nismod_decision_module

NISMOD: A decision module for NISMOD 2

Design of a LCOE-based Planning Module

The challenge is to design a Planning Module for NISMOD2 that will pick interventions for the energy supply model in each planning timestep (yearly, five-yearly or decadal). Initially, interventions include the construction, upgrading and retiring of transmission and distribution connected power plants and upgrade of existing or construction of new AC power lines. Interventions are specified with a location and technological and financial attributes.

The Planning Module will consist of two agents � a Power Producer and a Regulator. The Power Producer invests in generation plant using a heuristic approach based around the levelized cost of electricity (LCOE) to prioritise investments in interventions for each planning timestep. The Regulator will constrain the investments of the Power Producers based on performance in the previous planning period.

LCOE is calculated for a generator by finding the price of electricity which solves the equality such that the discounted stream of net revenues from operating a plant equals the discounted stream of costs (capital and operating). This requires assumptions on the future operation of the plants (load factor), financial variables (such as costs and a discount rate), and the lifetime of the plant.

The Power Producers would seek to select the combination of technologies from a list of interventions ranked according the their LCOE, subject to various constraints. Initially, these constraints could be limited to supply capacity exceeding projected demand, and meeting carbon emissions targets, but could include more advanced constraints that could be computed at the planning level (i.e. prior to running the energy supply model) such as generation diversity, a proxy of risk, metrics around affordability and so on.

Constraints are imposed by a Regulator, based on performance indicators from the previous planning period. For example, if cumulative emissions in the previous period exceeded the carbon budget, a market intervention (e.g. emissions cap, carbon price or renewable portfolio incentive) would be imposed. In addition, market-based instruments, including taxes (e.g. a carbon tax), subsidies (e.g. feed-in tariff) or contracts-for-difference could be included through modifying the LCOE calculation.

This approach presents an opportunity to represent myopic decision making, where the assumptions used in the Planning Module are gauged to represent a particular type of decision maker e.g. a private company, or a government agency. In addition, LCOE is an imperfect (sub-optimal, heuristic) method for choosing interventions, and we should expect to see some difference in actual operation costs than those assumed in the LCOE calculation. Likewise, the lag between out-turn and market intervention of the Regulator could help explore the relationship between markets and agents.

Extensions and Further Work

A more advanced optimisation approach would require running iterations over the energy supply model, which while possible, would require greater computational resource.

Multiple decision makers could be represented through pooling interventions by regions, or by intervention type (e.g. along transmission or distributed connected generation, or gas/electricity).

Work Plan

September � Develop formulation of Power Producer and Regulator decision agents and prototype using minimal version of energy supply model (3 busbars, gas nodes and energy hubs)

October � Complete implementation and test with full version of the energy supply model

November � Explore parameter space of decision module

Data Requirements

Generation Data these data need to be incorporated into the energy-supply interventions list Capital Cost � Value, Unit Economic lifetime Technical lifetime Location � ideally points, or lines (otherwise areas/spatial constraints availability) Start year (available from) (Construction time) � important for nuclear given the size of the projects (Decommissioning costs) � also particularly important for nuclear

Financial Assumptions

Rate of return Discount rate Inflation rate

Stories

Calculate emissions cap/tax from cumulative emissions

Labels:

• nismod_decision_module
• nismod2

Construct LCOE from power plant interventions dict

Labels:

• nismod_decision_module
• nismod2

Tasks:

• Compile example list of interventions
• Implement LCOE algorithm
• Write test handle for EnergyAgent decision module

Develop this within NISMOD2 together with smif v1.0.0-beta

The intervention generator creates a package of discrete sector model specific interventions

Labels:

• nismod_decision_module
• nismod2
• smif

There are potentially an infinite number of interventions that could be defined in each sector. A tool could be used to define discrete interventions through definition of rules or constraints:

• realistic spatial constraints (no wind farms in SSSI)
• realistic size constraints (nuclear power stations must be 1GW; reservoirs are between 1km^2 and 15km^2 in size)

This could be implemented using a script in a SectorModel wrapper, or decision module which is parameterised by global parameters, and modified by narratives.

Decide if interventions must have values for capital cost, economic lifetime, operational lifetime

Labels:

• nismod_decision_module
• nismod2

Design question for 'interventions' - should capital cost, technical lifetime and operational lifetime always be required?

Transport currently has an intervention for vehicle electrification, which adjusts the mix of diesel/petrol/electric vehicles directly. Capital expenditure and lifetimes for this intervention might not make sense.

More generally, when modelling policy interventions (taxes, subsidies, tariffs..) we might not have costings and lifetimes.