Energy Strategy Reviews 28 (2020) 100447

Contents lists available at ScienceDirect

Energy Strategy Reviews

journal homepage: http://www.elsevier.com/locate/esr

Review

Investment in renewable energy considering game theory and

wind-hydro diversification
Claudia María García Mazo a, *, Yris Olaya b, Sergio Botero Botero b
a
Facultad de Minas, Universidad Nacional de Colombia, Sede Medellín, Polit�ecnico Colombiano Jaime Isaza Cadavid, Medellín, Colombia
b
Facultad de Minas, Universidad Nacional de Colombia, Sede Medellín, Colombia

A R T I C L E I N F O A B S T R A C T

Keywords: This paper explores the relationship among resource complementarity, diversification and strategic investment in
Power market a power market with two energy sources: water and wind. A review of diversification studies shows that the
Diversification strategic aspects of investment, which arise from the oligopolistic structure of power markets, are often over­
Complementarity
looked. Similarly, studies of strategic investment usually focus on price formation and do not analyze resource
Theory games
complementarity. This paper proposes an investment game with two players simultaneously installing new hydro
Wind power
Hydro power or wind power generation capacity to satisfy demand in the same market. Individual payoffs depend on resource
availability, and the equilibrium outcomes are evaluated for different wind-hydro correlations. The high negative
correlation between resources increases diversification and investors’ payoffs. The results suggest that diversi­
fication has a strategic value for individuals and for the security of supply.

1. Introduction the security of supply [5]. To achieve the benefits of diversification,

resources need to be complementary; that is, resources need to have
The restructuring of electricity markets around the world has created negative correlations of prices, availability or other desired quality [6].
a competitive environment in which both public and private investors An example of complementarity is the negative correlation of hydro and
participate and invest. These reforms intended to increase economic wind resources, which has been reported worldwide, including Norway,
efficiency and social welfare and have had mixed results, from increased Pakistan [7], New Zealand [8], Canada [9], Colombia [10], and Brazil
competition [1] to lower prices [2]. Currently, restructured electricity [11] among others.
markets face the challenges of integrating renewable generation and Applications of minimum variance portfolios (MVP) to find optimal
assessing its impact on power system operation, market prices and mixes of power generation technologies are criticized because genera­
investment. tion assets (unlike financial assets) are long-lived, energy is traded in
Power generators in competitive markets are exposed to revenue imperfect markets, and historic variance measurements may not apply
risks arising from fluctuating prices, volume risks and balancing pen­ as unforeseen events cause structural changes [12]. Regarding compe­
alties [3]. Investing in renewable generation has additional risks due to tition, electricity markets are usually oligopolies and generators can
the variability of wind, solar and hydro resources. Additionally, the low influence price and output. Then, strategic tools are needed to under­
marginal costs of renewables reduce prices and returns, increasing the stand the behavior of rivals [13] and to improve a company’s capacity to
risks of future investments [3]. propose an adequate competitive strategy based on knowledge of the
Portfolio theory has been proposed to address these risks by looking expected reactions of its rivals [14].
for profitable investments that can satisfy the demand of energy while This paper analyzes how resource complementarity and strategic
minimizing risk and maximizing returns [4]. Diversification is a key behavior influence technology choices in power generation. The litera­
element of this theory. A diversified portfolio for power generation ture has approached diversification as a way to reduce risk but not as a
balances its mix of energy sources to cope with external shocks and determinant of competitive strategy. This paper contributes to the
adverse events. Depending on the planning objective, diversification literature by highlighting new research lines in the intersection of game
mitigates fuel price volatility, contributes to price stability and increases theory and diversification theory. A stylized strategy game is presented

* Corresponding author.
E-mail addresses: [email protected], [email protected] (C.M. García Mazo).

https://doi.org/10.1016/j.esr.2020.100447
Received 22 May 2019; Received in revised form 6 November 2019; Accepted 6 January 2020
Available online 25 January 2020
2211-467X/© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

to illustrate how resource complementarity affects generators’ choices shocks [18], and minimizes cost risk [19]. In addition to fuel price un­
and the final energy mix. Results suggest that diversification has a certainty, portfolios combining renewable technologies consider other
strategic value, which is reflected in the strategic choices of agents. uncertainties such as technological changes [20], financial incentives
This document is organized as follows: Section 2 describes concept [21], and availability and distribution of solar and wind resources [22,
diversification and its application in electricity markets. Section 3 dis­ 23]. Some studies model the risk of wind and solar resource availability
cusses complementarity between hydro and wind energy and shows using capacity factors [20,23] while others model variations in seasonal
some applications of this concept in the electricity sector. Section 4 re­ wind speed and solar radiation [21]. Some authors find energy-based
views and classifies game theory applications to the electricity sector. instead of capacity based optimal portfolios [22].
Section 5 proposes a theoretical investment game with complementary An important aspect of diversification (and energy supply diversifi­
resources, and section 6 presents conclusions. cation) is the search for assets that can be complemented, i.e., optimal
portfolios that exhibit greater profitability and lower risk [5]. The more
2. Diversification and portfolio optimization in electricity negative the correlation between the assets, the greater the risk reduc­
markets tion [25] or the complementarity of assets.
In electricity markets, the diversification of resources in renewable
The diversification of power generation portfolios can be analyzed energy is a coherent strategy to increase energy security and offset CO2
from a financial and strategic point of view. Financially, diversification emissions [26]. Renewable wind and solar generation are not exposed to
is associated with risk. Specifically, a diversified portfolio is structured fuel price risks, but are intermittent and do not guarantee an uninter­
in such a way as to reduce its volatility. In the context of strategy, rupted supply of electricity. Wind and solar generation, however, are
diversification means doing something new; this implies the consider­ distributed resources which can increase security of supply and resil­
ation of new industries or markets as interesting opportunities to grow ience [5].
and achieve benefits. Regarding the ability to meet peak demand with intermittent re­
In electricity generation, diversification arises as the correct reaction sources, the literature reports a positive correlation between solar re­
to the underlying uncertainty of fuel prices, environmental impacts, sources and load, and a positive, higher correlation between combined
resource availability, and climate variations. Diversification increases solar-wind resources and load in Brazil [27], USA [28], Australia [29].
flexibility and protects energy production from abrupt interruptions and Correlations between load and wind availability depend on specific
adverse events [6,15,16]. sites, but have been found to be lower than solar-load correlation [27],
Portfolio theory has been applied to finding optimal power genera­ and in some cases, negative e.g. Refs. [28,30].
tion portfolios that ensure the stability of electricity systems, satisfying Complementarity of resources is one of the factors influencing
demands without any setbacks. From the perspective of investors, P�erez, resilience of power supply. In order to assess opportunities to increase
Watts and Negrete-Pincetic [17] identifies that portfolio applications diversification and resilience, it is important to understand how inter­
address: capital allocation among generation technologies, timing of mittent resources are distributed and how they complement variable
investment, energy trade management, and diversification to other resources, particularly hydro.
sectors (ancillary services, demand response, capacity markets).
Table 1 summarizes the main uncertainties addressed in the litera­ 3. Complementarity between hydro and wind resources
ture. Volatility of fuel, electricity and carbon prices is one of the most
commonly used risk factors in studies finding a conventional technology Correlations between hydro and wind resources in different locations
portfolio that maximizes revenue [4], reduces impact of external price have been studied to identify the value of different hydro/wind

Table 1
Summary of applications of portfolio theory to energy portfolio diversification.
Uncertainty Objective Decision Resources/Technology Representation of Area/country Constraints
variable renewable resources
availability

