October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Design
of
a
harmonised
reactive balancing
market with cross
zonal optimisation
of
frequency
restoration
between LFC Blocks
Esther van Wanrooij
TenneT NL
Frank Nobel
TenneT NL
Bob Hebb
Elia
Jan Voet
Elia
1|Page
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Disclaimer
The preliminary proposals for exchanging balancing energy that are developed as part of the
present Pilot Project are based on working assumptions and subject to a favorable cost-benefit
analysis, regulatory approvals required etc.These are all prerequisite before any possible crossborder balancing cooperation analysed as part of the present Pilot Project can be put into effect.
2|Page
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Executive Summary
In February 2013 ENTSOe launched a call for balancing pilot projects in order to test the feasibility
of the balancing target model of the Framework Guidelines on Electricity Balancing (FGEB), to
evaluate the implementation impact as well as to gather and report on the experience gained. In
response, Elia and TenneT applied on 29 March 2013 for a nomination for a cross border
balancing pilot project called "Design and evaluation of a harmonised reactive balancing market
with XB optimisation of Frequency Restoration while keeping control areas, bidding zones, and
Regulatory oversight intact". The nomination was accepted by the ENTSO-E Market Committee on
25 June 2013.
The pilot project consists of 3 steps:

Step 1: Current processes and functionalities

Step 2: Market design scenarios

Step 3: Cost benefit analysis (CBA) and Go/No Go decision for implementation
The first step compares the present arrangements in Belgium and the Netherlands. The results of
this study, performed with the support of DNV KEMA, were presented in a public workshop on 1
February 2013. The full report can be found on the website of both TSOs.
Based upon these results, the exchange of balancing energy for contracted and non-contracted
automatic Frequency Restoration Reserves (FRR) bids and for non-contracted manual FRR bids
has been investigated in step two of the pilot project. This report represents the results of this
second step.
The report provides a detailed description of the underlying key features and philosophy of a reactive balancing market design. TSOs with a re-active market design only use the real-time
Frequency Restoration Control Error (FRCE) or Area Control Error (ACE) as an input for the
deployment of Frequency Restoration Reserves. BRPs are incentivized and enabled, at all times,
to balance their portfolio and/or to reduce the ACE Open Loop (system imbalance) of the Load
Frequency Control (LFC) Block via a single imbalance pricing mechanism. Countries with longer
Imbalance Settlement Periods (ISP) will typically have difficulty applying a reactive balancing
concept, and will typically require some proactive design elements in order to deal with the
discrepancy between the ISP and the time to restore frequency (15 minutes). The lessons drawn
from this pilot project can benefit all TSOs alike, as most TSOs are neither fully pro-active nor fully
re-active, but operate somewhere in between those extremes.
The report also investigates the impact of the exchange of balancing energy on the individual TSO
responsibilities as defined by NC LFCR. TSOs responsible for a LFC Block should comply with the
dimensioning rules for Frequency Restoration Reserves, and with the Frequency Restoration
Quality Target. This report reveals a close link between the required
FRR Capacity (reserve
dimensioning), the original system imbalance (“ACE Open Loop”) and the FRCE (ACE) regulation
quality of the LFC Block, since the process for exchanging FRR affects either the ACE Open Loop
3|Page
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
and/or the FRCE regulation quality of the connecting and requesting TSO. In addition there is an
indirect link between the automatic FRR ramp rates and the required volumes of balancing
capacity, as well as between the activation process (pro rata vs. merit order) and the required
volumes of balancing capacity. Balancing products harmonisation for the sake of the cross-border
exchange of balancing energy might therefore affect the need for –and the contracted amount of-,
and the local supply potential for aFRR. This might impact the level of the local access tariffs of
either the connecting and/or requesting TSO. All these aspects must be considered in the CostBenefit Analysis of step 3.
This report defines the main working assumptions for the upcoming cost-benefit analysis, i.e.
assumptions on the main design features for the exchange of FRR balancing energy via cross
zonal arrangements and the related harmonisation prerequisites. Such assumptions are in line
with the (draft) Network Codes NC EB and NC LFCR and take into account the above findings:

For Standard Products for Balancing Energy for automatic FRR and manual FRR – to be
offered by BSPs and used by both TSOs:

o
Harmonisation of the ramp rate of aFRR product to 7.5 minutes;
o
A scheduled energy product for mFRR.
Both TSOs transfer –in a firm way- the bids submitted by the BSPs for aFRR and noncontracted mFRR to a common merit order list;

TSOs request an activation of Balancing Energy to the activation optimisation function that
operates the common merit order list (CMOL). The requested volume by one TSO may
exceed the volume of bids individually made available on the CMOL as long as other

TSOs don’t need these reserves (in which case they would get prior access).
The exchange of FRR is organised via a virtual tie-line for automatic FRR (control request)
and manual FRR (power profile).

Reservation of cross zonal capacity for X-Border Balancing is initially not considered;

Allocation of remaining available between potential concurring processes is possible after
monitoring
Elia and TenneT also agreed on other working assumptions for the cost benefit analysis of the
cross-border exchange of balancing energy :

A first assumption is that manual FRR is only activated based on degree of saturation of
automatic FRR. In other words only technical and no economic consideration are involved
to trigger the activation of both products. This report demonstrates in the settlement
section that a harmonisation of the activation principle of the different products is
fundamental.

Different ways for exchanging automatic aFRR balancing energy were investigated:
o
The exchange of aFRR control target, i.e. exchanging the desired aFRR
activation without considering the finite ramp rates of the aFRR bids; hence the
exchanged energy cannot be physically delivered by BSPs in the LFC Block of
the Connecting TSO (such as implemented in the German GCC scheme); or
4|Page
October 17, 2014
o
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
The exchange of aFRR control request, i.e. exchanging the volume of aFRR that
can be physically activated by BSPs in the LFC Block of the Connecting TSO by
considering finite ramp rates of the aFRR bids.
An important conclusion is that this exchange of aFRR between LFC Blocks needs to be
organised via a control request. This report demonstrates that exchange of aFRR control
target between LFC Blocks conflicts with the individual responsibilities of the TSOs of the
LFC Blocks. Hence such a solution –as used in the GCC arrangements- can only be
implemented for the exchange between LFC Blocks in case a solution is found to resolve
this issue.

