8.1. Adding a new technology

8.1.1. Input Files

MUSE is made up of a number of different input files. These, however, can be broadly split into two:

• Simulation settings

• Simulation data

Simulation settings specify how a simulation should be run. For example, which sectors to run, for how many years, the benchmark years and what to output. In this context, benchmark years are the years in which the model is solved. In the examples following, we solve for every 5 years, ie. 2020, 2025, 2030…

Whereas, simulation data parametrises the technologies involved in the simulation, or the number and kinds of agents.

To create a customised case study it is necessary to edit both of these file types.

Simulation settings are specified in a TOML file. TOML is a simple, extensible and intuitive file format well suited for specifying small sets of complex data.

Simulation data is specified in CSV. This is a common format used for larger datasets, and is made up of columns and rows, with a comma used to differentiate between entries.

MUSE requires at least the following files to successfully run:

• a single simulation settings TOML file for the simulation as a whole

• a file indicating initial market price projections

• a file describing the commodities in the simulation

• for generalized sectors:

  • a file descring the agents

  • a file descring the technologies

  • a file descring the input commodities for each technology

  • a file descring the output commodities for each technology

  • a file descring the existing capacity of a given sector

• for each preset sector:

  • a csv file describing consumption for the duration of the simulation

For a full description of these files see the input files section. To see how to customise an example, continue on this page.

8.1.2. Addition of solar PV

In this section, we will add solar photovoltaics to the default model seen in the example page. To achieve this, we must modify some of the input files shown in the above section. To get the default model and be able to edit the files use:

python -m muse --model default --copy PATH/TO/COPY/THE/MODEL/TO

You can modify the files in your favourite spreadsheet editor or text editor such as VSCODE, Excel, Numbers, Notepad or TextEdit.

8.1.3. Technodata Input

Within the default folder there is the settings.toml file, input folder and technodata folder. To add a technology within the power sector, we must open the technodata folder followed by the power folder.

At this point, we must note that we require consistency in input and output units. For example, if capacity is in PJ, the same basis would be needed for the output files CommIn.csv and CommOut.csv. In addition, across sectors a commodity needs to maintain the same unit. In these examples, we use the unit petajoule (PJ).

Next, we will edit the CommIn.csv file, which specifies the commodities consumed by solar photovoltaics.

The table below shows the original CommIn.csv version in normal text, and the added column and row in bold.

ProcessName	RegionName	Time	Level	electricity	gas	heat	CO2f	wind	solar
Unit		Year		PJ/PJ	PJ/PJ	PJ/PJ	kt/PJ	PJ/PJ	PJ/PJ
gasCCGT	R1	2020	fixed	0	1.67	0	0	0	0
windturbine	R1	2020	fixed	0	0	0	0	1	0
solarPV	R1	2020	fixed	0	0	0	0	0	1

We must first add a new row at the bottom of the file, to indicate the new solar photovoltaic technology:

• we call this technology solarPV

• place it in region R1

• the data in this row is associated to the year 2020

• the input type is fixed

• solarPV consumes solar

As the solar commodity has not been previously defined, we must define it by adding a column, which we will call solar. We fill out the entries in the solar column, ie. that neither gasCCGT nor windturbine consume solar.

We repeat this process for the file: CommOut.csv. This file specifies the output of the technology. In our case, solar photovoltaics only output electricity. This is unlike gasCCGT which also outputs CO2f, or carbon dioxide.

ProcessName	RegionName	Time	Level	electricity	gas	heat	CO2f	wind	solar
Unit		Year		PJ/PJ	PJ/PJ	PJ/PJ	kt/PJ	PJ/PJ	PJ/PJ
gasCCGT	R1	2020	fixed	1	0	0	91.67	0	0
windturbine	R1	2020	fixed	1	0	0	0	0	0
solarPV	R1	2020	fixed	1	0	0	0	0	0

Similar to the the CommIn.csv, we create a new row, and add in the solar commodity. We must ensure that we call our new commodity and technologies the same as the previous file for MUSE to successfully run. ie solar and solarPV.

Please note that we use flat forward extension of the values when only one value is defined. For example, in the CommOut.csv we only provide data for the year 2020. Therefore for the benchmark years, 2025, 2030, 2035… we assume the data remains unchanged from 2020.

The next file to modify is the ExistingCapacity.csv file. This file details the existing capacity of each technology, per benchmark year. For this example, we will set the existing capacity to be 0. Please note, that the model interpolates between years linearly.

ProcessName	RegionName	Unit	2020	2025	2030	2035	2040	2045	2050
gasCCGT	R1	PJ/y	1	1	0	0	0	0	0
windturbine	R1	PJ/y	0	0	0	0	0	0	0
solarPV	R1	PJ/y	0	0	0	0	0	0	0

Finally, the technodata.csv contains parametrisation data for the technology, such as the cost, growth constraints, lifetime of the power plant and fuel used. The technodata file is too long for it all to be displayed here, so we will truncate the full version.

