Imprimer cette page

First step with level control

Start the model with all parameters to default

  • Startup: start the pump at index 8, and when liquid level reach 40%, open LIC control valve at 50%. Look at the level stabilization on trend (put the mouse over this button if it is red): output flow tends to be equal to input pump flow rate at steady state.
  • Command step change while in open loop: open level control valve at 70%, clicking on it with the two mouse button to enter a value. Look at the process stabilization .
  • Level control in manual mode: try now to set the level at 50%, acting only on the LIC control valve. When it's done, change the pump index to 3 and try to set back the level at 50%. When it's done, try to set the level at 25%, pump still at index 3. Estimate the time you took to set the level at each try.
  • Level control in automatic mode: set now the level controller LIC in automatic mode (right clic on its mode), with a setpoint at 50%. Look at the level stabilization, and the actions of the controller on the valve. Modify the pump index at 8 and wait for stabilization. Open the leak control valve at 10%, and look at the process and the controller response to a process disturbance.
  • Setpoint step change while in closed loop: wait for steady state with controller in autmatic mode, setpoint at 50%. Change the setpoint to 25% with the two mouse button to enter a value. After process stabilization, change the setpoint to 50%. These response are said process response in closed loop to a setpoint step change.
  • Saturation: open leak valve at 30%, with pump index at 8 and level controller in auto setpoint at 50%. The controller close the valve without succeding in taking control of the level. Its output is said saturated, here at 0% but it could be at 100%.

Process identification on an operating point

  • Start the level control model with all parameters to default
  • Start pump at index 8, and when level reach 40%, open level control valve at 50%
  • When level is stable, increase controller output by +10%, and look at measure trend for 2-3 minutes. With the mouse on the trend to read the process values, try to estimate the process identification parameters, i.e. process gain, pure lag and first order time constant in seconds.
  • Open now control valve at 40% and wait for process stabilization, with level at ~75%. Increase the controller output by +10% and estimate again process parameters: gain, lag and first order time constant.
  • Calculate controller PID tuning (controller gain, integral and derivative time constants) for each of the two open loop process identification done previously. Give some conclusion.

Closed loop control

Start the level control model with all parameters to default. Start pump at index 8, and set the level controller in auto with setpoint at 50%.

  • Proportional mode only: set the controller gain to kp=1, integral Ti=0s (no action), derivative Td=0.1s (with on headband or for ‘configuration' on controller). From level at steady state at 50%, change the setpoint to 70%, then to 30%, then back to 50%, waiting for stabilzation at each step. Do the same experience with controller gain kp=2, then kp=4. Process start swinging, due to excess of proportional action. Proportional action tends to prevent measure from moving apart the setpoint.
  • Integral action: set the controller gain to kp=1, integral Ti=20s, derivative Td=0.1s. From level at steady state at 50%, change the setpoint to 70%, then to 30%, then back to 50%, waiting for stabilzation at each step. Do the same experience with controller integral action Ti=10s, then Ti=7s. Process start swinging, due to excess of integral action. Integral action tends to set back measure equal to setpoint, as quickly as possible, as a function of process response time.
  • Derivative action: set the controller gain to kp=1, integral Ti=20s, derivative Td=3s. From level at steady state at 50%, change the setpoint to 70%, then to 30%, then back to 50%, waiting for stabilzation at each step. Do the same experience with controller derivative action Td=8s, then Td=15s. Process start swinging, due to excess of derivative action.
  • Controller tuning: do three open loop process identifications at 30, 40 and 50% level for a command step change of 5%. Calculate for each identification the resulting PID action with the following formula. Test each PID action set in closed loop control, and give a conclusion to the results.
  • PID action formula: controller gain : kp=0.85×θ/(Gp×τ), Integral : Ti=θ, Derivative : Td=0,4×τ, with Gp process gain in %/%, θ first order time constant (s), and τ pure lag (s). These formula can be different depending on process, they are adapted at this case only (see Command law and PID tuning for more info.