3

I asked a version of this question before and got some unhelpful answers, I think because 1) the minutiae of my question distracted from the core concept, and 2) I didn't adequately describe the operation of my system. Original question can be found here.

In my system, I have a single process variable that I would like to control, concentration (C). The equipment I'm using has two ways to influence C. One way is by changing the amount of stuff (A) or by changing the speed at which stuff is introduced (S). A and S are controlled independently. C is measured by a single sensor. I originally assumed that A was a sort of "coarse" control and S was a "fine" control, but I no longer believe that to be the case.

Currently, I'm choosing A at the start of an experiment and letting a single PI controller change S as needed. This seems like a waste, since the equipment was designed for the tandem use of A and S to control its output. I only have a single sensor to monitor C.

Is there a control scheme that's fitting for this situation? The sole purpose of this system is to control and measure C. I've heard about cascading PI controllers, but I don't have a second sensor or even a second variable to measure, and ultimately I'd like to control both A and S.

Ben S.
  • 71
  • 3
  • 5
    There is a good reason we use diagrams and dump the over-complicated mind-pictures painted somewhat carelessly when using words. – Andy aka Jan 13 '22 at 15:45
  • I think there are also two process variables : Concentration and Rate of Production. Though you may be relatively uninterested in it, acknowledging it may make the control system clearer. –  Jan 13 '22 at 16:02
  • Would be great if the question wouldn't be so abstract. Understanding what are you actually doing is like having a model without you building a model. Anyway, sounds like you might want two controllers, one of higher hierarchy. So towards the world it's still one, but inside the system it's like position over speed, or speed over acceleration. – TQQQ Jan 13 '22 at 16:10
  • 1
    How is that different than controlling the speed of a car by how much pressure you put on the pedal and how far you continue to press further to reach a speed? Multiple controls and multiple feedback. The error must be derived from independent feedback and compared with variable setpoint. If you don't have feedback, you must create it , sense and process it to create it.~ – Tony Stewart EE75 Jan 13 '22 at 16:10
  • It seems like A could be quantified as either too much for the system to accommodate or too little for S to satisfy the system. If you can get a "nearly too much (H)" and a "nearly too little (L)" signal, you can perhaps control A using by: turn on if L and run until H. –  Jan 13 '22 at 16:16
  • 1
    The same for a car with a speedometer to track acceleration (not normally done) , velocity and distance travelled ( I wish some bus drivers also had jerk feedback, ;) – Tony Stewart EE75 Jan 13 '22 at 16:16
  • I'm confused on A and S. Is A the total amount of solute, and S the rate of added solute? Or is A the total amount of solution and S the rate of added solute? Could you lay out the dynamics of the system, i.e. the relationship between the different variables? – Some Sorta EE Jan 14 '22 at 15:54
  • @Clipboard_Waving_Enginerd A is the amount of "stuff" added and S is the rate at which it is added. You can think of it as solute (and it would be, if I were doing wet chemistry), but the system creates an aerosolized mass concentration of "stuff" per unit volume of air. Basically A controls how much stuff a hopper drops on a moving belt, and S controls how fast that belt moves to the aerosolizer, which operates at a constant flow rate into the chamber. The manufacturer doesn't specify how A and S relate to real-world measurables, they just specify A and S go from 0 to 100, without any units. – Ben S. Jan 14 '22 at 16:09
  • Thanks that helps. So A can be thought in terms of kg/s, added to a belt running at S m/s. So however the physics of pouring stuff works out you end up with an effective "stuff added" rate is some function of A and S, like what you were saying about Coarse and Fine control. Does this sound right? – Some Sorta EE Jan 14 '22 at 16:43
  • @Clipboard_Waving_Enginerd Yes, that's accurate. – Ben S. Jan 15 '22 at 00:48

0 Answers0