Merge pull request #1071 from NREL-Sienna/jdgk/cost_functions

Initial refactor of cost functions

docs/src/modeler_guide/cost_functions.md

@@ -0,0 +1,16 @@
+# PowerSystems Cost Functions
+
+PowerSystems.jl provides an extensive type hierarchy to explicitly express relationships between power production and cost. This lets the modeler represent cost functions as linear, quadratic, or piecewise input-output curves, potentially piecewise marginal heat rates, average heat rates, and more, as best fits the input data.
+
+To represent a cost for a particular [`Component`](@ref), the modeler first chooses one of the variable cost representations in the table below. Then, they wrap this [`ProductionVariableCost`](@ref) in either a [`CostCurve`](@ref) to indicate a cost in currency or in a [`FuelCurve`](@ref) to indicate a cost per unit of fuel plus a fuel cost. Finally, the user creates a domain-specific [`OperationalCost`](@ref) that contains this variable cost as well as other costs that may exist in that domain, such as a fixed cost that is always incurred when the unit is on. For instance, we may have `RenewableGenerationCost(CostCurve(TODO), 0.0)` to represent the cost of a renewable unit that produces at TODO, or `ThermalGenerationCost(; variable = FuelCurve(TODO), fixed = TODO, start_up = TODO, shut_down = TODO)` to represent the cost of a thermal unit that produces at TODO. Below, we give the options for `ProductionVariableCost`s. Information on what domain-specific cost must be provided for a given component type can be found in that component type's documentation.
+
+## Variable Cost Representations
+For more details, see the documentation page for each type.
+| Type alias | Description | Constructor parameters | Example |
+| --- | --- | --- | --- |
+| `LinearCurve` | Linear input-output curve with zero no-load cost (constant average rate) | Average/marginal rate | `LinearCurve(3.0)` |
+| `LinearCurve` | Linear input-output curve with nonzero no-load cost (constant marginal rate) | Marginal rate, cost at zero production | `LinearCurve(3.0, 5.0)` |
+| `QuadraticCurve` | Quadratic input-output curve, may have nonzero no-load cost | Quadratic, proportional, and constant terms of input-output curve | `QuadraticCurve(1.0, 1.0, 18.0)` |
+| `PiecewisePointCurve` | Piecewise linear curve specified by cost values at production points | Vector of (production, cost) pairs | `PiecewisePointCurve([(1.0, 20.0), (2.0, 24.0), (3.0, 30.0)])` |
+| `PiecewiseIncrementalCurve` | Piecewise linear curve specified by marginal rates (slopes) between production points, may have nonzero initial value | Cost at minimum production point, vector of $n$ production points, vector of $n-1$ marginal rates/slopes of the curve segments between the points | `PiecewiseIncrementalCurve(20., [1.0, 2.0, 3.0], [4.0, 6.0])` |
+| `PiecewiseAverageCurve` | Piecewise linear curve specified by average rates between production points, may have nonzero initial value | Cost at minimum production point, vector of $n$ production points, vector of average rates at the $n-1$ latter points | `PiecewiseAverageCurve(20., [1.0, 2.0, 3.0], [12.0, 10.0])` |

docs/src/modeler_guide/market_bid_cost.md

@@ -32,7 +32,7 @@ generator = ThermalStandard(
         time_limits = (up = 0.015, down = 0.015),
         ramp_limits = (up = 5.0, down = 3.0),
         operation_cost = MarketBidCost(
-            no_load = 0.0,
+            no_load_cost = 0.0,
             start_up = (hot = 0.0, warm = 0.0, cold = 0.0),
             shut_down = 0.0,
         ),
@@ -79,8 +79,8 @@ the `Dict` with the data as follows:
 
 ### Step 3a: Adding Energy Bid TimeSeriesData to the device
 
-To add energy market bids time-series to the `MarketBidCost`, the use of `set_variable_cost!`
-is recommended. The arguments for `set_variable_cost!` are:
+To add energy market bids time-series to the `MarketBidCost`, use `set_variable_cost!`. The
+arguments for `set_variable_cost!` are:
 
 - `sys::System`: PowerSystem System
 - `component::StaticInjection`: Static injection device
@@ -93,8 +93,8 @@ set_variable_cost!(sys, generator, time_series_data)
 
