Assessing the value of increasing GB

interconnection
Waqquas Bukhsh, Callum MacIver, Keith Bell
Institute of Energy and Environment,
University of Strathclyde, Glasgow, United Kingdom.
{waqquas.bukhsh, callum.maciver, keith.bell}@strath.ac.uk

Abstract—The liberalisation of electricity markets and in- plans to increase current GB interconnection capacity by
creasing penetration of renewable generation are encouraging approximately 150% by the mid-2020’s with new intercon-
trading opportunities to be identified and developed. In Europe, nection projects under development with France, Ireland,
the importance of interconnection is well recognised. Intercon-
nections are particularly important for islanded nations, like Norway and Denmark [4]. Given such a large potential
Great Britain, which provides access to generation and demand increase in the GB interconnection capacity, the following
across the national boundaries via sub-sea HVDC cables while key questions arise:
providing means to share reserves enhancing the security of
supply and reducing overall costs. There are plans in place • What impact will future GB interconnection have on
to significantly increase the current 5 GW of interconnection GB electricity prices?
capacity between GB and neighbouring European countries. The • What impact will new GB interconnection have on
paper investigates the expected impacts that a large increase in the nature of the utilisation of existing and future GB
interconnection capacity could have on key electricity market
interconnection projects?
parameters such as marginal prices, carbon emissions and the
nature of utilisation of existing and future GB interconnections. • What role will new interconnection investments play in
Several scenarios have been considered to model future uncer- the decarbonisation of the GB and European electricity
tainty for the years 2020 and 2025. system?
Index Terms—HVDC interconnection ; linear programming; This paper aims to address the above questions whilst
European electricity markets; security of electricity supply; de- considering credible GB interconnection scenarios. To in-
carbonisation of electricity system; uncertainty quantification.
vestigate the impact of increasing GB interconnection on
electricity prices and carbon emissions, a modelling frame-
work is required that can adequately model the behaviours
of Europe wide electricity transmission system. It is also
I. I NTRODUCTION
important to model the spatial and temporal variation of
The interconnection of electricity markets is considered renewable sources and the transmission constraints that
a key component to unlock the potential of renewable gen- exists between various electricity markets in the European
eration in Europe. Several European scale case studies have electricity system.
demonstrated that increasing interconnection capacity can The academic literature focusing on the impact of future
enhance security of supply, increase overall social welfare GB interconnection is sparse. In [5], a regression model is
and help achieve decarbonisation targets through optimal used to study the impact of adding extra interconnection
utilisation of installed renewable generation [1], [2]. For the capacity between France and Great Britain. This approach
aforementioned reasons, the European Commission (EC) relies on historic data of flows between France and GB
has been promoting the development of interconnections and does not consider other GB interconnection or power
in parallel to market liberalisation and has set an ambitious exchanges between other European countries. An approx-
target of 15% import capacity as compared to the installed imate load flow model of the European interconnected
generation capacity for all member countries by 2030 [3]. system is presented in [6], which make use of publicly
For island nations like Great Britain (GB), meeting these available data of trades between countries. The methods
interconnection targets entails the use of long-distance based on historic data provide good approximations on
HVDC subsea transmission technology to enable cross flows and trades for a given topology but does not provide
border power flows. Currently, the GB electricity system has accurate results on the impact of new investments. Impact
an interconnection capacity of 5 GW (4 GW to mainland of selected new interconnection investments is presented
Europe and 1 GW to the island of Ireland). There are in [7]. The results demonstrate that new interconnection
investments increase overall welfare and reduce carbon
emissions.
The research leading to these results has received funding from the
Engineering and Physical Sciences Research Council under grant number In this paper, a European scale transmission system
EP/R021333/1. model is proposed that models the behaviour of a coupled

21st Power Systems Computation Conference Porto, Portugal — June 29 – July 3, 2020

PSCC 2020
European electricity market. Each European country is
represented by a single node and constraints are imposed
on the maximum net transfer capacity (NTC) of electricity
trades that can take place between connected countries.
The proposed model is tested on a range of GB inter-
connection scenarios and for the generation background
scenarios for the years 2020 and 2025. The generation
and demand scenarios are obtained from the ENTSO-e
Ten Year Network Development Plan [1]. The benefits of
each scenario are quantified in terms of its impact on the
marginal price, reduction in overall carbon emissions and
facilitation of the renewable generation resources.
Furthermore, the paper highlights the importance of
modelling interconnection losses and proposes a way to
capture it using equivalent hurdle costs. It is demonstrated
that without using hurdle costs or appropriate interconnec-
tion losses, the interconnection utilisation is overestimated
and may lead to erroneous conclusions.
The main contributions of this paper are as follows:
• use of a European transmission system model which
is spatially diverse (30 European countries with NTCs)
and temporally detailed (1-year of operation with
hourly resolution) to quantify the impact of new GB
interconnection;
• assessment of the impact of new GB interconnection
considering a range of future GB interconnection ca- Fig. 1. A representative European electricity transmission network. Each
pacity scenarios country is represented by a single node (except for the UK and Denmark,
which are represented by two nodes, respectively) and the Net Transfer
• highlighting the importance of modelling interconnec- Capacities (NTCs) are used to model the interconnection limits between
tion losses for assessing the value of future intercon- the countries. Data for the European network is obtained from the Ten
nections. Year Network Development Plan (TYNP) of ENTSO-e.

