Transcript Energy Exemplar & PLEXOS

Physical Energy Assets Valuation and Portfolio
Optimisation with PLEXOS®
Dr Christos Papadopoulos
Regional Manager Europe
Energy Exemplar (Europe) Ltd
5th Annual Quantitative Analysis in Commodity and
Energy Trading∙-Berlin, Germany
AGENDA
 About Energy Exemplar
 Modelling the intrinsic value of a generation asset via
realistic power plants operations
 Spreads and Call “Real” options valuation of energy
assets
 Physical energy assets’ portfolio optimisation
Energy Exemplar - PLEXOS® Integrated Energy Model
 PLEXOS® Integrated Energy Model - Released in 1999
 Continuously Developed to meet Challenges of a Dynamic Environment
 A Global Leader in Energy Market Simulation Software.
 Offices in Adelaide, AUSTRALIA; London, UK; California, USA-WC; Connecticut, USA-EC,
Johannesburg, SOUTH AFRICA.
 High Growth Rate in Customers and Installations
 30% staff with Ph.D. level qualifications spanning
Operations Research, Electrical Engineering,
Economics, Mathematics and Statistics
 European Office:
 Software Sales
 Customer Support
 Training
 Consulting
 European Systems/Markets &
Countries Datasets
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
About Energy Exemplar
Clients’ Portfolio in all five continents
Regional Office
Regional Office
Regional Office
Regional Office
Opening 2014
Regional Office
Head Office
Commercial and Academic experience
Academic (only) experience
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
PLEXOS® Integrated Energy Model for Energy (Power & Gas)
Systems & Markets Simulation, Optimisation & Analysis.
 Proven power market simulation tool &
Integrated Energy Model
 Uses cutting-edge Mathematical Programming
based Constrained Optimisation techniques
(LP/MILP/DP/SP),
 Robust analytical framework, used by:

Energy Producers, Traders and Retailers

Transmission System /Market Operators

Energy Regulators/Commissions

Consultants, Analysts and Research Institutions

Power Plant Manufacturers and Construction companies
 Integrated (Power & Gas) Systems/Markets
Models scalable to thousands of Generators,
transmission lines, Gas fields, Storages, Pipelines
& Nodes
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
PLEXOS® around the Globe
As of April2014 worldwide installations of PLEXOS
exceeded 820 at over 150 sites in 34 countries.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Bringing fundamentals back into the game to enhance
Modelling the intrinsic value of a generation asset via
model application and asset performance
realistic power plants operations
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
7
Co-optimisation of Physical (and financial) Power and Gas Assets with
Advanced Computational Tools (PLEXOS® Integrated Energy Model)
 In the era of Uncertainty and Complexity in Energy markets,
Power and Gas Utilities have been forced to operate into
extremely challenging operating environments, characterised by
increasingly uncertain and complex market conditions.
 Inevitably, there is a need and a strong call for advanced
computational tools that will be able to capture and handle this
complexity in the most efficient manner and to provide viable
strategic long-term and operational short-term solutions.
 PLEXOS® is such a computational tool that has been designed
and developed to provide this kind of solutions in complex
power and gas systems & markets conditions.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Modelling the intrinsic Value of a Generation Asset
 Modelling the intrinsic value of a generation asset involves
the use of fundamental modelling and the optimisation of
an appropriate Value or Payoff Function that dictates the
associated Cash Flows .
 The problem and the crucial difference of the electricity
markets though, when compared to other Commodity
(financial) markets, is that generators exercise their options
of selling power to “Cash Markets” that are not traded on a
forward (known prices) basis, but rather the dispatch of a
unit happens before all the relevant prices are known ...
(Under Prices’ uncertainty).
 This facts poses significant problems to a Proper Valuation.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Historical price analysis slowly becoming less relevant



Fundamental changes in the energy markets are already effecting
prices
 Changing government policies (EMR)
 Change in market design (coupling of markets)
 Renewable Integration/Subsidies
 Drop in energy demand and growth due to economic crisis
 Falling CO2 price
 Gas Spark spreads going to negative and falling (Expensive Gas)
 Dark spreads going positive (Cheap imported coal)
What do we have to consider next?
