Steve Miller, Technical Marketing Manager, Physical Modeling, MathWorks, Natick, Mass., mathworks.com
Simulating a new turbine piece by piece allows working out the bugs
where it’s easiest – in simulations.
Wind turbines are more than a set of integrated systems working together to produce electrical power. They must also correctly interpret a range of environmental conditions and react accordingly. As a commercial power source, unscheduled downtime is unacceptable. Turbine manufacturers must provide reliable and robust units capable of promised availability.
As for all equipment development, testing is an important phase of development. However, testing on full sized turbine prototypes is challenging because of the cost of the test unit and the range of test inputs based on wind conditions and turbine placement. Using Model-Based Design can reduce the number of tests involving prototypes, make those tests more effective, and enable engineers to maximize the performance of the entire system.
More complex all the time
There are more than 20,000 wind turbines in operation in Germany generating nearly 24,000 MW, which is roughly 7% of the electricity consumed in Germany. The rapid expansion of wind power is partly the result of increased turbine size and the addition of new wind farms. It is also due to technological improvements in the wind turbines themselves. Better rotor-blade aerodynamics, more efficient generators, and improved supervisory control systems make today’s wind turbines more efficient than ever before.
This enhanced technology presents engineers with new challenges. A wind turbine is a complex system in which a variety of subsystems must work together as efficiently as possible. The subsystems include mechanical devices such as rotor blades, gearboxes, hydraulic or electric drives for setting blade pitch angles, electrical yaw drives, generators, and the connection to the electrical grid. All of these are monitored by a complex supervisory control system that must respond in a specific way to varying environmental conditions, such as changing wind speeds.
In a traditional development process, the various subsystems are created in separate software and simulation environments, with requirements captured in other environments, such as text-based documents. This can result in several problems. Because the requirements have not been incorporated into the development process, it is difficult to compare the design with the specifications. Engineers are unable to determine if the changes to the design still enable the system meet requirements. Worst of all, requirements that are incorrect or incomplete will not be detected until the subsystems are combined in the final development phases when errors are expensive or impossible to fix.
Errors found late in the development process – when they are expensive to correct – can result when engineers do not integrate different designs early in the development process. For example, if the teams developing the generator and the supervisory control system work separately, it is difficult to predict what will happen when the subsystems are integrated. Engineers working in different software tools and simulation environments may not have the option of testing the integrated design in simulation. The result is that subsystems can only be tested together when hardware prototypes have been produced. Since the wide range of weather conditions and failure analyses prohibit exhaustive testing on hardware prototypes due to cost, safety, and feasibility, parts and systems may be overdesigned to make sure the turbine does not fail.
Various control systems with competing goals must interact with each other, and this creates a difficult design challenge. For example, the turbine monitoring system must ensure it generates electricity upon demand. At the same time, it must protect individual parts from unnecessary wear and tear. The supervisory control system must observe both of these contradictory goals.
Normally, the generator is only switched on when the wind reaches a speed of 2 to 4 m/s. Lower wind speeds fail to generate enough power and unnecessarily wear turbine parts. In high winds, controls shut down the generator and set the rotor spin slowly (or not at all) so as to reduce load on the drive train. The system that controls the blades attempts to keep the generator’s speed in a relatively narrow range so it can generate the maximum amount of power. Additionally, it must also react to imminent power failures to control or actuation systems within the turbine to prevent the turbine from becoming unstable and destroying itself.
Proper yaw control keeps the turbine facing the wind during normal operation. The yaw controller must also ensure that the nacelle doesn’t turn in the same direction all the time so that the cables in the tower do not twist beyond their limit. Again, it is up to the supervisory controller to reconcile these potentially conflicting goals. Adding to the challenge is the fact that the system is non-linear, including backlash in the gearbox and friction in its large ball bearings.
Smooth and continuous development
All subsystems in a turbine can be simulated and tested as part of an integrated system at an early development stage by using Model-Based Design. Controller hardware can be tested before building prototypes of the full system. Systems that must eventually work together– such as the pitch and yaw actuators – can be simulated together and matched for best performance.
Wind turbine developers who use this design philosophy profit from a smooth and continuous development process. The models and simulations exist in a single environment and are linked directly to the requirements and specifications. In addition, embedded software can be generated directly from the model. Doing so simplifies communication between various teams and enables integration issues to be discovered earlier in the design process.
