The Effect of Wind Generation on Combined Heat and Power
The thesis is submitted to University College Dublin in part fulfillment of the
requirements for the degree:
ME in Energy Systems
School of Electrical, Electronic & Mechanical Engineering
Supervisor: Prof. Mark O‘Malley
I would like to thank my supervisor, Prof. Mark O‘Malley, for his assistance in getting
me started on this project and arranging meetings. My supervisor at Dalkia, Patrice
Berthaud, was a great help in providing the financial model for the CHP plant. This
report would not have been possible without the kind assistance and data provided by
Kevin Hagan and Aoife Crowe from the Commission for Energy Regulation. I also extend
my gratitude to Pádraig Fleming from Bord Gáis for providing data and guidance. John
Fingleton and Michael Peters from Fingleton White gave me good perspective on CHP
during the project. Any mistakes are my own.
Many thanks go to Veolia Environment and the Ireland Fund of France for their generous
sponsorship of this effort. I would like to thank Dr. David Timoney for co-ordinating the
collaboration with Dalkia and for organising the energy systems course. I extend thanks
to the French ambassador, Yvon Roe D‘Albert, for hosting an unexpected sponsorship
ceremony at the French Embassy.
Whilst this report strives to ensure as high a level of accuracy as possible, I make no
warranties or representations of any kind with respect to: (a) the content of this
document (including without limitation its quality, accuracy, and completeness), or (b)
the content of any other document [or website] referred to [or accessed by means of a
hypertext link through this document]. Use of this document and the information it
contains is solely at the user's own risk. I accept no liability for any loss or damage
arising from use of this document or reliance on the information it contains. Where
other work is cited, readers should refer to the disclaimers attached to that work.
Combined Heat and Power (CHP) offers substantial savings on carbon emissions. A
megawatt of CHP can reduce CO2 emissions by the same amount as 1.5MW of wind.
However the large scale introduction of wind into the electricity sector can reduce
CHP‘s profits. Policy makers should be aware that by supporting wind, there may be an
increase in CO2 if a marginal CHP unit shuts down. This report uses modelled 2020
electricity price data from a previous study to examine the effect of wind on CHP
profitability. The cost-effectiveness of a 1MWe CHP was tested with different natural
gas and electricity price assumptions. The effect wind has on electricity prices was
found to be strongly dependent on the natural gas price and the generation margin. As
a share of all-island generation, 30% wind was found to reduce daytime electricity
prices by an average of 21% compared with a 10% wind penetration under a conceivable
2020 gas and carbon price assumption. It was found that a 30% wind penetration
reduces the gross savings of a CHP plant by 19% compared with a 10% wind penetration.
This decrease could be offset by regulators levying capacity charges on CHP based on
the volume of electricity they import, and not the capacity of their grid connection.
Table of Contents
List of Figures ........................................................................................ v
1. Introduction .................................................................................... 1
2. CHP Background ................................................................................ 2
2.1. CHP Targets ............................................................................... 2
2.2. Spark Spread .............................................................................. 4
2.3. CHP Barriers ............................................................................... 6
3. Single Electricity Market ...................................................................... 9
3.1. Capacity Payments ...................................................................... 10
3.2. System Marginal Price .................................................................. 11
4. Effect of Wind on Electricity Market ....................................................... 14
4.1. CER Market Impact Study............................................................... 15
4.2. CER Wind Data ........................................................................... 27
4.3. Wind Variability.......................................................................... 28
5. Natural Gas .................................................................................... 30
5.1. Shale Gas ................................................................................. 30
5.2. European Natural Gas Prospects ...................................................... 36
6. Effect of Wind on Natural Gas .............................................................. 38
7. Financial Analysis ............................................................................. 45
7.1. Effect of REFiT ........................................................................... 47
7.2. Effect of Wind on Startup Costs ....................................................... 50
7.3. Effect of Wind on Capacity Payments ................................................ 50
7.4. Financial Savings ........................................................................ 51
7.5. Payback Period .......................................................................... 57
7.6. Carbon Emissions With and Without CHP ............................................ 58
8. Conclusion ..................................................................................... 59
References .......................................................................................... 60
List of Figures
Figure 2.1: CHP capacity 1991-2008 including and excluding Aughinish Alumina  .... 2
Figure 2.2: CHP deployment scenarios and government 2010 and 2020 targets  ..... 3
Figure 2.3: CHP installed capacity by sector (2008)  ...................................... 3
Figure 2.4: Industrial sector spark spread 2004 to 2008  ................................. 5
Figure 2.5: Effect of spark gap on payback period of a 1MWe industrial CHP  ........ 5
Figure 2.6: Treatment of CHPs when applying for grid connection  .................... 7
Figure 3.1: Breakdown of SMP between uplift and revenue  ............................. 9
Figure 3.2: Decision paper example of capacity payment breakdown  ................ 10
Figure 3.3: Annual capacity payment pot 2007-2011 [17, 19] ............................... 11
Figure 3.4: Merit order for the start of 2008  ............................................. 12
Figure 3.5: Half hourly SMP data February 2008 to May 2010 in €/MWh ................... 13
Figure 3.6: Average SMP February 2008 to May 2010 ......................................... 13
Figure 4.1: Merit order effect at different times of the day  ........................... 14
Figure 4.2: Effect of a carbon price on the differential between coal and gas ........... 15
Figure 4.3: Additions to 2005 generation portfolio in AIGS by 2020  .................. 16
Figure 4.4: All-island peak demand extrapolated out to 2030 from 2010 GAR ............ 17
Figure 4.5: Actual wind generation vs. forecast ............................................... 18
Figure 4.6: Wind Forecast Error Duration Curve ............................................... 19
Figure 4.7: Correlation of natural gas price with SMP (assumes €30/tCO2)................ 19
Figure 4.8: Average SMP in Central Scenario vs. Low Growth Scenario .................... 21
Figure 4.9: Surplus capacity 2010-2016  ................................................... 22
Figure 4.10: Simple spark spread for CER scenarios .......................................... 23
Figure 4.11: Lowest 90% Price Duration Curve for selected 2020 portfolios............... 24
Figure 4.12: Top 10% Price Duration Curve for selected 2020 portfolios .................. 24
Figure 4.13: Annual average SMP by half hour of day for Low Fuel Scenario.............. 25
Figure 4.14: Annual average prices by half hour for Low Demand Scenario ............... 26
Figure 4.15: Difference in SMP by hour of day (P1-P5 Low Fuel Scenario)................. 26
Figure 4.16: Half hourly wind generation data for Portfolio 5 .............................. 27
Figure 4.17: Wind generation in CER study ..................................................... 28
Figure 4.18: 10 year monthly wind output variation  .................................... 29
Figure 5.1: Shale gas production 2000-2009  .............................................. 30
Figure 5.2: Potential future shale gas production  ....................................... 31
Figure 5.3: Shale production rate and breakeven US production prices  .............. 32
Figure 5.4: Annual average US wellhead natural gas prices  ............................ 33
Figure 5.5: Shale gas exploration sites in Europe  ........................................ 35
Figure 5.6: Shale gas land footprint  ....................................................... 35
Figure 5.7: Shale gas in context of global natural gas production  .................... 36
Figure 5.8: EU supply and demand projection to 2030 ....................................... 36
Figure 5.9: Historical and projected natural gas prices ...................................... 37
Figure 6.1: Annual Irish gas demand projections by sector .................................. 38
Figure 6.2: Objective function for a CEGN ................................................ 39
Figure 6.3: CCGT power output for high wind, low wind and base case scenarios  . 40
Figure 6.4: Linepack for low wind, high wind and base case scenarios  .............. 40
Figure 6.5: Wind Generation on 7th January 2010  ....................................... 41
Figure 6.6: 1 in 50 winter peak day demand and max supply capacity  ............... 42
Figure 6.7: Daily total Irish gas demand in 2010 and 2030  ............................. 43
Figure 6.8: Daily gas flows in the Irish market in 2020 from various sources  ........ 43
Figure 6.9: Flow flex duration curve for Ireland (mscm)  ............................... 44
Figure 7.1: Site electrical and natural gas demand with and without CHP ................ 45
Figure 7.2: Average SMP by category............................................................ 47
Figure 7.3: Average revenue received by wind generators in 2020 ......................... 49
Figure 7.4: Component energy cost comparison with and without CHP.................... 52
Figure 7.5: Savings and percentage savings with and without CHP ......................... 53
Figure 7.6: Sensitivity analysis Portfolio 5 Low Fuel .......................................... 55
Figure 7.7: Savings vs. Natural Gas Prices ...................................................... 56
Figure 7.8: Correlation of savings with wind penetration .................................... 57
Figure 7.9: All-island 2009 fuel mix ............................................................. 58
Utilising the exhaust heat from a power plant affords substantial carbon savings
compared to generating that heat separately. The penetration of CHP in Ireland has
been far below the technical potential for a variety of reasons discussed in section 2.
There is uncertainty as to what effect wind may have on the viability of CHP. The two
components which affect CHP are the price of electricity and the price of natural gas.
This report looks at how each may be affected by high levels of wind.
Section 2 discusses the current status of CHP in Ireland. Section 3 describes the
important aspects of the Single Electricity Market. Section 4 looks at how wind may
affect the electricity sector using data from the Commission for Energy Regulation‘s
(CER) ―Impacts of High Levels of Wind Penetration on the Single Electricity Market in
2020‖. Section 5 looks at the most salient aspects of the natural gas market in 2020.