 ### Step 3b: Adding Service Bid TimeSeriesData to the device
 
-Similar to adding energy market bids, for adding bids for ancillary services the use of
-`set_service_bid!` is recommended.
+Similar to adding energy market bids, for adding bids for ancillary services, use
+`set_service_bid!`.
 
 ```@repl market_bid_cost
 service = VariableReserve{ReserveUp}("example_reserve", true, 0.6, 2.0)

docs/src/modeler_guide/time_series.md

@@ -2,9 +2,9 @@
 
 ## Introduction
 
-The bulk of the data in many power system models is time series data, in order to
-organize the data the potential inherent complexity, `PowerSystems.jl` has a set of definitions
-to enable consistent modeling.
+The bulk of the data in many power system models is time series data. Given the potential
+inherent complexity, `PowerSystems.jl` has a set of definitions to organize this data and
+enable consistent modeling.
 
 - **Resolution**: The period of time between each discrete value in the data, all resolutions
   are represented using `Dates.Period` types. For instance, a Day-ahead market data set usually

docs/src/quick_start_guide.md

@@ -88,7 +88,7 @@ new_renewable = RenewableDispatch(
         prime_mover_type = PrimeMovers.WT,
         reactive_power_limits = (min = 0.0, max = 0.0),
         base_power = 100.0,
-        operation_cost = TwoPartCost(22.0, 0.0),
+        operation_cost = RenewableGenerationCost(CostCurve(LinearCurve(22.0))),
         power_factor = 1.0
     )
 

src/PowerSystems.jl

@@ -42,25 +42,25 @@ export PhaseShiftingTransformer
 export FunctionData
 export LinearFunctionData
 export QuadraticFunctionData
-export PolynomialFunctionData
-export PiecewiseLinearPointData
-export PiecewiseLinearSlopeData
+export PiecewiseLinearData
+export PiecewiseStepData
 export get_proportional_term
 export get_quadratic_term
 export get_constant_term
-export get_coefficients
 export get_slopes
 export get_x_lengths
 export is_convex
 export get_points
 export get_x_coords
-export get_y0
+export get_y_coords
 
-export ThreePartCost
-export TwoPartCost
-export MultiStartCost
-export MarketBidCost
-export StorageManagementCost
+export ValueCurve, InputOutputCurve, IncrementalCurve, AverageRateCurve
+export LinearCurve, QuadraticCurve
+export PiecewisePointCurve, PiecewiseIncrementalCurve, PiecewiseAverageCurve
+export ProductionVariableCost, CostCurve, FuelCurve
+export OperationalCost, MarketBidCost, LoadCost, StorageCost
+export HydroGenerationCost, RenewableGenerationCost, ThermalGenerationCost
+export get_function_data, get_initial_input, get_value_curve, get_power_units, get_fuel_cost
 
 export Generator
 export HydroGen
@@ -340,6 +340,12 @@ export get_data
 export iterate_components
 export get_time_series_multiple
 export get_variable_cost
+export get_no_load_cost
+export get_start_up
+export get_shut_down
+export get_incremental_offer_curves
+export get_decremental_offer_curves
+export get_ancillary_service_offers
 export get_services_bid
 export set_variable_cost!
 export set_service_bid!
@@ -453,7 +459,7 @@ import Logging
 import Dates
 import TimeSeries
 import DataFrames
-import DataStructures: OrderedDict
+import DataStructures: OrderedDict, SortedDict
 import JSON3
 import CSV
 import YAML
@@ -543,19 +549,18 @@ import InfrastructureSystems:
     FunctionData,
     LinearFunctionData,
     QuadraticFunctionData,
-    PolynomialFunctionData,
-    PiecewiseLinearPointData,
-    PiecewiseLinearSlopeData,
+    PiecewiseLinearData,
+    PiecewiseStepData,
     get_proportional_term,
     get_quadratic_term,
     get_constant_term,
-    get_coefficients,
     get_slopes,
+    running_sum,
     get_x_lengths,
     is_convex,
     get_points,  # TODO possible rename to disambiguate from geographical information
     get_x_coords,
-    get_y0,  # TODO reevaluate whether this should be exported