Pinheiro Neto et al. Debt, incentives Min. Risk Max. Capacity Hydro, Wind, Solar Seasonal indexes Brazil N/A
[21] Expected return share
Ramos et al. [24] N/A Min Risk, Max. Profit N/A Hydro, Wind N/A Brazil N/A
Sunderk€otter and Risk aversion Min. cost, Min. risk Capacity Nuclear N/A Germany N/A
Weber [19] Fuel prices share Hard coal
including CO2 Lignite
Natural gas (CCGT,
OCGT)
Roques, Hiroux, Resource Optimal wind N/A Wind Capacity factors Austria Transmission
and Saguan [23] availability portfolio Geographic Denmark France Resource
diversification Germany
Spain
Roques, Newbery, Electricity, CO2 Min risk-Max. Capacity Natural gas (CCGT) N/A Based on U.K. N/A
and Nuttall [4] and fuel prices Expected return share Coal data
Nuclear
Jansen, Beurskens, Fuel prices, CO2 Min cost risk Energy Natural gas (CC, CHP), N/A Netherlands Resource
and van Tilburg price, wind generated coal, nuclear, wind, (wind)
[22] availability biomass, other
renewables
Awerbuch and Fuel prices, Min risk-Max. Capacity Natural gas, coal, crude N/A European Union Existing
Berger [15] construction Expected return share oil, nuclear nuclear
period, O&M capacity
Humphreys and Fuel Price Max. Expected return Fuel Oil, Natural gas, Coal, N/A U⋅S. N/A
Mcclain [18] Reduce impacts of consumption
external Price shocks

2
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

portfolios with the same energy inflows in Canada [31], and to identify generation to reliability in hydropower dominated systems [10]. In
promising sites for wind generation in New Zealand [30,32], and Brazil Bruno, Ahmed and Shapiro [40] a combination of real options and
[33]. Other authors [34,35] study hydro and wind correlations to stochastic programming is proposed to find the optimal investment
evaluate the operational impacts and economic value of wind penetra­ strategies in electricity generation combining water and wind resources.
tion, to estimate costs of hybrid wind-hydro systems [36], and to The literature review in Table 2 shows the benefits of combining
compare costs of wind/hydro and thermal/hydro generation consid­ hydro and wind power generation due to their complementarity, leading
ering wind learning curves [37]. to a diversified portfolio. Among the tools used to model the in­
Results show that there is no complete complementarity between terrelationships between climatic variations, diversification, and
wind and hydro resources because wind availability depends on specific complementarity are time series, portfolio theory, and real options.
sites and has inter-annual in addition to intra-annual variations [31,37]. These works, however, do not consider the strategic behavior of gen­
Despite this, a wind-hydro power generation mix has been found to eration agents.
contribute to water reserves management [35], and to improve risk
profiles [31] at costs competitive with thermal and nuclear power 4. Applications of game theory in the power sector
generation [37].
Given the uncertainty in water and wind availability, several authors Any situation in which individuals make strategic decisions where
have applied a real options approach to estimate the value of hybrid results depend on what each individual does, constitutes a game of
wind – hydro systems [38,39] and to value contributions of wind strategy. A game is composed of players, information, actions, strategies,

Table 2
Summary of studies in hydro and wind power complementarity.
Technologies Region Objective Modeling of complementarity Findings

Tande and Vogstad Hydro, wind Norway Estimate implications of Expected wind power supply modeled with time Greater variation of annual
[34] power wind generation series supply in hyo power plants than
in wind power plants.
Seasonal wind variation matches
demand.
Vogstad [35] Hydro, wind, Mid Calculate value of large- EMPS. Stochastic dynamic programming and Value of wind generation
thermal, Norway scale wind generation simulation. estimated as value of generated
dispatchable loads Time series model for wind as run-of-river power plus reservoir
management.
100 MW wind avoid more
spillage than 3000 GWh
Jaramillo, Borja, Hybrid hydro- M�
exico Estimate firm energy cost Reservoir levels used to model hydro capacity Hybrid system has a competitive
and Huacuz [36] wind system from wind-hydro system factor, Weibull PDF to model wind capacity factor. levelized cost
Levelized costs of energy for hybrid system Stored water compensates wind
seasonal variations.
40.46 MW of nominal power
needed to supply 20 MW of firm
power
Denault, Dupuis, Hydro and wind Canada Minimize the long-term risk Copulas were used to model dependence of annual Evidence of cycles with long
and Couture- system (Qu�
ebec) (energy deficit in 50 years) energy inflows of water and wind. Portfolios periods of hydro and wind
Cardinal [31] combining 0%, 1%,2%, …,100% hydro and wind inflows much above and below
energy inflows were simulated. average.
Wind penetrations up to 30%
improve risk profile.
Branson [32] Hydro, wind New Find correlations between Weekly and monthly correlations are examined for Positive correlation of wind
Zealand wind/water availability 13 sites using 1990–2008 data. speed across sites.
Water resources more diversified
than wind (lower correlation).
Chade Ricosti and Hydro, wind Brazil Economic evaluation of Multi-annual complementarities of wind and hydro Evidence of inter-annual
Sauer [37] wind power as a substitute average normalized energy inflows are simulated complementarity of wind and
of thermal power to find expected capacity factors. hydro resources.
Lower than forecasted capacity
factors in wind power plants ask
for further research.
Jong et al. [27] Hydro, Wind and Brazil Analyze the ability to meet Average monthly data used for calculating Solar-load correlation higher
solar peak load with wind and Pearsons’ correlation coefficient than wind-load correlation.
solar power Combined solar and wind has
highest positive correlation with
load.
Combined solar-wind has higher
negative correlation with
reservoir levels.
Suomalainen et al. Hydro, Wind New Identify correlations not Fourier regression to separate seasonal variations Wind sites with strong negative
[30] Zealand related to seasonal from anomalies. seasonal correlations with hydro
variations. Correlation analysis of hydro reservoir levels and storage are identified.
wind speed/power, and of resources and demand/ Wind sites positively correlated
prices to demand are identified.
Witzler et al. [33] Wind Brazil Analyze the Reconstructed daily wind time series from 1948 to Significant inter-year
complementarity of wind 2014 to estimate power generation series. Cross- complementarity between south
generation sites. correlation analysis to assess complementarity and northeast regions.
among 14 sites with wind potential

