You are on page 1of 11

Renewable Energy 36 (2011) 97e107

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Solar and wind resource complementarity: Advancing options for renewable


electricity integration in Ontario, Canada
Christina E. Hoicka a, *, Ian H. Rowlands b
a
Geography and Environmental Management, University of Waterloo, 200 University Avenue West, Waterloo, Ontario, Canada N2L 3G1
b
Environment and Resource Studies, University of Waterloo, Waterloo, Ontario, Canada

a r t i c l e i n f o a b s t r a c t

Article history: In Ontario (Canada), the integration of renewable power is a priority policy goal. Since 2004, the
Received 17 March 2010 circumstances under which the integration of renewable power is evaluated have changed due to
Accepted 2 June 2010 successive changes in price as well as concerns that its over-production may add to grid congestion. This
research investigates the value of increasing complementarity (both proximate and geographically
dispersed) of wind and solar resources as a means by which electricity planners and researchers might
Keywords:
advance electricity sustainability in Ontario. More specifically, this paper asks the following questions: 1)
Complementarity
Does the combination of solar and wind resources in selected locations in Ontario serve to ‘smooth out’
Solar
Wind
power production, i.e., decrease instances of both high and low values, as compared to either resource
Renewable electricity producing individually? 2) Can this ‘smoothness’ be further improved by dispersing these resources
Canada (Ontario) geographically amongst locations? and 3) Does increasing the number of locations with solar and wind
resources further ‘smooth out’ power production? Three years (2003e2005) of synchronous, hourly
measurements of solar irradiance and wind speeds from Environment Canada’s Canadian Weather
Energy and Engineering Data Sets (CWEEDS) are used to derive dimensionless indices for four locations
in Ontario (Toronto, Wiarton, Sault Ste. Marie and Ottawa). These indices are used to develop three
transparent and accessible methods of analysis: (1) graphical representation; (2) percentile ranking; and
(3) using a theoretical maximum as a proxy for capacity. The article concludes that the combination of
solar and wind within locations and amongst two locations improves ‘smoothness’ in power production,
as compared to when each resource is produced on its own; moreover, it is further improved once more
than two resources and two locations are combined. However, there is neither further benefit, nor
drawback, associated with the geographic dispersion of complementarity between solar in one location
and wind in another, when compared to both resources in one location.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction For instance, many jurisdictions that have attempted to incorpo-


rate more renewable electricity are experiencing lengthening
Energy systems in their current form are unsustainable. While queues for connection to transmission and distribution systems;
the availability of modern energy, in particular electricity, is other disputes have also arisen. Renewable power integration
understood as a facilitating factor for economic development and poses unique challenges to conventional modes of electricity
many social aspirations [1], it is also responsible for environmental system operation, thus limiting the easy integration of these
and other sustainability impacts ranging from acid rain and smog resources. First, renewable resources tend to be geographically
to climate change [2]. While specifics vary, it is generally accepted dispersed and the greatest production may be far from load
that renewable resources for power generation need to play centres. Second, many forms of renewable electricity are ’inter-
a larger role. Thus, a growing number of governments are making mittent’, with temporal variance displayed hourly, daily, and
commitments to increase renewable energy in their power supply seasonally; hence, they may ‘over’ or ‘under’ produce from the
profiles, however, there are debates about how to go about this. perspective of the load or grid to which they are connected. These
characteristics pose significant challenges for system operators
who are tasked with providing reliable electricity and historically
* Corresponding author. Tel.: þ1 519 888 456732574. have developed markets or dispatched on the basis of
E-mail address: choicka@uwaterloo.ca (C.E. Hoicka). predictability.

0960-1481/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2010.06.004
98 C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107

