Control Design Tool

Our mission is to get new energy technology implemented in the industry. Therefore, we perform long- and mid-term research and development, and we help to remove technical barriers. ECN has been active in the wind energy research since 1975. For the last 10 years we have been significantly involved in control development. In the beginning the industrial interest was focused on consultancy for companies seeking to outsource the control design. In the last few years we have seen this changing. A recent market survey[1] shows that nowadays controls are an in-house activity. This explains why our focus has moved from consultancy to knowledge transfer. Therefore, we compiled our knowledge on control in design tools and made them ready-for-use. In 2007 the ECN Control Design Tool (CDT) was made commercially available.

Overview

The CDT contains feedback algorithms for rotorspeed and power control. The pitch motors and generator are used as actuators. The tool is tailored to pitch regulated, variable speed wind turbines, on a multi-megawatt scale. The format is suitable for wind turbine design, certification and implementation.

With the tool, we offer a control course. This is one way of knowledge transfer. The other is by distributing the tool as open source; both the design tools as well as the control algorithms. The design tools are implemented in MATLAB®[2], the algorithms in Simulink®, with which many control engineers are already familiar. An additional advantage is that the CDT can run on any platform that MATLAB can run on (Windows, Linux, Mac).

The CDT has a modular structure. This makes it easy to adapt the tool to your specific wishes. The tool is split up in three stages that will be presented below: synthesis, analysis and evaluation.

Synthesis

In the synthesis stage all the modules are run in sequence in order to parameterize the control algorithms. The example shows how the CDT deals with non-linear aerodynamic behavior. One module translates the rotor input data into aerodynamic sensitivities in many operating points. In a later module a gain scheduling table is derived from the sensitivities. Scheduling of the control gains provides equal control strength all over the operating range. The results are shown as graphs to the designer for inspection.

After all modules have been run, we already have a rather good view on the validity of the design per module. In the other stages, we check if these results also play well together.

Analysis

In the analysis stage we investigate the stability margins of the controlled wind turbine in all operating points. This is done by means of the well known Nyquist plot. This plot shows the open-loop frequency response of the system.

A control engineer immediately sees that the system is robust if the response passes the two critical points at the right hand side. Control features like dynamic inflow compensation or estimated wind speed feed-forward control can easily be switched on or off in the analysis stage. This is very handy and educational.

The two Nyquist plots in the example show the effect of wind speed feed-forward in one operating point. It can be concluded that it enhances the robustness because the red line lies further to the right.

Evaluation

For the final stage we move from the frequency domain to the time domain. In this stage we evaluate the controller performance through simulations in Simulink.

The simulation model contains three major parts:

• wind, providing turbulent and deterministic wind speed profiles
• turbine, containing a relatively simple, non-linear wind turbine model and
• control, which holds the complete control algorithm.

The flexibility of Simulink is ideal for quick development and testing. For high computational speed, the wind and turbine models are kept as simple as allowed.

Design iteration

Before the algorithm can be implemented in the turbine, a load analysis needs to be performed. Load calculations are beyond the scope of the Simulink model and so need to be performed with an accredited aero-elastic code. Fortunately, the Simulink-controller can be automatically compiled into executable code with the Mathworks tool Real-Time Workshop®. The control design tool can currently create dll's for ECN’s PHATAS or for GH Bladed. Through a similar mechanism, the Simulink-controller can be implemented in the hardware controller. Products like Bachmann’s “M-target for Simulink” can make the process completely seamless.

Unfortunately, no controller is ever optimal on the first design; often much iteration is required. The CDT offers three ways to reduce the optimization time.

• Graphical feedback is offered at each stage, so undesirable results can be identified early in the design process.
• Automated design and code generation allows the generation of a dll from altered input specifications in under a minute.
• The resulting dll can create signal logs and allows for plain text parameter tuning, enabling fast batch-testing of parameter sweeps in the final optimization.

Outlook

The current version of the CDT is limited to collective pitch control. Although ECN is active in research on individual pitch control (IPC), the implementation into the CDT is not straightforward. The effects of IPC can only be examined in the simulation stage if a more elaborate wind model is present. Also, a more detailed structural model for stability assessment is needed. In spite of these enhancements, the simulations in Simulink should still run fast.

To this end TURBU was developed. The near future will see the integration of TURBU results into the CDT, allowing the integration of IPC algorithms and more.

Conclusion

• The Control Design Tool is a proven tool for production quality wind turbine controllers.
• It is a strong platform for education and development thanks to its modular structure and open source implementation.
• Design iteration loops are short and fast due to the many visualization options and high level of automation.
• Easy code export to aero-code and PLC is possible thanks to the supplied dll interfaces and third party tools.

For more information, please download the latest paper on the CDT.

www.ecn.nl/publicaties/default.aspx

[1] source: Emerging Energy Research

[2] MATLAB and Simulink are products by The Mathworks.