3
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

and payoffs [41]. In cooperative games, it is possible for players to reach Generation output and investment in new capacity are the two
agreements to achieve a common goal [42], while non-cooperative strategic decisions for the majority of non-cooperative game theory
games rule out such possibility. models. The expansion problem can be formulated as an open-loop
Cooperative game theory has been used to analyze power system’s model in which agents simultaneously choose capacity and output, or
problems such as: loss allocation, optimal resource allocation in net­ as a closed-loop problem, which has an investment stage followed by a
works, allocation of transmission expansion costs, energy management market competition stage [61]. Both formulations can find equilibrium
frameworks for smart grids, system reliability performance, manage­ solutions for strategic expansion problems. Bunn and Oliveira, Genc and
ment and control of distributed systems and scheduling of renewable Sen, Pineau et al., Filomena, Campos-Na �n~ ez and Duffey [16,54,55,60]
energy [43]. Other applications of cooperative game theory address the model simultaneous capacity and output choices while [60] models
integration of renewable resources, particularly the aggregation of wind capacity and output choices as a two-stage game. Studies focusing on
resources and profit sharing [44,45], end-user coordination, and control optimal technology choice and welfare-maximizing expansion evaluate
of micro-grid distribution networks in hybrid solar and wind systems exogenous capacity scenarios [57] and least-cost expansion [57].
[45]. Regulation and incentives for distributed generation and Regarding market equilibrium, Cournot games are used to model
micro-grids have also been studied using cooperative games [46]. The competition and market clearing price in expansion games [16,54,59,
cooperative solution concepts applied are the Shapley value, nucleolus, 60] and agent-based simulation models [16,56]. Other studies find
dynamic programing (DP) equivalent method and Nash-Harsanyi market clearing prices using competitive [62] and Stackelberg models
schemes [43]. [55]. Some theoretical studies and studies presenting solution methods
Non cooperative game theory, particularly Cournot and Stackelberg do not specify generation technologies [42,54,60]. Bunn and Oliveira,
models, have been widely used to find equilibrium outcomes in Pineau et al., Wang et al., Murphy and Smeers [16,55,57,62] differen­
oligopolistic electricity markets [47,48]. Among the applications of tiate between peak and off-peak technologies while [56,58,59] charac­
game theory models of power markets are design and analysis of regu­ terize several generation technologies, including solar [56] and wind
lations considering bidding behavior of market participants [49]. Some [56,59]. So far, no examples of game theory models that examine the
market models assume general cost functions and do not include impact of resource variability in strategic generation and output choices
transmission constraints, while other models include specific technolo­ have been found.
gies, step-cost functions and transmission constraints for a more realistic This paper proposes an application of game theory for decision
representation of power markets and equilibria [47]. making on resource diversification through a stylized example including
Khare, Nema and Baredar [45] review applications of game theory to two generators in a hydro and wind power generation mix.
study the integration of renewables in energy systems. Among the issues
addressed in Ref. [45] are analysis of incentives for users, scheduling of 5. Investment game with complementary resources
large power systems, optimal bidding and contracting strategies.
Non-cooperative games have been applied to study integration of A simplified investment game illustrates how the complementarity of
renewable energy to power systems. Among the applications of resources influences equilibrium outcomes and diversification in power
non-cooperative games to the analysis of renewable energy integration generation. The game has the following components:
are dynamic pricing for demand side management in smart grids and Players: Two power generators decide to invest in new hydro or
analysis of mechanisms such as feed in tariffs and portfolio standards wind capacity to satisfy the totality of demand in a competitive market.
[50]. Other applications include integration of distributed resources Actions: Generators can invest in a new wind (W) or hydropower (H)
[51] and prosumers [52], and demand response management, [43]. plant, each with a nominal capacity of 100 MW (0.1 GWh). If market
supply is below demand, then power is imported (I) at a price of US$c
4.1. Models for capacity expansion 15,0/kWh.
Strategies: As the game only has one stage, the pure strategies of the
Before market reforms, the coordinated generation and transmission players correspond to the actions of investing in hydro or wind tech­
expansion planning was done in a centralized way, using reliability and nology. The strategic space of each player is:
least cost criteria. From a welfare perspective, capacity expansion in Player 1 (Hydro) ¼ {Invest in Hydro (H), Invest in Wind (W)}
power systems has been planned using optimization models with tech­ Player 2 (Wind) ¼ {Invest in Hydro (H), Invest in Wind (W)}
nical and economic parameters. After market reforms, expansion of The possible outcomes of the game are as follows: {H, H} {W, H} {H,
generation and transmission capacity is undertaken by individual firms W} {W, W}.
responding to price and profitability signals. In this context, models for The available energy for meeting the demand depends on the total
capacity expansion need to consider strategic competition and its impact capacity installed and on water/wind availability. If the available en­
on prices and investment, as well as the strategic aspects of investment. ergy is lower than the demand, suppliers pay a fine representing a ra­
The development of intermittent and distributed generation technolo­ tioning price or import fees.
gies adds new uncertainties to the planning process and the approaches Information: Players know the capacity and costs of each technol­
to generation expansion planning have been modified accordingly. ogy. This is a static, one-stage game where agents play simultaneously.
Sadeghi, Rashidinejad and Abdollahi [53] presents a comprehensive That is, players invest without knowing their opponents’ decision.
review of models for generation expansion planning from the perspec­ Payoffs: Annual payments depend on the generation costs, market
tives of power industry environment, environmental issues, new gen­ price for electricity, and the amount generated. Generation is a function
eration and control technologies, energy policies, mathematical of the technology mix and seasonal water and wind availability. There
modeling approach and solution methods. are two seasons rainy (R) and dry (D).
We focus our review on competitive game theory models for strategic To find the players’ utilities or payments the following equation is
expansion of power generation capacity, summarized in Table 3. These used:
models address a variety of problems from investment analysis under X
Ui ¼ ðITit CTit Þ (1)
uncertain conditions, technology mix and evaluation of GHG emission
t2fD; Rg
policies, expansion of generation capacity in oligopolistic markets, and
coordination of generation expansion. Among the uncertainties Where ITit is the total income of player i in season t ¼ D; R and is
considered in the literature are demand [54–56]; actions from other calculated as:
players [16,56–58]; policy targets [59] and regulations [56]; marginal
costs [56,60]; capital costs [59] and availability of resources [56].

4
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

Table 3
Applications of game theory for investment and capacity expansion in power generation.
Case study Problem Uncertainty Strategic Constraints Technologies Model Equilibrium concept
decisions formulation

Chattopadhyay Australia Quantitative Policy, Generation Resource Hydro, wind, Optimization, Nash Cournot for
[59] assessment of reduction supplied to the availability, geothermal, linear market clearing.
policies for targets, capital market restrictions on Biomass, programming for
reducing costs nuclear Nuclear, coal, inter-temporal
greenhouse expansion CCGT, IGCC- least cost
emissions CCS, IGCC, expansion,
considering least OCGT, CCGT- Cournot
cost expansion CCS competition
and market
bidding
scenarios.
Pineau, Rasata, Finland Evaluate the Uncertain Investment and Capacity Not specified Variational Nash model:
and Zaccour impact of demand, output peak and base inequalities (VI), simultaneous choice
[55] elasticity, generation non-linear of investment and
planning mix. programming output. Stackelberg
horizon, and (NLP), stochastic model: a leader
depreciation rate programming announces
on investment. (SP), and investment and
mathematical output choices and
programming competitors follow.
with equilibrium
constraints
(MPEC).
Bunn and Generic Price Other players’ Output sold at Capacity Not specified. Agent-based Nash-Cournot
Oliveira [16] system equilibrium in a actions baseload, peak Step cost simulation
pool market, and and shoulder functions
in a bilateral markets. Firms represent
trade market. can reorganize portfolios of
their plant baseload,
portfolio through shoulder and
acquisition of peak plants.
existing plants.
Chuang, Wu, Generic Expansion under Other players’ Simultaneous Reserve Existing nuclear Iterative Nash-Cournot
and Varaiya system 1) Cournot actions expansion, commitment and thermal, algorithm for
[58] competition all output and and capacity thermal finding
generators, 2) reserves constraints. expansion (oil, equilibrium
Collusion and 3) comittment natural gas and
Cournot coal).
competition one
generator and a
coalition of the
rest.
Voropai and Generic Coordination of Demand, Coordinated Coal and Coalitional game, Shapley value
Ivanova [42] system independent exports expansion natural gas additive multi-
companies for generation plants. criteria utility
generation function
expansion
Genc and Sen Ontario Generation Demand Production and Production Not specified. Stochastic Cournot model,
[54] expansion investment and capacity programming Open-loop
quantities that dynamic game equilibrium
maximize
discounted
expected profits.
Murphy and Theoretical Investment Time-varying Production and Capacity Peak and base- Problem Perfect competition,
Smeers [62] problem in demand investment load formulated as a closed-loop Cournot
asymmetric two-stage and open-loop
oligopolies with stochastic Cournot
technological program
diversity
Oliveira and Iberian Capacity Uncertain Firms adjust their Capacity CCGT, Coal Agent based and Repeated game with
Costa [56] market expansion in demand and generation ASC-FGD, monte-Carlo learning
oligopolistic marginal cost. portfolio by Nuclear PWR, simulation, based
markets production Investing/ Offshore and on an indirect
uncertainty divesting in onshore wind, reinforcement-
from technologies. Hydro >5 MW, learning
intermittent Price is formed in Solar PV > 50 algorithm.
energy sources an oligopolistic MW. Load
and regulatory market where factors for
uncertainty. firms compete in Hydro, solar,
quantities. wind.
Wang et al. [57] Generic Optimal Other players’ Expansion and Capacity, Not specified. Evolutionary Nash equilibrium
system planning actions bids for reserve reserves and algorithm to find
strategies and market.
(continued on next page)