A harmonisation towards the number of regulation states in the imbalance pricing structure
is required. The reason for this is that the occurrence of double marginal pricing –currently
applied by TenneT in some cases of bi-directional regulation within an ISP- might increase
due to the implementation of the exchange of balancing energy. This would significantly
reduce the incentives given to BRPs to support the system balance. Both TSOs should
apply a single pricing mechanism with only 2 regulation states. Other main features of
imbalance settlement are already harmonised between both countries.
The settlement and pricing chapters show that different pricing mechanisms can be used for the
settlement of balancing energy. After analysis the local pricing mechanism is recommended as the
best solution. In this mechanism the settlement price in one control area might not be influenced
by the prices of bids activated for other LFC Blocks. Indeed Imbalance prices should always reflect
the local imbalance situation of an LFC Block in order to respect the control strategy of both TSOs.
TSOs are responsible to balance their LFC Blocks. BRPs are –in case of single marginal pricing
per LFC Block- incentivized to support this process. A uniform imbalance price over several LFC
BLocks jeopardises the important principle that BRPs must be balanced per LFC Block as in such
case they are then incentivized to be balanced over several LFC Blocks only. Furthermore this
would jeopardise the principle that BRPs must nominate the exchange of energy between 2
different bidding zones.
Implementing the market design changes currently retained as working assumptions would imply
significant changes in both markets. The implementation of a standard aFRR product with a
(stricter) ramp rate of 7.5 minute might affect the liquidity and aFRR capacity procurement costs in
the Netherlands. The feasibility and impact of a move from a pro-rata to merit order aFRR
activation scheme in Belgium must be investigated (impact on FRCE regulation quality, required
aFRR capacity volumes,...). The current Belgian implicit unit-based bidding system for mFRR
energy needs to be replaced by an explicit portfolio-based bidding system to enable the crossborder exchange of mFRR energy. The number of regulation states used for the determination of
the Dutch imbalance prices must be reduced to two, affecting tariffs.
Beyond the implementation costs and efforts, the working assumptions mentioned in this report
also have their own welfare implications which still to be fully understood. There are potential
implications for the access tariffs to be borne by the local stakeholders. Hence a detailed analysis
as part of a cost-benefit analysis is required to assess the benefits, costs and redistribution effects
of such a cross zonal balancing market, and of the market design changes required under the
working assumptions.
5|Page
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
The findings of step 2 of our pilot project were presented to the stakeholders of both countries
during a public workshop. After this workshop a public consultation was organised. The feedback
received from stakeholders will be used as an input for step 3 of our pilot project which is: a cost
benefit analysis.
The results of step 2 of our pilot project are a good starting point for discussions with other TSOs
and other pilot projects on the harmonisation requirements for balancing markets which are
needed to enable the exchange of Frequency Restoration Reserves. Present findings are also an
important contribution for the implementation of the Network Code on Electricity Balancing.
6|Page
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Table of Contents
Executive Summary ........................................................................................................................... 3
Table of Contents .............................................................................................................................. 7
Introduction and background.......................................................................................................... 11
CHAPTER 1: Frequency restoration in the European context ......................................................... 12
1.1
Introduction ..................................................................................................................... 12
1.2
Frequency restoration process ........................................................................................ 12
Control structure and TSO responsibilities.............................................................................. 12
Control processes and reserves .............................................................................................. 13
FRR dimensioning rules ........................................................................................................... 15
Frequency Restoration Process in Belgium and the Netherlands ........................................... 15
Important considerations for cross-border exchange of balancing energy ............................ 16
CHAPTER 2: Reactive balancing ....................................................................................................... 17
2.1
Introduction ..................................................................................................................... 17
2.2
Division of responsibilities ............................................................................................... 17
2.3
Imbalance settlement period .......................................................................................... 19
2.4
Incentives and pricing within a reactive balancing design .............................................. 20
2.5
Effectiveness and implementation of reactive balancing ............................................... 22
2.6
Influence on project scope .............................................................................................. 23
CHAPTER 3: IGCC ............................................................................................................................. 24
3.1
IGCC Concept ................................................................................................................... 24
3.2
IGCC Results..................................................................................................................... 25
CHAPTER 4: Exchange of aFRR balancing energy ............................................................................ 26
4.1
Evolutions in aFRR framework ........................................................................................ 27
4.2
Technical functioning of aFRR activation schemes ......................................................... 27
Main blocks of a secondary controller .................................................................................... 27
aFRR control target and aFRR control request ........................................................................ 27
Pro-rata and merit order activation ........................................................................................ 28
4.3
Implementation of aFRR schemes in Belgium and the Netherlands............................... 30
4.4
Pro-rata and merit order aFRR schemes ......................................................................... 31
Compatibility pro-rata and merit order aFRR scheme for aFRR balancing energy exchange . 32
7|Page
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Impact of and conditions for transition to merit order aFRR scheme in Belgium ................... 34
4.5
Working assumption for aFRR Full Activation Time ........................................................ 36
Link between aFRR Full activation time, FRR dimensioning and FRCE (ACE) regulation quality
................................................................................................................................................. 36
aFRR Full Activation Time at Elia and TenneT ......................................................................... 37
Working assumption for aFRR Full Activation Time ................................................................ 38
Survey on aFRR ramp rate per bid........................................................................................... 43
4.6
Exchange of aFRR balancing energy: process.................................................................. 43
Bidding process and constitution of Common Merit Order List ............................................. 43
Activation process ................................................................................................................... 45
Exchange process .................................................................................................................... 47
Settlement process.................................................................................................................. 53
CHAPTER 5: mFRR exchange ........................................................................................................... 54
5.1
Overview and scope ........................................................................................................ 54
5.2
Working assumption schedule activated mFRR product ................................................ 55
5.3
Activation philosophies ................................................................................................... 56
5.4
Introduction to exchange of mFRR.................................................................................. 57
5.5
Common merit order list product ................................................................................... 57
5.6
Bidding process ............................................................................................................... 58
5.7
Activation......................................................................................................................... 59
5.8
Cross-border exchange.................................................................................................... 60
5.9
Settlement ....................................................................................................................... 61
CHAPTER 6: Settlement and pricing – considerations regarding harmonization............................ 62
6.1
Requirements by NC EB ................................................................................................... 62
6.2
Harmonisation for market functioning ........................................................................... 62
6.3
Financial impact –BSP vs. BRP Settlement ...................................................................... 63
6.4
Financial Impact – Local vs. Cross Border Settlement ..................................................... 64
6.5
Local TSO responsibility ................................................................................................... 64
CHAPTER 7: Pricing and settlement - Imbalance pricing ................................................................. 68
7.1
Similarities between Netherlands and Belgium .............................................................. 68
7.2
Differences between Netherlands and Belgium: regulation state .................................. 69
7.3
Differences between Netherlands and Belgium: additional components ...................... 71
7.4
Differences between Netherlands and Belgium: iGCC .................................................... 72
8|Page
October 17, 2014
7.5
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Differences between Netherlands and Belgium: price determination .......................... 74
CHAPTER 8: Pricing and settlement - Settlement of balancing energy ........................................... 76
8.1
Description current pricing mechanism for settlement .................................................. 76
8.2
Options for the settlement of balancing energy ............................................................. 77
8.3
Local pricing principle ...................................................................................................... 78
8.4
Considerations on cross product & cross border pricing ................................................ 79
Firmness of Common Merit Order lists ................................................................................... 79
Different control strategies ..................................................................................................... 80
Wrong local incentives to BRPs ............................................................................................... 81
Extendibility of the pilot project.............................................................................................. 82
CHAPTER 9: Use of Cross Zonal Capacity......................................................................................... 84
9.1
Availability of Cross Zonal Capacity ................................................................................. 84
9.2
What is congestion of the border? .................................................................................. 85
9.3
Capacity needs of different processes ............................................................................ 85
9.4
Allocating capacity to different processes ...................................................................... 89
9.5
Cross zonal capacity reservations.................................................................................... 90
9.6
Cross-border balancing processes in practice ................................................................. 91
9.7
Pricing and Imbalance Settlement in situations with congestion ................................... 91
Conclusions...................................................................................................................................... 93
References ....................................................................................................................................... 94
List of abbreviations ........................................................................................................................ 95
Annexes ........................................................................................................................................... 96
Annex 1: FRR dimensioning rules ................................................................................................ 96
Annex 2: IGCC .............................................................................................................................. 97
IGCC settlement....................................................................................................................... 97
Improved ACE quality .............................................................................................................. 97
Annex 3.1: Changes to Table 3 of Balancing Comparison ........................................................... 98
Annex 3.2 Functioning of a secondary controller ....................................................................... 99
Annex 3.3: Examples pro-rata and merit order activation ........................................................ 102
Annex 4.1: aFRR Full Activation Time, aFRR dimensioning, aFRR activation scheme and FRCE
(ACE) regulation quality............................................................................................................. 103
Annex 4.2: exchange of aFRR control target ............................................................................. 106
Annex 4.3: Exchange of aFRR control request .......................................................................... 108
9|Page
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 5.1: Counteractivation during 2013-H2.......................................................................... 109
Annex 5.2: Product definition schedule activated mFRR product ............................................ 110
10 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Introduction and background
In December 2012, ACER published its FGEB with the aim to define the principles of the future NC
EB. Within this context Elia and TenneT began to analyse the potential for cross-border
collaboration between their balancing markets. In June 2012 the two TSOs launched a study
together with DNV Kema with the objective to compare the present balancing arrangements in
both countries. On 1 February 2013, the results of this study (“Step 1 of the pilot project”) were
presented to stakeholders during a workshop.
The comparison study showed a large degree of similarity of high-level market design principles in
the procurement of operating reserves and real-time balancing in both countries. A closer analysis,
however, reveals some important differences in the specification, remuneration and use of different
products. The report also showed that further synergies than those already achieved by Elia and
TenneT (i.e. imbalance netting and pooling of tertiary reserves) will require some substantial
changes to products, processes, and to the regulatory framework.
In February 2013 ENTSO-E launched a call for pilot projects on balancing in order to test the
feasibility of the balancing target model as explained in the FGEB to evaluate the implementation
impact and to gather and report on the experience gained.
In response to this call and based on the above study, Elia and TenneT applied on 29 March 2013
for a nomination for a cross border balancing pilot project called "Design and evaluation of a
harmonised reactive balancing market with cross-border optimisation of frequency restoration
while keeping control areas, bid zones, and regulatory oversight intact".
On 25 June 2013
ENTSO-E accepted and approved the nomination.
The present report represents the results of Step 2 of the pilot project. Different working
assumptions were defined regarding the exchange of balancing energy for contract and noncontracted aFRR bids and for non-contracted mFRR bids. More specific in this step working
assumptions were defined regarding the following aspects:

Exchange of aFRR balancing energy (products, processes, merit order, exchange,…);

Exchange of mFRR balancing energy (products, processes, merit order, exchange,…);

Use of cross-zonal transmission capacity for the exchange of balancing energy; and

Settlement of balancing energy and imbalance settlement
On 13 June 2014 TenneT and Elia organised a stakeholder workshop to present and discuss the
intermediate findings of Step 2.
As a next step of the pilot project it is foreseen to perform a cost benefit analysis in order to assess
the different working assumptions and to support a go/no-go decision for implementation.
11 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
CHAPTER 1: Frequency restoration in the European context
1.1 Introduction
This pilot project is taking place within changing European markets for electricity. The EU goal of
completing the IEM has led to the process of developing a number of European Network Codes
defining requirements for TSOs and others to operate the European electricity system in a safe
and secure way, while integrating both wholesale and balancing markets. For this pilot project,
specifically the expected Network Codes NC LFCR and NC EB are of importance.
NC LFCR describes the technical requirements of the frequency restoration process, including the
frequency restoration target parameters that each TSO has to abide by. This is described further in
paragraph 1.2 below. NC EB describes the target model for the integration of markets for
balancing energy and capacity, assuming a stepwise integration of these markets and
harmonisation of principles. It also describes the way to define standard products to be used in
cross-border balancing.
The European context for this pilot project is important, as it shows that there are many
developments, and any findings of this research can be used to further the knowledge on
balancing markets within Europe and smoothen the transition process towards integrated markets
for balancing energy.
1.2 Frequency restoration process
The aim of load-frequency control is to maintain a satisfactory frequency quality within the
synchronous area and to restore the balance of each LFC Area within the time to restore
frequency (TTRF, 15 minutes for the synchronous area continental Europe). Table 1 provides an
overview of the parameters used to assess the regulation quality.
Table 1: parameters to assess the regulation quality
Area
Synchronous Area
LFC Block
Parameter
System frequency
Frequency Restoration Control
Error (FRCE); this is the ACE
Target
50 Hz
0 MW
The FRCE is equal to the residual imbalance of the LFC Block (or LFC Area) after balancing
actions performed by the TSO.
Control structure and TSO responsibilities
NC LFCR defines different types of areas for which TSOs are responsible to fulfil certain
obligations.
12 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 1: Process responsibility structure (simplified) of NC LFCR
The areas and responsibilities defined in the NC LFCR that are relevant for this Pilot Project are:
 LFC Area [Article 32(4)]:
o The TSO operating an LFC Area must operate a Frequency Restoration Process
(see below); and
o Endeavour to fulfil the FRCE (ACE) Quality Target Parameters of the LFC Area.
These Target Parameters are to be discussed between the TSOs of the LFC
Block. Each TSO has the obligation to respect these criteria to ensure good
balancing quality.
 LFC Block [Article 32(5)]:
o The TSOs operating an LFC Block must endeavour to fulfil the FRCE (ACE)
Quality Target Parameters of the LFC Block. These are binding Target
Parameters, discussed on European level and to be respected by all TSOs; and
o Comply with the FRR Dimensioning Rules and the RR Dimensioning Rules.
Elia and TenneT respectively operate the Belgian and Dutch LFC Block (LFC Area). The scope of
the pilot project is limited to the “Design and evaluation of a harmonised reactive balancing market
with XB optimisation of Frequency Restoration while keeping control areas, bid zones, and
Regulatory oversight”. Any discussion on the control structure (organization of LFC Areas and LFC
Blocks) is therefore out of the scope of this pilot project.
The responsibilities of both Elia and TenneT to be considered throughout this Pilot Project
are:


Both TSOs are responsible for dimensioning their own Reserve Capacity to control their
respective LFC Blocks;
Both TSOs must ensure that the FRCE of their respective LFC Blocks respects the FRCE
Target Parameters set in the NC LFCR [Article 20].
Control processes and reserves
For the process activation structure the NC LFCR defines the three different processes shown in
Table 2 and Figure 2.
13 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 2: Load frequency control processes and activation structure
Table 2: Overview of control processes
Process
Frequency
Containment
Mandatory
/ Optional
Mandatory
Goal


Frequency
Restoration
Mandatory


Reserve
replacement
Optional

Joint process by all TSOs of the Synchronous
Area
Stabilize frequency after a deviation to a value
different than the nominal frequency
Process operated by the TSOs of an LFC Area
Restore the balance of the LFC Area within Time
to Restore Frequency (TTRF), being 15 minutes
for Regional Group Continental Europe (RG CE)
Optional process aiming to restore sufficient
Frequency Restoration Reserves to cope with
future imbalances
In
scope
of pilot
No
Yes
No
The reserve replacement process is optional as a number of TSOs including TenneT TSO BV and
Elia believe the energy balance up to real time is the responsibility of market parties and the TSO
should not proactively attempt to restore this balance. The balancing markets in both Belgium and
the Netherlands actively incentivise BRPs to restore the energy balance of their portfolio until realtime. The BRPs can do so by adjusting their portfolio or by exchanging energy (HUB, ID
market,…). The TSOs in both countries therefore only perform actions to resolve the residual
power imbalance they face in real-time. As such their actions, along with the incentives given to
BRPs and market parties, reflect the imbalance in real-time. Such market design ensures that
financial benefits that otherwise could be realized by operating a reserve replacement process are
captured by market players (intraday markets,...).
The frequency restoration process aims at restoring the balance of an LFC Area within the TTRF
(15 minutes) by activation of Frequency Restoration Reserves (FRR). NC LFCR identifies 2 types
of FRR. These are shown in Table 3.
Table 3: Types of Frequency Restoration Reserves
Name
Automatic
Restoration
(aFRR)
Frequency
Reserves
Description
Reserves activated automatically by a
controller operated by the TSO which
continually looks
at
the
Frequency
Restoration Control Error (FRCE)
Former name
Secondary Control
Power
14 | P a g e
October 17, 2014
Manual
Restoration
(mFRR)
Frequency
Reserves
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Reserves activated upon a specific ad-hoc
(manual) request of the dispatching of the
TSO
Tertiary
Power
Control
FRR dimensioning rules
The following FRR dimensioning rules set forth in the NC LFCR are relevant for this project:

The FRR dimensioning rules [NC LFCR Art 46] reveal an important link between the
amount of required FRR Capacity on the one hand and the historical upward and
downward imbalances in an LFC Block on the other hand.

The rules also state that the TSOs of the LFC Block must define the FRR-split between
aFRR and mFRR, as well as the aFRR and mFRR Full Activation Time, in order to respect
the Frequency Restoration Control Error Quality Target Parameters (ACE quality). Annex
1 provides a more detailed overview of the FRR dimensioning rules.
This is explained later on in this section when discussing the individual TSO responsibilities.
Frequency Restoration Process in Belgium and the Netherlands
The scope of this pilot project is limited to the Frequency Restoration Process. This process is
performed by the activation of aFRR and mFRR. Furthermore many TSOs have recently set up an
(optional) Imbalance Netting Process [NC LFCR art. 2]. Both Elia and TenneT are participating
(and will continue to do so) in the same Imbalance Netting Process with other TSOs (the so-called
iGCC process). Hence benefits of the Imbalance Netting Process are already captured by Elia and
TenneT. The working assumptions for the cross-border exchange of aFRR and mFRR balancing
energy made in this pilot project must be compatible with the Imbalance Netting Process.
Imbalance Netting minimizes the simultaneous activation of upward and downward aFRR
Balancing energy subject to availability of remaining cross-border transmission capacity after the
intraday market (i.e. no transmission capacity is reserved for imbalance netting). This reduces the
activated amount of FRR balancing energy and therefore also the further benefits that can be
accomplished by the cross-border optimized exchange of FRR balancing energy via a Common
Merit Order List.
The frequency restoration process in Belgium and the Netherlands is performed by:
 Imbalance Netting (separate process);
 Automatic Frequency Restoration Reserves; and
 Manual Frequency Restoration Reserves.
The above list also represents the chronological steps of the Frequency Restoration Process.
TenneT and Elia do not perform economic optimisation between the use of aFRR and mFRR. The
activation trigger for both products is based on technical considerations only (regardless of the
underlying energy prices of these products).
15 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Important considerations for cross-border exchange of balancing energy
The above sections result in two very important considerations for the implementation of the crossborder exchange of FRR balancing energy:
1. Harmonisation of aFRR and mFRR energy products for the sake of cross-border
exchange of FRR balancing energy may have impact on the FRR capacity procurement
costs and in extension on local TSO access tariffs.
2. In a cross-border exchange of balancing energy between LFC blocks the individual
responsibilities of TSOs at LFC block level need to be taken into account.
These considerations must be taken into account when developing the working assumptions for
such cross-border exchange throughout the entire pilot project.
The impact of harmonization of FRR energy products on the FRR capacity procurement costs
(price and volume) to be borne by the involved TSOs (and their access tariffs) is often neglected:

The Full Activation Times of aFRR and mFRR are important parameters in the FRR
dimensioning process, determining the required volumes of aFRR and mFRR reserve
capacity.

The FRR energy product definitions themselves impact the FRR reserve capacity prices
(impact on market liquidity, resources capable of delivering the service,...).
Harmonization of aFRR and/or mFRR energy products can result in a cost increase for the FRR
reserve capacity, potentially outweighing the resulting benefits of the cross-border energy
exchange itself. FRR capacity costs impact local TSO access tariffs.
The TSOs of an LFC Block are, for their LFC Block, responsible for the FRR dimensioning and for
achieving a satisfactory FRCE regulation quality. The cross-border exchange of FRR balancing
energy between LFC Blocks must respect these individual TSO responsibilities.

The impact of the cross-border exchange of balancing energy on the FRCE regulation
quality of the individual LFC Blocks must be kept limited. Indeed it would be unacceptable
that a TSO cannot respect its legal obligation to achieve a good FRCE regulation quality
due to the cross-border exchange of FRR balancing energy;

The impact of the cross-border exchange of FRR balancing energy on the LFC Block
imbalances of the Connecting TSO must be kept limited. Given the link between the
historical LFC Block imbalances and the FRR dimensioning, such an impact can result in
increased FRR reserve capacity volumes (and costs) for the Connecting TSOs.
The above consideration is very important to consider for the cross-border exchange of balancing
energy between LFC Blocks, but is not relevant for the exchange of balancing energy within an
LFC Block (e.g. the German GCC) due to common dimensioning and shared FRCE quality targets
within an LFC block.
16 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
CHAPTER 2: Reactive balancing
2.1 Introduction
This chapter sets out to sketch the reactive balancing context within which the pilot project for
cross border balancing between Elia and TenneT is taking place. The functioning and concepts of
a reactive balancing design will be illustrated through comparison to a proactive design.
In current times the changing electricity markets and integration of renewable energy is creating a
challenging playing field for TSOs. This is e.g. illustrated in the TenneT study "White paper on a
1
sustainable design of the electricity market" . This is a time in which it is critical to focus on the
functioning and incentives in the balancing market and its interaction with wholesale markets.
Decisions on market designs should not be taken lightly and it is important to understand the
underlying intricacies before making any changes.
2.2 Division of responsibilities
Re-active TSO design
ISP = time to restore frequency
Technical
Economical
Frequency Containment Frequency Restoration
FCR
aFRR, mFRR
MW
System Imbalance
ISP + 1
Individual TSO or
Market:
IntraDay Trade,
Energy Dispatch
System Imbalance
time
Figure 3: Reactive balancing system: market and TSO responsibilities
The key feature of a reactive balancing market is the division of responsibilities between market
players and the TSO in relation to balancing the positions of market players. The underlying
principle behind reactive balancing is that market players not only bear the full financial
responsibility for their imbalances, but, more importantly, they are also allowed to and incentivized
to manage these risks. This will be explained in more detail below.
The responsibility of the TSO applying a reactive concept (see Figure 3) is restricted to maintaining
the power balance based on actual system imbalances. In regards to the TSO processes to
maintain balance, this means that the TSO bears the responsibility for the frequency containment
1
E-Bridge consulting and UMS group, White paper on a sustainable design of the electricity market,
Challenges from Renewable Energies and Guidelines for a Sustainable Market Design
(http://www.tennet.eu/nl/fileadmin/downloads/News/White_Paper_on_a_Sustainable_Market_Design__
1_.pdf)
17 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
process and the frequency restoration process, but does not make use of a reserve replacement
process. The intraday market takes the place of the replacement process and market players
maintain as much as possible the system energy balance.
The frequency containment process, an obligatory process used in both reactive and proactive
designs, is a reactive process, triggered by a frequency deviation. It has a definite start and
termination, based on the system frequency. The frequency restoration process is also an
obligatory process. When used in reactive designs, it is also a reactive process, in this case
triggered by a power deviation, with a definite start and termination.
The reserve replacement process is a proactive process, triggered not by an actual power
imbalance, but by an expected energy imbalance or the expected continuation of the current
energy imbalance towards the future, and is aimed at economic re-optimisation. It does not have a
definite start or termination, but can be applied based on predictions of what will happen. In
proactive designs, the frequency restoration process can also be used proactively, in order to
counteract expected imbalances, not in real-time. The proactive design is illustrated in Figure 4;
the TSO starts acting on the system energy balance at an earlier point in time, taking the
responsibility away from the market.
Pro-active TSO design
ISP > time to restore frequency
Technical
Frequency Containment Frequency Restoration
FCR
aFRR, mFRR
MW
System Imbalance
Economical
Reserve Replacement
RR
Individual TSO
Or Market
ISP + 1
time
Figure 4: proactive balancing market: market and TSO responsibilities
In a reactive balancing system such energy imbalances are the financial responsibility of market
participants, and the TSO shall not use a reserve replacement process. When a TSO applies a
reactive balancing concept, this means that market players play a more proactive role. In a
reactive market design the TSO doesn’t exclude market parties from taking their responsibility to
balance. Therefore the TSO shall avoid activating reserves for long time periods. Activation of
reserves for long time periods in the future would be inefficient if market parties restore their
imbalance (TSO then activated reserves in the wrong direction).
In practice the reactive balancing concept implies that market parties carry a greater responsibility
to maintain their own energy balance, and could even be incentivized to support the system
18 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
balance as a whole. In order for market parties to be able to make the right choices to help the
system, the design of the balancing market needs to be such that they are given the right
incentives and the right information.
2.3 Imbalance settlement period
Figure 5: imbalance settlement periods (ISP) throughout Europe (source: ENTSO-E)
Figure 6: imbalance pricing schemes throughout Europe (source: ENTSO-E)
Figure 5 and Figure 6 show that the duration of the imbalance settlement period (ISP) in
combination with time to restore frequency (15 minutes for Continental Europe) is a crucial in
deciding on an imbalance pricing methodology. Imbalance prices must incentivize BRPs to
maintain the energy balance within the imbalance settlement period. In case the ISP exceeds time
19 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
to restore frequency (15 minutes), BRPs cannot be incentivized to contribute to the restoration of
the frequency within 15 minutes. In such case most TSO apply replacement reserves to maintain
the energy balance within ISP.
Countries with longer ISPs will have difficulty applying a reactive balancing concept, and will
require some proactive design elements in order to deal with the discrepancy between the ISP and
the time to restore frequency.
2.4 Incentives and pricing within a reactive balancing design
Tools for the market
In order for market participants to successfully balance themselves and support the system, they
need the following:

Ex post notification within the bidding zone;

Position changes of BRP portfolio not restricted to ex-ante notification; ex-post notification
allowed;

Absence of internal congestions; in such case TSO can allow BRPs to change their
position up to real-time;

Easy access to intraday markets;

Timely information of imbalance prices and system regulation states;

Consistent imbalance prices (mostly single pricing); and

Ability to adjust balancing energy bids in intraday.
Single imbalance pricing
In order to give the right incentives and tools to market participants, some market design aspects
that are typical to a reactive balancing design. Single imbalance pricing is an example that implies
that imbalances of balance responsible parties that aggravate system imbalances are penalised.
Reactive markets have this in common with proactive market designs. Typical for the reactive
designs is that imbalances of balance responsible parties that support the restoration process are
rewarded.
Incentive compatibility requires such "penalties" respectively "rewards" to be related to the system
imbalance respectively the volume of actions undertaken by the TSO to restore the power
imbalance, i.e. the volume of frequency restoration reserves activated by the TSO to restore the
imbalance created by the BRPs in its area of responsibility.
A reactive market design requires a strong and consistent interdependency between the price of
imbalance and the power error to be restored on the one hand and the price of imbalance and
price of balancing energy activated to restore the power error to be restored on the other hand.
The price of imbalance needs to be elastic with respect to system imbalance of the corresponding
LFC Block (see Chapters 6, 7 and 8).
Passive contributions
The first step towards a reactive market design implies giving market parties the means and
incentives to maintain their own balance. One step further towards a market base balancing
design, use incentives enable BRPs to restore the system balance in case of large system
20 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
imbalances. Apart from incentivizing BRPs through penalizing respectively rewarding aggravating
respectively reducing imbalances, BRPs should also be allowed to act proactively on these
incentives, thereby contributing to the system balance in a positive manner. This passive
contribution is an aspect of some, though not all, reactive market designs.
This requires timely information of imbalance and balancing energy prices and regulation states to
the market (close to real time).
And it requires easy access to ID markets, including the possibility to notify intra Zonal positions
after delivery, and the ability to adjusts Balancing Energy Bids close to the potential ISP of
requested delivery. All such design aspects result in minimizing, through market processes, the
volume of balancing energy to be activated by the TSO.
Proactive control
In contrast, within proactive designs the BRPs should not be incentivized to support the system
balance by the imbalance prices, lest the reserve replacement process would be compromised.
When a TSO uses a reserve replacement process because they have a longer imbalance
settlement period, avoiding counter-activation (infra ISP activation of balancing energy in both
directions) requires a dual pricing system, at least for dispatchable resources, thus denying all
exploitation of flexibility unless through centrally dispatched bids. Proactive designs aim at
centrally controlled price optimization through the TSO.
Summary
Re-active Concept
Pro-active Concept
All TSOs
Curative Objectives

Contain Frequency Deviation

Return to Control Program/Frequency

Release Activated FCR
Some TSOs
Preventive Objectives

Non-activation FRR

Create Margin

Optimization
Input
Input

Actual Frequency deviation, ACE
Means



Frequency Containment Reserves
Frequency Restoration Reserves
Market Incentives

Means


TSO Expected System Imbalance
Replacement Reserves
Eliminate Market Incentives
Incentive Based Concept
Pro-active Concept
Some TSOs
Some TSOs
Objectives

limit TSO control actions
Objective

low activation prices
Means

Means





Induce Incentives to BRPs to reduce System
Imbalance
Only Fast FRR reserves needed
ID GCT close-to-real time
Short ISP
Single Imbalance pricing




Eliminate Incentives to BRPs to self balance or
reduce System Imbalance
Replacement Reserves needed
ID GCT early before real-time
Long ISP
Dual Imbalance pricing
21 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
2.5 Effectiveness and implementation of reactive balancing
A different accent in a reactive balancing design in comparison to a proactive balancing design, is
the focus on aspects on the costs of balancing. The cost of balancing consists of two variables: the
prices of flexibility, and the volumes of flexibility. In proactive market designs there is a focus on
limiting the prices of the flexibility used. In contrast, within a reactive market design the focus is on
limiting the volumes of balancing energy activated by the TSO.
In a reactive market design, the actions not taken proactively, are in practice taken over partly by
market parties, and are partly not necessary because they would have led to counteractivations.
The market makes use of their own flexibility in order to maintain their own energy balance, and
possibly support the system energy balance. This leads to a reduction of the imbalances in the
system and accordingly also to an improved ACE quality, as is illustrated by Belgian data over the
last couple of years, in Figure 7 and Figure 8. These numbers are rather unique, as they show
directly what the impact is of a TSO switching to a more reactive system; Belgium has switched
from a proactive design to a reactive design in late 2011, for instance introducing single marginal
imbalance pricing in order to incentivise market parties, and the system imbalance (ACE Open
Loop) has been steadily decreasing and at the same time –as a result- the ACE quality improved
significantly.
The Netherlands has been applying a reactive design for a longer time, since the beginning of
2001. Since then minor improvements have been made, in disseminating real time information to
the market, allowing the market to improve decision making, and in a modification of the
methodology to define the regulation state, aimed at reducing financial exposure to BRPs.
Some differences between the ways the reactive design is implemented in Belgium and the
Netherlands remain. These differences are explained in Chapters 6, 7 and 8, and relate amongst
other things to the way in which the regulation state in each country is determined, with all the
resulting effects on the distribution of risks, and of financial results between BRPs, BSPs, TSOs
and tariff payers.
Figure 7: evolution of absolute imbalance energy volume in the Elia LFC Block. Transition to re-active
design and single marginal pricing as of 2012 (red blocks). Values for 2014 represent extrapolation of
Jan – August 2014 period.
22 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 8: Evolution of standard deviation of system Imbalance and ACE Belgium
2.6 Influence on project scope
The fact that both Belgium and the Netherlands apply a reactive balancing concept, means that it
their opportunities for cross border balancing are limited to imbalance netting and exchange of
aFRR and mFRR. The former is already implemented via iGCC, so the focus of this pilot project is
on the exchange of balancing energy from aFRR and mFRR. Compared to some of the other pilot
projects, this leaves a relatively small market for balancing energy, especially because the reactive
market design also reduces the amount of imbalances themselves in the system, and hence the
required volume of balancing energy activation. However, it is the final step to complete a cross
border cooperation between two separate LFC blocks that are separate bidding zones, and both
apply a reactive balancing concept. This makes the pilot project unique within Europe.
The fact that there are fewer opportunities to gain cross border synergies through exchange of
balancing energy because the TSO does not use a reserve replacement process, does not mean
there are no cross border opportunities in that timeframe. Market participants balancing their
portfolios or helping to balance the system can use the intraday market for such purposes. In this
way, within a reactive balancing design the intraday market in a way takes the place that the
reserve replacement process plays in a proactive balancing design. Well-functioning intraday
markets are important in a reactive balancing design, especially with an increasing amount of
renewable energy sources connected to the grid.
Since within a reactive balancing design it is very important to give the right incentives to market
parties to balance their own portfolios and help maintain system balance, exchange of balancing
energy between two different LFC blocks both applying a reactive balancing concepts is
complicated. However, the lessons drawn from this pilot project can serve to benefit reactive and
proactive TSOs alike, especially since most TSOs are neither fully proactive nor fully reactive, but
operate somewhere on the spectrum between those extremes.
23 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
CHAPTER 3: IGCC
3.1 IGCC Concept
IGCC is International Grid Control Cooperation aimed at the avoidance of counteracting aFRR in
separate Control Blocks, within the Synchronous Area Continental Europe. Continental Europe
has a (decentralized) Load Frequency Control concept, as opposed to a Frequency Control
concept (Nordic, UK, Ireland). It is an international extension of part of the German collaboration
concept elaborated by the 4 German TSOs and the German regulator.
Presently (2014) IGCC includes: Energinet (DK-W), TenneT TSO (NL), SwissGrid (CH), Ceps
(Czech Republic) + Seps (Slovakia), Elia (Belgium); other TSOs have also shown interest in
acceding. The process is presently hosted by TransNetBW. In NC LFC-R it is referred to as the
Imbalance Netting Process.
The principles of the IGCC process are explained in the Figure 9. IGCC is a continuous dynamic,
real time process with no ramp rate limitations. It is performed by incorporating into ΔP a virtual tieline exchange value, representing the control target avoided by netting all market imbalances of
the participating partners.
Figure 9: functioning of the IGCC process
The virtual exchange due to IGCC is limited to remaining ATC after Intra-Day closure, and occurs
only within that capacity that has not been used by the market in previous, intraday markets, or by
other already committed exchanges of mFRR and/or aFRR. The process is interruptible at any
time; it does not affect local control reserve capacity requirements, nor does it require
harmonization of local imbalance pricing arrangements.
Avoiding X-Zonal counteractivation aFRR is beneficial as apart from reducing local aFRR
activation, it in addition will reduce potential inefficiencies as local counteractivation of aFRR due
to ramp rate limitations and/or dynamic Merit Order Lists.