Section 6 examines any possible effects high wind levels may have on natural gas supply
in Ireland. Finally, section 7 brings together insights from previous chapters and with a
financial model of a typical CHP plant explores what impact varying levels of wind may
have on CHP.
2. CHP Background
2.1. CHP Targets
A range of reports on Combined Heat and Power (CHP) in Ireland have been published
by the Sustainable Energy Authority of Ireland (SEAI) [1-11]. A 2001 report outlined the
government‘s targets for CHP under the National Climate Change Strategy . This
target was for a tripling of the then installed 122MWe to 377MWe installed by 2010,
which would abate 0.25MtCO2. This target was to exclude the large 310MWe plant
proposed for Aughinish Alumina, which was included in the Business as Usual (BAU)
scenario. At the end of 2008, 282MWe of CHP was considered to be operational .
However this includes the 160MWe addition to Aughinish Alumina  which means that
the adjusted installed figure is approximately 122MWe; almost the same figure as when
the National Climate Change Strategy target was set. This is shown in Figure 2.1 ,
however the estimate of installed capacity was lowered slightly in SEAI‘s 2010 report
Figure 2.1: CHP capacity 1991-2008 including and excluding Aughinish Alumina 
In 2007, the government‘s Energy White Paper  set targets for CHP deployment by
2010 and 2020. The 2010 target was set at a total installed CHP capacity of 400MWe,
with the provisional 2020 target described as ―at least‖ 800MWe pending further
analysis. Figure 2.2  shows these targets and SEAI‘s future deployment scenarios.
Figure 2.2: CHP deployment scenarios and government 2010 and 2020 targets 
From Figure 2.2 it can be seen that the 2010 target will likely be missed, and SEAI does
not envisage the provisional 2020 target being met.
The CHP installed capacity in Ireland is predominantly made up of industrial units. The
total number of installed CHP units was 167 in 2008, and industrial CHPs number 31
. While industrial CHPs only made up 19% of the number of units, they comprised
80% of the capacity. This is shown in Figure 2.3 .
CHP Installed Capacity by Sector (2008)
Figure 2.3: CHP installed capacity by sector (2008) 
It is clear that the majority of future potential capacity growth lies in the industrial
sector due to the much higher average capacity per unit compared to the services and
other sectors [3, 10]. This report will deal principally with industrial CHPs.
2.2. Spark Spread
The spark spread is the difference between what an industrial facility pays for a unit of
electricity and what it pays for a unit natural gas. Equation 1 shows the equation for
the spark spread.
Equation 1: Spark Spread
SEAI prefer to use the term ‗spark gap‘. Equation 2 shows the equation for spark gap.
Equation 2: Spark Gap
Both the spark gap and spark spread measure the same values. Figure 2.4 shows the
spark gap for industrial CHPs from 2004 to 2008. Since CHPs can last up to 25 years, the
spark gap volatility shown in Figure 2.4 is an issue which affects the investment
Figure 2.4: Industrial sector spark spread 2004 to 2008 
It can be seen that the spark gap has narrowed significantly in recent years. This has
the effect of extending the payback period. This effect is shown in Figure 2.5.
Spark Gap = 3
Spark Gap = 4
Spark Gap = 5
Figure 2.5: Effect of spark gap on payback period of a 1MWe industrial CHP 
Figure 2.5 shows a spark gap of 3, 4 and 5. Taking an electricity price of €0.10/kWh,
the difference between a spark gap of 3 and 4 is sufficient to reduce the payback
period from over 5 years to under 4 years.
2.3. CHP Barriers
There are a range of barriers to wider industrial CHP deployment. These are explained
in detail in . The barriers that most pertain to industrial CHPs are outlined below.
2.3.1. Suitability of Thermal Loads
CHP economics typically dictate that a unit should be operational for at least 4,000
hours per year and the CHP site should have balanced electrical and thermal loads .
While Ireland has some large heating loads, much of them are not suitable for CHP. The
cement industry consumes 19% of industrial energy but there is a mismatch between the
thermal demand of the site and the heat output from the CHP . CHP is also not
suitable for the cement industry because of the grade of heat required to fire the kilns
(1000°C). The mining industry consumes most of its energy as electricity, and thus has a
low heat to power ratio, which means that the potential for CHP deployment is limited.
2.3.2. Grid Connection
Another barrier to CHP is regulatory. If a CHP intends to export to the grid, they must
secure a grid connection. The criteria for obtaining a grid connection were amended by
the Commission for Energy Regulation (CER) in July 2009 . The Group Processing
Approach (GPA) of the CER is the process by which renewable generators receive offers
for connection to either the transmission or distribution system. The current iteration is
Gate 3, and applications are processed in batches based on time of application. This
can lead to significant delays for generators who have just joined the queue. However
based on the discretion of the CER, applications can be expedited if they meet the
public interest criteria. The public interest criteria can be summarised as :
Diversity of Fuel Mix
Predictability and power system support
The CER‘s amendment allows for the bypass of the GPA if the CHP is less than 5MWe and
is found not to interact electrically with other applications in Gate 3. Alternatively if
the CHP is greater than 5MWe then it may only be expedited if it can demonstrate to
the CER that it meets the public interest criteria. In addition if the CHP is seeking a
connection of less than 500kWe then if it meets the public interest criteria it can bypass
the GPA . This is summarised in Figure 2.6 .
Figure 2.6: Treatment of CHPs when applying for grid connection 
2.3.3. PSO levy
Another potential barrier to CHP adoption is the Public Service Obligation (PSO) levy.
The PSO levy is used to recompense ESB Public Electricity Supply (PES) for its obligation
to purchase peat generated electricity. It is also used to guarantee minimum electricity
to certain renewable generators, principally wind generation. The PSO is calculated
based on a site‘s Maximum Import Capacity (MIC). While a CHP might have a reduced
electricity import requirement the majority of the time, it will typically require a full
import connection for the period when the CHP unit is not available. Industrial sites
with CHPs under current rules thus have to pay the full PSO, irrespective of how often
they import electricity.
2.3.4. Use of System Charges
Transmission Use of System (TUoS) charges and Distribution Use of System (DUoS)
charges are levied on generators and consumers that are connected to the transmission
and distribution systems respectively . For a CHP which does not export, the
charges are based on the MIC and the charges remain the same .
2.3.5. Industrial Energy Reduction Programmes
Large energy users may prioritise relatively capital light measures such as efficiency
improvements over capital intensive measures such as CHP . However this can be
thought of as a positive barrier, because Energy Service Companies (ESCos) can combine
an energy efficiency programme with a reduced size CHP.
2.3.6. Corporate Investment Criteria
SEAI found that while the typical payback for a CHP would be 5 years, some corporate
investors will only consider capital projects with payback periods in the order of three
to four years . SEAI‘s survey found that the investment was still not seen as
economically viable when assessed against corporate investment criteria, even when
the investment was financed by ESCos.
2.3.7. Management & Operation
CHP is typically not a core activity within industry and therefore CHPs require external
expertise to run the plant. However ESCos can address this lack of expertise.
These barriers are summarised in Table 1.
Area Barrier Status
Economic Economic Viability Remaining Barrier
Corporate Investment Criteria New Barrier
Grant Assistance Partial/New Barrier
Fuel Prices Spark Gap Partial Barrier
Availability of Heat Loads Cement Manufacture Remaining Barrier
Mining Remaining Barrier
Dairies Partial Barrier
Electricity Market Grid Connection Partial Barrier
Public service Obligation Levy Partial/Removed Barrier
Use of System Charge Partial Barrier
Support Prices for Sale of Partial Barrier
Competition with Energy Industrial Energy Reduction New Barrier
Saving Programmes Programmes
Other Management & Operation New Barrier
Interaction with Authorities Partial Barrier
Table 1: Barriers to industrial CHPs 
3. Single Electricity Market
The Single Electricity Market (SEM) is a mandatory pool where the vast majority of
electricity is bought and sold at a single price. All units with a maximum export
capacity over 10MW (the de-minimis threshold) are obliged to sell their electricity to
the pool. Those units under 10MW can decide whether to sell to the SEM or not .
Generating units under this threshold – principally CHPs – can arrange a private contract
with an electricity supplier which will treat the electricity exported as a ‗negative
demand‘. This ‗negative demand‘ will reduce the volume of electricity the supplier
must purchase from the SEM to meet its customers‘ demand. For this to be attractive to
the electricity supplier, it would be priced at a discount to the SEM price.
The SMP has two components: the shadow price and uplift. The shadow price is
determined by the most expensive generating unit during a given half hour. However
the shadow price does not account for the generators‘ start-up and no-load costs. The
SMP is increased, by way of an uplift, to reflect these costs. While there are a number
of ways the uplift could be applied, after consultation, the regulators decided on an
approach which minimises distortion of the SMP . The breakdown of the SMP from
 is shown in Figure 3.1.
Figure 3.1: Breakdown of SMP between uplift and revenue 
In the example shown in , the time-weighted average SMP was €62.30/MWh of
which the uplift accounted for €8.21/MWh or 13%.
3.1. Capacity Payments
Generators also receive capacity payments which are designed to recover the capital
costs of power plants. Capacity payments are allocated across the year to incentivise
the provision of generation to meet a certain security standard (8 hours Loss of Load
Probability (LOLP) per year). The regulator determines the capacity payment pot each
year based on projected demand. As demand is higher in winter, the capacity payments
are higher than in summer. Figure 3.2 shows an example breakdown of a hypothetical
€100m capacity payment pot .
