MPC test aborted: Downstream block reporting limited status

Hello,

I'm trying to control an air temperature loop, at first I was trying to manipulate the AO directly with the MPC block, but the response speed was very slow and the AO never went past 30% output. With a PID loop I get a decent response, so that's why I set a previously tuned PID block as slave of the MPC block. However, now that I'm trying to get the model for the MPC block I get the following error message:

Test was aborted because of the following reasons:

-Downstream block connected to "output_variable_name" is reporting limited status

The process is stabilized in 50°C, a 5% step change should be between 34.64°C and 65.35°C (AI zero=-99°C, max= 208°C; span=307°C).

I checked the SP_LIMS of the PID and everything seems to be ok, since the 5% step change I'm making is inside the limits, so what block or parameter does deltav refer as limited?

Please find below some screenshots of my module for further visualization of the problem. Many many thanks in advance.

Error dialog

Module diagram: (MPC block status is Out of service, I'm using the MPC pro block). DO is the fan switch, AI1 is a DP sensor (disturbance to MPC pro block), AO is output to a fan SCR. Connecting output of MPC to the PID RCAS.

Module diagram

PID and AO parameters:

PID parameters 1PID parameters 2   AO parameters

MPC parameters:

MPC parameters 1MPC parameters 2

MPC test setup

Edit 24/2/2020 16:26 EST

MPC block MV, CV and disturbance properties

MPC predictpro test configuration

  • In reply to Julian Navas:

    Julian,
    In the pictures you sent, the AO was low limited (at 0%). If this is the case while you are testing and the SP change to the PID causes the PID/OUT and AO to stay at 0%, the low limited status of the AO will carry into a limited status of the PID. Is this the case? I generally check "ignore errors" for my tests. Also, JD was referring to MPC/Local, not Auto/Man. It is correct that without a model, the MPC is stuck in MAN or IMAN. MPC means the MV(s) are in the model (RCAS in you example) to accept the SP from the MPC block. You can either hit the "MPC" button on PredictPro or MPCOperatePro or go to the PID faceplate and put the MV Loop(s) in the MPC related mode (RCAS in your example).

    Note there are also limits on the CV's and LV's in the Test Setup that will abort the test.
    James
  • In reply to Julian Navas:

    Do you need to make 5% step changes? Perhaps if the process is getting limited during step testing, you could try a smaller step size like 1% or 2%. You need a step size large enough to identify the dynamics, but ideally, you want a step size small enough to avoid disrupting the process or saturating against instrument limits or process constraints. Assuming the process is linear, the dynamics will be the same for large or small steps. This also sounds like an asymmetrical process, which are always challenging. If I understand correctly, you said the process cools at a different rate than when it heats. Without some kind of nonlinear adaptive gain, you are going to have to make a compromise, whether you use MPC or PID.