Here, we will only define the parameters: processName, RegionName, Time, Level,cap_par, Fuel,EndUse,Agent2 and Agent1

We shall copy the existing parameters from the windturbine technology for the remaining parameters that can be seen in the technodata.csv file for brevity. You can see the full file here.

Again, flat forward extension is used here. Therefore, as in this example we only provide data for the benchmark year 2020, 2025 and the following benchmark years will keep the same characteristics, e.g. costs, for each benchmark year of the simulation.

ProcessName	RegionName	Time	Level	cap_par	cap_exp	…	Fuel	EndUse	Agent2
Unit		Year		MUS$2010/PJ_a		…			Retrofit
gasCCGT	R1	2020	fixed	23.78234399	1	…	gas	electricity	1
windturbine	R1	2020	fixed	36.30771182	1	…	wind	electricity	1
solarPV	R1	2020	fixed	30	1	…	solar	electricity	1

8.1.4. Global inputs

Next, navigate to the input folder, found at

{muse_installation_location}src/muse/data/example/default/input

We must now edit each of the files found here to add the new solar commodity. Due to space constraints we will not display all of the entries contained in the input files. The edited files can be viewed here however.

The BaseYearExport.csv file defines the exports in the base year. For our example we add a column to indicate that there is no export for solar. However, it is important that a column exists for our new commodity.

It is noted, however, that the BaseYearImport.csv as well as the BaseYearExport.csv files are optional files to define exogenous imports and exports; all values are set to zero if they are not used.

RegionName	Attribute	Time	electricity	gas	heat	CO2f	wind	solar
Unit		Year	PJ	PJ	PJ	kt	PJ	PJ
R1	Exports	2010	0	0	0	0	0	0
R1	Exports	2015	0	0	0	0	0	0
…	…	…	…	…	…	…	…	…
R1	Exports	2100	0	0	0	0	0	0

The BaseYearImport.csv file defines the imports in the base year. Similarly to BaseYearExport.csv, we add a column for solar in the BaseYearImport.csv file. Again, we indicate that solar has no imports.

RegionName	Attribute	Time	electricity	gas	heat	CO2f	wind	solar
Unit		Year	PJ	PJ	PJ	kt	PJ	PJ
R1	Imports	2010	0	0	0	0	0	0
R1	Imports	2015	0	0	0	0	0	0
…	…	…	…	…	…	…	…	…
R1	Imports	2100	0	0	0	0	0	0

The GlobalCommodities.csv file is the file which defines the commodities. Here we give the commodities a commodity type, CO2 emissions factor and heat rate. For this file, we will add the solar commodity, with zero CO2 emissions factor and a heat rate of 1.

Commodity	CommodityType	CommodityName	CommodityEmissionFactor_CO2	HeatRate	Unit
Electricity	Energy	electricity	0	1	PJ
Gas	Energy	gas	56.1	1	PJ
Heat	Energy	heat	0	1	PJ
Wind	Energy	wind	0	1	PJ
CO2fuelcomsbustion	Environmental	CO2f	0	1	kt
Solar	Energy	solar	0	1	PJ

The projections.csv file details the initial market prices for the commodities. The market clearing algorithm will update these throughout the simulation, however, an initial estimate is required to start the simulation. As solar energy is free, we will indicate this by adding a final column.

Please note that the unit row is not read by MUSE, but used as a reference for the user. The units should be consistent across all input files for MUSE; MUSE does not carry out any unit conversion.

RegionName	Attribute	Time	electricity	gas	heat	CO2f	wind	solar
Unit		Year	MUS$2010/PJ	MUS$2010/PJ	MUS$2010/PJ	MUS$2010/kt	MUS$2010/PJ	MUS$2010/PJ
R1	CommodityPrice	2010	14.81481472	6.6759	100	0	0	0
R1	CommodityPrice	2015	17.89814806	6.914325	100	0.052913851	0	0
…	…	…	…	…	…	…	…	…
R1	CommodityPrice	2100	21.39814806	7.373485819	100	1.871299697	0	0

8.1.5. Running our customised simulation

Now we are able to run our simulation, with the new solar power technology.

To do this we run the same run command as previously in the anaconda command prompt:

python -m muse settings.toml

The output should be similar to the output here. However, expect the simulation to take slightly longer to run. This is due to the additional calculations made.

If the simulation has run successfully, you should now have a folder in the same location as your settings.toml file called Results. The next step is to visualise the results using the python visualisation package seaborn as well as the data analysis library pandas.

import pandas as pd
import seaborn as sns

Next, we will import the MCACapacity.csv file into pandas and print the first 5 lines using the head() command.

