6
$\begingroup$

In the past 14 years, when dealing with enterprise planning challenges, I've encountered multiple challenges of "repeated planning" and "planning agility" in general. At some point, I started classifying, defining them and documenting them in our manual. Some of these terms come from academic paper or conference talks, but some of them I just named myself. Last week I did a video on these, but I started wondering which of these are known under other names?

  1. Continuous planning: frequent planning of a sliding planning window. For example: in nurse rostering plan a window of 3 weeks every week.
  2. Real-time planning: the input problem can change at any time and a new plan taking into account those changes must be delivered within seconds or less (using a warm start). For example: airport gate scheduling or train platform scheduling
  3. Non-disruptive replanning: replan an already published plan with minimal disruption (minimal changes). Also known as non-volatile replanning or semi-movable planning. For example: in conference scheduling, a speaker needs a different timeslot to make his/her airplane, after the schedule has been published already.
  4. Overconstrained planning: plan with a shortage of resources. For example: in nurse rostering, there are more shifts than available nurses
  5. Backup planning: minimize the impact of a worst case scenario by adding additional constraints. For example: in train scheduling, pad an extra 10 minutes between connections to account for the first train being late

Which of these are already defined in the academic world by other names? I think backup planning is a superset of robust planning.

Improve this question
$\endgroup$
3
$\begingroup$

I've seen the first one under the name "moving horizon models". It seems to me that some people refer to the second with the term "online" (as in "online optimization"), which (confusingly) does not refer to the Internet but to reacting as new data points (orders, gate changes, accident reports) arrive. The fifth one is referred to as "inserted slack" in scheduling, but I don't recall any phrase that applies to the idea of padding in a more general context.

Improve this answer
$\endgroup$
4
  • $\begingroup$ Thanks Paul! I agree that "online optimization" is a terrible term, so I 'll happily suggest the new name of "real-time planning" (or "real-time optimization") ;-) $\endgroup$
    – Geoffrey De Smet
    Sep 8 '20 at 6:52
  • $\begingroup$ Have you heard the term "robuust optimization" before and is it the same as the fifth one (backup planning) including a generalization of such "inserted slack" approaches or other ways to minimize the impact of the losing the most critical resource? Other examples include "when optimizing a train schedule, for the resulting plan, what is the impact of losing the worst possible rail link due to an accident? And how do we optimize to minimize that impact?" Similar story for electricity networks, satellite bandwidth scheduling, etc. $\endgroup$
    – Geoffrey De Smet
    Sep 8 '20 at 6:57
  • $\begingroup$ I'd be tempted to suggest "incremental planning" as an alternative to "real-time", but I think the first item also would be considered incremental, causing some ambiguity. I have heard of robust optimization, but I am not terribly familiar with it. My sense, though, is that robustness is (or at least can be) more general than just defense against the worst case. So for electrical networks it might mean a solution that is feasible (reasonably good?) under a wide range of demand profiles, though not necessarily the worst possible profile. $\endgroup$
    – prubin
    Sep 9 '20 at 18:08
  • $\begingroup$ Good point. I also think that incremental is ambiguous, with the first case and also with incremental calculation of the constraints it can cause confusion. $\endgroup$
    – Geoffrey De Smet
    Sep 10 '20 at 7:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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