 Demand Side Management
 Energy Storage technologies
 Capacity markets or more importance on reserves and
balancing
 Increased electrification of rail networks
 New Government legislation and policies
Understanding renewables profiles and potential variations is
becoming more critical in forecasting costs and prices both for
optimum investment planning, valuation and portfolio optimisation.
20/05/2014
Strong decrease of the weight in the
peak hours in a typical daily profile
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Power Markets’ Models Classification – Quantitative
 Quantitative (Econometric, Reduced-form) models:
Quantitative models characterize the statistical properties of electricity
prices over time, with the ultimate objective of derivatives evaluation and
risk management. They aim to recover the main characteristics of
electricity prices, typically at the hourly/daily time scale and monthly time
horizons. Although in this context the models’ simplicity and analytical
tractability are an advantage, in accurately forecasting e.g. hourly prices is a
serious limitation, while the recovery of their main characteristics is an
excessive luxury.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Power Markets’ Models Classification – Fundamental
 Fundamental Models:
Fundamental methods are based on the most basic economic principles of
supply and demand describing price dynamics and modelling the impact of
important physical and economic factors on the market equilibrium price of
electricity. The fundamental inputs (loads, weather conditions, system
parameters) are independently modelled and predicted, often employing
statistical, econometric or non-parametric techniques. Because of the
nature of fundamental data which is typically collected over relatively long
time intervals and the data availability issues, pure fundamental models are
mostly used for medium to long-term analysis and predictions however,
recent advances in computing power have led to their adoption for shortterm predictions.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Fundamental Models’ Classification – Production Cost
 Production Cost (or cost-based) models:
Pure production-cost models simulate the operation of generating units
aiming to satisfy demand at minimum cost. They have the capability to
forecast prices on an hour-by-hour, bus-by-bus level, however, when ignore
strategic bidding practices are not well suited for today’s competitive
markets.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Fundamental Models’ Classification – Market Equilibrium
Equilibrium (Game Theoretic) approaches may be viewed as generalizations of cost-based
models amended with strategic bidding considerations. They may give good insight into whether
prices will be above marginal costs and how this might influence the players’ outcomes. But they
pose problems if more quantitative conclusions have to be drawn. Furthermore, a substantial
modelling risk is present as the players, their potential strategies, the ways they interact and the
set of payoffs have normally to be defined up-front.
Various types of equilibrium approaches have been proposed:
Perfect Equilibrium – firms are price-takers and possess no market power
Cournot-Nash Game – quantity is the strategic variable, and firms choose quantities
simultaneously, under the assumption that other firms’ quantities are fixed
Bertrand Game – price is the strategic variable, and firms choose prices simultaneously, assuming
that other firms’ prices are fixed
Supply Function Equilibrium (SFE) – entire bid functions are the strategic variables, and firms
choose their supply functions simultaneously, under the assumption that other firms’ supply
functions are fixed; a market mechanism, e.g. an ISO, then determines price and sets the
quantity. Cournot-Nash framework tends to provide higher prices than those observed in reality
and the supply function equilibrium framework requires considerable numerical computations
and consequently, has limited applicability in day to-day market operations.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Fundamentals vs Pure Quantitative modelling
Fundamental Models
 Prices are determined by supply and
demand principles
 Replicates actual market design and
intended behaviour meeting economic
and operational constraints
 Can capture technical constraints on
physical assets operating within the
market
 Allows any type of “what if” analysis into
the future
 Can allow co-optimisation of other
requirements such as ancillary services
and/or district heating load etc.
 Produce results that reflect future
structural changes e.g. carbon price
impacts, changes to market rules,
renewable integration
20/05/2014
Pure Quantitative Models
 Prices depend mostly on historical prices
and random processes
 Usually
probabilistic,
explore
the
distribution properties of prices
 Can suffer from in-sample bias of historical
data
 Scenarios only with parameters and/or
explanatory variables
 Most models cannot handle negative prices
 Result focuses on prices only
 Limited understanding of what particular
input could be causing the resulting price
5th Annual Quantitative Analysis in
Commodity & Energy Trading
PLEXOS® Sophistication mixes elements of both Quantitative and
Fundamental (both Cost & Game) Models - Hybrid Modelling
 Detailed modelling of dynamic characteristics of generators
 Technical constraints, emission constraints, fuel contracts, custom constraints etc.
 Optimal Power Flow model fully integrated with UC/ED.