All participating TSOs and their users benefit from these positive effects.

Non-participating TSOs in the same Synchronous Area benefit from overall improved ACE
quality of the participating TSOs.

The virtue of the concept is that the process can be integrated independently of the market
design of the TSOs involved.
24 | P a g e
October 17, 2014

[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Different markets design may affect valuation of avoided activation, and may introduce
distributional effects. This should be remedied by local market design reform rather than
by global redistribution through settlement.
3.2 IGCC Results
Global results
REGULAR REPORT ON SOCIAL WELFARE
PAGE 2
05/10/13
Summary
Avoidance of
Counteracting SCE
Activation
Benefit
ACE Quality of a
Control Block
Overall ACE Quality
Frequency Quality
The IGCC avoided approximately 1.4 TWh of positive, respectively
negative counteracting secondary control energy
The overall benefit of avoided counteracting activation accumulated to
about 72 Million €
The IGCC interchange improves the ACE quality of the single
participating Control Blocks
The IGCC interchange improves the joint ACE quality of the
participating Control Blocks
The IGCC interchange improves the joint frequency quality of the
Synchronous Area
Figure 10: Summary observations on Imbalance Netting through IGCC
Local results are already presented in the Balancing Comparison; in Annex 2 an update is given.
The highlights are that volumes of IGCC Interchange are of the same order or magnitude as
volumes of aFRR in both Control Blocks, and lead to improvement of Control quality. Due to
different aFRR characteristics in B and NL the effects on avoidance of aFRR activation are not
identical.
The automatic, permanent and real time character of IGCC strongly reduces volumes of activated
aFRR; as a result it reduces additional benefits that can be realized by any exchange of aFRR
balancing energy.
Currently the Imbalance netting performed by iGCC is limited to the available aFRR volumes per
LFC Block; this limit is currently under discussion (see chapters 6, 7 and 8).
25 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
CHAPTER 4: Exchange of aFRR balancing energy
This chapter deals with the exchange of aFRR balancing energy on the basis of a Common Merit
Order List between Elia and TenneT NL. Conclusions of this pilot project are therefore relevant for
the development of the European integration model. It builds further on the Balancing Comparison.
In a first stage more details on the current activation process of aFRR at Elia (pro-rata scheme)
and TenneT (merit order scheme) and the functioning of the secondary controller are given:

Section 4.1 discusses evolutions in the aFRR framework of respectively Elia and TenneT
since the publication of the phase 1 report.

Section 4.2 explains the working principles of a secondary controller. It focuses on:
o
Main blocks of the secondary controller;
o
The difference between the aFRR control target and aFRR control request;
o
The difference between a pro-rata and merit order aFRR activation scheme.
In order to set up the cross-border exchange of aFRR balancing energy a thorough
understanding of the functioning of the secondary controller activation principles of aFRR
is crucial.

Section 4.3 highlights the specific implementation details of the Elia pro-rata and the
TenneT merit order aFRR system.
In a second stage this chapter focuses on defining working assumptions for the future exchange of
aFRR balancing energy:

Section 4.4 investigates whether the exchange of aFRR balancing energy on the basis of
a CMOL is possible between a pro-rata (Elia) and a merit order (TenneT) aFRR activation
scheme. A working assumption for this pilot project is defined.

Section 4.5 describes the working assumptions for this pilot project with respect to
harmonisation of the aFRR products used in Belgium and the Netherlands with a focus on
the harmonisation of the full activation time of Belgian and Dutch aFRR bids.

Finally section 4.6 describes the working assumptions regarding the process for the
exchange of aFRR energy for this pilot project. It thereby investigates:
o
Bidding process
o
Activation process
o
Exchange process
o
Settlement process
The exchange process section explains the significant difference between the options to
either exchange the aFRR control target or the aFRR control request in detail. A working
assumption is taken.
26 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
4.1 Evolutions in aFRR framework
At TenneT there were no evolutions in the aFRR framework since the publication of the Balancing
Comparison.
As of 2014 Elia procures part of its required aFRR capacity on a monthly basis (20 MW for 2014).
Elia therefore tenders in month M-1 part of the aFRR capacity for month M. This must be changed
in Table 3 of the Balancing Comparison according to Annex 3.1. No other changes regarding the
aFRR framework were implemented.
4.2
Technical functioning of aFRR activation schemes
The main goal of this section is to provide a high-level description of the main blocks of a
secondary controller. The difference between the aFRR control target and request, as well as the
difference between a pro-rata and merit order activation scheme are explained in detail.
Main blocks of a secondary controller
Figure 11 shows the structure of a secondary controller.
Figure 11: schematic overview of a secondary controller
Table 4 provides a high level overview of the functions of the different modules of the secondary
controller. More information can be found in Annex 3.2.
Table 4: high level overview of the different secondary controller modules
Module
Function
Input module
 Calculates the FRCE (ACE) of the LFC Area which is the input signal
for the secondary controller.
AGC controller
 Defines the aFRR control target. Basically this module consists of
some filters and a PI-controller.
Bid activation
 Determines the aFRR control request for the aFRR energy bids.
module
 Either a pro-rata scheme (parallel and proportional activation of all
aFRR energy bids) or a merit order scheme (activation of cheapest
aFRR energy bids only).
Communication
 Communication module of aFRR control request to the BRPs.
module
aFRR control target and aFRR control request
27 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
The main difference between the aFRR control target and control request is that:

the aFRR control target reflects the desired aFRR activation in case the ramp rate of
aFRR bids would be infinite; hence given the limited (contracted) ramp rate of aFRR
energy bids, it is possible that this signal cannot be delivered physically;

the aFRR control request is determined on the basis of the aFRR control target, but
respects the limited (contracted) ramp rates of the aFRR energy bids and therefore, at
least in theory, can be physically delivered by the aFRR provider.
The left scheme in Figure 12 an example for the different signals for a single AGC cycle:
Figure 12: Left: aFRR control target and aFRR control request; Right: determination of aFRR control
request
The right scheme in Figure 12 shows how the aFRR control request for a bid is defined on the
basis of the aFRR control target. The blue circle indicates the aFRR control request of the bid in
AGC cycle (t-1). The green circle indicates the aFRR control target of the bid in the next AGC
cycle (t). However the physical ramping capabilities of the bid are limited (defined by the ramp rate
of the aFRR bid). This means that the aFRR control request for AGC cycle (t) must be restricted to
a value that respects the ramping capabilities of the bid (red arrow). Therefore the aFRR control
request in AGC cycle (t) is restricted to the purple value.
It is important to state that the above explanation is valid for Belgium and the Netherlands. For
example in Germany the situation is different as no control request is defined.
Pro-rata and merit order activation
Pro-rata activation scheme
In case of a pro-rata scheme the aFRR control target (output AGC module) is proportionally
allocated to all the participating aFRR energy bids. The algorithm thereby neglects the aFRR
2
energy price of the selected aFRR bids . This means that, regardless the aFRR energy price, all
the bids are always activated in parallel. The contribution of each bid to the global aFRR control
target is proportionate to the size of the bid compared to the total aFRR volume. This is shown in
2
In order to organize competition on aFRR energy prices, Elia selects only the 150 MW of cheapest aFRR
energy bids in D-1.
28 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
step 2 of Figure 13. In this particular example the aFRR control target equals half of the available
aFRR bids. Hence the aFRR control target of each bid (green shared area) is half the bid size.
Figure 13: functioning of pro-rata aFRR activation scheme
In step 3 the aFRR control request per bid is determined by taking the physical ramping
capabilities of the bid into account. The (minimum) ramping capability of the aFRR bids can be
imposed by the TSO (e.g. 7% per minute) or can be nominated by the BSP (e.g. higher ramping
capability than the minimum imposed by the TSO).
Annex 3.3 gives a concrete example of the functioning of a pro-rata bid activation module.
For a pro-rata aFRR activation scheme the maximum ramp rate of the aFRR activation is constant
regardless the required volume of aFRR activation. The total maximum ramp rate is the sum of the
maximum individual ramp rates of the bids. This is due to the fact that all bids are activated in
parallel.
In general one can say that for the same aFRR volume and bids the pro-rata scheme provides a
faster response (and hence a better regulation quality) than a merit order scheme (see below) but
comes at a higher cost (all bids are activated in parallel, regardless their aFRR energy price).
Merit order activation scheme
In case of a merit order scheme the aFRR control target (output AGC module) is allocated to the
cheapest aFRR bids only. This ensures that only the cheapest aFRR bids are activated to deliver
the aFRR control target. This is shown in step 2 of Figure 14 (identical situation as depicted for the
pro-rata scheme). Again the sum of the green shaded parts of the bids equals the total aFRR
control target (output of AGC controller).
29 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 14: functioning of merit order aFRR activation scheme
In step 3, as for the pro-rata scheme, the aFRR control target per bid is then limited to the physical
ramping capabilities of the bid to define the aFRR control request. As shown above a pure merit
order activation approach for aFRR doesn’t exist, there is always partial activation of aFRR bids.
Annex 3.3 provides a concrete example of the functioning of a merit order bid activation module.
For a merit order aFRR activation scheme the maximum ramp rate of the aFRR activation
depends of the aFRR control target and volume of the selected aFRR bids to fulfil this target. The
maximum ramp rate is the sum of the ramp rates of the actual selected bids for aFRR activation. In
case of a small number of selected aFRR bids for activation (e.g. small imbalance), the ramp rate
is lower than for a pro-rata scheme. In case all aFRR bids are selected for activation (e.g. large
imbalance), the total maximum ramp rate is comparable with a pro-rata scheme3.
In a merit order scheme the total maximum ramp rate can be increased by selecting an additional
aFRR bid for activation (thereby impacting the price).
In general one can say that for the same aFRR volume and bids the merit order scheme leads to a
more economic activation of aFRR (and hence lower balancing energy costs) than a pro-rata
scheme (see below) but is slower and therefore results in a worse regulation quality (less bids are
activated in parallel so the ramp rate is lower).
4.3 Implementation of aFRR schemes in Belgium and the Netherlands
In general the implementation of the pro-rata aFRR scheme of Elia is less complex than that of the
Dutch merit order aFRR scheme, due to the following:
3
Under the assumption of identical aFRR bids in the pro-rata and merit order scheme; furthermore the
total maximum ramp rate in a merit order scheme also depends on whether the selected bids for activation
are already fully activated or not. In case they are, the ramp rates of these bids must be excluded from the
total maximum ramp rate. This is not the case for pro-rata schemes (all bids are activated proportionally).
30 | P a g e
October 17, 2014

