|Contents|| ||Usage Advice|
Some of you have asked for advice on using our codes. Unfortunately, we do not have the staff or budget to help people build turbine models. Because there is still a need for some basic advice, I thought I would put this page together to tell you how I use the codes.
Some of you want to use our codes to design new turbines. Some want to analyze existing turbines. I use them only to analyze existing turbines. Generally, our codes are analysis codes and not design codes, but one can use them to check experimental concepts during the design phase of a new turbine. They are too complex for casual use. Even the simplest of programs takes days to learn to use. Useful models take weeks or months of full-time effort to build.
Whenever I start a new project to model a turbine, I spend a lot of up-front time collecting turbine properties. This includes turbine layout, mechanical properties of flexible components such as blades and towers, and aerodynamic properties of the airfoils. I find that this phase is generally the most difficult part of the project.
For most of our codes (those that use the AeroDyn subroutine package), you need 360-degree airfoil data. Usually, airfoil data tables include data only for the linear and stall portion of the lift and drag curves. You can use various tricks to extend this data to the full 360-degree range of angle of attack. We distribute a spreadsheet called AirfoilPrep that will help automate this procedure. The folks at Windward Engineering created it. Still, I get the help of an aerodynamicist to check the results. This is a complicated process and I will not go into details here.
After I get my layout and aerodynamic data together, I create rigid models of the turbine. I first use WT_Perf. It is a simple performance code that is fast and relatively easy to use. It can tell you how much power your turbine will produce and even what the root flap-bending aerodynamic load of the blades will be. It is a steady-state model that does not produce a time series.
Once I think I am getting reasonable numbers out of WT_Perf, I build a rigid YawDyn model. I use YawDyn to get more information about the structural response (loads) of my turbine. YawDyn is a dynamic simulator that can use complex wind-inflow data to excite your model. For two- bladed turbines, I add the teeter and yaw degrees of freedom to YawDyn once I get loads and performance similar to what WT_Perf predicts. If I am modeling a turbine with three blades, I add flap (and possibly) yaw degrees of freedom.
After I have a working YawDyn model, I move on to either FAST or ADAMS®1 for a more accurate prediction of structural response. These codes are much more complex and require a lot more detail for your turbine properties. They do use the same aerodynamic model as YawDyn, so once I have a working YawDyn model, I do not need to reproduce that effort for FAST or ADAMS.
If I do not have accurate mass and stiffness distributions for the blades and tower, I may as well stick with a simpler code like YawDyn. If I am modeling a variable-speed turbine, YawDyn will not help, so I need to use one of these two programs. The advantage of ADAMS is that it is not a turbine-specific code and can model virtually any configuration. FAST will work only for turbines with more-traditional configurations. For instance, FAST cannot currently model turbines with preswept or precurved blades. It also cannot model blade torsion, so flap/twist-coupled blades must be modeled with a more-complex tool such as ADAMS. The advantages of FAST over ADAMS are that it is free, it is simpler to use, and it is much faster. A ten-minute ADAMS simulation will take a half hour or so to run on the fastest PC. Whichever code I use, I start with a simple model and add degrees of freedom systematically. I check my results before adding more complexity.
I once started with a fully featured ADAMS model of a turbine and could not get its predictions to agree with test data. Because my model was so complex, I could not find my error. I finally decided to try the PROP-PC code (before I wrote WT_Perf) and found that my power predictions were wrong. After careful checking, I found that the blades on the test turbine were pitched differently than what I had thought. After that experience, I swore I would always start out with a simple model.
You can use FAST to help you build your ADAMS model with its FAST-to-ADAMS preprocessor. It uses the same input files with some extra columns for flexible-body properties not used by FAST. It also has an additional ADAMS file for a few ADAMS-specific parameters. Please see the FAST user's guide for instructions on using this feature. Unfortunately, it is limited to standard turbine configurations, although it does include preswept and precurved blades and flap/twist coupling. If you have an unusual configuration, you can use FAST to build a basic turbine model and then use ADAMS/View to modify the configuration afterward. You can also edit the ADAMS dataset directly.
It is my personal opinion that ADAMS is so complex and requires so much time to learn and use, that it is usually only useful after the turbine is built. I need to be able to accurately measure turbine properties for the additional complexity to be of benefit. With accurate properties, I can tune my model to agree with test results. If I have an unusual configuration that the other codes cannot model, I have no choice but to use ADAMS. I think a complex ADAMS model is a great tool for finding ways to improve an existing design.
In addition to the turbine simulators, we also have several programs that we use in conjunction with them. I use IECWind to create hub-height winds that follow the IEC standard wind cases. They produce files that can be read by all the AeroDyn-based programs (YawDyn, FAST, and ADAMS). I use TurbSim to produce full-field wind files. These are planes of three-component winds that march past the turbine at a mean wind speed. These wind files work with the AeroDyn-based simulators and even Garrad Hassan's BLADED. I use Crunch, GPP, and GenStats to process the results of the simulations or test data. They do many types of analyses. I usually use Microsoft Excel to plot my results. These postprocessors produce files that are easy to import into Excel.
For running a large number of cases, I use RunIEC to run all the IEC distrete wind events or RunNTM to run many cases with the normal turbulence model.
Modeling wind turbines is a long and arduous process. Start out with simple models and work your way up. Take your time and check your work as you go.
Marshall L. Buhl Jr.
This page was last updated by B. Jonkman on 13-November-2012.
|NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency & Renewable Energy,
operated by the Alliance for Sustainable Energy, LLC
Webmaster | Security & Privacy | Disclaimer | NREL Home