5
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

Table 3 (continued )
Case study Problem Uncertainty Strategic Constraints Technologies Model Equilibrium concept
decisions formulation

energy/reserve transmission equilibrium

bidding constraints. strategies
strategies.
Markets are
competitive, the
ISO minimizes
the cost to
supply the load.
Filomena, Analytical Impact of Marginal cost. (i) the portfolio Capacity Not specified. Two-stage game. Open-loop problem
Campos- solution of marginal costs of technologies, Firms decide first formulated as a
N�an
~ ez, and theoretical on selection of (ii) each on a portfolio of complementarity
Duffey [60] model. generation technology’s technologies and problem. Profit
technologies capacity and (iii) capacities and in maximization to find
portfolio. each the second stage, symmetric and
technology’s decide their asymmetric subgame
production level production level perfect equilibria in
for every for each the two-stage closed-
scenario. technology in a loop model.
Cournot game.

ITit ¼ PSt *CGit (2) hydropower plants of Colombia are located. The Rio Grande con­
tributes its flow to “La Tasajera” hydroelectric power station and
PSt are market-clearing prices, which depend on the season and CGit is with other rivers, contributes to the hydroelectric chain of the Porce
the amount of energy generated by player i in season t ¼ D; R. II and Porce III power plants.
The total costs ðCTit Þ of player i in season t are calculated as: � The series for average daily wind speed in m/s (from average hourly
data) is from the high Guajira peninsula, at the north east on the
CTit ¼ MCit *CGit (3)
Caribbean coast, which has the best wind regimes in the country and
Where MCit are marginal costs of player i in season t. With the results where a pilot wind farm, named Jepirachi, is located (see Fig. 1b).
obtained from (1), we obtain the payoff matrix in millions of dollars.
Fig. 1 a shows average daily flows and wind speed from January
6. Data sources and conversion 2000 to December 2017. Daily water flow data were obtained from XM
[63] and IDEAM provided wind speed data from January 2000 to 2014.
To estimate the amount of energy generated by the players and the For years 2015, 2016 and 2017 we used a Monte Carlo simulation to
payoffs of the games, the following reference series of water and wind estimate the remaining data. The correlation coefficient between water
availability were used: flow and wind speed series is 0.24553.
In order to obtain hydro power generation, a conversion factor for
� Water flows are represented by the average daily flow (m3/s) for the the Tasajera power plant of 0,0077468 kW/m3/s [65] was used.
Río Grande, located in the Andean Region, west-center of Colombia The primary wind speed data series (Jan. 2000 to Dec. 2014) was
(see Fig. 1b). This is a mountain area with high rain where the main used to do a monthly forecast for years 2015, 2016 and 2017 using

Fig. 1. Reference water flow and wind speed series (a) and location of reference water and wind resources (b) from Ref. [64].

6
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

Table 4 and winds are negatively correlated and therefore, the wind and hydro
Estimated capacity factor and operation parameters for wind and hydro tech­ technologies are complementary.
nologies during the rainy (R) and dry (D) seasons. In this market, generators bid their available capacity at their mar­
Parameters Wind Hydro ginal cost. The levelized generation costs in Table 5 are estimated for
power each technology and season, based on capacity factors in Table 4 and the
Capacity factor Hydro 58,77% average capital and variable costs from Refs. [69–71].
power (R) Market demand is perfectly inelastic and equal to 270 GWh per se­
Wind (R) 36,16% mester. A market operator determines the equilibrium price by means of
Hydro 42,26%
a merit order based on the marginal costs of the plants.
power (D)
Wind (D) 57,29% Fig. 2 shows the supply and demand curves for the rainy and dry
Semester hours of operation (each Hours 4.380 4.380 seasons and each of the possible generation mixes, {H,H} {W,H} {H,W}
semester corresponds to a season). {W,W}, resulting from the investment game. Because wind and water
resources are complementary, the generators’ profits are greater with a
hydro/wind mix than with a hydro- or wind-dependent system.
During the rainy season, investors in a 100% hydro system obtain
Monte Carlo simulation. With the complete 2000–2017 wind speed se­
zero profits because the electricity price is equal to their marginal cost,
ries, wind power output was computed, assuming a Nordex N117/3000
but there is security in the system because the hydro resources are
kW unit and wind with 20 turbines. IDEAM wind speed data was ob­
enough to satisfy the required demand at a competitive price. During the
tained at a height of 10 m above ground. To transform this data to the
dry season, however, a 100% hydro system does not have enough re­
height at which the wind turbine works (117 m), the model “Profile of
sources to satisfy the demand, which creates shortages and losses for
Lysen” was used [66] with the following equation:
investors. Similarly, despite the high capacity factors assumed for wind
� � � � � ��
Z Zref plants, a 100% wind system does not generate enough energy to satisfy
V ¼ Vref LN LN : (4) demand during a rainy season but, during a dry season, a 100% wind
Z0 Z0
power system satisfies the totality of demand.
Where: As Fig. 2 shows, a combination of hydro and wind generation sat­
isfies the demand in the dry and rainy seasons. The generator with the
Vref is the known wind speed, lowest marginal cost during the season makes non-zero profits, whereas
Z is the height at which you want to know the speed V the other makes zero profits. Therefore, payoffs in this game are higher
Zref is the known height for outcomes with a diversified generation matrix than for a 100% hydro
Z0 is the equivalent roughness length, which depends on the char­ or wind generation system.
acteristics of the terrain. Table 6 summarizes results in Fig. 2. For each strategic profile, the
annual payoffs in Table 6 correspond to the total generators’ profits ðUi ;
After obtaining wind speeds at 117 m is obtained, the power curve i ¼ 1; 2Þ. These payments are calculated from the price and quantity
equation of the wind turbine is used (see Appendix A) to find the wind obtained in the dispatch for each season (Fig. 2), discounting the costs
generation data: (Table 5). Notably, for diversified outcomes (HW, WH) each player’s
profits during the season with high resource availability balance their
Y¼ 9; 2016X 3 þ 219; 64X 2 1:229; 3X þ 2:063; 6 (5) losses during the seasons with low resource availability.
Y is the power generated by the turbine in kWh. The Nash equilibrium in the previous game consists of each player
X is the wind generated at a height of 117 m above sea level. choosing the best possible strategy, given the strategies chosen by the
Capacity factors for each technology and season (rainy (R) and dry others. In a pure strategy equilibrium, only one action has a positive
(D)) are determined and shown in Table 4. Results in Table 4 show a probability of being selected. The game features two symmetric Nash
high capacity factor for the wind power plant. High wind capacity fac­ equilibria: {W–H} and {H–W}, in which the players obtain non-negative
tors are explained by the characteristics of the site selected for simula­ profits. Climatic changes and the intermittency of the resources affect
tion, which has high average wind regimes [67] with wind speed the system’s security when the players install the same type of
between 5 m/s and 11 m/s throughout the year [68]. The rainfall level technology.
In addition to the pure-strategy {W, H} {H, W} equilibria, there is a
mixed strategies equilibrium in which the players assign a probability
profile to their pure strategies such that each player is indifferent to its
Table 5 two pure strategies (see appendix A). In equilibrium, Player 1 assigns a
Fixed and variable cost of wind and hydro power generation technologies. probability pðHÞ ¼ 0; 548 and pðWÞ ¼ 0; 451 to its Hydro and Wind
Parameters Unit of Wind Hydro power strategies, and player 2 assigns a probability
measurement qðHÞ ¼ 0; 548 and qðWÞ ¼ 0; 451 to his Hydro and Wind strategies for a
Installation cost $US 162.400.000 290.000.000
correlation coefficient of 0.24 (see Appendix A for calculations). The
Fixed cost (R) $US/Semester 875.355 1.704.244 best response curves of players 1 and 2 in Appendix A intersect at three
Fixed cost (D) $US/Semester 1.386.872 1.225.540 points: pure strategy equilibrium: ðq ¼ 0; p ¼ 0Þ and ðq ¼ 1; p ¼ 1Þ, and
Variable cost O&M (R) $US/Semester 316.756 785.069 mixed strategy equilibrium: ðq ¼ 0; 548 ; p ¼ 0; 548Þ:
Variable cost O&M (D) $US/Semester 501.853 564.551
Table 7 shows results from the same game, assuming different cor­
Levelized cost Wind-Rain $US/kWh 0,05879
Levelized cost Wind- Dry $US/kWh 0,03710 relation coefficients for water and wind resources. The two pure-
Levelized cost Hydro $US/kWh 0,0562 strategies diversified equilibria {W,H}, {H,W} are maintained when
power-Rain resources are negatively correlated. Regarding mixed strategies, for a
Levelized cost Hydro $US/kWh 0,0782 hydro/wind correlation coefficient of 1, player 1 assigns a probability
power-Dry
Financing period Semesters 40 80
ðHÞ ¼ 0; 636 and pðWÞ ¼ 0; 364 to its Hydro and Wind strategies, and
Annual effective interest Percentage 10% 10% player 2 assigns a probability qðHÞ ¼ 0; 636 and qðWÞ ¼ 0; 364 to his
rate Hydro and Wind strategies. Payoffs for diversified outcomes ({W,H}, {H,
Biannual effective interest Percentage 4881% 4881% W}) increase for wind and hydro resources with perfect negative cor­
rate
relation. Positive correlations, on the other hand, result in a unique non-

7
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

Fig. 2. Merit dispatch and market equilibrium for rainy and dry seasons and game outcomes.

Table 6
7. Conclusions
Payoffs matrix for diversification investment game: The payments correspond to
the annual profits of each player assuming a correlation coefficient of 0.245 for
hydro - wind resources. A literature review shows that investment decisions in the electricity
market are exposed to new risk and uncertainty sources. Therefore, from
Player 2 Investment
the investor’s point of view, technology choices for new power gener­
Strategies ation capacity are strategic and their profitability depends on the sys­
Mixed strategies q ¼ 0; 54 1 q ¼ 0; 46 tems’ energy mix. Diversification and resource complementarity are one
H W of the least studied problems in the literature on investment in power
Player 1 Investment p ¼ 0; 54 H 6,4; 6,4 0,85; 1,7 systems and it is a potential line for future research. Furthermore, this
1 p ¼ 0; 46 W 1,7; 0,85 9,0; 9,0 review sheds light on how diversification and game theory complement
to represent complex interrelationships between agents and energy re­
sources. Understanding the relationship between technology mix,
resource availability and profits is key for analyzing strategic technology
diversified equilibrium {H,H} note that positive correlations between choices in power generation.
hydro and wind resources are associated with lower payoffs for diver­ The proposed investment game shows that, when resources are
sified outcomes {H,W}, {W,H}. This indicates that the value of wind/ complementary, a diversified generation matrix results in higher profits
hydro diversification is higher when resources are negatively correlated. for generators and lower supply risk than non-diversified outcomes. For
The two pure-strategy equilibria found when resources are nega­ the system operator, a diversified outcome with complementary tech­
tively correlated are solutions that increase each agent and the system’s nologies increases the security of supply.
diversity. Game theory, however, does not predict which one is going to Results from the game show that the complementarity between
be played. In a real decision environment, players can reach an agree­ hydro and wind resources can be exploited to benefit both investors and
ment or build coordination mechanisms to send public or private signals the operation of the system. Wind/hydro power diversification is then a
about the strategy to be followed to reach a more egalitarian solution or suitable combination in markets with a large hydro power share because
a solution that benefits everyone but does not render anyone worse. This it reduces the dependence on hydrology, decreases risk, and improves
is the idea of correlated equilibrium, developed by Aumann [72], ac­ reliability of supply among other factors.
cording to which agents respond to a signal produced by a joint random The proposed static, simultaneous game has several limitations that
mechanism operated by an mediator, and choose one of their pure show future research lines. First, after investing, the two players are
strategies in such a way that they have no incentives to deviate from the assumed to behave competitively and the oligopolistic and dynamic
signal suggestion. nature of the market is not considered. For future research, models such

8
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

Table 7
Diversification game equilibrium for different wind-hydro correlation coefficients (ρ).

as Cournot and Stackelberg can be adapted to study the impact of and diversity of players and generation technologies, and allow for in­
resource availability and diversification in strategic choices. Second, vestment and disinvestment decisions.
although payoffs and game equilibrium are analyzed for dry and rainy Despite its limitations, the proposed game represents the value of
seasons, investment decisions do not consider the uncertainty of the diversification appropriately and gives insight into the strategic value of
climate. Then, future research can develop real options models that resource complementarity for investors in electricity generation.
include the volatility of resource availability.
Third, no transmission and operative constraints are considered. Acknowledgements
Other future research can develop improved models with capacity and
transmission constraints. In addition, future work can analyze other Claudia García received financial support from Colciencias, Con­
stochastic variables such as demand and fuel prices, increase the number vocatoria N� 727 -2015-Doctorados Nacionales.

