researchers Participation in groups is possible. If all group members are students, the first category applies. 3 Evaluation and awards In every discipline (see below) we will rank the contributions for each of the two categories mentioned above depending on the cost: for the 30 s sprint and the 18 m/s Hurdles, the lowest cost will win (minimization problem). The results will be shared on the IEA Task LinkedIn page and participants will also have the opportunity to present their solution to the IEA Wind Task 52 LAC working group, which consists of a selection of academic and industry experts. Our Disciplines In this section, you will find the different disciplines where you can participate with your proposed lidar-data processing algorithms and lidar-assisted controllers. Regardless of your chosen category, we encourage you to submit outcomes for as many disciplines as you believe align with your capabilities. In the following subsections, each of the disciplines are presented together with the evaluation methods used for evaluating the performance of the algorithms you send. The 30 s Sprint In the first challenge, like a 100-meter sprint, we'll test how well your controller handles an Extreme Coherent gust with Direction change (ECD) at 12.5 m/s following the Design Load Case (DLC) 1.4 of the IEC 61400-1 international standard on wind turbine design requirements, see Figure 1. The challenge is addressing mainly control experts and the intention is to introduce you to LAC and to trigger new control concepts. In this discipline, your controller response will be tested under the most challenging conditions for any controller evaluation. The objective is to cancel out the effect of the gust and to have a smooth rotor speed and tower-top motion. For this purpose, the wind preview and therefore the rotor-effective wind speed are generated through lidar-simulations, representing realistic measurement conditions. Two different lidar configurations are considered as baseline cases: a four-beam pulsed system (4BeamPulsed) and a continuous-wave circular scanning system (CircularCW). Both setups provide different spatial averaging characteristics, allowing a comparison of their impact on the control performance. The scan distance has been optimized for the Summer Games 2024 to provide the best preview quality (smallest detectable eddy size) for frozen turbulence simulations. The script RunExample.m in the Sprint folder (within the “SummerGames2025” branch) provides an example using a baseline collective pitch feedforward control [1], which is designed to compensate the impact of the gust on the rotor speed and works well for high wind speeds but provides plenty of room for improvements for wind speeds close to rated wind, see Figure 3. The corresponding ECD wind field characteristics – including its magnitude, lateral, and vertical component – as defined in IEC 61400-1, are also illustrated in Figure 1. To simplify the task, we reduced the degrees of freedom of the wind turbine to the rotor motion (GenDOF), the first foreaft tower bending-mode (TwFADOF1), and the first side-side tower bending-mode (TwSSDOF1). The evaluation criterion for this discipline is the minimization of the combined relative reduction of the rotor and tower motion defined as: 𝐽 = max (|Ω−Ω0|) Ω0+ max (|𝑀𝑦T−𝑀𝑦T0|) 𝑀𝑦T0 , where Ω is the rotor speed and Ω0 its steady-state value. Further, 𝑀𝑦T is the fore-aft tower base bending moment and 𝑀𝑦T0 its steady-state value. 4 The results for the two baseline lidar setups are: • 4BeamPulsed: Cost = 0.731996 • CircularCW: Cost = 1.217946 The following rules apply: 1. You can replace the feedback controller (ROSCO) and use your own combined feedback and feedforward controller, model predictive control etc. adjusting collective/individual pitch angle and generator torque. 2. Changes in the OpenFAST input files other than adjusting the BLADED INTERFACE section in the ServoDyn input file are not allowed. 3. The lidar data processing and lidar simulation files can be changed, but should correspond to commercially available lidar systems. In addition to using the information of the lidar for continuously improve the control (e.g. feedforward control or model predictive control), we encourage participants also to enable extreme event detection and mitigation strategies. Besides the control challenge, another purpose of this discipline is to showcase the difficulties in simulating DLC 1.4. The following assumptions have been made: 1. The propagation wind speed (in the lidar input files LidarFile_4BeamPulsed.dat and LidarFile_CircularCW.dat) is set to 20 m/s, the mean of the wind speed magnitude over the simulation. 2. This propagation wind speed is also used to determine the buffer times in FFP_v1_4BeamPulsed.IN and FFP_v1_CircularCW.IN which assumes that the propagation wind speed is known to the controller. 3. The wind field components are rotated individually and propagated with the propagation wind speed, see Figure 2. No turbulence is added so far. 4. While the DLC 1.4 should be simulated at rated wind speed and rated wind speed ±2 m/s, we only selected the rated wind speed + 2 m/s case. We also encourage participants to comment on the following questions: 1. Is the DLC 1.4 as simulated here a valuable scenario for LAC? 2. If so, how should we simulate the DLC 1.4 more realistically? 3. If not, what should we simulate instead? 5 5. Figure 1: Wind-field components (magnitude, lateral, and vertical) for the ECD simulation for lidar-assisted control over different heights over ground. The time corresponds to the tower bottom center. The wind field and the figure are generated by the script WriteECD2BladedWind.m in the Sprint\Wind folder (within the “SummerGames2025” branch). Figure 2: Possible simulation options for the ECD. The left one (used in the Summer Games 2025) rotates the wind vectors in a wind field with horizontal inflow angle 𝛼H and propagates them with the mean wind speed 𝑢. Another option (right) would be to rotate the wind field around a certain rotation center (here the wind turbine). 