Make sure to change the file path of "../tutorial-code/1-add-new-technology/introduction/Results/MCACapacity.csv" to where the MCACapacity.csv is on your computer, otherwise you will receive an error when you import the csv file.

mca_capacity = pd.read_csv(
    "../tutorial-code/1-add-new-technology/1-introduction/Results/MCACapacity.csv"
)
mca_capacity.head()

	Unnamed: 0	agent	capacity	dst_region	index	installed	region	sector	technology	type	year
0	2	A1	10.0	R1	0	2020	R1	residential	gasboiler	retrofit	2020
1	0	A1	1.0	R1	0	2020	R1	power	gasCCGT	retrofit	2020
2	0	A1	15.0	R1	0	2020	R1	gas	gassupply1	retrofit	2020
3	2	A1	5.0	R1	0	2020	R1	residential	gasboiler	retrofit	2025
4	7	A1	19.0	R1	16	2020	R1	residential	heatpump	retrofit	2025

We will only visualise the power sector in this example, as this was the only sector we changed. Therefore, we filter for this sector, and then visualise it using seaborn:

power_capacity = mca_capacity[mca_capacity.sector == "power"]
sns.lineplot(data=power_capacity, x="year", y="capacity", hue="technology")

<Axes: xlabel='year', ylabel='capacity'>

We can now see that there is solarPV in addition to windturbine and gasCCGT, when compared to the example here! That’s great and means it worked!

The difference in uptake of solarPV compared to windturbine is due to the fact that solarPV has a lower cap_par cost of 30, compared to the windturbine. Meaning that solarPV outcompetes both windturbine and gasCCGT in the electricity market.

8.1.6. Change Solar Price

Now, we will observe what happens if we increase the price of solar to be more expensive than wind in the year 2020, but then reduce the price of solar in 2040. By doing this, we should observe an increase in wind in the first few benchmark years of the simulation, followed by a transition to solar as we approach the year 2040.

To achieve this we have to modify the Technodata.csv, CommIn.csv and CommOut.csv files.

First, we will amend the Technodata.csv file as follows:

ProcessName	RegionName	Time	Level	cap_par	cap_exp	…	Fuel	EndUse	Agent2
Unit		Year		MUS$2010/PJ_a		…			Retrofit
gasCCGT	R1	2020	fixed	23.78234399	1	…	gas	electricity	1
gasCCGT	R1	2040	fixed	23.78234399	1	…	gas	electricity	1
windturbine	R1	2020	fixed	36.30771182	1	…	wind	electricity	1
windturbine	R1	2040	fixed	36.30771182	1	…	wind	electricity	1
solarPV	R1	2020	fixed	40	1	…	solar	electricity	1
solarPV	R1	2040	fixed	30	1	…	solar	electricity	1

Notice that we must provide entries for 2040 for the other technologies, gasCCGT and windturbine. For this example, we will keep these the same as before, copying and pasting the rows.

Here, we increase the cap_par variable by 10 for solarPV in the year 2020, to be a total of 40, and then reduce cap_par by 10 in 2040, again for solarPV.

MUSE uses interpolation for the years which are unknown. So in this example, for the benchmark years between 2020 and 2040 (2025, 2030, 2035), MUSE uses interpolated cap_par values. The interpolation mode can be set in the settings.toml file, and defaults to linear interpolation. This example uses the default setting for interpolation.

Next we will modify the CommIn.csv file.

For this step, we have to provide the input commodities for each technology, in each of the years defined in the Technodata.csv file. So, for this example we are required to provide entries for the years 2020 and 2040 for each of the technologies. For now, we won’t change the 2040 values compared to the 2020. Therefore, we just need to copy and paste each of the entries for each of the technologies, as shown below:

ProcessName	RegionName	Time	Level	electricity	gas	heat	CO2f	wind	solar
Unit		Year		PJ/PJ	PJ/PJ	PJ/PJ	kt/PJ	PJ/PJ	PJ/PJ
gasCCGT	R1	2020	fixed	0	1.67	0	0	0	0
gasCCGT	R1	2040	fixed	0	1.67	0	0	0	0
windturbine	R1	2020	fixed	0	0	0	0	1	0
windturbine	R1	2040	fixed	0	0	0	0	1	0
solarPV	R1	2020	fixed	0	0	0	0	0	1
solarPV	R1	2040	fixed	0	0	0	0	0	1

We must do the same for the CommOut.csv file. For the sake of brevity we won’t show you this, but the link to the file can be found here.

We will now rerun the simulation, using the same command as previously and visualise the new results.

We must import the new MCACapacity.csv file again, and then visualise the results.

mca_capacity = pd.read_csv(
    "../tutorial-code/1-add-new-technology/2-scenario/Results/MCACapacity.csv"
)
power_capacity = mca_capacity[mca_capacity.sector == "power"]
sns.lineplot(data=power_capacity, x="year", y="capacity", hue="technology")

<Axes: xlabel='year', ylabel='capacity'>

From the results, we can see that windturbine increases rapidly between 2025 and 2040. However, between the years 2030 and 2050 solarPV increases. This is because of the changing cost of solarPV during these years. A crossover can be seen around the year 2038.

For the full example with the completed input files see here

8.1.7. Next steps

In the next section we will add a new agent to the simulation. Note, that we will keep the additional solarPV technology, as well as the changing costs in 2040.