[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
In the Dutch merit order scheme the activation of an aFRR bid directly impacts the
imbalance settlement price.

TenneT applies dual imbalance pricing in some cases when there is bi-directional FRR
regulation within a single ISP. The secondary controller is tuned in such a way that it
avoids bi-directional regulation when not really needed . This in order to preserve optimal
balancing incentives to market parties (see Chapters 6, 7 and 8.
In the Elia system there is only one aFRR price for respectively upward and downward aFRR
activation taken into account in the imbalance pricing, regardless of the activated aFRR volume
itself. The imbalance price itself is also determined by the price of mFRR activations. In case of
aFRR activation only the direction of activation of aFRR and not the amount of activated aFRR
bids themselves set the imbalance settlement price. This is due to the pro-rata (parallel) activation
of all aFRR bids and the application of average weighted pricing for aFRR in the imbalance tariffs.
Elia always applies single pricing in the imbalance settlement.
In both systems the BSPs are responsible for switching within the bid portfolio. Elia and TenneT
only send an aggregated aFRR control request per BSP. It is the responsibility of the BSP to
decide which units/portfolios to activate in order to comply with the control request.
The resolution of the aFRR bids is 15 minutes in both systems. Basically this means that every 15
minutes another set of aFRR bids can be activated.
4.4 Pro-rata and merit order aFRR schemes
Figure 15: Map showing where pro-rata and merit order activation is applied; source: ENTSOe survey
2013The map of Figure 15 gives an overview of which EU countries apply a pro-rata (dark purple)
31 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
and merit order (light purple) scheme for aFRR activation. The next sections investigate whether
harmonization is required or not.
Compatibility pro-rata and merit order aFRR scheme for aFRR balancing energy exchange
The previous sections explained that Elia and TenneT apply different activation schemes for aFRR
balancing energy (resp. pro-rata and merit order activation). This section investigates whether the
exchange of aFRR balancing energy according to the European integration model can be
performed between a pro-rata (Belgium) and merit order (Netherlands) aFRR scheme.
NC EB requires –subject to a positive outcome of the CBA of the target model- the activation of
aFRR balancing energy on the basis of a Common Merit Order List and an Activation Optimization
Function. The Common Merit Order List is a list ranking the aFRR energy bids (for standard
products) submitted within the CoBa in order of ascending (descending) aFRR energy prices for
upward (downward) aFRR. The Activation Optimization Function only activates the cheapest
aFRR energy bids on the CMOL.
In a pro-rata system all the aFRR energy bids are activated in parallel, regardless of their (usually
regulated) aFRR energy bid price. Hence a pro-rata scheme itself cannot provide for the activation
of aFRR energy on the basis of aFRR energy bid prices.
A possible solution for the exchange of aFRR energy between a pro-rata and merit order aFRR
scheme is to submit the entire aFRR volume of the pro-rata system as one single
upward/downward ‘merged bid’ to the Common Merit Order List (and with a single price). This
price can be e.g. the price of the most expensive aFRR bid of the pro-rata system or the average
price of the aFRR bids. An example is shown in Figure 16.
Figure 16: CMOL construction for pro-rata and merit order aFRR scheme
Such ‘merged bid’ consists of different aFRR bids with (possibly) different prices. The integration
of such a bid into the CMOL can result in situations where expensive aFRR energy bids are
activated before less expensive ones.
32 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
In addition there are some other shortcomings related to the exchange of aFRR balancing energy
between a pro-rata and merit order aFRR scheme:

The advantage of a pro-rata compared to a merit order aFRR scheme is its better
regulation quality due to its constant maximum ramp rate for aFRR activation [see section
4.5]. Its drawbacks are worse economic efficiency and lack of price progressivity related to
the activated aFRR volume.
Any exchange of aFRR energy between TSOs with pro-rata and merit order aFRR
schemes introduces an asymmetry in aFRR ramp rates, activated aFRR volumes and in
price information exchanged between the involved TSOs.
o
E.g. the TSO operating the pro-rata aFRR scheme might receive a ‘slower’ merit
order contribution if the merit order bids are cheaper on the CMOL. As such the
main benefit of operating a pro-rata scheme (better regulation quality) disappears.

Potential incompatibility in pricing schemes as a pro-rata scheme is (often) based on
regulated aFRR energy prices and pay-as-bid (incompatible with pay-as-cleared) while a
merit order scheme is (often) based on free prices and pay-as-cleared or pay-as-bid.

The exchange of aFRR between a pro-rata and merit order aFRR scheme does not create
a level playing field for BSPs within the CoBa.
o
E.g. some bids will be activated more (deliver more aFRR energy) than others,
thereby impacting the aFRR energy (and capacity) prices. A simple example is
that the cheapest bid in a pro-rata system is relatively less activated (always prorata) than the cheapest bid in a merit order system (often fully activated).

The ‘merged bid’ of the pro-rata system decreases the price progressivity of the aFRR on
the CMOL. This might result in a deterioration of balancing incentives given to market
parties as the aFRR energy price –and the resulting imbalance price- remain artificially
constant for a large volume of aFRR activated energy.
The exchange of aFRR energy between a pro-rata and merit order system according to the
European integration model is sub-optimal from economic point of view.
The working assumption for this pilot project is the exchange of aFRR energy between two
merit order aFRR schemes as it is economically more optimal and most in line with the
(current) target model of NC EB.
If case of negative outcome of the CBA for this working assumption, the option for aFRR exchange
between the Belgian pro-rata and Dutch merit order scheme can be considered as it is a
technically viable option with lower implementation costs. However in such case the potential
negative effects on the economic efficiency must be kept under control.
33 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Impact of and conditions for transition to merit order aFRR scheme in Belgium
The working assumption for this pilot project is the exchange of aFRR energy between two merit
order aFRR schemes. Elia is willing to investigate a transition from a pro-rata scheme to a merit
order activation scheme for aFRR balancing energy in case the cost benefit analysis performed
in phase 3 of the pilot project turns out to be positive and some conditions are fulfilled.
Some positive and negative impacts of the transition from a pro-rata towards a merit order aFRR
scheme for Elia are (non-exhaustive):
 Negative: Potential decrease in aFRR regulation speed and required increase of
aFRR capacity or decrease in aFRR Full Activation Time
o
The regulation speed for an aFRR merit order system is slower than for a pro-rata
system (especially for small activated volumes of aFRR). This slower regulation
speed potentially might result in a worse FRCE (ACE) regulation quality. Such
decrease must be offset either by:

Increase of the volume of costly aFRR capacity; or

Decrease of the aFRR Full Activation Time.
Both solutions potentially result in a significant increase in aFRR capacity
procurement costs (direct link with access tariffs) for Elia. This is explained in
more detail in Annex 4.1.

Positive: Potential increase in aFRR market attractiveness
o
A merit order aFRR scheme enables free pricing for aFRR energy bid, intraday
updating of aFRR energy bids, unlimited volume of free aFRR bids,...
o
This might attract new aFRR resources and have a positive impact on the aFRR
capacity procurement costs.

Positive: Potential increase in efficiency of balancing incentives
o
A merit order aFRR scheme introduces better price progressivity in the imbalance
tariffs as the imbalance price depends on the volume of activated aFRR. In a prorata aFRR scheme the imbalance price is independent of the volume of activated
aFRR energy.
o
Better balancing incentives tend to reduce the imbalances introduced in the
system by market parties.