The rest of the paper is organised as follows. Section II

provides a brief description of the modelling framework. respectively). The to and from net transfer capacities (NTCs)
Section III provides a discussion on scenarios for the year between the countries are obtained from the 2018 Ten Year
2020 and 2025. The results for all the scenarios are pre- Network Development Plan (TYNDP) of ENTSO-e [1].
sented and discussed in Section IV and the paper concludes
The European dispatch model is simulated for 1-year
with conclusions and discussion in Section V.
with a time resolution of 1-hour. This time resolution
II. M ODELLING FRAMEWORK requires hourly time-series of input parameters like demand
and available renewable generation as well as information
A European scale electricity dispatch model is used to
on the availability of hydro resources and the planned and
determine hourly expected imports/exports to the GB sys-
unplanned outage rates of the fossil fuel generation. This
tem. The electricity dispatch model is built using a platform
is used alongside assumptions on the market bid price of
provided by the French transmission system operator (RTE)
different generation types to determine the lowest overall
called ANTARES [8]. A particular feature of this platform is
cost hourly dispatch of generation.
its modelling of the hydro resources: reservoir, run of the
river and pumped storage. The penetration of renewable B. Generation and demand data
generation is modelled using historic data. The overall
mathematical formulation takes the form of a unit com- The generation capacities and demand data are obtained
mitment problem with weekly blocks that are coupled by from the TYNDP, which provides best estimates for the
the constraints on reservoir capacities [9]. Some important generation capacities in each country for 2020 and 2025.
features of the European dispatch model are discussed in The generation types represented in the European dispatch
the following subsections. model are presented in Table I alongside the assumed plant
efficiency and CO2 contribution by generation type.
A. The spatial and temporal scale of the model The data available through TYNDP of ENTSO-e report
Figure 1 shows the spatial scale of the model: each a single number for gas generation capacity and does not
country is represented by a single node (except for Den- differentiate between different levels of efficiency that the
mark and United Kingdom which are split into two nodes, gas generation units may have in each European country.

21st Power Systems Computation Conference Porto, Portugal — June 29 – July 3, 2020

PSCC 2020
 TABLE I D. Modelling of generation planned and forced outages
G ENERATION TYPES REPRESENTED IN THE E UROPEAN DISPATCH MODEL .
Planned and forced outages of generating units are mod-
Type CO2 Emission Efficiency elled using the reliability statistics provided by ENTSO-
(kg/GJ) % e for each generation type. Planned outages in Europe
Biofuels 0 40 normally occur outwith the winter months to avoid plant
Gas - Low 57 44 shutdowns when the demand for electricity is high. To
Gas - Med 57 52 model this, forced outages are assigned randomly across the
Gas - High 57 58 year whereas planned outages are given a lower probability
Hard Coal 94 40 of occurring over winter months.
Lignite 101 40 E. Modelling of renewable generation
Nuclear 0 33
Wind and solar generation are modelled as input hourly
Oil 100 35
time series in ANTARES. The profiles used are based on
CHP 57 58
the historical weather year of 2007 as obtained from re-
Other RES 0 40
newables ninja [11], [12]. This aligns with the demand year
Other NonRes 100 35
from which ENTSO-E derives its future demand time-series
estimates. The profiles are scaled in line with the estimated
TABLE II capacity in each scenario year with the solar generation
P LANNED AND COMMISSIONED INTERCONNECTION TO GB. being modelled as a fixed generation infeed to the system,
Name Country Capacity Length Loss while wind generation can be curtailed at a price.
(GW) (km) (%) F. Modelling of hydro generation
IFA France 2.0 73 2.34 Hydro generation is split into three categories: run-of-
Existing

Moyle N. Ireland 0.5 64 2.36 river (ROR), pumped storage and reservoir storage. Historic
BritNed Netherlands 1.0 259 3.00 data on realised hydro generation in each European country
EWIC Ireland 0.5 262 4.68 is used alongside knowledge of the capacity of each type
NEMO Belgium 1.0 140 2.60 within each country to model the three hydro generation
Eleclink France 1.0 51 2.50 types. The ROR is modelled as a fixed generation for every
IFA2 France 1.0 240 3.03 hour in a month but output varies across each month
Planned

