Investment Optimization

SpineOpt offers numerous ways to optimise investment decisions energy system models and in particular, offers a number of methologogies for capturing increased detail in investment models while containing the impact on run time. The basic principles of investments will be discussed first and this will be followed by more advanced approaches.

Key concepts for investments

Investment Decisions

These are the investment decisions that SpineOpt currently supports. At a high level, this means that the activity of the entities in question is controlled by an investment decision variable. The current implementation supports investments in:

• unit
• connection
• node storage

Investment Variable Types

In all cases the capacity of the unit or connection or the maximum node state of a node is multiplied by the investment variable which may either be continuous or integer. This is determined, for units, by setting the unit_investment_variable_type parameter accordingly. Similary, for connections and node storages the connection_investment_variable_type and storage_investment_variable_type are specified.

Identiying Investment Candidate Units, Connections and Storages

The parameter candidate_units represents the number of units of this type that may be invested in. candidate_units determines the upper bound of the investment variable and setting this to a value greater than 0 identifies the unit as an investment candidate unit in the optimisation. If the unit_investment_variable_type is set to :unit_investment_variable_type_integer, the investment variable can be interpreted as the number of discrete units that may be invested in. However, if unit_investment_variable_type is :unit_investment_variable_type_continuous and the unit_capacity is set to unity, the investment decision variable can then be interpreted as the capacity of the unit rather than the number of units with candidate_units being the maximum capacity that can be invested in. Finally, we can invest in discrete blocks of capacity by setting unit_capacity to the size of the investment capacity blocks and have unit_investment_variable_type set to :unit_investment_variable_type_integer with candidate_units representing the maximum number of capacity blocks that may be invested in. The key points here are:

• The upper bound on the relevant flow variables are determined by the product of the investment variable and the unit_capacity for units, connection_capacity for connections or node_state_cap for storages.
• candidate_units sets the upper bound on the investment variable, candidate_connections for connections and candidate_storages for storages
• unit_investment_variable_type determines whether the investment variable is integer or continuous (connection_investment_variable_type for connections and storage_investment_variable_type for storages).

Investment Costs

Investment costs are specified by setting the appropriate *_investment\_cost parameter. The investment cost for units are specified by setting the unit_investment_cost parameter. This is currently interpreted as the full cost over the investment period for the unit. See the section below on investment temporal structure for setting the investment period. If the investment period is 1 year, then the corresponding unit_investment_cost is the annualised investment cost. For connections and storages, the investment cost parameters are connection_investment_cost and storage_investment_cost, respectively.

Temporal and Stochastic Structure of Investment Decisions

SpineOpt's flexible stochastic and temporal structure extend to investments where individual investment decisions can have their own temporal and stochastic structure independent of other investment decisions and other model variables. A global temporal resolution for all investment decisions can be defined by specifying the relationship model__default_investment_temporal_block. If a specific temporal resolution is required for specific investment decisions, then one can specify the following relationships:

• unit__investment_temporal_block for units,
• connection__investment_temporal_block for connections, and
• node__investment_temporal_block for storages.

Specifying any of the above relationships will override the corresponding model__default_investment_temporal_block.

Similarly, a global stochastic structure can be defined for all investment decisions by specifying the relationship model__default_investment_stochastic_structure. If a specific stochastic structure is required for specific investment decisions, then one can specifying the following relationships:

• unit__investment_stochastic_structure for units,
• connection__investment_stochastic_structure for connections, and
• node__investment_stochastic_structure for storages.

Specifying any of the above relationships will override the corresponding model__default_investment_stochastic_structure.

Impact of connection investments on network characteristics

The model parameter use_connection_intact_flow is available to control whether or not the impact of connection investments on the network characteristics should be captured. If set to true, then the model will use line outage distribution factors (LODF) to compute the impact of each connection investment over the flow across the network. Note that this introduces another variable, connection_intact_flow, representing the hypothetical flow on a connection in case all connection investments were in place. Also note that the impact of each connection is captured individually.

Creating an Investment Candidate Unit Example

If we have model that is not currently set up for investments and we wish to create an investment candidate unit, we can take the following steps.

• Create the unit object with all the relationships and parameters necessary to describe its function.
• Ensure that the number_of_units parameter is set to zero so that the unit is unavailable unless invested in.
• Set the candidate_units parameter for the unit to 1 to specify that a maximum of 1 new unit of this type may be invested in by the model.
• Set the unit_investment_variable_type to unit_investment_variable_type_integer to specify that this is a discrete unit investment decision.
• Specify the unit_investment_lifetime of the unit to, say, 1 year to specify that this is the minimum amount of time this new unit must be in existence after being invested in.
• Specify the investment period for this unit's investment decision in one of two ways:
  • Define a default investment period for all investment decisions in the model as follows:
    • create a temporal_block with the appropriate resolution (say 1 year)
    • link this to your model object by creating the appropriate model__temporal_block relationship
    • set it as the default investment temporal block by setting model__default_investment_temporal_block
  • Or, define an investment period unique to this investment decision as follows:
    • creating a temporal_block with the appropriate resolution (say 1 year)
    • link this to your model object by creating the appropriate model__temporal_block relationship
    • specify this as the investment period for your unit's investment decision by setting the appropriate unit__investment_temporal_block relationship
• Similarly to the above, define the stochastic structure for the unit's investment decision by specifying either model__default_investment_stochastic_structure or unit__investment_stochastic_structure
• Specify your unit's investment cost by setting the unit_investment_cost parameter. Since we have defined the investment period above as 1 year, this is therefore the unit's annualised investment cost.

Model Reference

Variables for investments

Relationships for investments

Parameters for investments

Related Model Files