electricity system congestion problems is on the minds of system


Nomenclature operators. In addition to changing economic policies, and increased
technical issues, there is also increasing opposition around siting
Apv area of photovoltaic array (m2) decisions, augmented tension about the economic support accor-
Aw cross sectional area of the wind rotor (m2) ded renewable and growing hostility towards energy as a tool for
H0 initial hub height (m) economic development. These concerns are being voiced more
H hub height (m) loudly, resulting in changing social concerns. It is evident that in
K1 dimensionless constant for solar power Ontario the integration of renewable power is a priority policy goal,
K2 dimensionless constant for wind power but the shifting social, economic and technical concerns are
hpv photovoltaic system efficiency resulting in a rapidly changing policy environment. It is thus
hw wind turbine system efficiency equally clear that electricity planners are in need of tools to address
Pj power at hour j (kWh) questions of intermittency and geographic dispersion of renewable
Ppvj power from photovoltaic array at hour j (kWh) power generation under complex and changing circumstances,
Pwj power from wind turbine at hour j (kWh) particularly tools that might be used to inform decision makers and
Sj solar insolation at hour j (W/m2) to enlighten the public.
r air density (kg/m3) One potential answer is to examine benefits of complemen-
V average wind speed at hub height H (m/s) tarity, which is the combination of two or more resources whether
V0 average wind speed at initial hub height H0 (m/s) proximate or geographically dispersed, and is considered by some
Vj wind speed at hour j (m/s) researchers as an option to manage the variability of typically
intermittent renewable resources. Researchers seek to understand
whether complementarity ‘dampens’ variability, and further,
1.1. Ontario context whether geographic dispersion of combined resources produces an
improvement by further ‘dampening’ or ‘smoothing out’ renewable
Interestingly, many of these conflicts are playing out in Ontario power production. In the case of wind and solar, it has been noted
(Canada), a jurisdiction at the forefront of North America with that it is possible that wind power can be generated when solar
respect to an examination of the sustainability of its electricity power is low and vice-versa [7], and that geographic dispersion
supply system [3]. Targets and plans to increase the use of may thus provide further benefits than either resource on its own.
renewable resources in Ontario’s electricity supply have involved, Jurisdictions are also beginning to see analysts investigate more
in turn, requests for proposals, then a standard offer program formally where complementarity exists; to date, however, there
(<10 MW) and now a revised feed-in-tariff program e the latter has not been a rigorous investigation of complementarity among
two programs explicitly encourage the connection of renewable resources in Ontario. Drawing upon the literature on complemen-
electricity production facilities to the province-wide distribution tarity of wind and solar resources, this article presents a case study
system. Each new program has offered a substantial change in the of solar and wind power production in Ontario, aiming to deter-
price offered for renewable electricity, particularly to solar elec- mine whether resource complementarity of solar and wind serves
tricity which now attracts between 44.3 and 80.2 cents (Canadian) to ‘smooth out’ renewable power production, as compared to either
per kWh generated for the grid [4]. Despite economic policies solar or wind on its own. Given that an issue in Ontario is the
becoming more favourable for renewable electricity integration, experience of transmission and distribution constraints, particu-
following the general experience, Ontario is experiencing conflicts larly given the OPA’s stated concerns that all resources are ‘on’
and challenges in deploying these plans. According to the program simultaneously, rather than simply focusing on considering
delivery agency the Ontario Power Authority (OPA), an excess of complementarity to reduce instances of zero power production, it
grid-connected renewable power would affect the transmission seems appropriate to also investigate the extent to which both
system. More specifically, less than two years after the launch of the wind and solar resources generate ‘full power’ at the same time.
standard offer program, the OPA announced that applications “to A brief glance at solar and wind resource maps in Ontario
date have outstripped transmission availability (as well as Hydro confirms the conventional wisdom that solar irradiance varies little
One’s [Ontario’s transmission system operator] capacity to process across the province [8] while wind production varies greatly across
connection impact assessments). There are hundreds of projects the region [9], with areas close to the province’s lakes generally
applying for connection where there is no space to accommodate having higher wind speeds. Some locations in Ontario may be
them.” [5] Citing economic concerns, the OPA then stated: “It would windier with lower population density, while others may have
be imprudent and financially irresponsible for the OPA to procure a weaker wind resource, but a larger population (hence, more
new generation that will effectively constrain off other generators rooftops for solar power production). It follows that the geographic
e where they are already under contract with the OPA, and in some dispersion between complementary resources could also be
cases also renewable generation e and pay twice for the genera- examined for potential benefits.
tion.” [5] Thus, the OPA identified regions of the province for which For these reasons, this study seeks to answer the following three
it would no longer entertain renewable electricity contracts, questions: 1) Does the combination of solar and wind resources in
arguing that such regions were already constrained by the already- selected locations in Ontario serve to ‘smooth out’ power produc-
existing demands placed on the transmission system. These are tion, i.e., decrease instances of both high and low values, as
identified as ‘Orange zones’ (high constraints, no new renewable compared to either resource acting individually? 2) Can this
electricity projects allowed), ‘Yellow zones’ (moderate constraints, ‘smoothness’ be further improved by dispersing these resources
renewable electricity projects would be allowed in restricted geographically amongst locations? and 3) Does increasing the
amounts subject to review and approval), and ‘Green zones’ (no number of locations with solar and wind resources further ‘smooth
constraints, renewable electricity projects would be allowed to out’ power production?
proceed) [5]. In addition, the newest feed-in-tariff program Given the changing nature of policies and the electricity system
subjects all proposed projects to reviews for capacity for both in Ontario, this study seeks to avoid the social and economic
transmission and distribution before any agreement is reached [6]. elements of technology choice or the effects of supply and demand
Clearly, the potential for renewable electricity projects to add to of renewable power production and consumption. Acknowledging
C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107 99