NSL Norway 1.4 730 4.92 to reflect the changing flow rate in the rivers over the
GreenLink Ireland 0.5 160 2.64 year. Pumped storage is controllable and the optimisation
FABLink France 1.4 220 2.88 decides on the amount that is pumped or discharged
VikingLink Denmark 1.4 760 5.04 from pumped storage facilities in each country. The hy-
dro generation from the reservoirs are scheduled weekly
using ANTARES’ in-built heuristic method, with a monthly
To model this, the gas generation capacity is split into three constraint on availability [8].
categories of high, medium and low efficiency, as shown in G. Assumptions regarding the fuel costs and UK Carbon
Table I. Price Support
The marginal cost of a generator depends on a number of
C. Modelling of HVDC and HVAC losses
things: fuel cost, plant efficiency, start-up cost, shut-down
Losses are an inevitable consequence of transporting cost and cost of CO2 emissions per MWh. The data for these
power from generation to demand. In this work, losses parameters are taken from the TYNDP report [1]. In reality,
are split into two categories: transmission losses within the the price offered by the fossil fuel generators varies based
HVAC system and interconnection losses on HVDC links. on the location and over time due to changes in the global
In Europe, the transmission losses in the HVAC system are fuel prices. To approximate this, a stochastic daily price
typically in the order of 2% [10]. In this work, the demand modulation of ±2% is applied to all thermal generation
in each country is increased by 2% to represent the losses to reflect locational price variations while a ±5% variation
within the system. on the central marginal cost assumption is applied linearly
HVDC losses are modelled for the GB links using a across the year to reflect higher winter and lower summer
loss factor. The loss factors applied to GB interconnection fuel prices.
projects are presented in Table II using either publicly The fossil fuel electricity producers in the UK pay an
stated or estimated percentage loss figures. The estimated additional tax which is called UK Carbon Price Support
losses on an HVDC link depend on various factors including (CPS) [13]. Currently, the CPS is £18/ton of carbon emis-
the length of the link, the type of HVDC technology used sions. The CPS is imposed in the model for the 2020
(LCC or VSC), and the system voltage. scenario, which makes UK fossil fuel generation more

21st Power Systems Computation Conference Porto, Portugal — June 29 – July 3, 2020

PSCC 2020
expensive than the equivalent mainland European gener- S 32020

ation which is subject only to a carbon price set by the +IFA2

S 12020 S 22020 7.0 GW
EU Emmissions Trading Scheme (EU-ETS). The baseline
Existing +Eleclink
assumption is that the CPS is likely to be removed and
5.0 GW 6.0 GW S 42020 S 52020
carbon prices across Europe levelised for 2025 scenarios, +NSL +IFA2
however, in this paper, the results are provided with and 7.4 GW 8.4 GW
without the UK Carbon Price Support for the year 2025. (a) Five interconnection scenarios for the year 2020. The scenario tree
starts with the existing GB interconnection capacity of 5.0 GW.
III. S CENARIOS FOR 2020 AND 2025
S 22025 S 52025 S 72025 S 82025
A set of scenarios is developed to model the uncertainty
+FABLink +Viking +Green +Grid
in the future generation mix and GB interconnection ca- 9.8 GW 11.7 GW 13.1 GW
11.2 GW
pacity. The scenarios for the generation mix are taken from
the ENTSO-e TYNDP [1] for the year 2020 and 2025. The
S 12025
S 32025
interconnection scenarios are developed following consul- Best
tation with industry and experts about the plausible GB +Green
estimate
8.9 GW
interconnection that can be expected to come online in 8.4 GW
the coming years.
The results presented in this paper are for the scenarios S 42025 S 62025