Figure 3.2: Decision paper example of capacity payment breakdown 
The capacity payment pot is determined by the product of the annual projected peak
demand and the annual cost of a peaking plant.
Equation 3: Capacity payment pot calculation
The cost of peaking plant is determined based on a forecast of the cost of a Best New
Entrant (BNE). In 2009, it was determined that for 2010 the cheapest peaking plant
would be a distillate fuelled plant in Northern Ireland . This type of plant was
determined to have a levelised cost per kW of €85.58. The projected revenues from
ancillary services were €4.84/kW. Subtracting the latter from the former resulted in a
BNE cost per kW of €80.74 for 2010.
Based on a model of projected demand and based on modelling to ensure 8 hours LOLP,
a projected capacity requirement of 6,826MW was assumed. This resulted in a capacity
payment pot of €551m. The capacity payment pot has decreased from 2009 due to both
a reduction in the BNE price (9% higher in 2008, when the pot was determined) and
reduced projected demand (8% higher in 2008) . For 2011, it has been determined
that a capacity requirement of 6,922MW is required and that the BNE price is
Figure 3.3: Annual capacity payment pot 2007-2011 [17, 19]
3.2. System Marginal Price
The System Marginal Price (SMP) is set for every half hour of the day by the variable
costs of the most expensive unit which is generating during that time plus an uplift. The
variable costs of a unit are dominated by the cost of fuel, and so the SMP has a strong
correlation with fuel prices. Electricity generating units are turned on until demand is
met starting with the cheapest units and extending to the more expensive units. This is
called the merit order. The merit order for the start of 2008 is shown in Figure 3.4.
Renewable generation (principally wind) has almost no variable costs and so the wind is
used when it is blowing. Coal is next cheapest, followed by peat and then baseload gas,
mid-merit gas and peaking plant.
C ost (€ /MWh)
50 mid-merit gas
0 1000 2000 3000 4000 5000 6000 7000 8000
C a pa city (MW)
Figure 3.4: Merit order for the start of 2008 
The SMP varies across the day based on the units required to meet the load. The SMP
data for February 2008 to May 2010 are shown in Figure 3.5. Data was obtained from
the Single Electricity Market Operator (SEMO) website .
The periods above €200/MWh are relatively few, and they typically represent times
when Ireland has purchased electricity at short notice from Britain across the
Figure 3.6 zooms in on the area where the majority of data points lie. The five day
moving average is shown as the continuous white line. It can be seen that there is a lot
of volatility in the average SMP both monthly and yearly. This can be explained to a
large extent by natural gas price movements and changes in demand.
Figure 3.5: Half hourly SMP data February 2008 to May 2010 in €/MWh
Figure 3.6: Average SMP February 2008 to May 2010
4. Effect of Wind on Electricity Market
Wind tends to reduce the SMP by shifting the supply curve to the right. This is called
the merit order effect (MOE) and is illustrated in Figure 4.1. Wind has very low variable
costs, and so is always in merit when it blows. The demand curve for electricity is
relatively inflexible, so the reduction in cost doesn‘t tend to increase consumption.
Figure 4.1 shows the MOE effect to be greatest at peak electrical demand because it
displaces expensive peaking plant. The effect is lower during the day, and lowest at
night because wind may only be displacing the relatively inexpensive coal and nuclear.
Figure 4.1: Merit order effect at different times of the day 
Various studies have attempted to quantify the MOE of wind on electricity prices [23-
27]. The studies use different assumptions and look at different countries with different
market structures, so the results may not be comparable. The qualitative results may
be more meaningful.
In  it was found that the MOE in 2006 for Germany was 16% lower when the carbon
price was increased from €0 to €40 per tonne CO2. This is explained by the fact that in
Germany at times of low demand coal sets the SMP, while at high demand gas sets the
SMP. A carbon price increases the cost of coal more than the cost of gas generation.
This decreases the shoulder between the lower cost coal and higher cost gas. When
wind generation causes coal to set the SMP, where otherwise it would be gas, this then
decreases the MOE. This is illustrated in Figure 4.2.
Figure 4.2: Effect of a carbon price on the differential between coal and gas
Looking at Figure 4.2, it can be observed that if the price of coal increases but the gas
price does not, the shoulder will decrease and vice versa.
Delarue et al.  showed that for the Belgian system the MOE was relatively
insensitive to wind forecast error. Munksgaard  concluded that taking account of the
MOE, Danish consumer prices were €0.05—0.06/kWh more expensive than they would be
without wind. This was attributed to the feed-in tariff structure paid to wind
4.1. CER Market Impact Study
The CER conducted a study which calculated the SMP in 2020 for various penetrations of
wind . The study varied the natural gas and carbon price and computed the SMPs in
2020. The five generation portfolios studied were similar to those used in the All Island
Grid Study (AIGS) [31-36], with a few small changes. These changes were the inclusion
of 190MW of CHP, the addition of the Huntstown Power station, a higher maximum load
(10,407MW in the central scenario vs 9600MW in the AIGS), some different assumptions
about which plants were retired by 2020, and slightly different plant characteristics.
Apart from these, the substantive changes to the 2005 Irish portfolio were the same as
assumed in the AIGS Workstream 2B . These additions are shown in Figure 4.3 for
the five respective portfolios.
Figure 4.3: Additions to 2005 generation portfolio in AIGS by 2020 
Portfolio 1 is useful as a benchmark, since the proportion of wind is similar to that
presently on the Irish system (1.4GW) . Portfolio 2 is similar to Portfolio 1 but has
an extra 2GW of wind. Portfolios 3 and 4 are characterised by a high reliance on OCGTs
and coal, respectively. Portfolio 5 is similar to Portfolio 2 but has 2GW extra wind.
Portfolio 5‘s installed capacity of 6GW of wind in 2020 may now overshoot Ireland‘s 40%
renewable electricity target, with Eirgrid‘s latest Generation Adequacy Report (GAR)
projecting that 40% renewables may translate to 4.6GW wind .
The principal difference between Portfolios 1, 2 and 5 is the level of wind penetration
and for the purposes of this report this shall be the main consideration.
Figure 4.4 shows the all-island peaks extrapolated out beyond 2016 from the 2010 GAR.
It is clear that the AIGS  and the CER study  have projected 2020 demand far
ahead of the likely level. Electricity demand may rise at a faster pace after 2020 if
there is a concerted transfer of transport and thermal loads to electricity (e.g. electric
vehicles and electric heating). A scaled down Portfolio 5 offers perhaps the best post-
recession view of the generation mix in 2020. The exact numerical results are not as
important as the relative values of the portfolios, and for this the AIGS and CER‘s study
are very useful.
Figure 4.4: All-island peak demand extrapolated out to 2030 from 2010 GAR
The CER study computed the average SMP across the year using a market simulation
software model called PLEXOS. This model has been successfully calibrated for the
current SEM and is accurate. The CER took an annual wind profile and dispatched the
generation units assuming perfect foresight. In reality the wind forecast will be revised,
and the decision to dispatch units will be based on probabilistic methods. This is an
important caveat when interpreting the results. In 2020, if, instead of committing units
based on a day ahead schedule at present, a rolling unit commitment schedule is
implemented, the costs are projected to be similar . Based on the AIGS portfolios,
the cost increase comparing perfect wind forecast and that accounting for wind‘s
uncertain output has been estimated as 1.2% . The results presented here should
thus be broadly comparable to what can be expected under real circumstances. To
illustrate, Figure 4.5 shows the relative accuracy of wind forecast and wind output for
the first 8 months of 2010 by 15 minute interval .
Figure 4.5: Actual wind generation vs. forecast
Figure 4.6 presents the data in Figure 4.5 in a different manner. The error forecast
duration curve represents the percentage of time the (absolute) error was below a
certain threshold. For January to August 2010, 80% of the time the error has been
within 100MW and 50% of the time the error has been within 50MW. A small percentage
of the time the forecast is off by a large margin.
Figure 4.6: Wind Forecast Error Duration Curve
The CER used three different price scenarios and computed the SMP for each. The study
was performed in July 2008 when gas prices were very high and since then the price has
reduced considerably. The low price scenario is closest to the current price. Figure 4.7
shows the results of the study, and also shows the 2010 price.
Figure 4.7: Correlation of natural gas price with SMP (assumes €30/tCO2)
Figure 4.7 shows a high degree of correlation between the natural gas price and the
SMP for the respective portfolios. The lowest SMP can be seen in the high coal Portfolio
4. The next cheapest is the 6GW wind Portfolio 5. Third cheapest is Portfolio 2, which
adds 4GW wind and extra CCGTs. Next is Portfolio 1, which represents business as usual
with 2GW of wind. The most expensive is Portfolio 3, which relies heavily on inefficient
OCGTs to balance 4GW of wind, so this is not surprising.
The CER‘s result that Portfolio 5 and Portfolio 2 result in lower average SMPs than
Portfolio 1 is evidence that increased wind lowers the SMP. The higher the wind
penetration the lower the SMP. The effect is greater at higher natural gas prices as the
divergence of the lines in Figure 4.7 shows.
The three scenarios also varied the carbon price from €15/tCO2 in the low fuel scenario
to €30/tCO2 in the central scenario and €45/tCO2 in the high fuel scenario. In the low
fuel scenario this resulted in the carbon price making up half the price of coal
generation. As a result, coal was utilised much less in the low fuel than in the central
and high scenarios when the carbon price was a proportionately lower share of coal‘s
fuel costs. The cost of carbon was only approximately 10% of the cost of natural gas in
the low fuel scenario.
