year_vtg and year_act

[shreeyashn20]

I am using the MESSAGEix framework to build a model. My model works only when year_vtg is equal to year_act. Following are two examples

Example : Not Working

Technology	year_vtg	year_act
Grid	                2013	2021
Grid	                2013	2022
Grid	                2013	2023

Example : Working

Technology	year_vtg	year_act
Grid	                2021	2021
Grid	                2022	2022
Grid	                2023	2023

The Westeros tutorial uses vintage_and_active_years() function which gives pair of same year_vtg and year_act.

Now, The Grid is already installed in 2013. Why would I install new grid every year ? Please explain.

P.S. I am using MESSAGEix 3.4.0

[khaeru]

The key concept at the root of your question seems to be similar to the one in #679.

• year_vtg, "vintage year/period", or $y^V$ is the period in which capacity (CAP) of a particular technology is constructed or created (CAP_NEW). If it helps, think of a single substation in the electrical grid.
• year_act, "active year/period", or $y^A$ is the period in which the technology operates (ACT).
• As you note, a technology constructed in $y^V = 2021$ can be active in $y^A \in (2021, 2022, 2023, ... )$—which periods depends on its technical_lifetime and the duration_period of those periods.
• Equally, a technology constructed in $y^V = 2013$ could be active in $y^A \in (..., 2021, 2022, 2023, ...)$.
• A technology constructed in $y^V = 2023$ can never be active in $y^A = 2020$; this doesn't make any sense.

In MESSAGE, both ACT and CAP are decision variables: the solver chooses an optimal value of each one. The solver does this for only the periods $y$ (thus $y^V$ and $y^A$) that are in the model horizon or optimization years. A MESSAGE Scenario can contain historical periods before the model horizon. For instance, our global model contains historical periods like 1950, 1960, …, 2010, 2015; but the "first model year" or $y_0$ is 2020. So the solver can choose CAP and ACT values for $y_0 \le y$, in other words $y \in (2020, 2025, 2030, ...)$ but it cannot choose $CAP > 0$ for $y^V = 2015$, because this period is outside the model's horizon.

You haven't given a complete example, so I can only guess. But perhaps in the model you have constructed, the year 2013 is not part of the model horizon. If you give technology parameters only for $t=\text{“grid”}, y^V = 2013$, then the solver can never choose to construct this technology, and there will be no way to meet demand.

If that's correct, you could try things like:

• Adjust the model horizon, for instance by changing the first model year.
• Add values to historical_new_capacity to tell the model how much CAP was constructed in the historical periods before $y_0$.