Previously, the default startup cost and fuel consumption parameters applied to "hot" starts, which are impossible unless the hot start window is defined. Now the default startup parameters affect the "cold" startup, which is the startup always available to units with online variables.

Fixed an issue where the value of the storages at the end of the solve was considered regardless of the "storageValue" feature.

Avoid needlessly redefining fSolve for each solve, and make ts_length and tFull independent of the m dimension.

Fixing an issue where time series form reserve requirements were not generated if no time series found in data (but were generated based on forecast data read in the loop). Now determines the need for reserves based on constant parameters or the "use_time_series" flag.

NOTE! This flag is redundant with "useTimeseries" and is pending removal...

Model definition files included in the VCS now only include the GAMS definition (list of included equations), all of which are read by Backbone to allow for easier swapping of models without having to tamper with the "3e_solve.gms" fole. Rest of model parameters are expected to be given as input through the renamed "modelsInit.gms" in the input folder (by default). "periodicLoop.gms" and "periodicInit.gms" now treated as integral parts of the Backbone model.

Example input parameters provided by the "scheduleInit_temp.gms", "buildingInit_temp.gms" and "investInit_temp.gms" included in the defModels folder within the VCS.

Merge trtopi_develop. Major changes include changing the time step indexing to get rid of ft_dynamic for simplicity, distinction between directOnLP and directOnMIP efficiency approximations and a quick fix for the lambda approximation, as well as including the model-dimension to all model definition related parameters. Unit ramping constraints are currently disabled and pending changes.

NOTE! "invest_temp" model probably doesn't work at the moment due to changes in the model definitions, and reworking of the equations.

Added conditionals to the investment variables in the objective function to avoid generating unnecessary variables.

Fixing an issue where the reserves were fixed one time step too early at the start of the simulation.

Implementing most of the results arrays from trtopi_develop_RealValue due to their overall usefulness.

However, the intervals and forecast trees are currently not affected by the samples at all. How this should be implemented (if at all necessary) should be discussed with Juha.

Updated schedule model template which works with the model definition set changes. Removing unused set definitions.

All parameters in the model definition files now include the mType dimension to avoid overwriting of parameters if multiple models are declared.

Tweaks to the efficiency approximations, removing old unused sets and parameters, as well as minor readability improvements.

Again renaming and adding time and forecast subsets in an attempt to increase model performance. The "uftt" sets need serious rethinking to avoid excessive computations during each solve.

Old tt set renamed as tCurrent, and new tt now described as temporary subset of timesteps. Various performance upgrades, mainly to calculating the startup types and minimum up/downtime.

Implemented new "uftt" sets for reducing the time required to generate startuptype and minimum up-/downtime equations. Started using the tt(t) set to limit the time required for looping over t in some places, but could probably be utilized further.

There are still some old uses of the tt_ set as temporary subset of t, which is not really ideal when it comes to naming conventions.

Rewriting the efficiency approximation calculations, as well as reimplementing diagnostic COP and efficiency results.

Some minor readability and performance upgrades as well.

Readability improvements to 1a_definitions, as well as implementing the distinction between directOnLP and directOnMIP.

Rewrite of the boundCyclic equation to reduce the time it takes for Backbone to generate it. This might need rethinking in the future if more complex sample bounds need to be considered.

df changed to record all displacements in order to make it work properly with references to historical startup/shutdown information.

Rewrote 4a_outputVariant to match the new forecast-time structure. Model should be able to run now, but the structure still has to be excessively debugged in order to make the outputs make sense.

Also moved ft_nReserves definition inside the interval loop in order to possibly save an insignificant amount of execution time.