Figure 4.7 shows results for a peak load of 10.4GW; however the CER also performed a
low electrical demand growth sensitivity of 2.7% against 3.5% in the other scenarios.
This resulted in a peak demand of 9.4GW. Each portfolio in the low growth sensitivity
contained 1GW of generation (10% of peak) over and above what was required by
system security standards. The impact of this on the SMP is shown in Figure 4.8.
Figure 4.8: Average SMP in Central Scenario vs. Low Growth Scenario
Figure 4.8 shows for each portfolio the average SMP reduces for a given portfolio if
there is lower demand. The most significant reduction is in Portfolio 1 where the
€20/MWh reduction occurs because expensive peaking generation is used less because
of the high generation margin (installed plant capacity above the annual peak).
Portfolios 2 and 5 show lower reductions due to the higher wind capacity. A large
reduction occurs in Portfolio 3 due to the reliance on OCGTs, and the high coal Portfolio
4 shows a small reduction.
Figure 4.8 shows that a high generation margin leads to a lower average SMP. Currently,
and at least until 2016, Eirgrid projects a rising surplus capacity on the system .
This is shown in Figure 4.9.
Figure 4.9: Surplus capacity 2010-2016 
A surplus capacity can be expected to dampen average SMPs due to the reduced use of
peaking plant. The large generation margin is markedly different from pre-recession
GARs where deficits were forecasted as recently as the 2008-2014 GAR . Eirgrid‘s
median scenario is based on an average GDP growth of 5% between 2011 and 2016,
while the low scenario is based on 4.2% GDP growth. These scenarios respectively
envisage electricity demand growth of 2.2% and 1.9%. To the extent these figures are
optimistic the surplus capacity could be larger. Taking the low demand scenario, a
surplus capacity in 2016 of approximately 1550MW would exist. The low demand
scenario envisages the all island peak as 6,939MW in 2016. This equates to a substantial
22% surplus margin, which based on Figure 4.8 will act to keep the SMP lower than if a
normal margin existed.
Figure 4.10 uses the same data as Figure 4.7 to generate spark spreads. The simple
spark spread is the difference between the time weighted average SMP and the natural
gas price. The real spark spread is the difference between the metered electricity price
(€/MWh) less the natural gas price.
In each case the spark spreads of Portfolios 2 and 5 are below the low wind Portfolio 1.
The smallest difference between P2 & P5 and P1 is €7/MWh for the low fuel price
scenario, with the largest difference being €25/MWh for the high fuel price scenario.
Figure 4.10: Simple spark spread for CER scenarios
An instructive way to view the price data is through a price duration curve. A price
duration curve shows the proportion of the year that the SMP is below a given
threshold. The highest level of variation is in the top 10% of prices which typically
represent imports and peaking generation. For clarity, Figure 4.11 and Figure 4.12
respectively show the price duration curve for the lowest 90% and highest 10%
separately. Portfolios 1, 2 and 5 are considered from here on, as Portfolios 3&4 are
highly unlikely future generation mixes. The low fuel (LF) and low demand (LD)
scenarios were selected for comparison. The actual SMPs for 2008 and 2009 are shown
Figure 4.11: Lowest 90% Price Duration Curve for selected 2020 portfolios
Figure 4.12: Top 10% Price Duration Curve for selected 2020 portfolios
Another useful way to view the price data is by time of day. Figure 4.13 shows the three
selected low fuel price scenarios and Figure 4.14 shows the three low demand
For CHP units which export or import electricity at night, the effect of increased wind
on the system will be negligible. However during the day, wind will adversely affect
CHP units by lowering the SMP. For example Figure 4.15 shows the difference in SMP
between Portfolio 1 and Portfolio 5 for the low fuel scenario. The average difference
between 9am and 10pm is €24/MWh. As a percentage of Portfolio 1 daytime prices, this
is a substantial 21% reduction. For the low demand scenario, the corresponding
reduction is €19/MWh and 14%. Thus for a system with a higher generation margin the
merit order effect is lower.
Figure 4.13: Annual average SMP by half hour of day for Low Fuel Scenario
Figure 4.14: Annual average prices by half hour for Low Demand Scenario
Figure 4.15: Difference in SMP by hour of day (P1-P5 Low Fuel Scenario)
4.2. CER Wind Data
The wind data used in the CER study was based on wind capacity factors assigned to
different regions of Ireland by Eirgrid. This data for Portfolio 5 is shown in Figure 4.16.
Figure 4.16: Half hourly wind generation data for Portfolio 5
As Figure 4.16 shows, there is substantial variation in the 48 hour wind data. However it
can be seen that the moving average never reaches zero and only once touches
5,000MW while 6,000MW is the installed capacity. All the portfolios use the same wind
data, and so the output follows the same pattern regardless of installed capacity.
Figure 4.17 shows the wind output for the five scenarios.
Figure 4.17: Wind generation in CER study
From Figure 4.17, it can be seen that the average wind output for January was 3,000MW
for P5, or a capacity factor of 50%. This contrasts with the average August output of
1,000MW, or a capacity factor of 17%. The annual capacity factor for the wind data
used was 32%.
4.3. Wind Variability
Wind is variable over hourly, daily, monthly and yearly timescales. Figure 4.16 and
Figure 4.17 show the variation in the CER study wind output data on a half hourly, daily
and monthly timescales. The CER used a specific pattern of wind data for their study, it
is worth comparing this to historical patterns. Wind data going back 10 years is held by
the Electricity Research Centre. Figure 4.18 shows the monthly variation in wind data
for the last ten years. The monthly average capacity factor and max and min for each
month are shown. The capacity factor for the CER‘s study is also overlaid. The largest
historical variation in the wind data is for the month of February, where over the last
ten years the capacity factor has ranged from 30% to 57.5%. The variation is lowest in
July, ranging from 15% to 23%. The CER study‘s wind data tracks the average reasonably
5. Natural Gas
The cost of natural gas is a key determinant of CHP economics. Shale gas is the most
dynamic element of the natural gas market and is the most uncertain element of future
natural gas supply. The current shale gas situation is described, followed by a brief
discussion of the outlook for European natural gas.
5.1. Shale Gas
The large scale commercialisation of shale gas is a recent phenomenon, although the
first gas well in America was actually from a shale formation in Fredonia, New York in
1821 . The principal reasons behind this expansion are :
Advances in horizontal drilling
Advances in hydraulic fracturing
The rapid increase in natural gas prices as a result of significant supply and
Figure 5.1 shows the expansion of shale gas from 2000 to 2009 .
US Shale Gas Production 2000-2009
Marcellus Haynesville Woodford
Trillion Cubic Feet per Year
As % of US 2009 Demand
Fayetteville Barnett Antrim 8%
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Figure 5.1: Shale gas production 2000-2009 
There are a range of different shale ‗plays‟ in the US. The most important of these is
the Barnett shale, which accounts for over half of the current production. Future
increases will depend on the other plays, with the most promising being the Marcellus
and Haynesville shales.
In June 2010, MIT released an interim report entitled The Future of Natural Gas .
This was a multi-disciplinary study of the future supply and demand of natural gas, with
a focus on shale gas. The report includes sensitivities to account for the uncertain
nature of natural gas resources. Figure 5.2 show their projections of shale gas out to
Figure 5.2: Potential future shale gas production 
Figure 5.2 is in bcf/day (billion cubic feet). Converting that to tcf/yr, MIT predict shale
production rising from 2.8tcf in 2009 to 9.8tcf in 2020 and 10.6tcf in 2030. The key
reason why the potential production rate reaches a steady state is because of high
decline rates. Production from shale gas wells drop off precipitously within a couple of
years. Figure 5.3 shows the drop off in production from certain plays and additionally
the breakeven gas price for the different categories of US gas .
From Figure 5.3, it can be seen that the production from wells in the Haynesville drop
off approximately 75% within a year while Marcellus and Barnett wells drop off by
approximately 70%. Shale gas has thus been described as a „treadmill‟ where production
rates are limited by the number of rigs drilling for gas.
Although there is uncertainty as to the cost of shale gas, Figure 5.3 shows the MIT
study‘s cost curves for shale and the other types of US natural gas.
Figure 5.3: Shale production rate and breakeven US production prices 
The MIT study assumes there is 500tcf of shale gas available below $8/MMBtu (a MMBtu
is equivalent to 10 therms or 0.96mcf). Five hundred tcf is enough for 25 years of US gas
supply. Years of supply is a flawed metric, however, since based on Figure 5.2, the
maximum sustainable rate of production would be in the region of 10tcf. So US shale
could potentially supply approximately 40% of US 2009 demand for 50 years at under
However others dispute the assumed shale cost curve shown in Figure 5.3 [46-48]. The
disagreement centres on whether companies are including the ‗full cycle‘ costs. In
March John Dizard wrote in the Financial Times :
“Industry advocates forward consultants‟ studies talking of “learning curves” and $5
gas as far as the eye can see. Not many neutrals.
So I worked people in the energy service industry, and gas producers to try and refute
Ben Dell‟s numbers. I couldn‟t. My industry sources‟ numbers all converged close to $8
per mcf. They do not believe the producers are covering their all-in costs.”