 Stochastic Modelling
 User defined samples or an expected profile with an error function on most properties in
PLEXOS
 Scenarios or
 Monte-Carlo or
 Two Stage and Multi-stage SP (Decomposition options)
 Competition modelling





Perfect Competition (SRMC)
Bertrand Competition (Price the strategic variable)
Nash-Cournot Equilibrium
Residual Supply Analysis (Residual Supply Index -RSI)
LRMC Recovery algorithm (with adaptable generator peak/off peak bid-mark-ups)
 Ancillary Services
 Full energy and ancillary services co-optimised dispatch result
 Multiple reserve classes supported spinning, regulation and replacement
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Simulation Phases in PLEXOS®
LT Plan – Optimal investment
New Builds/retirements
PASA – Optimal reserve share
Maintenance Schedule
MT – Resource Allocation
Operating Policies
ST – Chronological
Unit Commitment
Detailed by-period results
1 year
20/05/2014
4 years
5th Annual Quantitative Analysis in
Commodity & Energy Trading
30 years
17
Modelling of Price Formation: What Timeframes and at
what Resolution?
PLEXOS®
LT
MT
ST
20/05/2014
• Optimal Expansion Plan
Energy prices
Capacity payments (prices)
•LT prices
• LRMC Recovery Method
• RSI
• Nash-Cournot Game
• Cost-based Efficiency
• Bertrand Game
• Nash-Cournot Game
• Uplift ex-post price
Company (player) revenue targets
Adjust bids: Mark-ups
•MT prices
Hourly (period) energy price forecast
(RT) Energy & Ancillary Services prices
•ST prices
5th Annual Quantitative Analysis in
Commodity & Energy Trading
LT to MT analysis tools in PLEXOS®
Energy Price Forecasts (Power/Fuels) & Energy Assets Valuations (Power Plants, Gas
Storages, Financial Contracts etc.).
Long & Medium Term Pricing & Valuation Methods
 LRMC Recovery Algorithm (Recovery of LT Investment Costs)
 Shadow Pricing (Bertrand game)
 Nash-Cournot Competition Prices
 Residual Supply Index (RSI)
 Uplift mechanisms
Decomposed to Short Term (Day ahead/Real Time/AS) Markets Pricing & Assets
Valuation
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
LT to MT analysis tools in PLEXOS®
Market Analysis - Resource allocation
PLEXOS® solves the operational problem in each phase (i.e. LT,
MT or ST).
MT phase solves “long term” constraints problem such as:
Storage release policies.
Emission allowance
Fuel-Energy limits.
Allocate any resource via custom constraints.
ST solves detailed period-by-period chronological Unit
commitment problems (considering the resource allocation
from MT)
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
PLEXOS® common model workflow
Uncertain Forecast Inputs
Fuel
Prices
Demand
Renewable
Forecasts
Outages
•
•
•
•
Generator technical
properties (heat rate,
ramp rates, MSL etc.)
Interconnector NTCs
System constraints
Ancillary Services
requirements
Market mechanisms
20/05/2014
Results
Hourly Price
forecasts
Technical Characteristics
•
Price Process
Replications
Statistical
Models
PLEXOS®
Fundamental
Model
Inputs for
Risk Models
Export to
External
Software
Multiple Scenarios
Mote Carlo runs
5th Annual Quantitative Analysis in
Stochastic Optimisation
Commodity & Energy Trading
Price Formation Mechanism
How prices are formed in Competitive Markets?
In a typical Pool Market, the price is determined through an auction
mechanism, usually from an ISO who determines the Electricity (Marginal
Cost) Price and schedules generation dispatch.
1. The generators and power marketers submit power supply bids to the
ISO. The bid is a set of pairs (Price, Quantity), from which a bid/supply
curve for a particular power supplier is constructed. The bid curve is a
function of the form Price = F(Quantity), which determines at which
price a generator is willing to supply a given quantity.
2. The ISO collects the bids from all the generators and sorts them by price
(merit order) to obtain the system Bid stack. Simultaneously the Demand
is determined either from respective Demand bids or by other means.