Positive: Facilitation of border-crossing exchange of aFRR energy
o
As described in the previous section it is sub-optimal to perform exchange of
aFRR balancing energy between pro-rata and merit order aFRR schemes.
o
Collaboration with Germany and the Netherlands is facilitated by a transition to a
merit order aFRR scheme.
o
Border crossing exchange might have a positive impact on the aFRR capacity
procurement cost.
A transition towards a merit order aFRR scheme for Elia can only take place if following
conditions are fulfilled:

A positive cost benefit analysis in phase 3 of the pilot project;
34 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]

No significant impact on the FRCE (ACE) regulation quality;

Confidence that the transition to a merit order aFRR scheme will increase the aFRR
market liquidity (new entrants), both for energy and capacity products;

Impact on aFRR procurement costs, activation costs and according impacts on tariffs
(access tariffs, imbalance tariffs) are fully recognized and accepted by CREG and
market parties;

The decision for a pricing scheme for aFRR energy (pay-as-bid or pay-as-cleared)
must take the aFRR energy market liquidity into account and might be complemented
with a cap and floor to ensure reasonable prices;

Sufficient resources and budget must be available to Elia in order to implement the
transition; and

Elia must have sufficient comfort that the chosen merit order scheme and aFRR
framework are sustainable towards the future. Indeed it must be avoided that the
chosen solution must be changed within some years. This is very important, given e.g.
the significant differences in merit order schemes between the Netherlands and
Germany.
35 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
4.5 Working assumption for aFRR Full Activation Time
One of the main questions for the exchange of aFRR balancing energy is whether or not to
harmonize the aFRR balancing energy products. It is important to notice that NC EB requires the
definition of a harmonized standard aFRR product.
In this perspective discussions regarding possible harmonization of the aFRR Full Activation Time
- which is the (maximum) time in which an aFRR bid must be able to fully deploy - are key. This is
due to the inherent links of this parameter with:

required aFRR capacity volumes (cfr. dimensioning section 1.2) and procurement costs;

aFRR market liquidity and prices (energy and capacity);

aFRR balancing energy prices;

FRCE (ACE) regulation quality;

...
Today the aFRR Full Activation Time is different for most EU countries; some examples are:

Netherlands: ≤ 15 minutes

Belgium: 7,5 minutes

France: 6,5 minutes

Germany: 5 minutes

...
First the impact of a change of aFRR Full Activation Time on the required aFRR Capacity Volumes
and FRCE Regulation Quality is explained.
Next the current aFRR Full Activation Time of TenneT and Elia is discussed. Finally the working
assumption for the aFRR Full Activation Time for this pilot project is elaborated.
Link between aFRR Full activation time, FRR dimensioning and FRCE (ACE) regulation quality
The FRR dimensioning rules (see section 1.2) reveal a link between

the aFRR Full Activation Time;

required aFRR capacity volume; and

the achievement of the FRCE (ACE) quality target parameters.
This can be explained by the fact that a faster aFRR energy product (lower aFRR Full Activation
Time) can better and faster offset an imbalance in the LFC Block resulting in a superior FRCE
regulation quality. The FRCE (ACE) regulation quality becomes worse in case the aFRR Full
Activation Time increases. This is shown in Figure 17. The faster the aFRR product, the earlier the
FRCE (ACE) can be brought back to zero MW and hence the better the FRCE (ACE) regulation
quality.
36 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 17: FRCE (ACE) regulation quality in case of slow (left) and fast (right) aFRR product. It is clear
that in case of the fast aFRR product, the FRCE (ACE) is brought back faster to 0 MW, resulting in a
better regulation quality.
When looking at a possible harmonization of the aFRR Full Activation Time within a CoBa it is
therefore very important to consider the impact on the FRCE regulation quality of the LFC Blocks.
Each LFC Block is responsible to respect the FRCE Quality Target Parameters as set forth in the
NC LFCR.
aFRR (automatic activation) generally has a higher regulation performance than mFRR (manual
activation). Reasons for this are that mFRR –in contrast to aFRR- has a constant setpoint over the
entire ISP and has generally a slower ramp rate than aFRR. Therefore a deterioration of the FRCE
regulation quality due to an increase of aFRR Full Activation Time might (at least partially) be
compensated by increasing the volume of available aFRR capacity. The more aFRR available in
the system, the less the need for activating less performant mFRR reserves.
Harmonisation of aFRR Full Activation Time for the sake of enabling cross-border
exchange of aFRR balancing energy can result in a cost increase for the aFRR reserve
capacity, potentially outweighing the resulting benefits of the cross-border energy
exchange itself. The Cost Benefit Analysis for the implementation of the cross-border
exchange of aFRR balancing energy must consider the full picture.
In addition to the aFRR Full Activation Time also the aFRR activation scheme (pro-rata or merit
order) is a determining factor for the aFRR deployment speed and hence FRCE (ACE) regulation
quality. As explained before in general a pro-rata scheme has faster aFRR deployment than a
merit order scheme (especially for small activated aFRR volumes).
More detailed information on the link between the aFRR Full Activation Time, aFRR activation
scheme, FRCE (ACE) regulation quality and aFRR dimensioning can be found in Annex 4.1.
aFRR Full Activation Time at Elia and TenneT
Table 5 summarizes the current aFRR full activation time requirements at Elia and TenneT as well
as the procured aFRR volumes. The aFRR ramp rate is the inverse of the aFRR full activation
time.
37 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Table 5: aFRR Full Activation Time requirements at TenneT and Elia
Elia
Contracted
aFRR
volume
140 MW
Maximum aFRR full
activation time
(ramp rate)
7,5 minutes
300 MW
(13,3% per minute)
≤ 15 minutes
Pro-rata
TenneT
Merit
order
(≥ 7% per minute)
Explanation
Elia determines the maximum aFRR ramp
rate to be 13,3% per bid. All submitted aFRR
bids must follow this requirement.
The BSP has to nominate the maximum
aFRR ramp rate of its aFRR bids. TenneT
withholds bids respecting the ‘≥ 7% per
minute’-criterion.
It is observed however that BSPs nominate
often the minimum ramp rate.
The above table shows that for Elia the maximum aFRR ramp rate is independent of the activated
aFRR volume (pro-rata system) and equals 18,7 MW per minute (140 MW * 7%/min). For TenneT
the maximum aFRR ramp rate depends of the activated aFRR volume (merit order system). In
case the full 300 MW of aFRR is activated, the maximum ramp rate is similar to the Elia system
(300 MW * 7% per minute). For lower activated aFRR volumes the maximum ramp rate is less in
the TenneT system. Additional non pre-contracted aFRR volumes can increase the ramp rates.
Elia’s control strategy is to have a relatively small band of fast aFRR volumes which is
complemented with slower mFRR volumes. TenneT relies on a bigger amount of aFRR volumes
with a slower ramp rate for small imbalances and a higher one for big imbalances, and as a result
largely avoids the activation of mFRR.
The above reasoning only provides some qualitative insights in the ramp rates, as for merit order
systems the ramp rates also depends of the aFRR bid size, the number of aFRR bids already
delivering their requested activation, the activation algorithm itself (activate bids in parallel under
some circumstances),...
The definition for aFRR ramp rate is different for Elia and TenneT (imposed fix value in Belgium –
maximum requirement in the Netherlands). The proposed definition of aFRR ramp rate for the rest
of this pilot project is:
The aFRR ramp rate is defined as the minimum ramp rate requirement (% per min) the TSOs
require. Faster ramp rates (reaction) than the minimum requirement are allowed and will be
used accordingly by the TSOs. However on the CMOL the bids are just ranked on the basis
of balancing energy price.
Working assumption for aFRR Full Activation Time
This section describes the options that Elia and TenneT identified to deal with the aFRR ramp
rates in the exchange of aFRR balancing energy between their respective LFC Blocks. It provides
a qualitative comparison between the options.
There are two main options for the aFRR ramp rates for the exchange of aFRR balancing energy:
 Option 1: harmonisation of aFRR full activation time; and
38 | P a g e
October 17, 2014

[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Option 2: no harmonisation of aFRR full activation time.
The implementation of the exchange of aFRR balancing energy is the simplest in case both TSOs
use the same local products, so no translation of products is necessary in the creation of the
common merit order list. This situation allows for a fully symmetric approach for the exchange of
aFRR balancing energy and a level playing field for BSPs.
However, the current ramp rates of local aFRR products are often the result of an optimisation
taking into account the historical environment:

Elia developed a relatively fast aFRR product and algorithm due to historical low liquidity in
the aFRR market. The Belgian mFRR market is very liquid. The combination of a relatively
fast aFRR product, a fast pro-rata activation scheme with slower mFRR reserves enables
Elia to achieve a satisfying ACE quality with a limited aFRR volume.

TenneT developed a slower aFRR product to attract as much liquidity as possible to the
aFRR market. Indeed, by lowering the requirements there are more resources able to
submit aFRR bids to the TSO (at lower price). Furthermore this allows BSPs to keep some
additional flexibility to perform reactive balancing by reacting on market prices.
Harmonisation prevents these local considerations from being fully taken into account, which could
have a negative impact on the efficiency of the balancing arrangements in one or both countries. It
is therefore important to consider all the impacts of a change in aFRR ramp rates in the costbenefit analysis of step 3 of this project.
In case of option 2 a ‘standard aFRR product response’ is exchanged in two directions over the
border. This ‘standard response’ is the response of the aFRR product with the fastest ramp rate.
The reason for this is that it would be unacceptable for the TSO with the faster aFRR ramp rate, to
receive a slower response in case the aFRR bids at the other TSO are cheaper.
There are two ways to implement option 2:

Option 2a: parallel partial activation of slower bids in the area of the TSO with the slowest