that electricity planners may encounter a variety of social and basis of power production equations, however, the Ontario situa-
economic issues pertinent to their decision-making, this article tion for renewable power has the characteristics of grid-connected
nevertheless explores methods that aim to apply easily attainable and a changing electricity price structure, which affects the value
measured data to practical problems, using a dimensionless of power produced. Further, Ontario has densely populated areas
biophysical measure of complementarity for wind and solar which are large load centres as well as much less densely popu-
resources. To accomplish this end, the authors examine lessons lated areas. Depending on location, loads vary hourly, daily,
learned from the literature and identify appropriate available data seasonally, and annually. Perhaps more appropriate to an analysis
for analysis to develop a dimensionless index. This dimensionless for Ontario, Tina et al. [15] size a solarewind system based on
index is examined in three ways: by means of graphical represen- long-term average performance which can both export and import
tation, by using a percentile ranking method, and, finally, by power from the grid. They describe their approach to hybrid sizing
developing a theoretical maximum as a proxy for a capacity as ‘versatile’ to decision-making in that it is adaptable for the
measurement. many pricing scenarios of buying and selling power in a grid-
connection, and hence compare power production values at
1.2. Learned lessons from the literature various temporal scales (i.e., between yearly and hourly) depen-
dent upon the situation. Taking geography into account, Notton
A growing body of literature on renewable power notes that et al. [10] employ dimensionless parameters to compare solar and
the combination of wind and solar resources improves reliability wind resources and rank reliability and optimal power production
and predictability of power production. Indeed, examinations of values to compare sites in different locations. Beluco et al. [16],
the complementarity between wind and solar resources have while focusing on hydroelectricity and solar power, show that one
been carried out since at least 1981 (Aspliden as cited by Notton useful solution rests in using a dimensionless index to compare
et al. [10]). The literature shows that methods and approaches to site options, including amongst different locations. On the other
define and measure complementarity vary according to the hand, Dihrab and Sopian [17] compare a grid-connected solar and
system configuration and the optimal solution(s) sought. System wind hybrid at three locations by applying eight years of meteo-
configurations may be any of: autonomous battery connected, rological data to a selected wind turbine and solar photovoltaic
autonomous and connected to a constant or variable load, or array. However, these methods of analysis which use dimension-
grid-connected. Optimal solutions sought might concern any less parameters and avoid social and economic elements show the
combination of: relative size of solar to wind capacity or most promise for applicability to the Ontario situation, a changing
production, cost minimization, or to compare site options within policy environment in which there are concerns to reduce
a geographic region. These assumptions affect the methodology instances in which renewable resources are either ‘on’ or ‘off’
that is ultimately selected by defining the granularity of data simultaneously.
used, whether social and economic elements (such as electricity
price and technology cost) are included, or whether technology is 2. Material
specified, or rather, whether biophysical and dimensionless
parameters are adopted. 2.1. Data selection
Within the literature, one key difference in methodology is that
non-grid-connected studies tend to focus on cost minimization, The ideal data to undertake this analysis are synchronous
while, for whatever reason, approaches to assess complementarity solar and wind power production measurements. A review of
of renewable resources that are grid-connected tend to avoid the solar and wind power production data in Ontario reveals that
economic element of cost minimization; finally, those that compare there are no publicly available solar power production data and
geographic locations tend to favour dimensionless parameters that the temporal period of available solar irradiance data (before
which avoids choice of technology or economic elements (such as 2005) does not correspond with the period of publicly available
cost) altogether. Further, most studies of complementarity ask wind power production data (after 2006). The next best available
whether the combination of intermittent solar and wind resources data are Environment Canada’s Canadian Weather Energy and
reduces variability in power production. That is, they explore Engineering Data Sets (CWEEDS), which consist of hourly
whether the combination reduces the instances of zero power measurements of weather conditions at 145 Canadian locations
produced. for up to 53 years (between 1953 and 2005), and include both
For example, studies that include battery storage generally solar and wind measurements. CWEEDS provide solar global
assume that this will take up any statistical variations of power horizontal irradiance measurements as the total of direct and
produced, hence, hourly variations are ignored and daily and diffuse radiant energy (in kJ/m2) received on a horizontal surface
monthly/seasonal variations are instead analysed [7,11]. These by a pyranometer. CWEEDS provide measurements of wind
assumptions dominated early studies which also tended to analyse speeds at the standard height of 10 m above ground in a flat,
off-grid (autonomous) configurations and size systems according to open exposure as measured by anemometers [18]. For the
minimizing costs of the combination of wind turbines, solar panels purposes of this study, CWEEDS measurements are considered an
and battery banks. Markvart [11] and Ai et al. [12] size hybrid appropriate source of biophysical data for an analysis of
systems assuming battery storage; Habib et al. [13] optimize to complementarity. This study uses hourly data for 2003, 2004 and
satisfy a constant load; Celik [7] and Diaf et al. [14] solve for variable 2005.1
loads, although Diaf et al. [14] increase the resolution of time and
compare various system sizes and costs by modeling hourly 2.2. Choice of location
production. All of these studies optimize system size by minimizing
cost, sometimes e as in the case of Ai et al. [12] e specifying for CWEEDS records are available to 2005 in ten locations across
trade-offs between solar panels and battery banks. Ontario. Of the ten available locations, four locations were chosen
The purpose of this paper is to take contributions from the because they are close to (or part of) Ontario’s transmission and
literature while innovating techniques according to the specific
situation of Ontario. The literature on solarewind complemen-
1
tarity is useful to compare system configurations and to derive the February 29th, 2004 was eliminated from analysis for the basis of comparison.
100 C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107

distribution systems (determined with the use of Hydro One’s A (flat-plate) photovoltaic array of an area Apv (m2) and solar
map of the province’s transmission system). In choosing loca- insolation Sj, produces solar power at hour j, Ppvj, according to
tions, consideration was also given to whether the area was
windy or a large load center. Given the types of complementarity Ppvj ¼ Sj $hpv $Apv (2)
sought amongst locations, four locations were chosen as follows:
where system efficiency is hpv. While it is recognized that effi-
Wiarton (44.75 N latitude, 81.1 W longitude) is a windy and
ciency, hpv, changes with temperature, it is assumed that the vari-
rural area on the shore of Lake Huron within a designated orange
ance is small enough that hpv can be assumed to be constant for all
transmission-constrained zone; it has already attracted close to
values of Ppvj, while panel area, Apv, also remains constant, hence,
500 MW of wind power development [19]; Toronto (43.67 N,
hpv$Apv can be replaced by a single constant, K1.
79.63 W), with a population of 5,531,300 [20], was chosen as
Adapting an equation from Gipe [24], we note that wind power
a large load center with little local power generation; given its
production is calculated as
high population density, however, it has many rooftops that are
potentially useful for solar power production. For the same
Pwj ¼ 0:5$hw $r$Aw $Vj3 (3)
reason, the moderately large urban area of Ottawa (45.38 N,
75.72 W), which has a population of 1,198,700 [20] was selected. where Aw is the cross sectional area of the wind rotor in m2,
Finally, Sault Ste. Marie (46.48 N, 84.50 W), a moderately remote Vj is the wind velocity in m/s at the hub height, r is the air
and industrialized location proximate to the windy Lake Superior, density in kg/m3 and hw is the system efficiency. The actual
which has attracted Ontario’s largest wind power project, the available wind energy that can be obtained depends on the
Prince I and II wind developments [19], was also chosen. Fig. 1 wind turbine’s power output which is usually illustrated by its
shows where these four locations may be found in the province characteristic curves [24]. Wind velocity increases with hub
of Ontario. height [24], and as hub height may vary anywhere upwards of
25 m [24], the social or economic element of wind turbine
2.3. Data preparation choice presents numerous possibilities. In order to estimate
wind speed at a height other than the height of measurement,
CWEEDS data measure wind velocity at the standard height of Gipe [24] adjusts values using a surface roughness exponent, a,
10 m. The CWEEDS solar data are measured at a horizontal tilt. The measured according to:
distribution of solar power varies temporally dependent upon
 