Appendix A. Power Curves Nordex N117/3000 [73]

9
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

Appendix B. Mixed strategies for a ρ ¼ ¡0.24 correlation of wind-hydro resources

P Q
In mixed strategies being ui ðsÞ ¼ ui ðaÞPrða jsÞ y Prða jsÞ ¼ sj ðaj Þ. The best response and Nash equilibrium are defined from the strategies
aeA jeN

actions: s*i 2 BRðs i Þ iff8si 2 Si ; ui ðs*i ; s i Þ � ui ðsi ; s i Þ. So the Nash equilibrium is: s ¼ ðs1 ; …:; sn Þiff8i; si 2 BRðs i Þ.
Before solving the game by mixed strategies, the following notation is required:
Player 1 has qðHÞthe probability to choose Hydro power and tqðEÞhe probability that it can choose Wind. Analogous to player 2 hypðHÞdraulic
probability and WinpðEÞd probability. Suppose that Player 1 uses a pure strategy either H o E and Player 2 uses some mixed strategy, tp ¼ ðpðHÞ;
pðEÞÞhen player 1 faces, that is to say:

p ¼ the probability of player 1 (line) that chooses H

1-p ¼ the probability of player 1 (line) that chooses W
q ¼ the probability of player 2 (column) that chooses H
1-q ¼ the probability of player 2 (column) that chooses W

Player 2

Player 1 Mixed Strategies

q 1-q
H W
p H 6,4; 6,4 0,85; 1,7
1-p W 1,7; 0,85 9,0; 9,0
With the information above, we calculate the pay-off function of each player.
Pay-off function of the player 1:
PP1 ¼ 6; 4pq þ 0; 85pð1 qÞ þ 1; 7ð1 pÞq 9; 0ð1 pÞð1 qÞ (B.1)

PP1 ¼ 6; 4pq þ 0; 85p 0; 85pq þ 1; 7q 1; 7pq 9; 0 þ 9; 0q þ 9; 0p 9; 0pq (B.2)

PP1 ¼ 17; 95pq þ 9; 85p þ 10; 7q 9; 0 (B.3)

Then we calculate the first-order condition:
∂PP1 = ∂p ¼ 17; 95q þ 9; 85 ¼ 0 (B.4)
Solving equation (A.4), we obtain:
q¼ 9; 85= 17; 95 ¼ 0; 548 (B.5)

1 q¼1 0; 54 ¼ 0; 451 (B.6)

Pay-off function of the player 2:
PP2 ¼ 6; 4qp þ 0; 85qð1 pÞ þ 1; 7ð1 qÞp 9; 0ð1 qÞð1 pÞ (B.7)

PP2 ¼ 6; 4qp þ 0; 85q 0; 85qp þ 1; 7p 1; 7qp 9; 0 þ 9; 0q þ 9; 0p 9; 0pq (B.8)

PP2 ¼ 17; 95qp þ 9; 85q þ 10; 7p 9; 0 (B.9)

Next, we calculate the first-order condition:
∂PP2 = ∂q ¼ 17; 95p þ 9; 85 ¼ 0 (B.10)
Solving the equation (A.10), we obtain:
p¼ 9; 85= 17; 95 ¼ 0; 548 (B.11)

1 p¼1 0; 548 ¼ 0; 451 (B.12)

The following graph shows the best-response curves for each player The Nash equilibria are shown where best responses intersect. The two pure-
strategy equilibria are signaled in black and the mixed-strategy equilibrium q ¼ p ¼ 0.548 in green.

10
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

References [16] D.W. Bunn, F.S. Oliveira, Agent-based analysis of technological diversification and
specialization in electricity markets, Eur. J. Oper. Res. 181 (2007) 1265–1278,
https://doi.org/10.1016/j.ejor.2005.11.056.
[1] M. Mulder, Competition in the Dutch electricity Market : an empirical analysis over
[17] R. P�erez Odeh, D. Watts, M. Negrete-Pincetic, Portfolio applications in electricity
2006 – 2011, Energy J. 36 (2015) 1–28, https://doi.org/10.5547/
markets review: private investor and manager perspective trends, Renew. Sustain.
01956574.36.2.1.
Energy Rev. 81 (2018) 192–204, https://doi.org/10.1016/j.rser.2017.07.031.
[2] M. Hyland, Restructuring European electricity markets - a panel data analysis, Util.
[18] H.B. Humphreys, K.T. Mcclain, Reducing the impacts of energy price volatility
Policy 38 (2016) 33–42, https://doi.org/10.1016/j.jup.2015.11.004.
through dynamic portfolio selection, Int. Assoc. Energy Econ. Stable U. 19 (1998)
[3] D. Bauknecht, G. Brunekreeft, R. Meyer, From niche to mainstream: the evolution
107–131, https://doi.org/10.5547/ISSN0195-6574-EJ-Vol19-No3-6.
of renewable energy in the German electricity market, in: F.P. Sioshansi (Ed.), Evol.
[19] M. Sunderk€ otter, C. Weber, Valuing fuel diversification in power generation
Glob. Electr. Mark. New Paradig. New Challenges, New Approaches, 1St ed., 2013,
capacity planning, Energy Econ. 34 (2012) 1664–1674, https://doi.org/10.1016/j.
pp. 169–198, https://doi.org/10.1016/B978-0-12-397891-2.00007-9. Amsterdam.
eneco.2012.02.003.
[4] F.A. Roques, D.M. Newbery, W.J. Nuttall, Fuel mix diversification incentives in
[20] A. van Zon, S. Fuss, Risk, embodied technical change and irreversible investment
liberalized electricity markets: a Mean-Variance Portfolio theory approach, Energy
decisions in UK electricity production, in: M. Bazilian, F. Roques (Eds.), Anal.
Econ. 30 (2008) 1831–1849, https://doi.org/10.1016/j.eneco.2007.11.008.
Methods Energy Divers. Secur., 1St ed., Elsevier Ltd, 2008, pp. 275–303, https://
[5] G. Escribano Franc� es, J.M. Marín-Quemada, E. San Martín Gonz� alez, RES and risk:
doi.org/10.1016/B978-0-08-056887-4.00014-.
renewable energy’s contribution to energy security. A portfolio-based approach,
