Input file calibration and update¶
The primary input file (see previous section) includes rough data as extracted from the GIS layers. Before proceeding with the electrification analysis, these values need to be conditioned and/or calibrated.
- Conditioning makes sure that physical values (e.g. GHI, Wind speed, land cover, elevation etc.) are within acceptable limits.
- Calibration makes sure some of the parameters (e.g. total population, urban/rural ration, electrification rate etc.) are in line with official statistics.
- New columns are also created and support later stages of the analysis (e.g. wind capacity factor, grid penalty ratio, electrification status etc.).
Calibration with GEP Generator¶
The GEP Generator is an interactive interface, developed as a jupyter notebook (.ipynb) in order to support calling functions in the gep_onsset model. The GEP Generator is located in the root directory of the repository. You may access it by simply navigating there via anaconda prompt using:
> cd ..\my_designated_local_directory
..\my_designated_local_directory> jupyter notebook
Jupyter notebook will open on your default browser; simply select to open the GEP Generator.ipynb and you are set to go!
The GEP Generator runs in 9 steps (or blocks). Steps 1-5 are used to define calibration parameters and conduct the conditioning/calibration process. In particular,
- Step 1 requires that the user provides the primary input file (see previous section)
- Steps 2 & 4 allow the user to interactively provide the calibrating parameters
- Step 5 conducts the conditioning and calibration process
Note
- Step 3 is related to the definition of scenario parameters and is discussed in more detail in the following section.
- The GEP generator does not store the calibrated results in a separate file but rather continues right away to the scenario runs. This makes the process faster on the one hand, but it means that the conditioning and calibration process runs anew everytime a scenario is executed (only one at a time).
Calibration with gep_runner¶
The gep_runner.py is an alternative way to call functions from the gep_onsset.py. You may execute gep_runner in any IDE of preference, we suggest PyCharm. Interaction with the code using gep_runner takes place in the python console of your IDE. Upon initiation, the code will prompt you to select one of the three following options:
- 1: To split countries in case of multiple country runs (used rarely)
- 2: To prepare/calibrate the GIS input file
- 3: To run scenario(s)
For calibration you may select option 2. Execution requires two files:
- The primary input file (see Malawi.csv from previous section)
- The specs file (see example of specs_mw_one_scenario.csv).
The specs file contains the parameters and their values against which the GIS data are conditioned or calibrated. The user shall fill in all necessary values in the SpecsData sheet. A description of the parameters is presented below.
| Parameter | Description | |
|---|---|---|
| Country | Name of the country | |
| CountryCode | ALPHA-2 country code as per ISO 3166 international standard | |
| StartYear | Base year of the analysis; usually selected based on data availability | |
| EndYEar | End year of the analysis | |
| PopStartYear | Official population at the base year | |
| UrbanRatioStartYear | Official urban population ratio in the base year | |
| UrbanCutOff | Population threshold above which a settlement can be considered urban (optional) | |
| UrbanRatioModelled | This value is provided by the model after calibration | |
| PopEndYearHigh | Expected population at the end year based on high growth rate | |
| PopEndYearLow | Expected population at the end year based on high growth rate | |
| UrbanRatioEndYear | Expected urban population ration in the end year | |
| NumPeoplePerHHRural | Number of people per household - rural settlements | |
| NumPeoplePerHHUrban | Number of people per household - urban settlements | |
| GridCapacityInvestmentCost | Expected investment cost per kW of additional capacity in the central grid system | |
| GridLosses | Expected transmission ans distribution losses in the grid network | |
| BaseToPeak | Average to peak load ratio for the grid; used for sizing additional capacity due grid extension and to accommodate reliability issues | |
| ExistingGridCostRatio | Persentage of capital cost increase in each grid extension iteration; used to accommodate reinforcement of grid and reliability of supply | |
| MaxGridExtensionDist | Maximum distance (in km) that MV lines can reach in each iteration loop | |
| NewGridGenerationCapacityAnnualLimitMW | Capacity that can be added to the central grid per year of analysis | |
| ElecActual | Official national electrification rate in the base year | |
| Rural_elec_ratio | Official national electrification rate in rural areas in the base year | |
| Urban_elec_ratio | Official national electrification rate in urban areas in the base year | |
| ElecModelled | This value is provided by the model after calibration | |
| urban_elec_ratio_modelled | This value is provided by the model after calibration | |
| rural_elec_ratio_modelled | This value is provided by the model after calibration | |
| MinNightLights | Nighttime light value above which a settlement can be consedered electrified; used to identify and calibrate electrification rate in the base year | |
| DistToTrans | Distance to transformers (in km) above which a settlement can be consedered electrified; used to identify and calibrate electrification rate in the base year | |
| MaxGridDist | Distance to T&D network (in km) above which a settlement can be consedered electrified; used to identify and calibrate electrification rate in the base year | |
| MaxRoadDist | Distance to road network (in km) above which a settlement can be consedered electrified; used to identify and calibrate electrification rate in the base year | |
| PopCutOffRoundOne | This value is provided by the model after calibration | |
| PopCutOffRoundTwo | This value is provided by the model after calibration |
With gep_runner the calibration process is separated from the scenario runs. That is, the code stops once the conditioning and calibration process is complete. The result is exported in a “calibrated” input file.