6 Figure 3: Simulation results of the baseline cases “4BeamPulsed” (top) and “CircularCW” (bottom), showing the response of the controller during the ECD. The figure illustrates the starting point of “The 30 s sprint” challenge, where the goal is to trigger new control concepts in order to reduce the impact of the ECD on rotor and tower motion. Generated by the script RunExample.m in the Sprint folder (within the “SummerGames2025” branch). 7 The 18 m/s Hurdles In the second challenge, we'll test how well your lidar-data-processing algorithm is able to detect obstacles at high wind speeds, here 18 m/s. The challenge is mostly addressing lidar experts and the intention is to introduce you to lidar-data-processing for control and to trigger new lidar-data processing algorithms and lidar scan configurations. The objective is to reduce the measurement error of the predicted rotor-effective wind speed (REWS) with two seconds of prediction time, typically needed for feedforward control for this turbine. The scripts RunExample.m/.py in the Hurdles folder provide examples using a baseline lidar-data processing [1] for a typical continuous wave lidar (circular scan) and a pulsed lidar (4 beams). Building up on the Summer Games 2024, we added blade blockage and wind evolution using the commercial lidar simulator solis. This simulator can either be run stand-alone with predetermined inputs or as an extension to various wind turbine simulation softwares such as OpenFAST or HAWC2. For the Summer Games 2025, the line-of-sight wind speed measurements include the effects of probe-volume averaging. Blade blockage effects are modelled based on a constant rotor speed and blade dimensions from the IEA 15 MW reference wind turbine (here, the isValid flag is set to false). The wind fields that have been sampled are based on Kaimal spectra turbulence generated using TurbSim. The effects of spatial turbulence evolution are included with an exponential coherence model as described in [6]. The evaluation criterion for this discipline is the minimization of the mean absolute error (MAE) between the REWS from the wind field shifted two seconds into the future (REWSWF) and the processed REWS estimate from the lidar (REWSLP). Additionally, we remove the mean values of the error to be independent of offsets: error = REWSWF − REWSLP 𝐽 = mean(|error − mean(error)|) . In the examples, the REWS is filtered by a low pass filter and then buffered. The MEA is averaged over the last 600 seconds of six simulations with different turbulent wind fields generated by TurbSim. The used baseline lidar data processing LDP_v3.m is based on the Summer Games 2024, ignoring measurements with a false isValid flag, causing the “blade ghost effect” [7]. The results for the two baseline lidar setups are: • 4BeamPulsed: Cost = 0.515998 m/s • CircularCW: Cost = 0.463842 m/s The following rules apply: 1. The lidar configurations are currently limited to the two mentioned above for simplicity. 2. New lidar data processing ideas should be implemented with the same interface as the function LDP_v3.m. 8 Figure 4: Starting point of “The 18 m/s hurdles” challenge for the baseline cases “4BeamPulsed” (top) and “CircularCW” (bottom) for the first seed: Goal is to reduce the measurement error. Here, the rotor-effective wind speed (REWS) from the wind field is shifted by two seconds into the future and compared to the processed lidar estimate (top and center). We would like to trigger new lidar data processing algorithms and scan configurations. Generated by the script RunExample.m in the Hurdles folder (within the “SummerGames2025” branch). 9 Our timeline The Summer Games 2025 have already started and will run until February 28, 2026 (end of winter semester of most universities). What should be provided: 1. For the discipline “The 30 s Sprint”: please upload the zipped outb files to the Teams folder Uploads (WG 2 – LAC>SummerGames2025>Uploads) in the Sprint folder using the following filename for your zipped file: <Name>_<YYYYMMDD>. 2. For the discipline “The 18 m/s Hurdles”: please upload MAT files with your REWS estimate from your lidar data processing algorithm. 3. If possible: In addition to the files above, DLLs and other files necessary to run simulations or lidar data processing algorithms (via a Git fork). Please check with someone independently to ensure that the scripts run and state clearly which script produces the submitted files. We will then evaluate the results together in our working group meetings and announce the rankings in the two disciplines and two categories. Any changes to the rules or submission dates can be discussed in the discussion forum: https://github.com/IEAWindTask52/LidarAssistedControl/discussions. How to get started To get started with our Lidar-Assisted Control open-source tools, you need to clone the GitHub repository to your local computer using either Git or GitHub Desktop. Please make sure you change to the “SummerGames2025” branch. In the repository, we provide various example scripts for LAC in both Matlab and Python. To get started with the Matlab scripts you need to navigate to the Sprint or Hurdles folder and then you can start working based on the example scripts (see disciplines above). The scripts use functions which are provided in a folder WetiMatlabFunctions and added to the MATLAB path. To get started with Python scripts unlike Matlab users need to install various modules like numpy, pandas, or scipy. To make it simpler we have provided a ‘setup.py’ script that installs all the necessary modules in your environment. Support If you have questions regarding the Summer Games in general or regarding the code which might be interesting for others as well, please use our Forum on GitHub: https://github.com/IEAWindTask52/LidarAssistedControl/discussions. If you require further support, please don’t hesitate to contact our support team via email: • Simon Weich [email protected] • David Schlipf [email protected]