angle of orientation for incidence of solar irradiation [21]. ESP-r, V H a
a modelling tool which can be used to predict solar power ¼ (4)
V0 H0
production, was used to convert horizontal diffuse and direct
components of solar irradiance data provided by CWEEDS to solar Where V is average wind speed at hub height H, V 0 is wind speed at
irradiance values at an angle of incidence of 45 for all locations2 anemometer height H0, and a is the wind shear exponent, which is
[22,23]. It should be noted that if a different angle of incidence generally be assumed to be constant [24]. Accordingly, if a constant
was chosen, this would alter the monthly ratios of wind to solar hub height is assumed, and a is assumed to be constant, then the
power production [21]. velocity measured by CWEEDS at 10 m will increase by the same
factor for all hourly wind values to give a velocity at some new hub
3. Method and results height of wind power production. Further, it is assumed that
although air density, r, varies with temperature and elevation, the
3.1. Dimensionless index effects of changes in wind velocity far outweigh the effects of these
changes in r [24]. Hence for this study, it is assumed that r is
The focus of this research is on the biophysical complemen- essentially constant for all wind values. Thus, decisions regarding
tarity between wind and solar resources as a proxy for power the social and economic elements are avoided by assuming
production. In order to examine the hourly distribution of solar constant hub height, H, constant a, and that all of the elements
and wind resources across a given year, a constant is derived to beyond wind velocity are constant for each wind speed across
compare dimensionless hourly values of solar and wind produc- 8760 h of the year; therefore, 0.5∙hw∙r∙Aw is replaced by a single
tion. To arrive at this constant, it is assumed that the wind and constant, K2.
solar resources produce equal amounts of electricity e namely, Taking into account the constant coefficients for both wind and
106 kWh per year (a value chosen arbitrarily for simplicity in solar power production for a solar array of a given size and a wind
calculation) e but that the electricity produced is distributed turbine of a given size, power produced at hour j is now assumed
differently across the 8760 h of the year, according to the to be
biophysical profile of each of the two resources.
The hourly solar/wind hybrid power production, Pj, is given by: Pj ¼ K1 $Sj þ K2 $Vj3 (5)

Pj ¼ Ppvj þ Pwj (1) To derive a dimensionless biophysical measure, as previously


described, annual production for each resource is assumed to be
where Ppvj is the hourly-calculated solar power at hour j in kW h, equal, and for simplicity, given an arbitrary value of 106 kW h per
Pwj is the hourly-based calculated wind power in the hour j in kW h. year. Hence, solar power produced across a given year is then
CWEEDS data are used to estimate solar and wind power represented as
produced using the following equations.
X8760
K1 1
Sj ¼ 106 kWh (6)
2
Due to erroneous solar radiation measurements provided by CWEEDS on
and wind power across a given year is represented as
a small number of days, Ottawa 2003, hours 11e15 of day 153 were replaced by the
average of each of the hours for days 152 and 154 for, and hours 8e 12 of day 161
were replaced by the average of each of the hours for days 160 and 162. Sault Ste.
X8760
K2 1
Vj3 ¼ 106 kWh (7)
Marie 2004 days 42 and 43 were also replaced with the average of days 41 and 44.
C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107 101

Fig. 1. Map of selected locations in Ontario.

Using hourly CWEEDS data for Sj and Vj for the 8760 h in each 3.2.1. Approach 1: graphical representation and visual inspection
3
year solves for K1 and K2. To begin to address the first research question, K1$Sj and K2$Vj
are plotted graphically to visually assess the distribution of power
3.2. Assessing complementarity production and compared to a plot of the combined resource,
calculated as (K1$Sj þ K2$V3j )/2 (divided by two in order to ‘weight’
This papers aims to answer the three questions set out in the total power production to 106 kW h per year for the combined
introduction, drawing from the lessons of Tina et al.’s [15] desire to resource). This display is useful in answering our research ques-
use data to provide ‘versatile’ analysis, and Notton et al. [10] and tions, and it can be used to show how the range of resource
Beluco et al.’s [16] approaches to use dimensionless values to production for any combination of solar and wind resource varies
compare complementarity of resources in different geographic according to which resources and locations are combined. Although
locations. Additionally, the importance of the specific needs of these values are calculated for all 8760 h of the year for each of
energy planners in Ontario e particularly the complex and unique 2003, 2004, and 2005 for all locations and location combinations
issues that arise as a result of renewable electricity integration in previously defined, Figs. 2e5 present the distribution of solar, wind
Ontario e was also kept in mind. To these ends, this study uses the and combined resources for one week in January and one week in
dimensionless variables K1, K2, and the hourly CWEEDS data for July in Toronto in 2004. Given that the combined resource is
2003, 2004 and 2005 in order to explore the answers to these three weighted, the high values of the combined resource are reduced
questions by three distinct approaches: (1) graphical representa- unless both resources are ‘on’ simultaneously. Other locations
tion; (2) percentile ranking, and; (3) using a theoretical maximum produce a similar pattern in reducing the maximum wind values,
as a proxy for capacity. thus producing a ‘smoothing’ of at least the high end of the
102 C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107

2500 1000
Solar (K1S)
Solar (K1S)
Wind (K2V^3)
Wind (K2S*V^3)
2000

K1S or K2V^3
K1S or K2V^3

1500

500

1000

500

0
0 4345 4512
1 168 Hour
Hour
Fig. 4. Hourly solar and wind resource patterns normalized to 106 kWh output per
Fig. 2. Hourly solar and wind resource patterns normalized to 106 kWh output per year for Thursday July 1 to Wednesday July 7, 2004 in Toronto.
year for Thursday January 1 to Wednesday January 7, 2004 in Toronto.

dispersion is represented by the following matching of locations:


