Optimization variables
EnergyModelsInvestments requires that variables are declared by the package which uses it. These variables are required for being able to extend the model with the potential for investments and to follow a given nomenclature utilizing a prefix symbol.
The individual variables can be differentiated in Cost variables, Capacity variables, and Auxiliary variables.
General structure of variables
As an example, consider the capex variables. The capex variables with a prefix :cap are given by $\texttt{cap\_capex}[n_\texttt{inv}, t_\texttt{inv}]$ for Node $n_\texttt{inv}$ with investments in investment period $t_\texttt{inv}$. They are extracted using the functions functions EMI.get_var_capex(m, prefix::Symbol) and EMI.get_var_capex(m, prefix::Symbol, element) in which m corresponds to the JuMP model, prefix to a prefix used in variable declaration (in this case :cap) and element to an instance of the element (in the previous example given as $n_\texttt{inv}$).
They are not declared within EnergyModelsInvestments, but have to be declared within the model using EnergyModelsInvestments. This is illustrated in the EMIExt of EnergyModelsBase.
Cost variables
EnergyModelsInvestments requires the introduction of variables that help extracting the cost of investments in an element at each investment period. One example is given through EnergyModelsBase:
- $\texttt{cap\_capex}[n_\texttt{inv}, t_\texttt{inv}]$: Undiscounted total CAPEX of
Node$n_\texttt{inv}$ with investments in investment period $t_\texttt{inv}$.
The total CAPEX takes into account the invested capacity to calculate the total costs as well as the end of horizon value of the individual technologies including discounting.
The variable is extracted using the functions EMI.get_var_capex(m, prefix::Symbol) (for all variables) and EMI.get_var_capex(m, prefix::Symbol, element) (for only the variable from a given element instance) in which m corresponds to the JuMP model, and prefix to a prefix used in variable declaratio.
Cost variables provide the absolute costs within an investment period $t_\texttt{inv}$. An example unit would be € or $.
Capacity variables
Capacity variables are variables that manipulate the installed capacity. In general, we can differentiate in installed capacity variables and change of capacity variables. The installed capacity variables are:
- $\texttt{cap\_inst}[n, t]$: Installed capacity of
Node$n$ with investments in operational period $t$ and - $\texttt{cap\_current}[n_\texttt{inv}, t_\texttt{inv}]$: Installed capacity of
Node$n_\texttt{inv}$ with investments in investment period $t_\texttt{inv}$.
The approach is similar to the Cost variables as variables are created for each of the individual elements. They are extracted using the functions functions EMI.get_var_inst(m, prefix::Symbol) and EMI.get_var_inst(m, prefix::Symbol, element) as well as EMI.get_var_current(m, prefix::Symbol) and EMI.get_var_current(m, prefix::Symbol, element).
The variable $\texttt{cap\_inst}$ is slightly redundant in this design. It is indexed over operational periods $t$ to allow for variations in the demand on an operational level within nodes without investments. This could have been also solved by using a profile, but it was decided to keep the current design. The introduction of a new variable through $\texttt{cap\_current}$ for the capacity at an investment period simplifies the indexing.
If you do not have nodes without investments, you still have to create the variable. In this case, we suggest to not use this variable at all and link the capacity usage directly to $\texttt{cap\_current}$
In addition, we introduce variables for investments in an investment period as:
- $\texttt{cap\_add}[n_\texttt{inv}, t_\texttt{inv}]$: Added capacity of
Node$n_\texttt{inv}$ with investments in the beginning of investment period $t_\texttt{inv}$.
The investments are available at the beginning of an investment period. They are extracted using the functions functions EMI.get_var_add(m, prefix::Symbol) and EMI.get_var_add(m, prefix::Symbol, element).
The model can also choose to retire technologies at the end of each investment period through removal variables given as:
- $\texttt{cap\_rem}[n_\texttt{inv}, t_\texttt{inv}]$: Retired capacity of
Node$n_\texttt{inv}$ with investments at the end of investment period $t_\texttt{inv}$.
The retired capacity corresponds to removal of capacity, either due to the end of lifetime or due to lack of usage. Capacity removal has an impact on the objective function due to removal of the fixed OPEX. It can hence be beneficial for the model to remove unused capacity to avoid fixed OPEX when the technology is not used in the future. Early removal of a technology, that is before the end of its lifetime, does not provide a rest value to the objective function. The variables are extracted using the functions functions EMI.get_var_rem(m, prefix::Symbol) and EMI.get_var_rem(m, prefix::Symbol, element).
The units of the capacity variables are defined by the user. Within EnergyModelsBase, we use both rates (normal nodes as well as charge and discharge capacity of Storage nodes) and energy/mass (level capacity of Storage nodes). Hence, it is important to consider the requirement of the model when deciding the unit.
Auxiliary variables
Auxiliary variables are variables that are required for certain investment modes. The model creates these variables only if the investment mode requires them. The meaning of the auxiliary variables changes depending on the investment mode.
These variables are:
- $\texttt{cap\_invest\_b}[n_\texttt{inv}, t_\texttt{inv}]$ and
- $\texttt{cap\_remove\_b}[n_\texttt{inv}, t_\texttt{inv}]$,
accessed through the functions EMI.get_var_invest_b(m, prefix::Symbol), EMI.get_var_invest_b(m, prefix::Symbol, element), EMI.get_var_remove_b(m, prefix::Symbol), and EMI.get_var_remove_b(m, prefix::Symbol, element).
The auxiliary variables are only created if the investment mode requires them through the application SparseVariables.
BinaryInvestment
The variable $\texttt{cap\_invest\_b}$ is a binary and used in calculating the current capacity. The variable $\texttt{cap\_remove\_b}$ is not included in any constraint.
DiscreteInvestment
The variable $\texttt{cap\_invest\_b}$ is an integer and used in calculating the added capacity. The variable $\texttt{cap\_remove\_b}$ is an integer and used in calculating the removed capacity.
SemiContiInvestment
The variable $\texttt{cap\_invest\_b}$ is a binary and used in calculating the added capacity by providing a lower bound as well as in the CAPEX calculation in the case of SemiContinuousOffsetInvestment. The variable $\texttt{cap\_remove\_b}$ is not included in any constraint.
FixedInvestment
The variable $\texttt{cap\_invest\_b}$ is fixed to 1 and used in calculating the current capacity. The variable $\texttt{cap\_remove\_b}$ is not included in any constraint.