The costs quoted by the shale companies are thought to exclude land costs, seismic
surveys and operating costs . They also rely on optimistic production forecasts, with
residual production lasting up to 40 years. Since shale is a recent phenomenon, these
forecasts cannot be proved one way or the other.
Figure 5.4 shows the inflation adjusted average US natural gas price from 1982 to 2009 .
Annual Average US Wellhead Natural Gas
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
Figure 5.4: Annual average US wellhead natural gas prices 
Natural gas prices have average $4.50/MMBtu for the first half of 2010 . If the true
cost of shale gas was closer to $8/MMBtu than $4/MMBtu, then shale producers would
be losing money. However there is an explanation as to why producers would continue
to drill even if their true costs are above current market prices. In , Mark Rowland,
Chief Financial Officer (CFO) of a major shale gas company, Chesapeake, states: ―I
believe at least half and probably 2/3 or 3/4 of our gas drilling is what I would call
involuntary. It's being incentivized by something other than the gas price.”
Two reasons why companies would continue to drill are:
They have hedged at a higher natural gas price, and so are not exposed to the
current uneconomic price
They acquired expensive leases prior to 2009 which will expire unless they drill
Much of the current drilling may thus be ‗involuntary‘. Shale gas is being overproduced
due to factors other than the gas price. As hedges and leases expire, drilling could be
expected to decline. With the high decline rates from shale gas wells, the price of
natural gas should converge to a price that incentivises enough shale gas production to
meet demand. Whether that price is the $4/MMBtu that the MIT study suggests, or
closer to $8/MMBtu, remains to be seen.
Shale gas exploration in Europe is still in the preliminary stages. Figure 5.5 shows the
sites where oil and gas companies are currently exploring. Exxon Mobil has found the
results of drilling in Hungary disappointing and has withdrawn from that area .
European geology is more complex and fragmented than America‘s which may make
developing shale more challenging. Potential additional barriers to shale include
Europe‘s higher population density, stricter environmental regulation and limited
numbers of drilling rigs . The fracturing of shale requires significant quantities of
water. Figure 5.6 shows a typical shale gas drilling site. The footprint of a shale gas
operation will be more significant if shale resources lie under a built up area. The pools
of water must be remediated as they contain chemicals.
The US Energy Information Administration (EIA) made a forecast for future shale gas
production in March 2010 . Figure 5.7 shows this projection in the context of global
natural gas production. As can be seen, in 2009 shale gas only produced 1.7% of global
natural gas production. The EIA shows global production of shale reaching
approximately 10 tcf (0.3 tcm) by 2030, or 6.4% of the projected global total. This 10
tcf is close to the 10.6 tcf the MIT study  considered to be the maximum sustainable
output of US shale gas. While shale gas production outside of the US is likely to prove
more challenging, if European (or Indian/Chinese) shale proves feasible, then this total
may rise. If the rest of the world was to match the US shale output from the MIT study,
then the total would rise to 21.2tcf or 13.5% of the projected total. If EU production of
shale developed quickly it could have a significant dampening effect on EU natural gas
Figure 5.5: Shale gas exploration sites in Europe 
Figure 5.6: Shale gas land footprint 
Figure 5.7: Shale gas in context of global natural gas production 
5.2. European Natural Gas Prospects
A projection of EU gas production and imports out to 2030 from the IEA‘s World Energy
Outlook 2009 Reference Scenario is shown in Figure 5.8 . The IEA project European
gas production to continue declining while from 2015 demand is expected to pick up.
The result is that by 2020 EU imports increase from 312bcm in 2007 to 425bcm in 2020
and 428bcm in 2030. The IEA makes a conservative estimate of 15bcm of EU
unconventional production (coal bed methane and shale gas) by 2030. This is the largest
uncertainty of future EU production. UK gas supply is projected to decline from 73bcm
in 2008 to 44bcm in 2015 and under 20bcm by 2030.
Figure 5.8: EU supply and demand projection to 2030
No one can predict the future gas price, however the IEA take a nuanced approach and
their forecast is useful for comparison. Figure 5.9 shows the historical gas price and the
IEA‘s projection for European gas prices to 2020. The IEA still think, on balance, that EU
prices will retain a link to oil. The rise shown is linked to the IEA‘s projected increase of
oil prices to $100/bbl in 2008 prices by 2020. Also shown are the CER‘s assumed gas
prices in 2020. It can be seen that the central and high fuel prices are well above what
is currently expected. This is because the CER study was carried out during the run up
of oil and gas prices in 2008.
Figure 5.9: Historical and projected natural gas prices
6. Effect of Wind on Natural Gas
Wind has the potential to affect the natural gas grid, which, as a gas consumer, could
impact CHP. Looking at Figure 4.16 it is clear that the output of wind can vary from
high output to low output. Flexible power plants, principally natural gas fired, must
counteract these fluctuations. The changing electrical output places a demand on the
natural gas grid. The question then arises as to whether the natural gas grid can supply
sufficient quantities of natural gas to these power plants. A large part (65% in the 2009
gas year) of Ireland‘s natural gas demand comes from power generation. Figure 6.1
shows this and also shows projections out to 2019 estimated in the CER‘s Joint Capacity
Statement . Very modest growth is expected principally due to the weakened
economy, energy efficiency, and increasing renewable penetration.
Figure 6.1: Annual Irish gas demand projections by sector
Few studies have been undertaken on the effect of wind on natural gas. Of those, the
approach generally taken is to model a Combined Electricity and Gas Network (CEGN)
[55, 56]. A CEGN takes account of the physical constraints on both the electricity and
natural gas networks. Qadran et al.  describe a CEGN which seeks to minimise total
system costs over a time horizon of two days. Figure 6.2 shows the cost function which
they seek to minimise.
Figure 6.2: Objective function for a CEGN 
The technical limitations of both networks are constraints on the least cost optimisation
model. The authors selected a high wind Great British system which approximated the
Gone Green scenario laid out by National Grid as a likely outcome by 2020 if EU targets
are to be met . Qadran et al. assume 25GW of wind installed in Great Britain by
2020, although the Gone Green scenario lists 29.4GW.
Based on January 2005 weather, the wind output in January 2020 in Britain was
determined. The fluctuations are balanced principally by CCGTs ramping up and down.
A two day period of low wind output and a two day period of high wind output were
selected to test the effects on the gas grid. Figure 6.3 shows the CCGT output over the
assumed days. The CCGT power output is shown for the low wind case, the high wind
case, and for 2009 as a reference case. At hours 42-44 in the low wind case, peak
electrical gas demand coincides with non-electrical gas peak demand. The pressure in
the gas pipeline was found to be insufficient to supply all the CCGTs, and for those few
hours the interconnector with France was utilised.
Figure 6.3: CCGT power output for high wind, low wind and base case scenarios 
Figure 6.4: Linepack for low wind, high wind and base case scenarios 
The fundamental difference between electricity and gas networks is that while
electrical energy in a wire travels at close the speed of light, gas in a pipeline flows
orders of magnitude slower at approximately 10m/s or 36km/hr. Hence gas injected at
Moffat, Shannon, or Kinsale will take several hours to reach Dublin. Deliverability from
the gas grid relies on maintaining high pressure in the gas pipeline. Gas withdrawals are
typically balanced with gas injections over a 24 hour period. If too much gas is
withdrawn during the day, and high-deliverability storage located close to the load
centre can‘t inject fast enough, then the pressure in the pipeline will drop. The
pressure drop is proportional to the drop in the amount of gas or ‗linepack‘ in the gas
pipeline. Figure 6.4 shows the drop that occurred between hours 36 and 44.
An intra-day study of the variation of linepack with wind output has not been performed
for Ireland. However the CER‘s Joint Gas Capacity Statement 2010 stress tests the gas
grid for a 1 in 50 severe winter . Additionally the high demand of this past winter
was a real world test.
A record gas demand was set on 7th January 2010. The wind generation on this day was
very low as is shown in Figure 6.5. Wind output averaged just 10% of rated capacity and
output was only 5% during the peak demand period of 5.00-6.00pm when electrical
demand peaked at an all-time record 4,950MW.
Figure 6.5: Wind Generation on 7th January 2010 
The electrical peak coincided with a 20.7 Degree Day (DD) at Dublin Airport, which was
just short of the 21DD for a 1 in 50 winter. The industrial/commercial demand was not
at its previous record due to the recession. The record combined demand of Ireland,
Northern Ireland and the Isle of Man was 30.2 mscm/day. The gas grid successfully met
this without the need to resort to curtailment. If the deliverability was not there, then
the power sector may have had its gas supply curtailed .
The Gas Capacity Statement models the future 1 in 50 day requirements of the grid for
each year until 2019. Figure 6.6 shows the components of demand and the ability to
meet the demand based on whether Ireland has no storage (Moffat only), whether there
is storage at Inch (Moffat + Inch), the addition of Corrib (Moffat + Inch + Corrib), and
the addition of storage at Larne (Moffat + Inch + Corrib + Larne). As is evidenced, with
the addition of Corrib in 2013, the maximum gas supply is comfortably above a 1 in 50
winter peak day requirement. The situation is further improved if the Larne storage
facility comes online in 2016. The addition of Corrib is highly likely; however the
commercial decision for Larne has yet to be taken. The Shannon LNG terminal would
further improve the maximum supply capacity. Shannon LNG has planning permission
however a commercial decision has yet to be taken.