production curve, the focus of the first research question. However, Combination B) Toronto solar, Wiarton wind; Combination C)
it is difficult to tell the extent to which the resource production Ottawa solar, Sault Ste. Marie wind. The same is done for the third
improves, especially the extent to which the occurrences of close to research question, using the following combinations: Combination
zero output are reduced. Hence a more specific measure of D) Toronto and Ottawa solar and Wiarton and Sault Ste. Marie wind
complementarity would be useful. (and weighted accordingly by dividing by 4); Combination E) solar
and wind for all of Ottawa, Toronto, Sault Ste. Marie and Wiarton
3.2.2. Approach 2: percentile ranking (and weighted accordingly by dividing by 8). The results are shown
A percentile ranking approach is used to provide analysis to in Fig. 6 and Table 1. Overall, the 20th and 50th percentiles follow
answer each of the three research questions. In this case, the a similar pattern: Combination A produces a marked rise compared
dimensionless values are ranked and the 80th, 50th and 20th to the singular solar or wind resource. With respect to the second
percentiles are noted and plotted for the individual resources by research question, there is little difference between the values
location and for each combination of locations chosen. Using this produced at the four locations of Combination A when compared to
method, an answer in the affirmative to the first, second and third Combinations B and C. With respect to the third research question,
research questions (in order) would be revealed by a graph where Combinations D and E do produce further improvement to the 20th
consecutive values of the 80th percentile should lower, and and 50th percentile values showing potential benefits of combining
consecutive values measured at the 20th percentile should rise. The multiple resources geographically. Meanwhile, analysis of the 80th
median value of resource production, the 50th percentile, should percentile shows a different pattern. First, the 80th percentile of
rise to demonstrate a rise in median production values. solar and wind as singular resources are remarkably different,
Accordingly, for singular locations the 80th, 50th and 20th hence Combination A produces an averaging of these resources.
percentiles of the singular resources (K1∙Sj for solar and K2∙V3j for With respect to the second question, Combinations B and C do not
wind) are plotted. For the first research question, the combined produce much improvement over Combination A. With respect to
resource by location is called Combination A, is weighted as the third question, Combinations D and E produce a slight
(K1∙Sj þ K2∙V3j )/2. For the second research question, geographic

2005 1000

2000
(K1S + K2V^3)/2
(K1S +K2V^3)/2

1005

500

1000

500

0 0
1 168 4345 4512
Hour Hour

Fig. 3. Hourly combined solarewind resource patterns normalized to 106 kW h output Fig. 5. Hourly combined solarewind resource patterns normalized to 106 kW h output
per year for Thursday January 1 to Wednesday January 7, 2004 in Toronto. per year for Thursday July 1 to Wednesday July 7, 2004 in Toronto.
C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107 103

Fig. 6. Average of years 2003, 2004 and 2005 percentile ranking for all combinations of resources and locations.

improvement (lowering of the 80th percentile) over Combinations economic element of technology choice. For this study, 15 m/s is
A, B and C. chosen as the theoretical maximum as it was the maximum value
for most selected wind sites for most years.3 Following these
3.2.3. Approach 3: use of a theoretical maximum as a proxy for decisions, the dimensionless constant, the highest K1 is just less
capacity than 1.0 and the highest K2 is less than 1.4. Hence K1max is arbi-
Another method to gather evidence to answer the three trarily set at 1.0, and K2max at 1.5. The theoretical maximum of the
research questions is to estimate a theoretical maximum produced combined resource is estimated to be 3280.3 (a dimensionless
by combined solar and wind resources at any given hour in the year value). While the theoretical maximum is measured somewhat
and then use it as a proxy for a ‘rated capacity’. Following Rowlands arbitrarily, of importance to the analysis is the extent to which the
and Jernigan’s [25] analysis of wind variability, three thresholds are number of instances within different thresholds change as our
defined: a) less than 2% of rated capacity, b) 2e50% of rated analysis attempts to answer the three research questions.
capacity; and c) above 50% of rated capacity. Less than 2% can be Table 2 displays the results of this method. In examining the
considered as instances when both resources are ‘off’ and above threshold greater than 50% of the theoretical maximum, it can be
50% can be considered as instances when both resources are ‘on’. seen that wind tends to exceed this level in a range of 8e35 h per
The same combinations as Approach 2 are used; however, rather year, while solar does not exceed this value at all. With respect to
than ranking values, the instances in which the singular (solar or the first research question, Combination A reduces the number of
wind) or combined dimensionless measure of the resource falls instances this threshold is exceeded to a range of zero to 8 h per
within these ranges are counted and compared. Using this method, year. With respect to the second research question, Combinations B
an answer in the affirmative to the first, second and third research and C do not produce much improvement in reducing the number
questions (in order) would be revealed as the number of instances of instances above this threshold, with zero to seven instances per
in the ranges of less than 2% and above 50% production should year which exceed it. However, with respect to the third research
decrease, while those in the 2e50% range should increase, showing question, Combinations D and E do show a further reduction in
a reduction in the spread of values of power production. instances above the 50% threshold, although not for the year 2005,
The theoretical maximum was difficult to choose and selected as in which Combination D produces four instances which exceed this
a somewhat arbitrary value based on CWEEDS data and the derived threshold. For the less than 2% production threshold, the unique
dimensionless constants, K1 and K2. Knowledge of an hourly solar resource occurs below this threshold for 5433 to 5641
maximum global insolation value in Ontario would be ideal. instances per year, and wind for 5134 to 5630 instances per year,
However, while there are solar maps of Ontario, such as the one amongst locations over the three years of analysis. With respect to
provided by Natural Resources Canada [8], these provide mean the first research question, Combination A provides a significant
daily, but not hourly, or maximum hourly, insolation values. improvement over the singular resources, with the instances below
Further, this map does not match the 45 tilt of the converted
CWEEDS irradiance data. The maximum solar irradiance value
found in the three years of CWEEDS data was 1498.4 W/m2, at the
3
location of Wiarton, chosen as the maximum solar irradiance value. It can be noted that wind velocities exceeded this in Toronto for a total of 23 h
across three days in 2003, for 7 h across three days in 2004, for 10 h across four days
In choosing a maximum wind value, wind turbines typically have in 2005. This was also exceeded in Sault Ste. Marie for 3 h over one day in 2004.
a ‘cut-off’ point, a speed above which zero power is produced; These velocities measured above 15 m/s were assumed to be 15 m/s rather than
however, the decision as to where this lies implies some social or zero.
104
Table 1
Results for percentile ranking for combinations of resources and locations, 2003, 2004 and 2005.

