Bladed Hardware Test ModuleDong Fang Steam Turbine Corporation 3-5 December, 2008Garrad HassanTechnical consultancy services and products for renewable energy projects.Wind EnergyMarine EnergySolar EnergyWhy choose Garrad Hassan? Were a recognised worldwide consultancy with 24 years of proven experience Were an independent company at the forefront of developments We offer industry standard software solutionsAbout us Founded in 1984 in UK Now have offices worldwide Local understanding informs global perspectiveWere Independent We never take an equity stake in any project or technology! so our clients always receive an objective serviceAbout Me Thomas Cook Bristol-based Experience throughout the project lifecycle Systems Engineering based on Runge Kutta explicit methods Step size and order are determined automatically in order to meet a predicted accuracy requirement Numerically efficient Rarely needs user interventionReal Time Fixed step size 4th order Runge Kutta explicit method User specifies step size No guaranteed accuracy; may easily become unstable!Normal and Real-Time Simulations contdNormal Simulation time advances at a variable pace Simulation rate depends on turbine dynamics and non- linearitiesReal Time Simulation time advances at a fixed rate (Windows permitting) Multi core CPU required Multimedia timers used for non blocking wait to keep to real time No guard against running slower than real time Repeatable simulation rateReal Time Simulation Two parts to preparing a simulation to run in real time: Achieving a reasonable/desirable simulated step size Making the real time step size match the simulated step size We will generally iterate back and forth between these two componentsReal Time Simulation Four way trade off: Model scope Solution accuracy Run in real time Available hardware Start with desired model scope Keep reducing model scope until its possible to find a step size which gives sufficient accuracy and is capable of running in real time Establish baseline step size from normal Bladed Software Performance outputReal Time Simulation contdImprove simulation fidelity Decrease step size Add high-frequency modesImprove numerical accuracy Decrease step size Remove high-frequency modesImprove execution speed Increate step size Remove high-frequency modes Buy better hardwareReal Time Simulation contd Some realistic limits: Windows timing constraints about 5ms minimum step size Watch out for some funny chipsets In practise, it is good to have a sample rate four times the highest frequency component So Windows 5ms limit leads to a maximum 50Hz mode in the turbine model, lower if the available hardware cannot sustain the simulation at this rate.Bladed Dynamics Modal blade bending Modal tower bending, including multi-member modal dynamics and foundation flexibility Drive train torsion Yaw degree of freedom Power converter time constant Pitch actuator dynamics static void Main(string args) protocolReady = new AutoResetEvent(false);Controller controller = new Controller(); controller.InitProtocol();protocolReady.WaitOne(); Console.WriteLine();for (; ; ) for (int i = 0; i 2.0StartupRebootPower_ProductionAny StateController_StateTimer 2.0Controller_Reboot = trueController Startup Strategy Presently, the controller starts up in full pitch-torque- speed control mode it can produce some large transients! We should wait for the wind to be within a specified range, then gently ramp down the pitch demand from 90 to operating pitch. The controller has existing internal values finePitch and featherFinePitch the pitch demand is clipped to these Add a variable Controller_ActiveFinePitch which can be used to raise the bottom of the pitch demand clipping range. Well also give the controller control over the generator contactor with a new variable: Controller_GeneratorContactorStartupRebootWaiting_For_Wind Controller_ActiveFinePitch = 90 deg; Controller_Contactor = falseController_StateTimer 2.0Controller_Reboot = true Any StatePower_RampingController_StateTimer 2.0Generator_ContactorController_MeasuredGeneratorSpeed 800 Controller_Contactor = trueController_ActiveFinePitch -= 0.5 degrees per secondPitch_RampingController_WindSpeed 13 repeating each rotor revolution (1P)120010008006004002000-200-400-6000306090120150180210240270300330360 风轮角度(度)叶根弯距=正常化频率周期性气动力学载荷(Periodic aerodynamic loads) 风剪切(Wind shear) 塔影(Tower shadow) 偏航误差(Yaw misalignment) 主轴倾角(Shaft tilt) 上吹风流(Upflow) 尾流速度分布(Wake velocity profile)风速付里叶分量风剪切(Wind shear)导致轴向风速及载荷的 周期性变化(Periodic variation in axial wind speed and therefore loading)变化包括 1P, 2P 和 3P 等分量(Variation includes components at 1P, 2P, 3P etc.)方位角正常化频率风速付里叶分量塔影(Tower shadow)导致轴向风速及载荷 的周期性脉冲变化(Periodic impulsive variation in axial wind下风speed and loading) 上风 变化包括1P的高频谐 波分量(Variation includes high frequency harmonics of 1P)方位角正常化频率谐波分量(The