GitLab

Added

• New model setting 't_perfectForesight' tells the number of time steps (from the beginning of current solve) for which realized data is used instead of forecasts. This value cannot exceed current forecast length, however. Setting the value lower than 't_jump' has no effect.
• Automated the calculation of sample start and end times if using long-term scenarios. Also setting number of scenarios to one, instructs the model to use central forecast for the long-term.
• Speedup for model dimension calculation (set msft etc.)
• Support long time intervals in the first block
• Possibility to limit v_online to zero according to time series
• Output for reserve transfer results
• Reserve provision limits with investments
• Constrain the set of units to which ramp equations are applied
• Piecewise linear heat rate curves
• Checks for reserves
• Allow to set certain value for v_gen at 't000000'
• New model setting mSettings(mType, 'onlyExistingForecasts') = 0|1 to control the reading of forecasts. Set to 1 to only read forecast data that exists in the file. Note that zeros need to be saved as Eps when using this.
• Proper stochastic programming for the long-term scenarios period. Possible also to create a stochastic tree from the original data.
• Clickable link to sr.log in the process window in case of SCENRED2 error
• New diagnostic parameter for timeseries scenarios d_ts_scenarios
• Enable long-term samples that extend several years by using planning horizon which is longer than one scenario (e.g. 3 years). Note: Cannot use all data for samples as last years need to be reserved for the planning horizon.
• New model setting dataLength to set the length of time series data before it is recycled. Warn if this is not defined and automatically calculated from data.
• Command line arguments '--input_dir=' and '--ouput_dir=<path' to set input and output directories, respectively.
• Added sample dimension to most variables and equations (excl. investments). Samples can now be used as long-term scenario alternatives (for e.g. hydro scehduling)
• Number of parallel samples can be reduced using SCENRED2. Activate with active('scenRed') and set parameters in modelsInit.

Changed

• Removed some old command line arguments
• Removed obsolete 'emissionIntensity' fuel parameter
• Speedup for timeseries calculations
• Automatic calculation of parameter dt_circular takes into account time steps only from t000001 onwards.
• Debug mode yes/no changed to debug levels 0, 1 or 2. With higher level produces more information. Default is 0, when no extra files are written (not even debug.gdx). Set debug level with command line parameter --debug=LEVEL.

Fixed

• Unit ramps during start-up and shutdown
• Refreshing forecast data in inputsLoop
• Aggregated groups that were not in use were included in the model
• mst_end not found for the last sample
• Start-up not working for units without start costs or start fuel consumption
• periodicInit will fail with multiple model definitions
• Reserves should not be allowed to be locked when the interval is greater than smallest interval in use
• Start-up phase and aggregated time steps do not work together
• In SOS2 unit cannot exceed the generation of p_ut_runUp
• Startup cost calculation
• Efficiency presentations
• p_uNonoperational not fully correct
• Major bug in state variable reserve equations
• Scenario smoothing alogirithm
• Probabilities were not updated after using scenario reduction
• Severe bug in setting node state level limits
• Only selects forecasts with positive probability for the solve
• Calculation of parameter df_central
• Readability of some displayed messages