From model to code generation
Engineers start the development process by modeling a wind turbine entirely in software such as MATLAB and Simulink. Various blocks represent the physical system with its mechanical, electrical, and hydraulic subsystems. These are supplemented by models of aerodynamic loads and are stimulated by various inputs, including wind speed and direction.
Mechanical and hydraulic systems modeled using SimDriveline and SimHydraulics. Blocks from standard libraries can be combined to create custom geartrain and actuation models.
Engineers can do system-level analyses to select technologies and determine system requirements. Ideal models of actuation systems can be gradually refined and replaced by realistic models to determine system performance. For example, an idealized pitch actuator can determine the force an actuator will need, letting the engineer size a hydraulic cylinder. Developers can then add a more detailed model of the selected hydraulic unit in simulations. A yaw actuator’s model can start as a single ideal torque source, and incrementally refined to include four individual motors, a model of the mechanical system including a gearbox, circuit diagram, and other details.
As the requirements for the design are refined, they can be added to the specification. Documents containing specifications and requirements connect directly to the model using bi-directional links. This lets designers easily navigate between the model and the specification to check if all requirements are still being satisfied at each stage of development. This gradual progression lets engineers test their design every step of the way, validating, and refining requirements.
Models for all subsystems can then be combined and simulated in the early stages of development. Subsystems developed by separate teams can be gradually added to the overall simulation to test system performance. At each step, the tradeoff of model fidelity and simulation speed can be balanced so designers can iterate quickly and check for integration issues. For example, when focusing on a yaw controller, the engineer can use a detailed model of the yaw system and quickly substitute a lower fidelity model for the pitch system into the overall model. This keeps simulation times short while making it possible to check for integration issues between these two systems.
Different simulations can now be conducted using an entire system-level model. A 3D animation of the system and plots displaying different values of relevance can show turbine developers how the design reacts under varying conditions.
Reducing need for physical prototypes
The simulation model can generate C code that can be used for two purposes. C code is generated from the control algorithms in the model for the supervisory control system and can be targeted to controller hardware. To test this control code and the controller hardware, hardware-in-the-loop (HiL) tests can be used instead of physical prototypes of the wind turbine. HiL involves using the model of the physical system (mechanical, electrical, and hydraulic) to generate C code and download it onto a real-time computer. The HiL real-time computer connects to the hardware controller and mimics the behavior of the actual wind turbine. Engineers can test the control system over a wider range of conditions than would be possible with a physical prototype. And, by using the same model of the physical system as used in the earlier phases of development, the engineer can verify that the generated code performs exactly as it did in during desktop simulation.
Using Model-Based Design allows testing the system and controller hardware before hardware prototypes are even made, as well as on-site power failures. This saves engineers from traveling to a turbine site to diagnose problems, which is particularly useful for turbines erected at remote locations. Engineers can streamline the wind turbine development process, optimize performance, and cut development times in half.
Filed Under: Featured, Software, Turbines
Viet Ngo Duc says
Dear Mr Paul Dvorvak,
I am very need your model for my job and for my thesis. I was work in renewable energy field and my thesis also about wind energy
Abstract: I will use SVC to survey the power grid with a wind turbine system.
I hope you can send me your model. It will very useful for me.
Thank you and best regards,
a very illuminating reading. Simple, holistic and deep enough to trigger the enthusiasm in the reader.
By that, is any part of the code shown in the pictures available in the Matlab Central under some sort of CC or BSD license?
Thanks a lot,
Paul Dvorak says
Glad you liked the article. I’m not familiar with the details of the software. I’d suggest calling The MathWorks. Tony Lennon there will be able to answer your questions. Call him at: 508 647-7696 or email@example.com.
I know that Vestas use MBD in their R&D in the UK. I know because I went for a job interview there. Sadly I didn’t get the job 🙁
Igor Braverman says
Great article, thank you.
Do you know of anybody in wind industry actually doing MBD?
Do they use commercial RTW Targets? What kind of controllers? (Bachmann/B&R)
Paul Dvorak says
Give MathWorks and Tony Lennon a call at 508 647-7696 or firstname.lastname@example.org
He knows the software better than I.