developed for the years 2020 and 2025. The TYNDP of +Viking +Green
9.8 GW 10.3 GW
ENTSO-e provides a single scenario for generation back-
(b) Eight interconnection scenarios for the year 2025. The scenario tree
ground for the years 2020 and 2025. This data is used
starts with the best estimate for GB interconnection capacity of 8.4 GW.
along with a set of GB interconnection scenarios for the Fig. 2. Two set of GB interconnection scenarios for the year 2020 and
two years. For the year 2025, the results are presented for 2025.
two different carbon price scenarios of Coal before Gas
(CBG) and Gas before Coal (GBC). The merit-order switch of
coal and gas is modelled using an assumed EU-ETS carbon which means that increasing GB interconnection capacity
price of 30 e/tonCo2e for CBG and 60 e/tonCo2e for GBC, by 1 GW decreases the average marginal price in GB by
respectively. 0.5 e/MWh.
Table II presents the existing and planned GB inter- Figure 3(b) presents the incremental impact of increasing
connection projects considered in this paper. The sub- GB interconnection on GB and European carbon emissions.
sea HVDC interconnection projects have long time-scales, A negative value in the graph means that the carbon emis-
are very expensive and are subject to a range of un- sions have decreased as compared to the previous scenario.
certainties that could delay or in some cases cancel the It can be seen that the GB carbon emissions improve
project altogether. In this context, it is important to capture significantly with increasing GB interconnection, signifying
the uncertainty in the realisation of such interconnection a large reduction in the use of fossil fuel plant in GB. How-
projects. Figures 2 presents GB interconnection scenarios ever, the results show that this is not reflected through to
for the years 2020 and 2025. Five interconnection scenarios overall European carbon emissions which remain relatively
are considered for the year 2020 that range from 5 GW to unchanged by new GB interconnection, with only small net
8.4 GW of interconnection capacity. Eight interconnection carbon reductions observed for additional interconnections
scenarios are constructed for the year 2025, with a base case to France. Indeed, in moving from scenario 2 to scenario 4,
capacity estimate of 8.4 GW and a maximum of 13.1 GW. the total emissions actually rise with the introduction of
a link to Norway. This can be explained by the fact that
IV. R ESULTS the reduction in GB fossil fuel is made possible by greater
The proposed framework of quantifying the impact of imports from neighbouring countries. While the displaced
increased interconnection is demonstrated for the years generation in GB is dominated by gas, the additional
2020 and 2025. The results for the two years are presented imports from other European countries are facilitated by
in the following subsections. increases in for example hydro and nuclear power in di-
rectly connected countries but also by a partial increase in
A. Results for the year 2020 the use of fossil fuel generation across several countries.
Figure 3 presents the impact of increasing GB intercon- Some of this is contributed by heavily polluting lignite and
nection on average GB marginal price and carbon emis- coal plants, particularly in countries like Germany and the
sions. It can be noted that the average GB marginal price Czech Republic with the net impact being that emissions
reduces with increasing GB interconnection capacity. This is reductions in GB are largely counterbalanced and in some
due to an increase in GB imports which displace expensive instances even surpassed by emissions increases elsewhere.
thermal generation within GB. The best-fit regression line Figure 4 presents the impact of increasing GB intercon-
presented in Figure 3(a) has a negative gradient of 0.5, nections on the utilisation of the existing links and the

21st Power Systems Computation Conference Porto, Portugal — June 29 – July 3, 2020

PSCC 2020
 60 average price difference between Norway and GB decreases
Expectd marginal price (e/MWh) by 1 e/MWh between scenario 1 and scenario 3. This is be-
cause of the convergence of marginal prices across Europe
S 12020
S 42020 due to increasing GB interconnection. In scenario 4, with
58 1.4 GW interconnection capacity added between Norway
S 52020
and GB, the average marginal price difference is 9.3 e/MWh
S 22020 S 32020
with 84% utilisation representing a 1.4 e/MWh convergence
in average marginal price. Scenario 5 models 1 GW link
to France coming online after NSL and it reduces NSL
56
Data utilisation by 1%.
Linear regression: -0.5190*X + 61.28
New interconnection investments impact the wider sys-
5.0 6.0 7.0 7.4 8.4 tem prices as well. Overall, in all scenarios presented in
GB interconnection capacity (GW) Figure 4, it can be noted that average price differences were
(a) Average GB marginal price for the 5 interconnection scenar- reasonably high and additional GB interconnection capacity
ios. Red line is the best fit line that shows the decreasing trend did not significantly impact utilisation of the existing links.
of the GB marginal price.

2 B. Results for the year 2025

GB only
Change in emissions (MTCO2e)