Figure 6.6: 1 in 50 winter peak day demand and max supply capacity 
A recent Poyry report investigated the issue of gas intermittency in Great Britain and
Ireland in 2030, assuming high levels of wind . Eight GW of wind were assumed to
be installed by 2030, which is substantially above the 4.6GW which will represent the
40% renewable target in 2020. Figure 6.7 shows the results of Poyry‘s modelling.
Substantially higher variation is visible compared to the present situation.
Figure 6.7: Daily total Irish gas demand in 2010 and 2030 
The study also modelled how the flows from Corrib, Inch, Shannon, Larne and Moffat
would look in 2020.
Figure 6.8: Daily gas flows in the Irish market in 2020 from various sources 
The study suggests that the provision of all these sources could result in an oversupply
of flexibility to the market. This would be favourable from a security of supply point of
view but the returns to the operators of these facilities may not be sufficient. Poyry
neglected to study the intra-day effects in their study as Qadran et al did. They
approximated this, however, by measuring the flow-flex. Flow-flex is defined as the
ratio of the average gas flow rate over the first 16 hours of the day to the average
across the full day. A high flow flex means more erratic gas demand. From Figure 6.9,
the green 2019 line shows approximately 140 days of higher flow-flex than 2009. Poyry
did not perform a hydraulic study of the Irish gas grid which would be required to
determine whether there would be periods of low pressure. However Poyry stress that
changes are likely to be gradual and that the System Operator will have time to adapt,
for instance by moving to shorter balancing periods.
Figure 6.9: Flow flex duration curve for Ireland (mscm) 
An intra-day study with a two hour time step such as in Qadran et al would clear up any
uncertainty, however based on Figure 6.6, the gas grid appears robust. Any disturbances
would only be for a very small number of hours and could be addressed by operator
vigilance. All power plants are required to hold reserve fuel oil which could be
employed during periods of high gas grid stress.
7. Financial Analysis
A financial model of a representative CHP plant, compiled by Dalkia, was used in Excel
to determine the effect of different levels of wind on the overall savings a CHP accrues.
The input data was the SMP for the selected scenarios from the CER study. The
modelled CHP plant was sufficient to meet the baseload thermal demand of a typical
pharmaceutical plant. The plant has a usual day time (8am-11pm) demand of 1.9MWe
which falls off at night to 1.3MWe. The thermal demand on the site ranges from 1.5MWth
to 1.9MWth. The CHP has been sized to provide a constant thermal baseload of 1.2MWth
and 1MWe. The availability of the CHP is 92%. All electricity is imported and backup
boilers employed for the 8% CHP downtime. The site has a heat load factor of 99%,
meaning 1% of the heat energy produced by the CHP is not put to a useful purpose.
The result of the CHP is to increase the site‘s natural gas requirements by 67% but to
reduce the electrical requirements by 45%. Figure 7.1 shows the monthly volumes of
electricity and natural gas required with and without CHP.
Figure 7.1: Site electrical and natural gas demand with and without CHP
The savings made by the reduced electricity imports must pay for the increased gas use
and capital costs. As Figure 4.13-14 show, wind will principally affect the daytime
electricity price by reducing the SMP.
The site‘s electrical demand was broken up into five categories:
Winter Peak (Jan, Feb, Nov, Dec 5-7pm)
Winter Day (Jan, Feb, Nov, Dec, 8am-11pm excl. 5-7pm)
Summer Day (Mar-Nov 8am-11pm)
Weekend day (all weekends 8am-11pm)
Night (all nights 11pm-8am)
The half hourly SMP data was arranged into these five bins for each of the categories
and the average price in each category was determined for the scenarios from the CER
study. The results are shown in Figure 7.2. The same trend observed in Figure 4.13 is
evident. The largest difference between 2GW, 4GW and 6GW of wind is seen on the
winter peak with a reduction of approximately €20/MWh per additional GW for the low
fuel scenario. The next largest change is in the summer day (which included the 5-7pm
peak) where in the low fuel scenario the average SMP drops from €110/MWh in Portfolio
1 to €94 in Portfolio 2 and €84/MWh in the high wind Portfolio 5. Figure 7.2 replicates
what Figure 4.15 showed - that there is no difference in the night time price regardless
of wind penetration. The same categorisation is shown for the low demand, 2008 and
2009 data for comparison.
Figure 7.2: Average SMP by category
The Excel model allowed for an input of SMP prices and natural gas prices to determine
the total energy bill with and without CHP. It includes the various components of the
final price including:
Gas commodity cost
Gas transmission charges
Gas distribution charges
The most direct impact wind could have on these charges is via the Renewable Energy
Feed-in Tariff (REFiT).
7.1. Effect of REFiT
REFiT was introduced in 2006 to facilitate the expansion of renewable energy to meet
the 2010 renewable targets. It currently provides support for 1,075 MW of wind
generation , or 74% of the 1,459MW installed as of July 19th 2010 . REFiT
guarantees an SEM payment of €57/MWh indexed to any Consumer Price Index (CPI)
increases since 2005. The CPI for 2005 was 2.5%, hence the REFiT for 2006 was €58.43.
The CPI for 2006 was 4% which resulted in a 2007 REFiT of €60.67 (58.43 * 1.04). CPI for
2007 was 4.9% which put the 2008 REFiT at €63.74 . The 2008 CPI was 4.1% which
increased REFiT to €66.35. The 2009 CPI was negative so REFiT did not increase.
Inflation forecasts vary but 1-2% is the general expectation for the coming years. At 2%
out to 2020, the REFiT rate would jump to €79/MWh (66.35*1.01^9). At 1%, it would be
While originally REFiT was restricted to 1,450MW of installed wind and projects
operational by 2010 , the government has made it clear that it will consider
projects which are not yet operational . It is also seeking EU approval for an
expansion of the 1,450MW cap on installed wind . It is unclear to what extent REFiT
support will be extended or how much of the prospective 4.6GW of wind by 2020 will be
covered by REFiT.
The average SMP for 2008 was €80/MWh, while the average for 2009 was €43/MWh.
Although REFiT facilitates other renewables as well as wind, wind makes up by far the
biggest segment. Wind power generation in 2008 was approximately 2,410 GWh and
2,955 GWh in 2009 . In 2008, if wind received the average SMP, no REFiT charge
would occur. In 2009 if wind received the average SMP, a subsidy of €23/MWh would be
required. This would result in a charge of €68m (2955*1000*23). In 2020, the 40%
renewable target would result in approximately 12,000 GWh of wind if Eirgrid‘s
projection of 4.6GW is correct and a 30% capacity factor is assumed (4.6*0.3*24*365).
If the current quota of wind with REFiT is assumed then 1,450MW, generating 3,800
GWh (1.45*0.3*24*365), would be eligible for guaranteed generation payment in 2020.
The average revenue received per MWh by wind was computed from the CER data.
Figure 7.3 shows the results. The high wind Portfolio 5 under the low natural gas price
scenario receives an average €62/MWh. The medium wind Portfolio 2 receives €68/MWh
and Portfolio 1 wind receives €78/MWh. The increasing wind tends to cannibalise its
own revenue. The lower SMP offsets, to an extent, any increase wind may impose on
system costs. The low demand, central scenario, and high fuel scenario would not
require REFiT as their gas price assumption of €47/MWh (low demand and central) and
€70/MWh (high) generate high enough electricity prices to clear the REFiT hurdle.
Assuming REFiT stays capped at 1,450MW, and an average inflation of 2%, wind in
Portfolio 5 would receive €17/MWh from REFiT (79-62). With 3,800 GWh generated per
year that would amount to compensation of €65m/year (3800*1000*17). If REFiT was
expanded to cover all wind generators on the system in 2020 under the same terms as
current wind generators, payments would increase to €204m (12000*1000*17).
Based on a PSO levy of €157m this year, the charge for industrial customers came to
€14/kVA (although domestic customers may cover industrial customers‘ share this year)
. A €204m charge for wind may thus result in an €18/kVA charge in 2020
(14*(204/157)). This site has a 3,200kVA MIC, so this would result in an annual charge of
€57,600 to subsidise wind. This will not directly affect CHP economics, as the MIC is the
same with or without a CHP.
Figure 7.3: Average revenue received by wind generators in 2020
The decision to index REFiT to CPI is questionable. Currently, the capital cost of Irish
onshore wind is estimated at €1.05m/MW by Poyry . In 2007, Eirgrid estimated the
levelised cost of wind electricity (LCOE) with €1m/MW capital costs and a 34.5%
capacity factor at €42.1/MWh .
The CER assumed capital costs for wind of €183,000 per MW per year with O&M of
€61,000. This resulted in additional Portfolio 5 new wind generation (approximately
5,000MW) losing €239m under the low fuel price scenario. This is similar to the €204m
calculated above; implying that the CER assumed the LCOE for wind was near €79/MWh.
The Eirgrid estimate with €1m/MW resulted in capital costs of €112,000 per year and
O&M of €17,000 was assumed. The cost of capital was slightly lower in the Eirgrid study
(7.32% vs. the CER‘s 8%) however the 15 year assumed lifetime was the same. Poyry‘s
assumptions of €1.05m/MW installed with 10% capital costs and a 20 year lifetime led to
an LCOE of approximately €72/MWh . The LCOE is very sensitive to interest rates,
which explains part of the difference between Eirgrid‘s and Poyry‘s numbers. Poyry also
assumed substantially higher O&M than Eirgrid, at approximately €20/MWh. The actual
cost range may thus lie between €42 and €72/MWh depending on capacity factor, cost
of capital and maintenance. The capital cost is subject to factors such as commodity
costs (e.g. steel) and international wind turbine demand.
