Short term storage

[flomnes]

Input file format changes

Time-series

• 5 additional TS per STS object
  • PMAX-injection.txt default values TBD
  • PMAX-withdrawal.txt default values TBD
  • inflow.txt, default values 0
  • lower-rule-curve.txt, default values 0
  • upper-rule-curve.txt, default values 1

Properties

One file list.ini per area, with properties

• name: str
• group: str
• injectionnominalcapacity: double
• withdrawalnominalcapacity: double
• reservoircapacity: double
• initialreservoirlevel: double
• efficiency: double
• storagecycle: unsigned int

Output file format changes

Values

• For each area, sum the reservoir levels in values-xxx.txt ?
• For each STS, save the reservoir level in details-ST-storage.txt

Pre-simulation checks

List incompatible parameters, incoherent data, etc. that would make it impossible to carry out a simulation. It's better to fail earlier than later.

Check on parameters

• 0 <= initialreservoirlevel <= 1
• 0 <= injectionnominalcapacity
• 0 <= withdrawalnominalcapacity
• 0 <= reservoircapacity
• 0 <= efficiency <= 1 (otherwise energy is created, makes no sense)
• 1 <= storagecycle <= simplexOptimizationRange (24 or 168)

Check on time-series

• 0 <= PMAX-injection(h)
• 0 <= PMAX-withdrawal(h)
• 0 <= inflow(h)
• 0 <= lower-rule-curve(h) <= upper-rule-curve(h) <= 1

Changes to the optimization problem

For each STS, add 3 hourly variables

• P-injection (h)
• P-withdrawal (h)
• Level (h)

[flomnes]

Development steps

[1] MVP

Reservoir level between 0 and 1. No inflows, no rule curves, cycle = simplex opt. range (weekly / daily)

[2] Add rule curves & inflows

Add rule curves & inflows
Rule curves induce additional constraints

[3] Add storage cycle

Storage cycle induce additional constraints

[4] Free initial level (calibration)

[5] Forbid simultaneous injection & withdrawal

[flomnes]

Optimization problem

Impose the initial level

Free initial level

[flomnes]

SIM

The plan is to use SIM only as an abstraction layer for the calendar & time-series.

• For all other data members (areas, short-term-storage, etc.), use C++ structures directly. Don't use SIM as a translation layer to C. These structures are not duplicated (multi-threading), but for read-only accesses it's OK.
• Add variable & constraint correspondance arrays

OPT

• Count variables & constraints opt_decompte_variables_et_contraintes.cpp
• Constraints
  • opt_gestion_second_membre_cas_lineaire.cpp
  • opt_construction_matrice_des_contraintes_cas_lineaire.cpp
• Variables
  • opt_gestion_des_bornes_cas_lineaire.cpp
  • opt_construction_variables_optimisees_lineaire.cpp
• Objective function No change in the objective function