What is the highest sampling time ever applied in Model Predictive Control (MPC)?

For an HAVC (Heat, Air Ventilation, and Cooling) it can go to hours.

But, I am looking for a sampling interval in the range of months or years.

Does any one know any practical application?

  • 1
    $\begingroup$ This question isn't very good for the SE format since this question arguably has multiple valid answers, while SE is meant for questions which only have one true answer (which is why you can only select one answer as correct). For example, Tobias has given you an example with a sample period in the scale of months, but there may be another example which is just as valid. Or maybe there's another example which is in the scale of years, such that his answer is actually wrong in terms of "the highest sampling time ever applied to MPC". $\endgroup$ – Wasabi Jul 12 '18 at 19:03

More of a comment, but a bit longer:

The sampling interval correlates with the time constant of the controlled system. You would need to find a system that is slow enough such that MPC with sampling time of a few months is still sufficient. Maybe you can control the movement of a glacier..

The chemical industry has some of the slowest controlled processes as far as I know (please correct me if there are slower controlled processes) which is also why it was the origin of MPC. Of course the time constant of the system is only an upper bound for the sampling time, whoever designs the controller might still decide to pick a shorter sampling time.


NMPC (nonlinear MPC) of oil recovery processes is an ongoing field of research, and it involves control intervals on the scale of months. It is also referred to as closed-loop reservoir management (CLRM).

This website has an introduction to NMPC for oil reservoirs and CLRM. It also contains links to other research groups and organizations, publications, and software.


Not the answer you're looking for? Browse other questions tagged or ask your own question.