3. Finally, the MCP is determined as the highest price on the system bid
stack at which the total generation will match the demand.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Price Formation Mechanism – (Example 1-Bid Stack)
2500
Offer Price ($/MWh)
20
25
30
35
50
Offer Quantity (MWh)
50
100
200
400
600
Incr. Rating (MW)
50
150
350
750
1350
Quantity (MWh)
Generator G1
Generator G2
Offer Price ($/MWh)
18
40
50
60
70
Offer Quantity (MWh)
18
40
300
400
500
Incr. Rating (MW)
18
58
358
758
2000
1500
G1 Bid
1000
G2 Bid
500
0
MAX CAP 0
1258
50
100
Price ($/MWh)
2608
Bid Stack
Offer Price ($/MWh)
18
20
25
30
35
40
50
60
70
Offer Quantity (MWh)
18
68
168
368
768
808
1508
1908
2408
G1
0
50
150
350
750
750
1350
1350
1350
G2
18
18
18
18
18
58
158
558
1058
If the forecasted demand at a given hour is 767MWh, the MCP for this hour will be 35 $/MWh, for which
price, Generator 2 will supply 18MWh and Generator 1 the rest 749MWh. Since, Generator 1 will supply the
last MW of power, meaning that he is in the margin (marginal generator), it is its bid that determines the
MCP. At a forecasted demand of 769MWh, Generator 1 will supply 750MWh, while Generator 2 becomes
now the marginal, supplying the rest 19MW and determines the price at 40$/MWh.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Bid
Stack
Supply (bid) Curves of multiple generators
Daily development of the supply curves submitted to the California Power
Exchange during a 24-hour period
Energy Laboratory Publication # MIT_EL 00-004
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Two fundamental modelling approaches in PLEXOS®
Bid based Stack model:
 Each generator is represented with offer
price/quantity files which must be known in
advance (“backasting - calibration”) or
inferred.
 Easier to setup, no need to calculate each
element that makes up a generators SRMC.
 Can link price/quantity files to an external
source
to
update
regularly
and
automatically
 Unit commitment decisions will still be
optimised by PLEXOS such as MUT, MDT,
ramp limits, start profiles etc. However min
stable level and max capacity of units need
to be defined.
BID STACK
20/05/2014
Generation Cost based Stack model:
Each cost element makes up a generators
offer into the market can be separately
inputted
Each
generator
offer
price/quantity
calculated based on SRMC plus any mark-ups,
so it is finally transformed to a Bid-stack model.
Allows more flexibility when modelling the
overall effect of changes of certain generator
values (fuel costs, heat rates, outage rates etc.)
Harder to gain accurate technical and
commercial characteristics on competitors
plant.
If inputs are realistic, a more useful model for
price “forecasting” when compared to a stack
model, at the cost of increased run time
GENERATION (SRMC) STACK
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Plants’ Bid Stack vs Generation (SRMC) Stack
Bid Stack
20/05/2014
Generation Stack
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Impact of Growing Generation from RES on Supply Stack & the Wholesale
Power Price
Supported RES generation brings volatile and less predictable
Demand
supply
The spot prices decline (not the final price for the consumer!)
damages non-RES generation
Negative impact also - Lower utilization of non-RES generation
Hard coal
Lignite
75
Gas
50
Nuclear
25
RES
0
20
20/05/2014
40
60
5th Annual Quantitative Analysis in
Commodity & Energy Trading
80
Source: CEZ
Spreads and Call “Real” options valuation of
energy assets
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Underlying physical drivers that affect power prices?
 Once we show how the procedure that determines the market prices
work, next step is to examine what constitutes the inputs into the
procedure or else what are those random underlying drivers that
affect power prices and the intrinsic value of a generation asset in the
end.
 A logical assumption of fundamental hybrid modelling is that in
determining their bids, the generators take into consideration the cost
of power generation for each unit of their portfolio.
 The cost is primarily depend on the price(s) of fuels (for thermal
plants), the variable operation & maintenance costs of running the
unit(s), emissions costs, specific technical characteristics & constraints
of the plants and last but not least plants’ outages (forced or random)
which may have a tremendous impact on prices. Outages are a critical
price driver in conjunction with demand, especially also in periods of
capacity shortage.
5th Annual Quantitative Analysis in
20/05/2014
Commodity & Energy Trading
Spark Spreads & Spark Spread “Real Options”
The concept of “Real Options” consists in considering the power plants and production
decisions alike other commodities (financial) options, and to apply to them the well
established Option Valuation techniques.