European
At the time of writing, electricity producers in the UK pay
an additional tax called carbon price support (CPS) which
0
is £18/ton1 of carbon emission [13]. For the year 2025, the
following four scenarios were considered:
• Gas before coal (GBC) without UK CPS
• Coal before gas (CBG) without UK CPS
−2
• Gas before coal (GBC) with UK CPS
• Coal before gas (CBG) with UK CPS
Figure 5 presents the impact of increasing GB intercon-
nection for the four chosen scenarios. In all four scenar-
0 0 0 0
202 202 202 202 ios, the total change in average GB marginal price be-
0 S2 0 S3 0 S4 0 S5
20 2 → 202 → 20 2 → 2 02 →
S1 S2 S2 S4 tween all interconnection scenarios is within approximately
(b) Impact of GB interconnection on carbon emissions. The impact 1 e/MWh. This is a lower impact than seen in the 2020
is calculated as an incremental change of adding an extra GB in- results, suggesting a diminishing incremental influence of
terconnector. Negative value means decrease in carbon emissions. new interconnection capacity as total capacity rises. Fig-
Fig. 3. Impact of increasing GB interconnection on GB marginal price and
ure 5(a) and Figure 5(b) presents average marginal prices
the carbon emissions.
for the two scenarios without UK Carbon Price Support. In
these two cases, the GB generation prices are assumed to
be equalised with European prices. This means that average
average absolute price difference between the neighbouring
GB prices are no longer significantly higher than most
electricity markets. It can be noted that the absolute price
other neighbouring countries, as in the 2020 case, which
difference and the utilisation of the links decrease as ca-
means that there is a greater balance between imports and
pacity increases. In Figure 4(a) the average price difference
exports. While in importing periods GB marginal price will
between GB and France is approximately 15 e/MWh and
reduce on average due to interconnection providing access
utilisation of the link is 96%. The next scenario adds
to cheaper generation in other markets, the opposite is
1 GW of additional capacity between GB and France,
true in exporting periods with GB marginal price being
which reduces the average price difference between the
increased by interconnection facilitating increased access
two countries to 12 e/MWh and also marginally decreases
to market for more expensive local generation. Figure 5a
the utilisation to 95%. The third scenario models additional
shows that average prices remain relatively similar for all
1 GW capacity between GB and France and with this link
interconnector scenarios showing that the impact on prices
the average marginal price decreases to 11 e/MWh and
of importing and exporting cases largely balance out in this
the utilisation of the link decreases to 94%. The price
scenario. Figure 5b shows an increase in average marginal
difference of 11 e/MWh is still high and it is higher than
prices in the Gas before Coal scenario as new intercon-
all other price differences shown in Figure 4(c). Due to this
nection is added which suggests that exports increasingly
high price difference between GB and France, additional
dominate with the change in merit order making GB a
interconnection capacity does not dramatically decrease the
utilisation of existing GB-France links. 1 UK Carbon Price Support of £18/ton is approximated to e21/ton for
The link to Norway NSL comes online in scenario 4. The the analysis provided in this paper

21st Power Systems Computation Conference Porto, Portugal — June 29 – July 3, 2020

PSCC 2020
 Annual average utilisation (%)

Annual average utilisation (%)

Annual average utilisation (%)

100 100 100
GB-BE GB-BE GB-BE
80 80 80
GB-FR GB-FR GB-FR
GB-NL
GB-NL
60 60 60 GB-NL

40 40 40
GB-IE GB-IE GB-IE
20 GB-NO 20 GB-NO 20 GB-NO

0 0 0
5 10 15 5 10 15 5 10 15
Average absolute price difference Average absolute price difference Average absolute price difference
(a) S 12020 - 5.0 GW GB interconnection capacity (b) S 22020 - 6.0 GW GB interconnection capacity (c) S 32020 - 7.0 GW GB interconnection capacity
Annual average utilisation (%)

Annual average utilisation (%)

100 100
GB-BE GB-BE
80 80
GB-FR GB-FR

60 GB-NL GB-NO 60 GB-NO

GB-NL

40 40
GB-IE GB-IE
20 20

0 0
5 10 15 5 10 15
Average absolute price difference Average absolute price difference
(d) S 42020 - 7.4 GW GB interconnection capacity (e) S 52020 - 8.4 GW GB interconnection capacity
Fig. 4. Results showing the relationship between annual average absolute price difference and utilisation of selected GB interconnections for the five
scenarios for the year 2020.