Reduced costs may provide scope for lowering the REFiT price which would reduce
payouts should REFiT be extended. If REFiT is expanded its level should be based on
actual costs, with the link to CPI rescinded.
7.2. Effect of Wind on Startup Costs
High levels of wind will likely lead to increased cycling costs. The magnitude of these
cycling costs is unclear. The following quote illustrates the lack of knowledge :
―It is estimated that cycling costs can range from $300 to $500,000 per single on-off
cycle, depending on the type of unit.‖
The CER included a sensitivity allowing for increased cycling costs. They approximated
this by increasing the variable startup costs by 50% for all units. This manifests itself as
additional uplift, a part of the SMP. Their sensitivity showed a relatively small effect –
with the average SMP increasing by €1.50 for the high wind Portfolio 5 in the central
7.3. Effect of Wind on Capacity Payments
The CER study used the same capacity payment methodology as currently. As Figure 4.4
showed, the CER‘s projection of required capacity of 11,400MW in 2020 is far above
what is now likely. Using a cost of capital for a peaking plant of €81/kW – similar to
today‘s figure – a capacity payment pot of €918m, €919m and €923m was generated for
Portfolio 1, Portfolio 2 and Portfolio 5 respectively. Despite having a substantially
higher installed capacity in Portfolio 2 and 5 (due to wind), the capacity requirement
did not reduce. Hence wind generation is not a major factor in deciding the capacity
If the capacity requirement was scaled back to a more likely 8,000MW in 2020 (Figure
4.4), the pot would be reduced to €650m. This is fairly close to 2010‘s €550m (Figure
7.4. Financial Savings
The model was set up with all the various tariffs which were relevant as of 2009. These
included the large energy user (LEU) rebate. The LEU offers a reduction of €15/MWh
per unit of electricity consumed by large energy users (the modelled plant here
qualifies). The LEU penalises CHP because, by importing less electricity, a given site
gets a lower rebate.
The data from Figure 7.2 was inputted to the model along with the associated gas
prices and the results computed. This scenario, with 2009 tariffs remaining in place, is
shown in Figure 7.4. While there are many components to the annual energy costs, the
majority of the costs are made up of the SMP, supplier capacity charges and gas
commodity costs. The most significant costs are shown in the legend. For the low fuel
Portfolio 1 in 2020, the savings with CHP amount to €425,000 or an 18.6% reduction on
the no CHP case. The savings would be €509,000 but for the LEU (21% reduction).
Figure 7.4: Component energy cost comparison with and without CHP
The costs with and without CHP were computed for all the scenarios in the CER‘s study.
The savings and the percentage savings were recorded. It was decided to exclude the
LEU from the analysis as it is unclear whether this subsidy to large energy users will still
apply in 2020. The PSO rebate was additionally excluded. Figure 7.5 shows the results.
Some clear trends are visible. In all cases, increasing wind leads to decreased savings
for CHP. For all scenarios, there is clear differentiation between Portfolios 1, 2 and 5.
The output of wind as a share of total all-island generation was computed to be 10% for
P1, 20% for P2 and 30% for P5. For the low fuel scenario, the difference between 10%
wind penetration and 20% wind penetration resulted in a drop in savings of
€55,000/year. The increment from 20% wind to 30% wind meant further reduced savings
For the other three scenarios, the same trend is observed, with the drop between 10%
and 20% wind being almost equal to the further drop between 20% and 30%.
Additionally, the effect of surplus generation capacity is seen in the large drop in
savings, particularly between P1 central fuel and P1 low demand. The drop is
€156,000/year. The difference between P2 central fuel and P2 low demand is €116,000
while for respective P5 data points the difference is €82,000. The effect of wind is
lowest when there is surplus generation capacity. The difference between 10% and 30%
wind was €80,000 for the low demand scenario. For the high fuel scenario, the savings
are reduced by €100,000/year for each 10% increment in wind. The percentage savings
are lower due to the larger numbers involved.
Figure 7.5: Savings and percentage savings with and without CHP
A tornado plot was compiled to examine the sensitivity of the results to changes in each
cost element. Portfolio 5 in the low fuel scenario was selected because this is the most
likely scenario, although all the portfolios show similar sensitivities. Each element was
independently increased and decreased by 50% and the reduction or increase in savings
recorded. The origin is placed at the original savings point of €411,725/year. This is
shown in Figure 7.6.
The results are very sensitive to the cost of gas and daytime electricity. This is to be
expected as they are the largest cost elements. If the price of gas remained the same
but the cost of daytime electricity was reduced by 50% (because of a large generation
margin, for example), savings would decrease by 34% to €272,000 per year. If the price
of gas increased by 50%, and the price of electricity remained the same, savings would
be reduced by 38% to €263,000 per year. This is unlikely, however, because the increase
in gas prices would result in a concomitant increase in electricity prices.
The results are also sensitive to night time and weekend electricity prices. However
these prices would not be as depressed by a large generation margin as they are
principally determined by baseload generation costs.
The SEM supplier capacity charge is derived from the capacity payment pot. If there is
overcapacity on the system, the capacity payment pot should reduce, resulting in lower
charges. Based on the forecast demand in 2020, the capacity payment pot, and supplier
capacity charges, should remain close to what they are today. However if charges are
increased by 50%, savings would be increased by €67,000 per year.
The gas transmission capacity charge is a relatively small element.
The carbon tax is based on a €15/tCO2 base level, and the range shown is €7.50-
€22.50/tCO2. The carbon tax is determined by the Irish government and is outside the
EU-ETS, which is also currently at €15/tCO2 . The low fuel scenario also includes a
level of €15/tCO2. The EU-ETS carbon price in 2020 was expected to double, however
barring EU intervention, there may now be an excess of permits due to the recession. In
August 2010 the first 2020 carbon futures were traded at €22.72/tCO2  which
illustrates current sentiment. To the extent that the Irish government increases the
carbon tax above ETS levels, this will disadvantage CHP, although the sensitivity is
The results are not sensitive to the other elements of the electricity and natural gas
price. There is no change for the other elements of the electricity price based on the
site‘s MIC. This is because the MIC remains the same as it is required for the small
percentage of the year when the CHP is down for maintenance. A fairer way of levying
the capacity charge for CHP sites would be based on the volume of electricity they
import. If this was done for this site, based on the reduction of 45% of electricity
imports, the TUoS and DUoS capacity charges would be reduced to €50,000 from
€90,000 for the no CHP case. This would result in 10% greater savings for the scenario
shown in Figure 7.6.
Substantial investment is required to upgrade transmission lines, partly to
accommodate wind . If this led to a doubling of the TUoS capacity charge, savings
would remain the same, however the percentage savings would reduce slightly. If based
on electricity volume, however, a doubling of TUoS would result in savings of €16,000.
The same applies to the PSO levy. If a €204m levy was applied in 2020, this may result
in a charge of €57,600. If based on electricity volume, though, it would be
advantageous for CHP and would lead to augmented savings of €26,000.
Figure 7.6: Sensitivity analysis Portfolio 5 Low Fuel
The savings show a high correlation with the price of natural gas. Figure 7.7 plots the
natural gas price against the savings for the low, central and high fuel scenarios. For
each portfolio the savings increase with increasing gas price. The slope of Portfolio 1 is
smaller than Portfolio 2 or 5, and the savings difference gets larger the higher the gas
price. Figure 7.7 shows €/MWh on the left axis and £pence/therm on the right axis.
There are 34.12 therms in a MWh, and the August 2010 conversion factor of €1=£0.82
was used to calibrate the right axis. The equations relating gas price to savings are
shown and, by inputting a gas price in €/MWh, the savings can be found.
Figure 7.7 shows that the effect of wind on the economics of CHP depends on the
natural gas price. At the low fuel price level, increasing wind penetration from 10% to
30% reduces savings by 19%. For the central fuel price level, the corresponding
reduction is 25% and at the high fuel price the reduction is similarly 25%.
Using the equations in Figure 7.7, if the fuel prices were to stay at the current level
(€16/MWh) then the move from 10% wind to 20% wind would reduce savings by
€51,000/year or an 11% reduction. Savings would be curtailed a further €37,000/year
moving from 20% to 30% wind. Although the three scenarios had different carbon price
assumptions, the high correlation of savings with natural gas price indicates that carbon
prices are a secondary consideration to the natural gas price. The equations in Figure
7.7 assume the same generation margin at each natural gas price. They do not apply to
the low demand scenario.
Figure 7.7: Savings vs. Natural Gas Prices
Figure 7.8 shows the savings in terms of the percentage of wind penetration. Some
caveats are required when interpreting this, however. As shown in Figure 4.17, the wind
penetration in Portfolios 2 and 5 was essentially a scalar multiple of the wind in
Portfolio 1. Figure 4.18 showed that there can be substantial variation from year to
year in wind output, and this will affect the results shown. From the CER data, the wind
output in each half hour was added up for each portfolio. The wind output was:
Portfolio 1: 5,587GWh
Portfolio 2: 11,174GWh
Portfolio 5: 16,759GWh
Total generation for the year was approximately 57,000 GWh. Transmission line losses
of 9.3% reduced the final electricity consumption to 54,000TWh. Portfolio 1 represented
approximately 10% wind (5,587/57,000), with Portfolio 2 at 20% and Portfolio 5 at 30%.
