When making investment decisions, entrepreneurs might only build power plants if they expect to at least recover all their variable of production and fixed costs, including build costs, debt costs and required return on investment.
Revenues can come from multiple sources including:
In a perfectly competitive electricity market generation plant will offer generation into the market at short-run marginal cost (SRMC), enough to recover only their variable cost of production: specifically fuel and certain operation and maintenance costs. Generators will be dispatched in 'merit order' of SRMC, with the market price set by the last dispatched generator.
The capacity expansion formulation in LT Plan models this production simulation and market price mechanism precisely but also includes the expansion formulation, building new resources at least cost or retiring plant as economics dictate. One might assume that the optimal solution to the combined production and expansion problem would result market prices that reflect not just SRMC but LRMC (long-run marginal cost) i.e. the incremental cost of capacity and production combined.
In reality though market prices in LT Plan tend towards LRMC only under particular circumstances i.e. when:
Capacity expansion decisions are rarely continuous. Instead plant must be built in discrete 'lumps'. The LT Plan problem then becomes a mixed integer programming problem (MIP) where the integer decisions control the building or otherwise of expansion options. The optimal solution to this MIP is often a solution where the 'marginal' investment's costs are not fully recovered by the market price. From the optimizer's viewpoint the two alternatives are worse: either build less capacity and incur more unserved energy, or build more capacity and incur greater losses. This is a classic example of a 'duality gap' in integer problems where there is an inconsistency between the primal (physical) solution and the dual (pricing) solution due to binding integrality conditions.
If constraints, such as minimum capacity reserve margins, mean that more capacity must be built than would be economic otherwise, these constraints bind and the shadow price forces down the market price leading to some new builds incurring losses.
In either case then there is no guarantee that all plant will recover their variable and fixed costs through the spot market alone, particularly marginal units which will receive little 'rent' from their production as market prices will be low relative to their implied LRMC.
Commonly generators will mitigate risk in the spot market by striking contracts with purchasers. The Financial Contract class is used to model these contracts. Contracts can be contract-for-difference (CfD) or floor or cap style. The contracts are settled ex-post and reporting flows through to the Company reporting.
Financial Contracts can be defined on generator projects that are expansion candidates and the contract settlement will be calculated by the simulator only when the expansion option is taken up.
Unless new generation projects are 100% contracted however, there will still be some revenue required from the spot market, and thus the underlying issue of revenue adequacy remains.
The long-run marginal cost of generation is reported by Region and Zone with the LRMC output property.
The capacity payments used in this calculation are calculated regardless of whether or not the LT Plan Capacity Payments Enabled setting is enabled thus the LRMC output is always available.
It should now be clear that the 'simplest' method to addressing revenue adequacy is to assume the existence of a capacity market or other out-of-market mechanism that would provide capacity payments. The Region and Zone outputs Capacity Price and Capacity Payments can then be used to estimate the price of capacity and the size of the capacity payment 'pool' required.
The simulator will calculate the capacity payments required to make the marginal investment break even given how much it makes in the energy market versus its annualised fixed costs.
The question now is:
What happens if there is no capacity market or other out-of-market mechanism that compensates investments made for capacity purposes?Generators will need to exercise some degree of market power in order to lift the electric price above SRMC until the price alone is sufficient for the marginal investment to breakeven.
The feature relevant here is the "LRMC Recovery". This is an option under Competition settings. This method will iteratively modify Generator Mark-up until both variable and fixed costs are recovered. The algorithm runs in MT Schedule and ST Schedule not directly in LT Plan. LT Plan first selects the optimal set of investments and passes to MT Schedule the annualised fixed cost charges on the Generator objects.
The result of the LRMC Recovery algorithm is electric market prices that are consistent with the investments selected by LT Plan. In summary then:
In addition to LRMC Recovery the following models of imperfect competition can be used in MT Schedule and ST Schedule:
These models do not take input from LRMC however, but derive competitive market prices directly from the market dynamics. Although Nash-Cournot will adapt as expansion occurs, RSI less so and may need substantial tuning of parameters to achieve useful outcomes.
The next question is:
How do I model non-marginal cost bidding directly in LT Plan?In the above scheme the recovery of fixed costs is done after the expansion plan is determined. You may want to see the impact of non-marginal cost bidding of generators on the expansion plan itself. You can set mark-ups a priori using either the Generator Mark-up or Bid-Cost Mark-up properties. The latter is easier to use in that it defines a percentage mark-up rather than an absolute value and thus it automatically changes the mark-up applied to generation as costs change over time.
LT Plan finds the expansion and production solution that maximizes net social benefit but now taking into account the impact of generator bidding behaviour.
How do I determine appropriate mark-ups?The mark-ups can be extracted from a run of the LRMC Recovery method. The relevant properties are Generator Mark-up and/or Bid-Cost Mark-up or mark-ups can be estimated from historical observation.