+    get_y_coords,
     get_raw_data,
     get_raw_data_type
 
@@ -598,7 +603,6 @@ include("models/static_injection_subsystem.jl")
 # PowerSystems models
 include("models/topological_elements.jl")
 include("models/branches.jl")
-include("models/operational_cost.jl")
 #include("models/network.jl")
 
 # Static types
@@ -611,6 +615,18 @@ include("models/dynamic_generator_components.jl")
 include("models/dynamic_inverter_components.jl")
 include("models/OuterControl.jl")
 
+# Costs
+include("models/cost_functions/ValueCurves.jl")
+include("models/cost_functions/cost_aliases.jl")
+include("models/cost_functions/variable_cost.jl")
+include("models/cost_functions/operational_cost.jl")
+include("models/cost_functions/MarketBidCost.jl")
+include("models/cost_functions/HydroGenerationCost.jl")
+include("models/cost_functions/LoadCost.jl")
+include("models/cost_functions/RenewableGenerationCost.jl")
+include("models/cost_functions/StorageCost.jl")
+include("models/cost_functions/ThermalGenerationCost.jl")
+
 # Include all auto-generated structs.
 include("models/generated/includes.jl")
 include("models/regulation_device.jl")

src/data_format_conversions.jl

@@ -90,23 +90,43 @@ function _convert_cost(points::Vector)
     @assert all(length.(points) .== 2)
     @assert all([all(typeof.(point) .<: Real) for point in points])
     # NOTE: old representation stored points as (y, x); new representation is (x, y)
-    return PiecewiseLinearPointData([(x, y) for (y, x) in points])
+    return PiecewiseLinearData([(x, y) for (y, x) in points])
 end
 
+# _convert_op_cost: take a component type, an old operational cost type, and old operational
+# cost data; and create the proper new operational cost struct. Some of these cost structs
+# no longer exist, so we dispatch instead on symbols.
+_convert_op_cost(::Val{:ThermalStandard}, ::Val{:ThreePartCost}, op_cost::Dict) =
+    ThermalGenerationCost(
+        CostCurve(InputOutputCurve(op_cost["variable"])),
+        op_cost["fixed"],
+        op_cost["start_up"],
+        op_cost["shut_down"],
+    )
+
+# TODO implement remaining _convert_op_cost methods 
+
 function _convert_data!(
     raw::Dict{String, Any},
     ::Val{Symbol("3.0.0")},
     ::Val{Symbol("4.0.0")},
 )
     for component in vcat(raw["data"]["components"], raw["data"]["masked_components"])
+        # Convert costs: all old cost structs are in fields named `operation_cost`
         if haskey(component, "operation_cost")
             op_cost = component["operation_cost"]
+            # Step 1: insert a FunctionData
             if op_cost["__metadata__"]["type"] in COST_CONTAINERS &&
                haskey(op_cost["variable"], "cost")
                 old_cost = op_cost["variable"]["cost"]
-                new_cost = IS.serialize(_convert_cost(old_cost))
+                new_cost = _convert_cost(old_cost)
                 op_cost["variable"] = new_cost
             end
+            # Step 2: convert TwoPartCost/ThreePartCost to new domain-specific cost structs
+            comp_type = Val{Symbol(component["__metadata__"]["type"])}()
+            op_cost_type = Val{Symbol(op_cost["__metadata__"]["type"])}()
+            new_op_cost = IS.serialize(_convert_op_cost(comp_type, op_cost_type, op_cost))
+            component["operation_cost"] = new_op_cost
         end
     end
 end

src/definitions.jl

@@ -4,6 +4,7 @@ const StartUpShutDown = NamedTuple{(:startup, :shutdown), Tuple{Float64, Float64
 const FromTo = NamedTuple{(:from, :to), Tuple{Float64, Float64}}
 # Exception to CamelCase convention for aliases due to confusssing reading of FromToToFrom
 const FromTo_ToFrom = NamedTuple{(:from_to, :to_from), Tuple{Float64, Float64}}
+const StartUpStages = NamedTuple{(:hot, :warm, :cold), NTuple{3, Float64}}
 
 "From http://www.pserc.cornell.edu/matpower/MATPOWER-manual.pdf Table B-4"