[21] D. Pinheiro Neto, E.G. Domingues, A.P. Coimbra, A.T. de Almeida, A.J. Alves, W.
Renew. Sustain. Energy Rev. 26 (2013) 549–559, https://doi.org/10.1016/j.
P. Calixto, Portfolio optimization of renewable energy assets: hydro, wind, and
rser.2013.06.015.
photovoltaic energy in the regulated market in Brazil, Energy Econ. (2017),
[6] K. Costello, Diversity of generation Technologies : implications, Electr. J. 20 (2007)
https://doi.org/10.1016/j.eneco.2017.03.020.
10–21, https://doi.org/10.1016/j.tej.2007.04.006.
[22] J.C. Jansen, L.W.M. Beurskens, X. van Tilburg, Application of portfolio analysis to
[7] S. Awan, M. Ali, M. Asif, U. Amjad, Hydro and wind power integration: a case study
the Dutch generating mix Reference case and two renewables cases : year 2030 - SE
of dargai station in Pakistan, Energy Power Eng. (2012) 203–209, https://doi.org/
and GE scenario, Energy Res. Cent. Netherlands (2006) 5–67. https://publicaties.
10.4236/epe.2012.44028, 04.
ecn.nl/PdfFetch.aspx?nr¼ECN-C–05-100.
[8] J. Branson, B. Layton, Exploring Wind-Hydro Correlation Report to New Zealand
[23] F. Roques, C. Hiroux, M. Saguan, Optimal wind power deployment in Europe-A
Steel and the Major Electricity Users ’ Group, New Zeal. Inst. Econ. Res., 2008. http
portfolio approach, Energy Policy 38 (2010) 3245–3256, https://doi.org/10.1016/
s://nzier.org.nz/static/media/filer_public/0f/00/0f00dc1a-56ee-4528-8c27-8ab4
j.enpol.2009.07.048.
637b8620/0810_exploring_wind-hydro_correlation_final.pdf.
[24] D.S. Ramos, L.A.S. Camargo, E. Guarnier, L.T. Witzler, Minimizing market risk by
[9] L.E. Benitez, P.C. Benitez, G.C. van Kooten, The economics of wind power with
trading hydro-wind portfolio: a complementarity approach, in: 10th Int. Conf. Eur.
energy storage, Energy Econ. 30 (2008) 1973–1989, https://doi.org/10.1016/j.
Energy Mark. 27-31 May 2013, Institute of Electrical and Electronics Engineers,
eneco.2007.01.017.
Stockholm, 2013, pp. 1–8, https://doi.org/10.1109/EEM.2013.6607300.
[10] C.F. Isaza, B.S. Botero, Wind power reliability valuation in a Hydro-Dominated
[25] D. Pinheiro Neto, E.G. Domingues, A.P. Coimbra, A.T. de Almeida, A.J. Alves, W.
power market: the Colombian case, Renew. Sustain. Energy Rev. 57 (2016)
P. Calixto, Portfolio optimization of renewable energy assets: hydro, wind, and
1359–1372, https://doi.org/10.1016/j.rser.2015.12.159.
photovoltaic energy in the regulated market in Brazil, Energy Econ. 64 (2017)
[11] O.A.C. do Amarante, D.J. Schultz, R.M. Bittencourt, N.A. Rocha, Wind/hydro
238–250, https://doi.org/10.1016/j.eneco.2017.03.020.
complementary seasonal regimes in Brazil, DEWI Mag. Nr. 19 (2001) 79–86. http
[26] J.N. Vald�es Lucas, G. Escribano Franc�es, E. San Martín Gonz� alez, Energy security
://www.dewi.de/dewi/fileadmin/pdf/publications/Magazin_19/13.pdf.
and renewable energy deployment in the EU: liaisons dangereuses or virtuous
[12] S. Awerbuch, A. Stirling, J.C. Jansen, L.W.M. Beurskens, Full-spectrum portfolio
circle? Renew. Sustain. Energy Rev. 62 (2016) 1032–1046, https://doi.org/
and diversity analysis, in: K.B. Leggio, M.L. Taylor, D.L. Bodde (Eds.), Manag.
10.1016/j.rser.2016.04.069.
Enterp. Risk, 1St ed., Elsevier Science, 2006, pp. 202–222, https://doi.org/
[27] P. de Jong, A.S. S�anchez, K. Esquerre, R.A. Kalid, E.A. Torres, Solar and wind
10.1016/B978-0-08-044949-4.X5032-6.
energy production in relation to the electricity load curve and hydroelectricity in
[13] D. Langary, N. Sadati, A.M. Ranjbar, Direct approach in computing robust Nash
the northeast region of Brazil, Renew. Sustain. Energy Rev. (2013) 526–535.
strategies for generating companies in electricity markets, Int. J. Electr. Power
http://www.sciencedirect.com/science/article/pii/S1364032113000981.
Energy Syst. 54 (2014) 442–453, https://doi.org/10.1016/j.ijepes.2013.07.031.
[28] M. Lave, A. Ellis, Comparison of solar and wind power generation impact on net
[14] H.T.J. Smit, L. Trigeorgis, Strategic Investment : Real Options and Games,
load across a utility balancing area, in: 2016 IEEE 43rd Photovolt. Spec. Conf.,
Princeton University Press, 2004. https://press.princeton.edu/titles/7727.html.
IEEE, Albuquerque, 2016, pp. 1837–1842, https://doi.org/10.1109/
[15] S. Awerbuch, M. Berger, Applying portfolio theory to EU electricity planning and
PVSC.2016.7749939.
policy-making, IEA/EET Work 3 (2003) 69. Pap. https://awerbuch.com/s
[29] Y. Li, V.G. Agelides, Y. Shrivastava, Wind-solar resource complementarity and its
himonpages/shimondocs/iea-portfolio.pdf.
combined correlation with electricity load demand, in: 2009 4th IEEE Conf. Ind.

