Units: | $ |
Mode: | Input Only |
Multi-band: | False |
Default Value: | 0 |
Validation Rule: | Any Value |
Key Property: | No |
Description: | Price for violating production non-anticipativity constraints |
Facility Production Non-anticipativity applies to multi-sample simulations with the Stochastic Method "Stochastic" i.e. stochastic optimization using scenario-wise decomposition. It is used to identify the non-anticipative Production decisions, and can optionally be set in combination with Production Non-anticipativity Time.
Defining the property adds the following constraints to the formulation:
Production(s,t) = Production(s+1,t) ∀s < S
where:
s is the sample number
S = Stochastic
Risk Sample Count
The property can take any value, but there are two special cases to consider:
The property is commonly used in medium or long-term Rolling Horizon models to ensure that Production decisions do not anticipate future information e.g. hydro inflows. Note that is generally not necessary to define this for the SDDP stochastic algorithm.
Facility | Property | Value | Units |
A | Units | 1 | - |
A | Max Operating Level | 200 | - |
A | Production Non-anticipativity | -1 | $ |
A | Production Non-anticipativity Time | 12 | h |
B | Units | 1 | - |
B | Max Operating Level | 100 | MW |
B | Production Non-anticipativity | 10000 | $ |
For the example in Table 1 Facility "A" Production is a non-anticipative (or first-stage in a two-stage stochastic optimization) decision for the first 12 hours and a recourse (second-stage) decision after that time, whereas Facility "B" is a non-anticipative decision for all periods but with a penalty of $10,000 if the constraint is violated.
See also: