Steam Turbine Valve Time Constant
I need to find the time constant of the valves that control the steam. So this is the time taken for the valves to go from fully closed position to 63% open position. I have tested this using the woodward 505 controller to tell the valves to open completely in one step. The only problem is that the 505 controller has rate limiters and the result was that I get a ramp response instead of a step response.I would like to know if there is another method to stimulate this kind of response. I have thought about using a signal generator to send a "fake" signal to the actuator making it open the valves instantaneously. The only drawback is, by using this method i am considering an open loop control so this may not provide correct information in modelling a closed loop control system.A simplified and linearized model for the steam valve spool position is K*omega^2/(s*(s^2+2*zeta*omega*s+omega^2))The limit you see is probably caused by the gain K which has units of steam spool velocity/control output. Therefore if the control output is 100% you are still limited to some velocity. You will not have a time constant in the sense that the spool position will approach the set point in a exponential manner. It is a little trickier than that.Omega is the natural frequency of the spool and hydraulic actuator, zeta is the damping factor. The damping factor is hard to calculate as it changes with respect. The controller is probably tuned very conservatively.To do the system identification you need to move the 4-20ma spool position reference to the controller or go into a manual or open loop mode where you can send a open loop control signal from the controller to the hydraulic servo valve that is controlling the spool. From this data one can determine the gain K, natural frequency and damping factor. Our controllers have an auto tuning feature that will compute this data for you. The controller collects control signal and hydraulic actuator position and uses the Levenberg_Marquardt algorithm to that finds the Gain, damping factor and natural frequency that minimizes the sum of error between the estimated position and the actual position. This isn't trivial. Get a math package like Scilab or Matlab to do this.
MORE NEWS