relatively lower-priced market on average compared to its C. Impact of modelling HVDC losses
neighbours in this scenario.
Modelling of losses over HVDC interconnection intro-
Figure 5(c) and Figure 5(d) presents results for the case duces a hurdle cost between connecting markets. it is
with UK Carbon Price Support still in place in the year important to capture this hurdle cost in the modelling to
2025. In the Coal before Gas case (Figure 5(c)), the average correctly quantify the utilisation and impact of an inter-
GB marginal prices are higher than in the case without connector project. Some case studies in the literature have
CPS included but the addition of interconnection again indeed used an assumption of lossless HVDC link e.g. in
decreases prices from this higher base level. This shows that [15].
the reinstatement of the CPS changes the relative price of In this section, a case study is presented that demon-
the GB market compared to its neighbours which return it strates the importance of modelling HVDC losses. Sce-
to the status of being a predominantly importing market nario 5 of the year 2020 with 8.4 GW of GB interconnection
again, as in the 2020 case. In the Gas before Coal scenario capacity is considered. In this scenario, there is 1.4 GW
including the CPS, (Figure 5(d)), it can be seen that the of interconnection capacity between Norway and GB. The
impact on GB average price varies depending on the type following four cases were simulated for the chosen scenario:
of interconnection investment. For example, increasing the (i) with 4.92% loss factor on the GB-Norway link, (ii) without
interconnection to Ireland increases GB marginal prices any loss on the link, (iii) hurdle cost of 4.0 e/MWh on
because of increased exports to Ireland, whereas increasing imports and exports, and (iv) hurdle cost of 2.5 e/MWh on
interconnection to France decreases the GB marginal prices. flows from GB to Norway and 2.9 e/MWh between Norway
Overall, in this scenario, the impact of various intercon- and GB, respectively. The choice of 4.0 e/MWh in case iii)
nection investments cancel out each others impact and the is taken from [16] where the authors have used this hurdle
linear regression curve is close to a constant. cost for all interconnectors in their model. The asymmetric
Due to the limited space in paper, the graphs of carbon hurdle costs used in case iv) are derived by considering the
emissions and impact on utilisation of interconnection for loss factor and average price in each market with detailed
the year 2025 are not discussed in the paper. These are explanation provided in Appendix A.
made available as a supplementary material here [14]. Figure 6 presents hourly flows between GB and Norway

21st Power Systems Computation Conference Porto, Portugal — June 29 – July 3, 2020

PSCC 2020
 61 73

Expectd marginal price (e/MWh)

Expectd marginal price (e/MWh)

S 42025 S 72025
2025 S 42025 S 72025 S 62025
60 S 3 72
S 62025 S 32025

S 22025 S 52025 S 82025

S 52025
S 82025 S 22025
59 S 12025 71
S 12025
Data Data
Linear regression: 0.035*X + 59.27 Linear regression: 0.1624*X + 69.98
58 70
8 9 10 11 12 13 8 9 10 11 12 13
GB interconnection capacity (GW) GB interconnection capacity (GW)
(a) 2025 Coal before Gas without UK Carbon Price Support (b) 2025 Gas before Coal without UK Carbon Price Support

77
Expectd marginal price (e/MWh)

Expectd marginal price (e/MWh)

64 S 2025
3 S 42025
S 42025 S 62025
76 S 32025 S 72025
S 62025
S 72025
63
S 12025 S 82025
S 22025 S 22025 S 52025
75 S 12025 S 82025
S 52025

62 Data Data
Linear regression: -0.1653*X + 64.76 Linear regression: -0.0030*X + 75.63
74
8 9 10 11 12 13 8 9 10 11 12 13
GB interconnection capacity (GW) GB interconnection capacity (GW)
(c) 2025 Coal before Gas with UK Carbon Price Support (d) 2025 Gas before Coal with UK Carbon Price Support
Fig. 5. Impact of increasing GB interconnection on GB marginal price. Four cases were considered – changing the merit order of coal and gas, and
with and without consideration of UK Carbon Price Support.

for the four cases while Table III reports the average case by 4.3%. With the asymmetric hurdle cost based on
marginal prices in GB and Norway, GB imports, GB exports the average prices of GB and NO, the model underestimates
and utilisation of the link. In Figure 6 the x-axis represents the utilisation of the link by just 0.8%. This suggests the
the price difference between GB and Norway and the y- asymmetric hurdle costs of 2.5 and 2.9 e/MWh are a good
axis is the interconnection flows. The Figure shows that approximation of modelling the losses explicitly. However,
modelling of losses using a loss factor of 4.92% introduces as Figure 6 highlights there is a significant difference be-
a non-uniform hurdle cost between GB and Norway that tween the calculated hurdle costs and those implied from
is asymmetrical, approximately 3.0 e/MWh for GB imports the explicit modelling of losses with apparently a slight
and 1.0 e/MWh for GB exports. In this case there was no underestimation of the GB import hurdle cost and an
interconnection flow between Norway and GB for 8.2 % of overestimation of the GB export hurdle costs. This means
the time (718 hours in a year) because the marginal price that flows in many hours are different under the two
difference between the two countries in these hours is low methods and suggests for greatest accuracy losses should
enough that making a trade is not worth the losses en- be modelled explicitly.
countered. Total utilisation in this case is 87.17%, calculated
as percentage of total annual absolute flows vs theoretical V. C ONCLUSIONS AND D ISCUSSION
maximum absolute flows. In this paper, the application of a framework to assess the
To show the value of modelling losses, case ii) shows value of increasing interconnection capacity is presented.
that when you ignore the loss on the link, the modelled The proposed framework is demonstrated on increased
utilisation increases from 87.17% to 95.04% (Table III). This GB interconnection capacity for the years 2020 and 2025
approach is likely to be least accurate with nonzero flow in with a range of different interconnection and generation
all hours and the potential for significant overestimation of background scenarios. It is shown that increasing inter-
GB imports and exports through the link. With the hurdle connection capacity has a positive impact on GB marginal
cost of 4.0 e, the model significantly underestimates the prices for the year 2020 as it facilitates imports. Whereas in
utilisation compared with the explicitly modelled losses the year 2025, the impact is lower and largely dependent on