Year 2003 2004 2005

Location Ottawa Sault Toronto Wiarton Ottawa Sault Toronto Wiarton Ottawa Sault Toronto Wiarton
Ste.Marie Ste.Marie Ste.Marie

C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107


Solar
80 percentile 236.0 222.0 240.1 222.5 233.2 214.7 228.7 221.2 232.5 224.9 243.3 236.8
50 percentile 7.9 7.7 9.0 8.5 7.7 7.7 9.6 8.8 7.4 8.3 8.7 8.7
20 percentile 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Wind
80 percentile 165.6 176.2 142.8 176.7 168.7 201.2 147.4 163.2 175.1 202.7 153.9 163.0
50 percentile 33.1 35.2 32.3 28.4 40.3 37.8 33.3 33.1 35.0 29.9 37.4 30.8
20 percentile 6.5 3.2 5.0 4.7 7.3 4.0 6.0 5.1 6.9 1.2 5.8 3.9
Combination Aa
80 percentile 211.9 214.5 210.7 211.9 220.9 215.6 212.5 205.1 215.5 216.4 212.4 218.2
50 percentile 55.6 57.6 54.5 55.0 58.2 59.2 56.3 57.9 59.2 60.0 58.4 62.1
20 percentile 8.0 4.9 7.0 8.5 8.4 4.4 8.0 8.6 9.0 4.0 6.9 4.8
Wind and solar, mixed locations
Combined Combination Bb Combination Cc Combination Dd Combination Ee Combination Bb Combination Cc Combination Dd Combination Ee Combination Bb Combination Cc Combination Dd Combination Ee
locations and
resources
80 percentile 218.2 211.4 213.4 212.4 216.6 204.2 207.6 209.2 217.4 214.3 211.6 211.2
50 percentile 63.4 57.0 67.8 67.6 64.9 60.2 71.6 73.2 60.3 63.0 68.1 72.4
20 percentile 5.5 8.3 11.5 13.4 5.1 9.2 15.2 19.5 4.0 4.8 8.0 14.3
a
Combination A: solar and wind combined for each location.
b
Combination B: Ottawa solar, Sault Ste Marie wind.
c
Combination C: Toronto solar, Wiarton wind.
d
Combination D: Ottawa and Toronto solar, Sault Ste. Marie and Wiarton wind.
e
Combination E: all locations and resources.
C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107 105

this threshold reduced to a range of 4470 to 4705 instances over the

Combination Ee
three years across all locations. With respect to the second research
question, Combinations B and C produce power below this
threshold between 4462 and 4640 h per year; again, this produces

0
4424

0
4336

4613

0
4147

4571
4189
significant benefits over the singular resources, but not necessarily
over Combination A. Further, as the number of instances reduces,
Combinations D and E show improvement over Combinations A, B
Combination Dd

and C, with production below this threshold between 4206 and


4336 h per year.
0
4425

0
4335

4554

4
4206

4452
4307
4. Discussion
Combination Cc

4.1. Analysis

This article has set out to answer the following questions: 1)


3
4117

0
4640

4132

2
4628

4284
4476

Does the combination of solar and wind resources in selected


Geographic dispersion

locations in Ontario serve to ‘smooth out’ power production, i.e.,


decrease instances of both high and low values, as compared to
Combination Bb

either resource acting individually? 2) Can this ‘smoothness’ be


further improved by dispersing these resources geographically
amongst locations? and 3) Does increasing the number of locations
0
4277

2
4483

4298

7
4462

4274
4483

with solar and wind resources further ‘smooth out’ power


production?
Wiarton

To answer these questions, this study has utilized three years of


3
4092

0
4665

4055

2
4705

4290
4470

data, from 2003 to 2005, for solar and wind resources for four
locations. In response to the first research question, combining
resources does ‘smooth out’ power production for solar and wind
Toronto

over the individual resource, as measured using a biophysical


8
4040

0
4712

4089

1
4671

4165
4595

indicator, in the following three ways. First, this research shows


that for this time period and for these locations, the combination of
Sault Ste.

wind and solar resources reduces the instances in which power


Marie

production is ‘off’. This is demonstrated in Figs. 2 and 4, which show


0

2
4130
4630

7
4157
4603

4217
4540
Combination Aa

a reduction in zero, or close to zero, values when solar and wind


Results for capacity of theoretical maximum for combinations of resources and locations, 2003, 2004 and 2005.

resources are combined. This is also demonstrated in Fig. 6 and


Ottawa

Table 1, with a rise in the 20th percentile value between single


3

0
4095
4662

0
4207
4553

4184
4576

resource and locations and either of Combinations A, B or C.


Further, Table 2 demonstrates a reduction in the number of recor-
Wind
27

20
3462
5271

20
3128
5630

3121
5474

ded instances below a threshold of 2% of a theoretical maximum


when comparing unique resources to Combinations A, B or C.
Wiarton

Second, this research also shows that this same combination of


0
Solar

0
3214
5546

0
3223
5537

3223
5557

resources reduces the instances in which both are above a peak


D: Ottawa and Toronto solar, Sault Ste Marie and Wiarton wind.

value, or in which power production is ‘on’. Again, this can be seen


Wind
35
3159

22
5566

3313

22
5447

3309
5441

in Figs. 2 and 4, which show a definite improvement in the high


values (particularly for wind) over those of Figs. 1 and 3. A reduc-
Toronto

tion in the 80th percentile value is found in Fig. 5 and Table 1 when
0
Solar

3266

0
5494

3301

0
5459

3301
5497

comparing the unique resources and Combinations A, B or C.


Further, Table 2 demonstrates a reduction in the instances above
A: solar and wind combined for each location.
Wind
16
3564
5180

20

the threshold of 50% of a theoretical maximum when comparing


3621
5134

20
3612
5391

unique resources and Combinations A, B or C.


Sault Ste.

B: Ottawa solar, Sault Ste Marie wind.

Third, this research shows that this combination of resources


Marie

Solar
0
3168
5592

0
3159
5601

0
3159
5535

improves the median value of output of power production. This can


C: Toronto solar, Wiarton wind.

E: all locations and resources.

be seen in Fig. 6 and Table 1, through the rise of the 50th percentile
Unique resources

Wind

value. This can also be seen in Table 2, as the instances in which


27
3462
5271

8
3321
5437

8
3323
5194

power is produced between the 2% and 50% thresholds rises when


Ottawa

comparing unique resources to Combinations A, B or C.