11
C.M. García Mazo et al. Energy Strategy Reviews 28 (2020) 100447

Electron. Appl., IEEE, Xi’an, 2009, pp. 3623–3628, https://doi.org/10.1109/ [51] W. Su, A.Q. Huang, A game theoretic framework for a next-generation retail
ICIEA.2009.5138882. electricity market with high penetration of distributed residential electricity
[30] K. Suomalainen, G. Pritchard, B. Sharp, Z. Yuan, G. Zakeri, Correlation analysis on suppliers, Appl. Energy 119 (2014) 341–350, https://doi.org/10.1016/j.
wind and hydro resources with electricity demand and prices in New Zealand, apenergy.2014.01.003.
Appl. Energy 137 (2015) 445–462, https://doi.org/10.1016/j. [52] N. Zhang, Y. Yan, W. Su, A game-theoretic economic operation of residential
apenergy.2014.10.015. distribution system with high participation of distributed electricity prosumers,
[31] M. Denault, D. Dupuis, S. Couture-Cardinal, Complementarity of hydro and wind Appl. Energy 154 (2015) 471–479, https://doi.org/10.1016/j.
power: improving the risk profile of energy inflows, Energy Policy 37 (2009) apenergy.2015.05.011.
5376–5384, https://doi.org/10.1016/j.enpol.2009.07.064. [53] H. Sadeghi, M. Rashidinejad, A. Abdollahi, A comprehensive sequential review
[32] J. Branson, Exploring Wind-Hydro Correlation. Report to New Zealand Steel and study through the generation expansion planning, Renew. Sustain. Energy Rev. 67
the Major Electricity Users’ Group, 2008. Wellington, https://nzier.org.nz/static/ (2017) 1369–1394, https://doi.org/10.1016/j.rser.2016.09.046.
media/filer_public/0f/00/0f00dc1a-56ee-4528-8c27-8ab4637b8620/0810_explo [54] T.S. Genc, S. Sen, An analysis of capacity and price trajectories for the Ontario
ring_wind-hydro_correlation_final.pdf. electricity market using dynamic Nash equilibrium under uncertainty, Energy
[33] L.T. Witzler, D.S. Ramos, L.A.S. Camargo, E. Guarnier, Reconstruction of wind Econ. 30 (2008) 173–191, https://doi.org/10.1016/j.eneco.2007.02.005.
generation historical series aiming at the analysis of energy complementarity: [55] P.O. Pineau, H. Rasata, G. Zaccour, Impact of some parameters on investments in
methodology and applications, Int. Conf. Eur. Energy Mark. EEM. 2016-July oligopolistic electricity markets, Eur. J. Oper. Res. 213 (2011) 180–195, https://
(2016), https://doi.org/10.1109/EEM.2016.7521324. doi.org/10.1016/j.ejor.2011.02.033.
[34] J.O.G. Tande, K.-O. Vogstad, Operational implications of wind, in: E. Petersen, P. [56] F.S. Oliveira, M.L.G. Costa, Capacity expansion under uncertainty in an oligopoly
H. Hensen, K. Rave, P. Helm, H. Ehmann (Eds.), 1999 Eur. Wind Energy Conf. using indirect reinforcement-learning, Eur. J. Oper. Res. 267 (2018) 1039–1050,
Wind Energy Next Millenn., James & James, Science Publishers) Ltd., Nice, 1999, https://doi.org/10.1016/j.ejor.2017.11.013.
pp. 425–429. [57] J. Wang, M. Shahidehpour, Z. Li, A. Botterud, Strategic generation capacity
[35] K. Vogstad, Utilising the Complementary Characteristics of Wind Power and expansion planning with incomplete information, IEEE Trans. Power Syst. 24
Hydropower through Coordinated Hydro Production Scheduling Using the EMPS (2009) 1002–1010, https://doi.org/10.1109/TPWRS.2009.2017435.
Model, Wind Power 21st Century, 2000, pp. 1–5. [58] A.S. Chuang, F. Wu, P. Varaiya, A game-theoretic model for generation expansion
[36] O.A. Jaramillo, M.A. Borja, J.M. Huacuz, Using hydropower to complement wind planning: problem formulation and numerical comparisons, IEEE Trans. Power
energy: a hybrid system to provide firm power, Renew. Energy 29 (2004) Syst. 16 (2001) 885–891, https://doi.org/10.1109/59.962441.
1887–1909, https://doi.org/10.1016/j.renene.2004.02.010. [59] D. Chattopadhyay, Modeling greenhouse gas reduction from the australian
[37] J.F. Chade Ricosti, I.L. Sauer, An assessment of wind power prospects in the electricity sector, IEEE Trans. Power Syst. 25 (2010) 729–740, https://doi.org/
Brazilian hydrothermal system, Renew. Sustain. Energy Rev. 19 (2013) 742–753, 10.1109/TPWRS.2009.2038021.
https://doi.org/10.1016/j.rser.2012.11.010. [60] T.P. Filomena, E. Campos-N� an
~ ez, M.R. Duffey, Technology selection and capacity
[38] W.H. Reuter, S. Fuss, J. Szolgayov�a, M. Obersteiner, Investment in wind power and investment under uncertainty, Eur. J. Oper. Res. 232 (2014) 125–136, https://doi.
pumped storage in a real options model, Renew. Sustain. Energy Rev. 16 (2012) org/10.1016/j.ejor.2013.07.019.
2242–2248, https://doi.org/10.1016/j.rser.2012.01.025. [61] S. Wogrin, B.F. Hobbs, D. Ralph, E. Centeno, J. Barquín, Open versus closed loop
[39] S. Weibel, R. Madlener, Cost-effective design of ringwall storage hybrid power capacity equilibria in electricity markets under perfect and oligopolistic
plants: a real options analysis, Energy Convers. Manag. 103 (2015) 871–885, competition, Math. Program. 140 (2013) 295–322, https://doi.org/10.1007/
https://doi.org/10.1016/j.enconman.2015.06.043. s10107-013-0696-2.
[40] S. Bruno, S. Ahmed, A. Shapiro, A. Street, Risk neutral and risk averse approaches [62] F.H. Murphy, Y. Smeers, Generation Capacity Expansion in Imperfectly
to multistage renewable investment planning under uncertainty, Eur. J. Oper. Res. Competitive Restructured Electricity Markets, 2003, pp. 1–56, https://doi.org/
250 (2016) 979–989, https://doi.org/10.1016/j.ejor.2015.10.013. 10.1109/PES.2011.6039563.
[41] D.E. Waldman, E.J. Jensen, Industrial Organization : Theory and Practice, fourth [63] XM, Hist� orico Oferta, Portal Bi. (2017).
ed., 2013. [64] J.J. Ramírez C, MERRA-based Study of the Wind/Solar Resource and Their
[42] N.I. Voropai, E.Y. Ivanova, Shapley game for expansion planning of generating Complementarity to the Hydro Resource for Power Generation in Colombia
companies at many non-coincident criteria, IEEE Trans. Power Syst. 21 (2006) External Advisor : Contents Motivation and Research Question Methodology
1630–1637, https://doi.org/10.1109/TPWRS.2006.873053. Results and Analysis Main Conclusions, 2015, pp. 1–170.
[43] S. Abapour, M. Nazari-Heris, B. Mohammadi-Ivatloo, M. Tarafdar Hagh, Game [65] Consejo Nacional de Operaci� on, Acuerdo 1046 Por el cual se aprueba la
theory approaches for the solution of power system problems: a comprehensive incorporaci� on de un cambio en el factor de conversi�on de las plantas de generaci�on
review, Arch. Comput. Methods Eng. 7 (2018), https://doi.org/10.1007/s11831- Guadalupe III y IV y La Tasajera, 2018, pp. 1–2.
018-9299-7. [66] E.H. Lysen, Introduction to Wind Energy, 1983. Amersfoort The Netherlands.
[44] L. Chinmoy, S. Iniyan, R. Goic, Modeling wind power investments, policies and [67] Unidad de planeaci� on Minero Energ�etica (UPME), Atlas de Viento y Energía E� olica
social benefits for deregulated electricity market – a review, Appl. Energy 242 de Colombia, 2006, pp. 1–168. http://www.minambiente.gov.co/images/camb
(2019) 364–377, https://doi.org/10.1016/j.apenergy.2019.03.088. ioclimatico/pdf/colombia_hacia_la_COP21/ABC_de_los_Compromisos_de_Colomb
[45] V. Khare, S. Nema, P. Baredar, Solar-wind hybrid renewable energy system: a ia_para_la_COP21_VF.pdf.
review, Renew. Sustain. Energy Rev. 58 (2016) 23–33, https://doi.org/10.1016/j. [68] J.F. Ruíz-Murcia, J. Serna-Cuenca, H.J. Zapata-Lesmes, Atlas de viento de
rser.2015.12.223. Colombia, 2017, pp. 1–158. www.imprenta.gov.co.
[46] C. Lo Prete, B.F. Hobbs, A cooperative game theoretic analysis of incentives for [69] IRENA, Renewable Power Generation Costs in 2017, 2018.
microgrids in regulated electricity markets, Appl. Energy 169 (2016) 524–541, [70] U.S. Energy Information Administration, Cost and Performance Characteristics of
https://doi.org/10.1016/j.apenergy.2016.01.099. New Generating Technologies, Annual Energy Outlook 2019, Annu. Energy
[47] M. Ventosa, A. � Baíllo, A. Ramos, M. Rivier, Electricity market modeling trends, Outlook 2018. 2019, 2019, pp. 1–3. https://www.eia.gov/outlooks/aeo/assumpt
Energy Policy 33 (2005) 897–913, https://doi.org/10.1016/j.enpol.2003.10.013. ions/pdf/table_8.2.pdf.
[48] S. Mei, W. Wei, F. Liu, On engineering game theory with its application in power [71] CAISO Operations and Maintenance Cost Report, Externally-Authored Report
systems, Control Theory Technol 15 (2017) 1–12, https://doi.org/10.1007/ Variable Operations and Maintenance Cost, 2018, pp. 1–21.
s11768-017-6186-y. [72] R.J. Aumann, Correlated equilibrium as an expression of bayesian rationality,
[49] G. Li, J. Shi, X. Qu, Modeling methods for GenCo bidding strategy optimization in Econom. Soc. Stable. 55 (1987) 1–18.
the liberalized electricity spot market - a state of the art review, Energy 36 (2011) [73] Nordex Energy, Noise Level , Power Curves , Thrust Curves. Nordex N117/3000,
4686–4700, https://doi.org/10.1109/PESGM.2015.7285840. 2016, pp. 1–81.
[50] Z. Yi, Z. Xin-gang, Z. Yu-zhuo, Z. Ying, From feed-in tariff to renewable portfolio
standards: an evolutionary game theory perspective, J. Clean. Prod. 213 (2019)
1274–1289, https://doi.org/10.1016/j.jclepro.2018.12.170.

12