For instance, you could consider a power plant (or a portfolio of them) as a complex Call
“Real” Option (e.g. Call “Real” Option) with the right (but not the obligation) to:
Invest in building/modifying/upgrading the plant
Consume a certain amount of resources and
Inject/Sell corresponding power output into the grid.
For this approach to work in practice, models must be flexible enough to take into account:
The detail technical characteristics & constraints of the equipment (such as startup
and shutdown profiles, ramping, minimum and maximum output, yield, CO2
emissions)
The stochastic dimension of most input data (such as electricity and fuel prices,
renewable energy production)
The possibilities of recourse (i.e. to modify our decisions as the uncertainty
decreases).
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Spark Spreads & Spark Spread “Real Options”
So in principle, a Power Plant can be described as complex Spark Spread “Real Option”, that is a Real
Option on the spread between Power and Fuel prices (mainly Gas, Coal, Oil).
If Q is plant’s load, C is plant’s built capacity, P is the forward (sell) price of electricity , Fu is the
forward price of fuel, HR is the heat rate, Vom is the variable O&M cost, Em, are the emission costs &
restrictions, Tr, transmission costs & restrictions, and Fix, all other Fixed Costs (FO&M costs, Equity
Costs, Debt costs) that have been already described:
  max  Q  P  HR* Fu  Q  Vom Em  Tr   C  Fix,0
T
t 0
Where Π is exactly the Payoff of the Call Option at maturity (ideally Lifecycle) on
the Spark Spread between Power and Fuel, with total OPEX plus CAPEX being
the strike payoff.
This is very important, since based on it and on PLEXOS capacity for PBUC/ED as described before,
someone may be able to built a proper investment and/or operations strategy of a Power plant or a
portfolio, evaluate more realistically its Investment and Operational Risks and/or assess its Financial
position, given markets and drivers uncertainty employing hybrid methods like Price Based
Stochastic Optimisation.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Spark Spreads & Spark Spread Options
What is simply needed are realistic stochastic input samples of Forward Power
and Fuel prices.
The Total value of the Real Option (Power Plant) consists of the intrinsic value
of the option (positive Clean Spark Spread) plus the time value which depends
on the lifecycle operation of the plant and market volatility. Total OPEX and
CAPEX constitute the plant’s LRMC, which is exactly the Strike Price for Payoff of
the Real Option (Power Plant).
In this sense, a power plant is said to be “unhedged” if it purchases fuel(s) on
the spot market and also sells the electricity on the spot market.
Since natural gas plants and/or coal plants are often on the margin (price
setting or marginal plants) and as long as the power markets become more
competitive and mature, the use of Spark Spreads for power plants’ valuation,
operational risk assessment and hedging and the related Spark Spread “Real
Options” are becoming more and more an extremely valuable risk management
tool.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Spark Spreads & Spark Spread Options
In addition, due to the complexity and difficulty involved in the
modelling of generation characteristics and constraints of a
power plant, the computation of the price sensitivities –
otherwise known as the “Greeks” of a power plant, essential for
a proper hedging and risk management strategy is a continuous
challenge.
Practically, every aspect of energy production and delivery or
every Energy Asset (e.g. power plants, storages, transmission
lines etc.) can be expressed using such Spread Options.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Spark Spread & Plant Revenue
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Spark Spread & Plant Generation Level
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Spark Spread & Net Generator’s Profit
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Spark Spread & Net Company’s Profit
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Clean Spark Spread by Company & Technology
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Physical Energy Assets’ Portfolio Optimisation
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
What is an Energy Assets Portfolio?
Within the energy industry a portfolio can be divided into two distinct parts:
 Physical Portfolio: represented by physical assets - Power Plants, Gas Fields,
flow (power/gas) lines, Storages etc.
 Financial Portfolio: consisting of contract assets - Financial and Physical
contracts such as futures, forwards, swaps, options and FTRs but also PTRs
contracts for electricity and FTS for gas.
While the optimisation of a contract portfolio in a traditional financial market
has been well debated, problems can arise when optimising the entire power
portfolio.