Solar
0
3308
5452

0
3327
5433

0
3327
5526

On the other hand, the results for geographic dispersion are


mixed. In response to the second research question, this analysis
shows that there is not much benefit produced from the combi-
% of theoretical

nation of one resource (either solar or wind) per location in two


maximum

locations over the benefits of combining two resources (both solar


2e50%

2e50%

2e50%

Combination
Combination
Combination
Combination
Combination
>50%

>50%

>50%

and wind) in one location. For instance, in Fig. 6 and Table 1, it can
<2%

<2%

<2%

be seen that Combinations B and C do not produce much


improvement in either the 80th percentile (by lowering it) or the
Table 2

2003

2004

2005
Year

20th percentile (by raising it) as compared to Combination A. The


d
a

e
b
c

same result can be seen in Table 2, whereby for each of the


106 C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107

measured thresholds, Combinations B and C do not demonstrate maximum values (i.e., are ‘on’ at the same time), it could be that
significant improvement over Combination A. a different combination of locations is more optimal in terms of
Turning to the third research question, there is improvement: lowering maximum production values in a similar manner. Further
when the number of locations is increased beyond two, geographic to this, multiple locations with multiple resources do produce
dispersion does improve the highs, lows and overall production of a more reliable resource, hence planners and decision makers
the combined resource by all approaches used. This can be seen in should continue to examine and encourage geographic dispersion
the drop of the 80th percentile and the rise of the 20th and 50th of renewable resources in Ontario. This could be taken into account
percentiles when comparing Combinations D and E to Combina- for multiple locations which are geographically dispersed on the
tions A, B or C in Fig. 5 and Table 1. This can also be seen in the same transmission or distribution system.
improvement in all thresholds (reduction in instances greater than
50% and below 2%) when comparing Combinations D and E to A, B
5. Conclusions
or C in Table 2.
Overall, this study found that the combination of solar and wind
4.2. Impacts of analysis
within locations and amongst two locations produces less vari-
ability in power production than when each resource is produced
A motivation for this research was to illuminate options for
on its own. Further, once more than two resources and two loca-
biophysical data to provide analysis of solar and wind resources
tions are combined, this ‘smoothness’ further improves. This
in Ontario. Further, given that Ontario can be characterized by
investigation was carried out using publicly available data; these
a changing policy environment, particularly with respect to the
data were used to develop a dimensionless measure, for equal
value of renewable power and options for connecting to the
annual amounts of solar and wind power. While the findings led to
grid, another motivation was to provide tools for electricity
specific policy recommendations (enumerated above), this area is
planners to better integrate renewable electricity to Ontario’s
certainly worthy of continued attention.
grid. In response, this study found that the easily obtainable
The authors suggest that CWEEDS data could be mined for other
CWEEDS data are a valuable asset in Canada as these biophysical
optimal location combinations according to the social and
data can be used to facilitate the analysis of solar and wind
economic elements that communities and planners find important.
resources. Whereas the literature demonstrates a fair bit of
Given that CWEEDS data are provided for locations across Canada,
complexity in analysis of complementarity between solar and
researchers and electricity planners in other provinces are
wind resources, the methods used in this analysis are meant to
encouraged to use these data to investigate the ways in which
allow planners to engage with the public and decision makers;
renewable resource complementarity could conceivably advance
hence the simplicity of the visual methods employed herein
their own policy goals.
might prove useful for such communication. Further, this anal-
As for Ontario, there is no reason why this method of analysis
ysis adds to the literature of grid-connected systems, as opposed
cannot be applied using more recent power production data; given
to autonomous systems, which require different decision-
Ontario’s three recent renewable power procurement programs,
making tools.
there are now more such data being collected. The OPA reserves the
As for decision-making, the analysis leads to a number of
right to audit these production data, hence, it is possible for this
conclusions. First, in order to meet the twin goals of renewable
planning agency to consider more analysis internally for improved
electricity provision and electricity system reliability, these results
decision-making for renewable power options.
indicate that an optimal policy solution for planners and decision
While CWEEDS data prove to be a very useful resource for
makers would be to encourage hybrid renewable electricity
investigation, much of this analysis is preliminary and points to
combinations in order to reduce overall variability of electricity
the need for more data collection and further research with which
produced for the grid in Ontario. In other words, electricity plan-
to make more reliable decisions. Further analysis could also
ners and decision makers in Ontario should strongly consider
compare these renewable resource (and/or electricity) data to
policies that favour the combination of renewable resources,
loads in different distribution areas within the transmission-con-
particularly solar and wind.
strained sections of the electricity system. Such an investigation
In terms of placing limits on awarding contracts for renewable
could determine whether peak production of the windesolar
electricity according to transmission constraint zones and distri-
combination coincides with peak demand, or, whether there is
bution constraints, planners should consider this method of anal-
further merit in pursuing options for geographic dispersion. This
ysis to ascertain the likelihood that both solar and wind power are
could most likely be done using the Independent Electricity
‘on’ simultaneously: this analysis shows that when resources are
System Operator’s (IESO) nodal (locational marginal) prices by
combined and weighted according to annual power production this
region.
likelihood is reduced, but not eliminated.
A better understanding of complementarity will allow renew-
Further to this, consideration should continue to be given to
able electricity to play a larger role in electricity production in
encouraging geographic dispersion of renewable electricity
Ontario, and should support the province in achieving its sustain-
production across Ontario. For instance, for this study, combina-
ability goals.
tions of locations were made assuming that windier regions and
regions with more urban density (i.e., more rooftops) could
produce better results. However, when comparing production Acknowledgements
values of combined solar and wind resources in the same or
different locations, for instance, comparing Combination A with This study was completed with support from the Solar Buildings
Combinations B and C, little difference was found. This could be Research Network under the Strategic Network Grants Program of
interpreted as geographic dispersion provides few benefits. On the the Natural Sciences and Engineering Research Council of Canada.
other hand, in a province with transmission constraints, this anal- The authors are grateful for this support. Further thanks are given
ysis still offers choice in decision-making as one combination is at to Dr. Ian Beausoleil-Morrison and Briana Paige Kemery for their
least no worse than the other. That is, in the instances where help in accessing and managing CWEEDS data. The authors,
locations are combined such that solar and wind still have several however, remain fully responsible for the contents of the article.
C.E. Hoicka, I.H. Rowlands / Renewable Energy 36 (2011) 97e107 107