21st Power Systems Computation Conference Porto, Portugal — June 29 – July 3, 2020

PSCC 2020
 1500
assumes perfect foresight of electricity demand and gener-
ation availability. In reality, demand, generation availabil-
ity and generation from renewable resources are subject
1000
to uncertainties. Furthermore, the interconnection trading
takes place in stages involving forward contracts, bilateral
500
trading, day-ahead market clearing and balancing. The
NO to GB flow (MW)

decisions made in these stages influence the eventual flow

0 of power on an interconnector. The different electricity
market structures across interconnection may result in less
−500 than optimal utilisation of the links, as demonstrated in [17]
for the interconnection between Ireland and Great Britain.
−1000
with loss However, further research is required to adequately capture
without loss
GB2NO = 4.0 €, NO2GB = 4.0 €
the suboptimal behaviour of the trades that take place at
GB2NO = 2.5 €, NO2GB = 2.9 € different time periods between the European countries.
−1500
−20 −10 0 10 20 30
GB-NO (€/MWh) A PPENDIX A
H URDLE COSTS
Fig. 6. A graph showing price difference vs flows on an interconnector
between Great Britain and Norway. Four different cases of interconnction
loss are considered.
p tA p tB
TABLE III X MW
VALUE OF MODELLING HVDC LOSSES ON UTILISATION AND AVERAGE PRICES A B

Hurdle costs Fig. 7. An illustrative example of an HVDC link connection two electricity
w loss wo loss 4.0 {2.5, 2.9} markets A and B. X MW is flowing from A to B. The prices are denoted
by p tA and p tB during the time period t in market A and B, respectively.
GB avg price 58.83 57.73 57.86 57.83
NO avg price 49.96 50.00 49.94 49.95
Let A and B denote two electricity markets that are
GB imports (TWh) 9.64 10.33 9.25 9.54
connected by an interconnector. Let p tA and p tB denotes
GB exports (TWh) 1.02 1.29 0.89 1.01
the marginal price in market A and B at time period t ,
Utilisation (%) 87.17 95.04 82.9 86.33
respectively. If X MW is flowing from market A to market B,
then according to the economic principles of power flowing
from regions of the high price to low price the following
the choice of price background and generation merit order
inequality must hold:
which determines the extent to which GB changes from
a predominantly importing market to a more exporting λ
p tA X < p tB X (1 −
) (1)
market. Similarly, the impact of increasing interconnection 100
capacity on carbon emissions is largely dependent on where λ is the percentage loss on the interconnector.
the choice of generation background and associated price Rearranging the above equation we have the following:
assumptions. The results for 2020 show reduction of the GB
λp tB
carbon emissions in all the five scenarios, whereas in one of p tB − p tA ≥ (2)
the scenario’s the overall carbon emissions increased, which 100
means that the GB thermal generation is being replaced by The inequality in (2) mean that the price difference
more carbon-intensive but cheaper thermal generation in between the two markets should atleast cover the cost of
Europe. losses on the link. If the power is flowing from B to A, then
The paper also highlighted the impact of modelling we have the following:
losses. It is shown that modelling of losses is important to
accurately model the utilisation of the links with significant λp tA
p tA − p tB ≥ (3)
potential to overestimate utilisation if losses are ignored. An 100
important outcome is that the implied hurdle costs from The right hand sides of inequalities (2) and (3) are hurdle
explicitly modelling the losses are non uniform and can be costs of power flowing from A to B and from B to A,
asymmetrical depending on flow direction. respectively.
Increasing interconnection capacity increases competi- The region of zero flows shown in Figure 6(a) is given by
tion, reduces market reserve requirements and facilitates the following interval:
utilisation of renewable energy sources. However, these " #
benefits can only be achieved under efficient trading be- λp GB
t λp tNO
− ,
tween markets. This paper has considered a model which 100 100

21st Power Systems Computation Conference Porto, Portugal — June 29 – July 3, 2020