The result file and the updated parameters in the specs file should be reviewed to decide whether the result is satisfactory or the process requires further calibration. Key outputs to cross-check include:
- Population projection
- Modelled urban/rural classification
- Modelled electrification rate (national, urban, rural)
Note
The conditioning & calibration process is driven by relevant functions located in gep_onsset.py. One can access and modify these functions in case their existing form does not serve the intended purpose. This requires some experience with the model; in case you are a new user you may experiment with the GEP Generator first before engaging in modification of the core code.
Example of the calibrated input file¶
The calibration process will add the following columns to the input file.
| # | Column | Unit | Description |
|---|---|---|---|
| 39 | PopStartYear | people | Calibrated population to match with official statistics in the base year |
| 40 | Pop<year>High | people | Projected population in the specified <year> based on high growth indicators; for intermediate and end years |
| 41 | Pop<year>Low | people | Projected population in the specified <year> based on low growth indicators; for intermediate and end years |
| 42 | Pop<base year> | people | Calibrated population to match with official statistics in the base year |
| 43 | RoadDistClassified | 1 to 5 | Classified value of distance to road used to calculate grid penalty factor |
| 44 | SubstationDistClassified | 1 to 5 | Classified value of distance to sub-station used to calculate grid penalty factor |
| 45 | LandCoverClassified | 1 to 5 | Classified value of land cover type used to calculate grid penalty factor |
| 46 | ElevationClassified | 1 to 5 | Classified value of elevation used to calculate grid penalty factor |
| 47 | SlopeClassified | 1 to 5 | Classified value of slope used to calculate grid penalty factor |
| 48 | GridClassification | 1 to 5 | Grid extension suitability index; Higher value indicates higher suitability; based on an Analytic Hierarchy Process (AHP) over the above parameters |
| 49 | GridPenalty | number | Grid extension cost multiplier based on above classification; default value 1 induces no additional costs |
| 50 | WindCF | from ~0 to 1 | Wind capacity factor estimated based on available wind speed and power rating of Vestas V-44 600kW turbine |
| 51 | ElecPopCalib | people | Number of people with access to (grid) electricity, calibrated to match official statistics in the base year |
| 52 | ElecStart | 0,1 | Electrification status in the base year; 0: non-electrified 1: electrified (by the grid) |
| 53 | GridDistCalibElec | km | Distance to nearest power infrastructure element (transformer, MV, HV); based on their availability. In case transformers are not available it will lookup the next available element (e.g. MV) |
| 54 | Elec_Initial_Status_Grid<base year> | 0,1 | Grid electricity status in the specified base year; 0: non-electrified 1: electrified by the grid |
| 55 | Elec_Init_Status_Offgrid<base year> | 0,1 | Off-grid electricity status in the specified base year; 0: non-electrified 1: electrified by an off-grid technology |
| 56 | Actual_Elec_Status_<base year> | 0,1 | Overall electrification status in the specified year; 0: non-electrified 1: electrified by any technology |
| 57 | FinalElecCode<base year> | 1 to 8 or 99 | Code of electrifying technology in the specified year (1: grid, 2: stand-alone diesel, 3: stand-alone PV, 4: Mini-grid diesel, 5: Mini-grid PV, 6: Mini-grid Wind, 7: Mini-grid Hydro, 8: Hybrid Mini-grid, 99: not-electrified) |
| 58 | GridReachYear | year | Estimated year that the grid might be able to reach this settlement; currently de-activated and not used in the GEP |
When the calibration process is complete you may proceed with running an electrification scenario (see next section)!