References [11] Markvart T. Sizing of hybrid photovoltaic-wind energy systems. Solar Energy
1996;57:277e81.
[12] Ai B, Yang H, Shen H, Liao X. Computer-aided design of PV/wind hybrid
[1] Owens S. Energy, planning and urban form. Pion; 1986.
system. Renewable Energy 2003;28:1491e512.
[2] Goldemberg J, Johansson T, Anderson D. World energy assessment: overview
[13] Habib MA, Said S, El-Hadidy M, Al-Zaharna I. Optimization procedure of
2004 update. United Nations; 2004.
a hybrid photovoltaic wind energy system. Energy 1999;24:919e29.
[3] Ontario Power Authority. Directives to OPA from Minister of Energy and
[14] Diaf S, Diaf D, Belhamel M, Haddadi M, Louche A. A methodology for optimal
Infrastructure. Available from: http://www.powerauthority.on.ca/Page.asp?
sizing of autonomous hybrid PV/wind system. Energy Policy 2007;35:
PageID¼122&ContentID¼6268; 2010 [Accessed 15.02.10].
5708e18.
[4] Ontario Power Authority. OPA feed-in tariff program: quick facts
[15] Tina G, Gagliano S, Raiti S. Hybrid solar/wind power system probabilistic
table. Available from: http://fit.powerauthority.on.ca/Page.asp?PageID¼122&
modelling for long-term performance assessment. Solar Energy 2006;80:
ContentID¼10195&SiteNodeID¼1098&BL_ExpandID¼259; 2009a [Accessed
578e88.
25.01.10].
[16] Beluco A, de Souza PK, Krenzinger A. A dimensionless index evaluating the
[5] Ontario Power Authority. Ontario’s standard offer program: transmission
time complementarity between solar and hydraulic energies. Renewable
constraints. Available from: http://www.powerauthority.on.ca/Page.asp?
Energy 2008;33:2157e65.
PageID¼829&ContentID¼4061; 2008 [Accessed 15.09.09].
[17] Dihrab SS, Sopian K. Electricity generation of hybrid PV/wind systems in Iraq.
[6] Ontario Power Authority. Feed-in tariff program. FIT Rules. Version 1.0.
Renewable Energy 2010;35:1303e7.
Ontario Power Authority; 2009b.
[18] Environment Canada-Atmospheric Environment Service and The National
[7] Celik AN. Optimisation and techno-economic analysis of autonomous photo-
Research Council of Canada. Canadian weather energy and engineering data
voltaic-wind hybrid energy systems in comparison to single photovoltaic and
sets (CWEEDS Files) and Canadian weather for energy calculations (CWEC
wind systems. Energy Conversion and Management 2002;43:2453e68.
Files) updated user’s manual. Available from: ftp://arcdm20.tor.ec.gc.ca/pub/
[8] Natural Resources Canada. Photovoltaic potential and solar resource maps
dist/climate/CWEEDS_2005/ZIPPED%20FILES/ENGLISH/CWEEDS%
of Canada. Available from: https://glfc.cfsnet.nfis.org/mapserver/pv/
20documentation_Revised%2020081023.txt [Accessed 25.01.10].
pvmapper.phtml?MapSize¼500%2C500&ViewRegion¼-57000%2C5830950%
[19] Ontario Power Authority. Wind power contracts. Available from: http://
2C1688900%2C7523650&minx¼-2508487.000000&miny¼5139927.500000&
www.powerauthority.on.ca/Page.asp?PageID¼924&SiteNodeID¼234; 2009c
maxx¼3080843.000000&maxy¼10729257.500000&imagewidth¼500&image
[Accessed 22.01.10].
height¼500&CMD¼ZOOM_IN&CMD¼ZOOM_IN&CHKBOX%5B4240%5D¼4240&
[20] Statistics Canada. Population of census metropolitan areas (2006 Census
CHKBOX%5B2057%5D¼2057&CHKBOX%5B2701%5D¼2701&CHKBOX%5B2700%
boundaries). Available from: http://www40.statcan.ca/l01/cst01/demo05a-
5D¼2700&CHKBOX%5B92163%5D¼92163&units¼0&tilt¼1&period¼13&title¼
eng.htm; 2009 [Accessed 22.01.10].
PV+potential+and+insolation&title_e¼PV+potential+and+insolation&title_f¼
[21] Page J. The role of solar radiation climatology in the design of photovoltaic
Potentiel+ photovolta%EFque+et+ ensoleillement&NEK¼e&LAYERS ¼2700%
systems. In: Markvart T, Castañer L, editors. Practical handbook of photovoltaics:
2C2701%2C2057%2C4240&SETS¼1707%2C1708%2C1709%2C1710%2C1122&
fundamentals and applications. Elsevier Science & Technology; 2003. p. 5e70.
RLAYER¼92163; 2009 [Accessed 20.01.10].
[22] Clarke JA. Energy simulation in building design. 2nd ed. Oxford, UK: Butter-
[9] Ministry of Natural Resources (Ontario). Wind resource atlas. Available
worth-Heinemann; 2001.
from: http://www.lio.ontario.ca/imf-ows/imf.jsp?site¼windpower_en; 2009
[23] Duffie JA, Beckman WA. Solar engineering of thermal processes. 3rd ed.
[Accessed 25.01.10].
Hoboken, USA: John Wiley and Sons, Inc.; 2006.
[10] Notton G, Lazarov V, Stoyanov L. Complementarity of solar and wind sources:
[24] Gipe P. Wind energy comes of age. John Wiley & Sons Inc.; 1995.
application in Bulgaria and Corsica. In: Sayigh A, editor. World renewable
[25] Rowlands I, Jernigan C. Wind power in Ontario: its contribution to the
energy congress (WRECX); 2008.
electricity grid. Bulletin of Science, Technology & Society 2008;28:436.

You might also like