With 6GW installed in Portfolio 5, the capacity factor for this year was 32%. A year with
a lower capacity factor may have a lower impact on the SMP, with increased savings for
The most interesting aspect of Figure 7.8 is the low demand line. Despite a natural gas
price that is double the low fuel price scenario, the savings are lower. A high generation
margin can be expected to shift the savings line to the left (reduce savings) and to
increase its slope (reduce the impact of wind on savings). This describes the movement
of the red central fuel line.
Figure 7.8: Correlation of savings with wind penetration
7.5. Payback Period
The capital cost of CHP is variable; however SEAI report costs of €1m/MWe for large
units . The modelled site used a 1MWe gas turbine. The IEA reports O&M costs of on
average €7/MWhe for CHP . Based on annual production of 2,606MWhe from the
site‘s CHP, this equates to an annual O&M estimate of €18,250. A 10% discount rate was
used to represent the opportunity cost of investment and a 15 year lifetime for the CHP
was assumed. The payback period was computed for the low fuel scenario. Under these
assumptions, Portfolio 1 had a discounted payback period of 5.0 years, Portfolio 2 took
5.6 years and Portfolio 5 had a payback period of 6.2 years. The deterioration in
payback period is a cause for concern, given that the recent SEAI report stated the
typical payback period as 5 years and that some corporations would only consider
investing in projects with a 3-4 year payback .
7.6. Carbon Emissions With and Without CHP
The carbon emissions of the site with and without the CHP plant were computed from
the Excel model. An emissions intensity of 0.55tCO2/MWh was used for grid electricity.
The emission intensity for 2009 all-island electricity was 0.504tCO2/MWh . Figure
7.9 shows the 2009 fuel mix. Wind accounted for 11%, which suggests the figure would
be approximately 10% higher if not for wind, or about 0.55tCO2/MWh. The savings from
1MWe of CHP with a heat to power ratio of 1.2 and a 92% capacity factor were
2,136tCO2/year. Wind at a 30% capacity factor will generate 2,606MWh/year. Based on
0.55tCO2/MWh, wind saves 1,433tCO2/MW. Hence the CHP for this site saves the
equivalent of 1.5MW of wind (2136/1433).
Figure 7.9: All-island 2009 fuel mix
The main effect high levels of wind will have on CHP is to decrease the savings
compared to a no CHP scenario. This will occur principally due to the merit order effect
reducing daytime electricity prices. The merit order effect of wind is strongly
dependent on the generation margin and on the natural gas price. If the generation
margin increases, wind will have less effect on CHP. As the natural gas price increases,
so does the impact of wind on CHP‘s savings.
Wind is unlikely to significantly affect the Irish natural gas market, although if any
problems arise, they can be addressed through operator vigilance. If the costs of
running the gas grid increase, the sensitivity of CHP‘s savings to such an increase is low.
Given the current natural gas price outlook, the low fuel scenario will offer the best
guidance for the likely effect of wind in 2020. However as a high generation margin is
forecast for the coming years by Eirgrid, the impact of wind on CHP may be reduced.
In most cases the references are a direct link to the relevant pdf file. All links were
tested and working as of 28th August 2010. Shortened hyperlinks were used for brevity.
 Sustainable Energy Authority of Ireland. (2001). An Examination of the Future Potential of
CHP in Ireland. Available: http://tinyurl.com/34pmr2y
 Sustainable Energy Authority of Ireland. (2004). Combined Heat and Power in Ireland
Trends and Issues 1991-2002. Available: http://tinyurl.com/2bb92mr
 Sustainable Energy Authority of Ireland. (2006). CHP in Ireland Options for a National
Policy to 2010. Available: http://tinyurl.com/349vhf5
 Sustainable Energy Authority of Ireland. (2006). Evaluation of Legislation and Regulation
Affecting New CHP Installations in Ireland. Available: http://tinyurl.com/336ma9v
 Sustainable Energy Authority of Ireland. (2006). Benchmarking Report: Status of CHP in
EU Member States. Available: http://tinyurl.com/36eyqp9
 Sustainable Energy Authority of Ireland. (2006). Irish Supply Chain Capability for CHP
Applications. Available: http://tinyurl.com/34zutyv
 Sustainable Energy Authority of Ireland. (2006). New Technologies for CHP Applications.
 Sustainable Energy Authority of Ireland. (2006). A Guide to CHP in Ireland. Available:
 Sustainable Energy Authority of Ireland. (2006). Micro CHP Field Trial - Consultation.
 Sustainable Energy Authority of Ireland. (2009). Combined Heat and Power (CHP)
Potential in Ireland. Available: http://tinyurl.com/32eg3ew
 Sustainable Energy Authority of Ireland. (2010). Combined Heat and Power in Ireland
2010 Update. Available: http://tinyurl.com/3765vt3
 Department of Communications Marine and Natural Resources. (2007). Delivering a
Sustainable Energy Future for Ireland. Available: http://tinyurl.com/2ahdhcd
 Commission For Energy Regulation. (2009). Treatment of Small Renewable and Low
Carbon Generators. Available: http://tinyurl.com/36896yj
 Poyry. (2007). Trading and Settlement Code - Helicopter Guide. Available:
 All Island Project. (2007). SMP Uplift Parameters Decision Paper. Available:
 All Island Project. (2006). Single Electricity Market Capacity Payment Factors Decision
Paper. Available: http://tinyurl.com/26gmvlx
 Commission For Energy Regulation. (2009). Fixed Cost of Best New Entrant 2010 Decision
Paper. Available: http://tinyurl.com/39jjhp6
 All Island Project. (2008). Annual Capacity Payment Sum, Best New Entrant Price, and
Capacity Requirement 2009. Available: http://tinyurl.com/28eelx7
 Commission For Energy Regulation. (2010). Decision Paper on Best New Entrant Peaker
for 2011. Available: http://tinyurl.com/3ajnuyx
 Economic Social Research Institute. (2009). Should Coal Replace Coal? Available:
 Single Electricity Market Operator. (2010). Available: http://tinyurl.com/2vvtvyn
 European Wind Energy Association. (2009). The Economics of Wind. Available:
 T. Jónsson, et al., "On the market impact of wind energy forecasts," Energy Economics,
vol. 32, pp. 313-320, 2010.
 Poyry. (2010). Wind Energy and Electricity Prices - Exploring the merit order effect.
 F. Sensfuß, et al., "The merit-order effect: A detailed analysis of the price effect of
renewable electricity generation on spot market prices in Germany," Energy Policy, vol.
36, pp. 3086-3094, 2008.
 H. Weigt, "Germany's wind energy: The potential for fossil capacity replacement and cost
saving," Applied Energy, vol. 86, pp. 1857-1863, 2009.
 J. A. a. H. Weigt, "Electricity Markets Working Paper WP-EM-38 Combininwwg Energy
 E. D. Delarue, et al., "The actual effect of wind power on overall electricity generation
costs and CO2 emissions," Energy Conversion and Management, vol. 50, pp. 1450-1456,
 J. Munksgaard and P. E. Morthorst, "Wind power in the Danish liberalised power market--
Policy measures, price impact and investor incentives," Energy Policy, vol. 36, pp. 3940-
 Commission For Energy Regulation. (2009). Impact of High Levels of Wind Penetration in
2020 on the Single Electricity Market (SEM). Available: http://tinyurl.com/25mppt8
 Department of Communications Marine and Natural Resources. (2008). All Island Grid
Study Overview. Available: http://tinyurl.com/33t3r7c
 Department of Communications Marine and Natural Resources. (2008). All Island Grid
Study Workstream 1 Renewable Energy Resource Assessment. Available:
 Department of Communications Marine and Natural Resources. (2008). All Island Gridy
Study Workstream 2A High Level Assessment of Suitable Generation Portfolios for the All-
Island System in 2020. Available: http://tinyurl.com/3yyy2qzc
 Risoe National Laboratory. (2007). Final Report for All Island Grid Study Work-stream 2(b):
Wind Variability Management Studies. Available: http://tinyurl.com/3xp26jt
 TNEI. (2007). Workstream 3 All-Island Grid Study Report. Available:
 Department of Communications Marine and Natural Resources. (2008). Available:
 Eirgrid. (2010). Available: http://tinyurl.com/2f52a99
 Eirgrid. (2010). Generation Adequacy Report 2010-2016. Available:
 P. Meibom, Barth, R., Hasche, B., Brand, H., Weber, C. and O´Malley, M.J., "Stochastic
optimization model to study the operational impacts of high wind penetrations in
Ireland," IEEE Transactions on Power Systems, In Press, 2010.
 Eirgrid. (2010). Wind Generation. Available: http://tinyurl.com/kslumk
 Eirgrid. (2007). Generation Adequacy Report 2008-2014. Available:
 E. Lannoye, "Electricity Research Centre, Personal Communication," ed, 2009.
 U.S. Department of Energy. (2009). Modern Shale Gas Development in the United States:
A Primer. Available: http://tinyurl.com/d2opdx
 U.S. Energy Information Administration. (2010). Shale Gas - A Game Changer for U.S. and
Global Gas Markets? Available: http://tinyurl.com/yhlhtqe
 Massachusetts Institute of Technology. (2010). The Future of Natural Gas. Available:
 J. Dizard, "Sleight of Hand Over Shale Gas Costs," in Financial Times, March 21 ed, 2010.
 D. Parkinson, "A contrarian makes another call - this time natural gas," in The Globe and
Mail, April 18 ed, 2010.