PSCC 2020
where p GB
t and p tNO are the marginal prices in GB and [15] F. Gaffney, J. Deane, S. Collins, and B. O. Gallachóir,
FR at time-perio t , respectively. The price in time period “Consumption-based approach to res-e quantification:
Insights from a pan-european case study,” Energy Policy,
t cannot be known apriori so in order to approximate the vol. 112, pp. 291 – 300, 2018. [Online]. Available:
hurdle cost to be used in section IVC, we have used average http://www.sciencedirect.com/science/article/pii/S0301421517306468
marginal price in GB and NO. [16] S. Collins, D. Saygin, J. Deane, A. Miketa, L. Gutierrez,
B. O. GallachÃşir, and D. Gielen, “Planning the european
power sector transformation: The remap modelling
framework and its insights,” Energy Strategy Reviews,
58.83 × 4.92
Hurdle cost for GB to NO = = 2.5 e/MWh vol. 22, pp. 147 – 165, 2018. [Online]. Available:
100 http://www.sciencedirect.com/science/article/pii/S2211467X18300798
49.96 × 4.92 [17] I. V. Lytvyn and N. Hewitt, “Barriers to increased electricity intercon-
Hurdle cost for NO to GB = = 2.9 e/MWh nection between neighboring markets,” in 2013 10th International
100 Conference on the European Energy Market (EEM), May 2013, pp. 1–
8.
R EFERENCES

[1] European Comission, “Scenario report, entso-e ten year network

development plan,” Tech. Rep., 2018. [Online]. Available:
https://tyndp.entsoe.eu/
[2] European Commission, “Study on the benefits of
additional electricity interconnections between iberian
peninsula and rest of europe,” 2016. [Online].
Available: https://publications.europa.eu/en/publication-detail/-
/publication/2ad84365-3451-11e6-969e-01aa75ed71a1/language-en
[3] European Commission, “Towards a sustainable and integrated
europe,” Report of the Commission Expert Group on Electricity
interconnection targets, Tech. Rep., 2017. [Online]. Available:
https://ec.europa.eu/energy/sites/ener/files/documents/report_of_the
_commission_expert_group_o n_electricity_interconnection_targets.pdf
[4] Office of Gas and E. M. (Ofgem), “Elec-
tricity interconnectors,” Tech. Rep., 2018. [On-
line]. Available: https://www.ofgem.gov.uk/electricity/transmission-
networks/electricity-interconnectors
[5] G. P. Swinand, A. Natraj, and A. O’Mahoney, “The european target
model and the impact of implicit auctions on interconnection ca-
pacity prices,” in 2016 13th International Conference on the European
Energy Market (EEM), June 2016, pp. 1–6.
[6] Qiong Zhou and J. W. Bialek, “Approximate model of european
interconnected system as a benchmark system to study effects of
cross-border trades,” IEEE Transactions on Power Systems, vol. 20,
no. 2, pp. 782–788, May 2005.
[7] V. Di Cosmo, V. Bertsch, and P. Deane, “Impact of new interconnec-
tion lines on the eu electricity market,” in 2016 13th International
Conference on the European Energy Market (EEM), June 2016, pp. 1–
5.
[8] M. Doquet, C. Fourment, and J. M. Roudergues, “Generation and
transmission adequacy of large interconnected power systems: A
contribution to the renewal of Monte-Carlo approaches,” in 2011 IEEE
Trondheim PowerTech, June 2011, pp. 1–6.
[9] K. Bruninx, Y. Dvorkin, E. Delarue, H. Pandzic, W. D’haeseleer, and
D. S. Kirschen, “Coupling pumped hydro energy storage with unit
commitment,” IEEE Transactions on Sustainable Energy, vol. 7, no. 2,
pp. 786–796, April 2016.
[10] Council of European Energy Regulators (CEER), “Ceer
report on power losses,” 2017. [Online]. Avail-
able: https://www.ceer.eu/documents/104400/-/-/09ecee88-e877-
3305-6767-e75404637087
[11] I. Staffell and S. Pfenninger, “Using bias-corrected reanalysis to
simulate current and future wind power output,” Energy,
vol. 114, pp. 1224 – 1239, 2016. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S0360544216311811
[12] S. Pfenninger and I. Staffell, “Long-term patterns of european pv
output using 30 years of validated hourly reanalysis and satellite
data,” Energy, vol. 114, pp. 1251 – 1265, 2016. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S0360544216311744
[13] The Future of Carbon Pricing, “Implementing an independent
carbon tax with dividends in the uk,” Policy Exchange, Tech.
Rep., 2018. [Online]. Available: https://policyexchange.org.uk/wp-
content/uploads/2018/07/The-Future-of-Carbon-Pricing.pdf
[14] W. Bukhsh and C. MacIver, “Supplementary data for the pscc 2020
paper: Assessing the value of increasing gb interconnection,” Sep.
2019. [Online]. Available: https://doi.org/10.5281/zenodo.3463674

21st Power Systems Computation Conference Porto, Portugal — June 29 – July 3, 2020

PSCC 2020