PLEXOS® “emulates” how the market operates but more importantly the real
electricity price formulation mechanism.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
40
Energy Assets Portfolio Optimisation
In PLEXOS®, Portfolio Optimisation accounts for both Physical and
Financial Assets.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
41
Physical Energy Assets’ Portfolio Optimisation with
PLEXOS®
 The return of an energy portfolio is affected by four
major sources of uncertainty;
 Power Spot Prices
 Demand (Power/Fuel)
 Inflows
 Fuel Prices
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
42
Physical Energy Assets’ Portfolio Optimisation with
PLEXOS®
 Traditional production cost simulation has the objective
function of minimizing total system costs, and the price of
energy is an outcome.
 In Portfolio Optimization, some or all of the commodity
prices e.g. electric prices and/or Reserve prices are 'known',
being defined either as deterministic or stochastic series.
 The objective function then changes to include revenue as
well as costs so overall the objective is to maximize profit
(value of sales less production cost).
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Company Class- Portfolio Optimisation & Risk
The Company Class is the one that collects all the physical energy (& financial)
assets of a Portfolio and reports all required financial metrics such as e.g. Net
Profits, Net Revenues et.al.
Company Formulate Risk is a flag indicating if constraints should be formulated to
bound Net Profit Risk.
These constraints apply when running a stochastic portfolio optimization (ST
Schedule Stochastic Method = "Scenario-wise Decomposition") and the simulation
is multi-sample (Risk Sample Count > 1).
A constraint is then formed on each Company so that the Net Profit in each
scenario is no less than Target Net Profit less the Acceptable Risk. These bounds
are usually determined by first running the stochastic optimization without these
risk constraints, and setting the Target Net Profit such that less desirable outcomes
are constrained away.
Company Target Profit is the target Net Profit for risk constraints.
Company Acceptable Risk is the acceptable risk around Target Profit for risk
constraints.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Present Values Calculation
Additionally, in PLEXOS 7, there is the ability to calculate present values (NPV) to any
desired discounting rate for any cash flow property (Profits, Costs, Revenues etc. or
else any reported properties with an '$' symbol in their unit field).
Net Present Value of a stream of Cash Flows:
Cash flow is reported as the amount of money that is either paid out or received,
differentiated by a negative or positive sign at the end of a period. Conventionally,
cash flows that are received are denoted with a positive sign (total cash has
increased) and cash flows that are paid out are denoted with a negative sign (total
cash has decreased).
The cash flow for a period represents the net change in money of that period.
Calculating the net present value (NPV) of a stream of cash flows consists of
discounting each cash flow to the present, using the Present Value Factor and the
appropriate number of compounding periods, and combining these values.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Risk Constrained Integrated Resource Planning with
SOLT Plan
 Recent developments in PLEXOS® allow the exploitation of Modern Portfolio
Theory in an Integrated Resource Plan (IRP) or Capacity Expansion Planning
(CEP).
 Risk, Budget & Physical Constrained Stochastic LT Plan provides to our
users with more risk analysis information extending the capabilities of
traditional IRP or CEP.
 The approach allows the analyst to explore the impact of variability in key
decisions, such as the type of generation technology built, on the overall
solution giving a deeper insight into the resource planning problem.
 With this technique, we can explore Feasible and Optimum solutions from
minimum cost on a portfolio with comparatively high CVaR, through to
solutions with higher costs but lowest risk.
 We can also identify the key investment decisions, whether the amount of
generation built, retired or the type of technology used that will have the
greatest impact on risk.
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Stochastically Feasible & Optimum Efficient Frontiers
20/05/2014
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Risk Adjusted Values
Measurement Issues:
 Deterministic Scenarios provides a
measure of value at given conditions:
 Value of portfolio given average
conditions
 Stochastic (MC) Simulation measures
values of all measured conditions
weighted by probabilities providing the:
 Exprected value of portfolio given
ALL conditions
 Stochastic (MC) Optimization measures
values of all measured conditions
weighted by probabilities providing the:
 Optimum value of portfolio given
ALL conditions
20/05/2014
Why use Risk in Planning Decisions?
 It is likely that decisions made under
deterministic planning, while optimal
for the deterministic case, yield a
decision which is costly under other
known risks
 What is the Risk Adjusted Value?
5th Annual Quantitative Analysis in
Commodity & Energy Trading
Thank you for your time and the
opportunity
Dr Christos Papadopoulos
Regional Manager Europe
[email protected]
Office: +44 (0) 208 899 6500
Mobile: +44 (0) 776 031 6905
energyexemplar.com