aFRR bids to achieve the fast ‘standard response’ (left scheme in Figure 18).
Option 2b: no parallel activation of slower bids in the area of the TSO with the slowest
aFRR bids. Hence the ‘standard response’ is delivered partly by activated aFRR and partly
as FRCE (ACE) of the Connecting TSO (right scheme in Figure 18).
39 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 18: Different implementation options for the exchange of aFRR balancing energy without
harmonization of aFRR Full Activation Time
It is clear that in case of option 2a no pure merit order activation takes place (partial parallel
activation). This reduces pricing transparency and increases the complexity of the Activation
Optimization Function. In case of option 2b the TSO with the slowest aFRR bids will face a
deterioration of FRCE (ACE) regulation quality. This impacts the (future) invoice for the settlement
of unintended deviations according to NC EB. Table 6 summarizes advantages and disadvantages
for option 1 and option 2.
On the basis of the table below the working assumption for this project is to harmonise the
Belgian and Dutch aFRR ramp rates. This option appears to be the most sustainable and robust
option and allows for a level playing field between BSPs.
On the other hand it is clear that, although sub-optimal, there are also options that do not require
harmonisation. Hence in case the outcome cost benefit analysis of phase 3 of the project is
negative for the harmonisation case, the non-harmonisation options might be further investigated.
Table 6: pros and cons for harmonization and non-harmonization scenario of aFRR ramp rates
Reciprocity
Harmonization of aFRR ramp rate
No harmonization of aFRR ramp rate
Optimal
Sub-optimal
However dependent on other factors
TSO
(bidding process, settlement,…)
reaction in case importing;
with
slower
bids
gets
faster
TSO with faster bids pays higher price
(parallel activation of bids) in case of
importing
40 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Different aFRR ramp rates result in
different activated volumes per aFRR
bid; impact on pricing and bidding
Level playing field
Yes
No
for BSPs
However dependent on other factors
Different ramp rates do not foster level
playing field.
More robust / less complex
Less robust / more complex
The algorithm must only take one
Activation Optimisation Function must
aFRR ramp rate into account
consider ≥1 aFRR ramp rate
OR
Robustness
complexity
/
No
Extendibility
Sustainability
difference
in
aFRR
activation
Increase of FRCE (ACE) at TSO with
speed depending on activation location
slowest aFRR bids
More difficult as it requires more
Simpler
TSOs to harmonize their aFRR ramp
becomes more complex if more different
rates
aFRR ramp rates
Compliant with NC EB (in line with
Incompliant definition of standard
the definition of a standard aFRR
products in NC EB
product and current target model)
Also parallel activation of slower bids
(no
harmonization)
but
undermines the pure merit order list
activation set forth in the current target
model of the NC EB
The working assumption for this pilot project is a harmonization of aFRR ramp rates in Belgium
and the Netherlands. The working assumption for the aFRR Full Activation Time has to be
determined. Most TSOs in Europe have an aFRR ramp rate between 5-10 minutes. Some
important examples are Germany (5 minutes) and France (6,5 minutes). Looking at the future with
increased volumes of intermittent RES, it is widely accepted that there might be a need for faster
reserves compared to today. As a conclusion the aFRR Full Activation Time should optimally be
between 0 – 10 minutes in order not to compromise extendibility or sustainability.
Elia performed some simulations to assess the impact on the FRCE (ACE) regulation quality for
Belgium (see Annex 4.1) for a transition to a merit order aFRR scheme, for different aFRR Full
Activation Times. These simulations show that it is no option for Elia to increase the current aFRR
Full Activation Time from 7,5 minutes to ≤15 minutes (TenneT value) given the drastic impact on
the FRCE (ACE) regulation quality in Belgium. In such case Elia must procure more than 100 MW
of additional aFRR capacity (>+70%!) to maintain the current FRCE (ACE) regulation quality. This
would come at unacceptable cost (reflected in the Belgian access tariffs).
On the other hand a decrease of the aFRR Full Activation Time for TenneT might introduce a
negative effect on their local aFRR capacity and balancing energy market. It is important to notice
that the border-crossing exchange of aFRR balancing energy might increase liquidity for aFRR
balancing energy, but not for aFRR capacity as the exchange of aFRR capacity requires avaiability
of border crossing transmission capacity.
41 | P a g e
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
October 17, 2014
Table 7 provides a high level overview of the effects of a change in the aFRR Full Activation Time.
Table 7: Impact of change in aFRR Full Activation Time
Evaluation criterion
aFRR capacity
procurement costs
aFRR balancing
energy prices
aFRR balancing
energy bids liquidity
Explanation
 Higher ramp rate requirements lead to lower liquidity as less
BSPs will be technically able to provide the service; less offered
volumes result in higher prices.
 The pilot project only investigates the exchange of aFRR
balancing energy; the exchange of aFRR capacity requires
transmission capacity. Hence there is no increase in aFRR
capacity bids due to cross-border exchange of aFRR balancing
energy.
 Higher ramp rate requirements might lead to less (same) aFRR
capacity for same (better) FRCE quality
 Lower ramp rate = less energy delivered by BSP (negative
impact);
 Lower ramp rate = more flexibility for BSP for passive
contribution (positive impact);
 Lower ramp rate = more competition (positive impact)
 Stricter ramp rate requirements lead to reduced market liquidity
(less non-contracted bids,...) as less resources might be
technically able to provide the service.

Border-crossing exchange of aFRR balancing energy might
increase the liquidity.
Based on the above it was decided that the working assumption for this pilot project is to
consider an aFRR Full Activation Time of ≤7,5 minutes (no increase for Elia and according
lowest possible decrease for TenneT). TenneT is prepared to investigate this if the outcome of the
cost benefit analysis of phase 3 of the project is positive. Furthermore discussions on EU-level
need to be followed closely (aim to only change ramp rate once). The CBA should at least focus
on following points:

Capacity costs and volumes (incl. link between FRCE Quality Target Parameters and
dimensioning)

aFRR energy prices

Market liquidity

Extendibility

Reciprocity and fair distribution of benefits

Robustness

Complexity / technical feasibility

Impact on TSO responsibilities

Sustainability

Compatibility with options for technical implementation

Competition between BSPs and level playing field
42 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Survey on aFRR ramp rate per bid
Given the actual differences in ramp rates in Belgium and the Netherlands, both Elia and TenneT
launched a survey towards resp. FEBEG and EnergieNed to get an idea of the impact of possible
harmonization of the aFRR ramp rates. Some participants to the survey required confidentiality of
their answers. Therefore they are not included in this report. In general the responses confirmed
the above analysis performed by Elia and TenneT. Nevertheless a more thorough analysis is
required in the cost benefit analysis of phase 3.
4.6 Exchange of aFRR balancing energy: process
This section describes the process for the exchange of aFRR balancing energy for this pilot
project. Different options are analyzed and working assumptions are defined. Table 8 provides a
high-level overview of the process.
Table 8: high level overview of process for the exchange of aFRR balancing energy
When?
Ex-ante
What?
Bidding process
Realtime
Activation process
Realtime
Ex-post
Exchange process
Settlement
Description
 From individual BSP offers for aFRR energy bids
towards a Common Merit Order List of aFRR bids for
the CoBa.
 Each TSO defines the need for aFRR activation for its
own LFC Block;
 The Activation Optimization Function optimizes the
activation of aFRR for all TSOs of the CoBa.
 The TSOs of the CoBa exchange cross-border aFRR
balancing energy over a Virtual Tie-Line.
 The TSOs of the CoBa settle the activated aFRR
energy on the basis of the TSO-TSO settlement
function (see settlement and pricing chapter).
Special focus is on following principle questions:

How will the Common Merit Order List and Activation Optimization Function work?

Which power profile is exactly exchanged between the Connecting and Receiving TSOs
over a Virtual Tie-Line?

How is the exchange of aFRR balancing energy taken into account in the secondary
controllers of the TSO?

How to organize the exchange of aFRR balancing energy in order to respect the individual
TSO obligations as described in section 1.2?
Bidding process and constitution of Common Merit Order List
This section focuses on the working assumptions for the aFRR bidding process characteristics, as
well as on the formation of the aFRR balancing energy Common Merit Order List.
NC EB requires that BSPs have to submit their aFRR bids to their Connecting TSO before aFRR
balancing energy market gate closure time. Hence the gathering of aFRR bids remains a local
TSO responsibility. The working assumptions for the bidding process characteristics are shown in
Table 9.
43 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Table 9: aFRR energy bidding process working assumptions for this pilot project
Parameter
Portfolio based /
unit based
Working assumption
Portfolio based
bidding
Rationale


More flexible for market parties
Allows for aggregation

Low value increases the amount of
resources able to offer aFRR energy
bids (market liquidity)
Avoid overlap with (local and/or crossborder) intraday markets
Ensures that all flexibility is offered first
on the intraday market before being
offered on the balancing market
Ensures market reflective bid prices of
aFRR energy (reflect at least the
intraday energy prices)
Ensures efficient balancing incentives to
market parties
Working assumption to introduce a
cap/floor still under investigation (at least
for Elia to ensure reasonable aFRR
prices)
Both contracted and non-contracted
aFRR energy bids will be incorporated in
the CMOL
All non-contracted bids will be accepted
(no cap on the volume)
Due to the short gate closure time the
BSPs have to withdraw their bids if they
cannot be activated anymore after an
activation
Information required for local congestion
management
bidding
Minimum bid size
1 MW
aFRR balancing
(Max.) 1 hour
before delivery
energy GCT



Price
Contracted /
To be defined by
BSP
Both



non-contracted

aFRR energy
bids
Quarantine time
Not applicable

Location
Connecting TSO

defines required
information
The gate closure time of maximum one hour before the market is a target. TenneT and Elia don’t
exclude that in a first stage collaboration can be set up without harmonizing this parameter. The
impact of such decision must be analyzed in more detail.
As already described in section 4.2 in a merit order aFRR scheme the aFRR bids are ranked upon
ascending (descending) aFRR energy prices for upward (downward) aFRR bids on the Merit Order
List. Basically the same principles apply for the constitution of the Common Merit Order List for
aFRR in a CoBa. The Connecting TSOs of the CoBa have to forward their local aFRR bids to the
Common Merit Oder list directly after the aFRR energy market gate closure time. The Common
Merit Order list consist of price-ranked bids of multiple TSOs (see Figure 19).
44 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 19: construction of aFRR balancing energy CMOL
For congestion management reasons as well as in case of outages, the TSOs of the CoBa can
qualify the bids on the CMOL unavailable. Such bids cannot be activated by the Activation
Optimization Function afterwards.
Activation process
The activation process of aFRR in a CoBa is performed by the Activation Optimization Function .
The goal of the Activation Optimisation Function is to ensure the most optimal activation of aFRR.
Section 4.2 shows that for a merit order aFRR scheme the aFRR control target defines which
aFRR bids need to be activated, as well as the aFRR control target per bid. This is indicated in
Figure 20.
Figure 20: determination of aFRR control target per bid
This is some kind of Activation Optimization Function as it ensures the economic most efficient
activation of aFRR energy. In a last stage the merit order scheme takes the limited ramping
capabilities of the activated bids into account and defines the aFRR control request on this basis.
In case of a CoBa the aFRR Activation Optimization Function becomes more complex as the
aFRR control targets of multiple TSOs must simultaneously be optimized. It is no longer operated
by a single TSO but now becomes a common function within the CoBa.
Different principles can be applied for the Activation Optimisation Function:
 First come first serve (FCFS) principle:
o
Each time (each 4-10s) a TSO forwards its aFRR control target to the Activation
Optimisation Function, the algorithm activates the cheapest available aFRR
resources at that time for that TSO on a FCFS basis.
o
There is no ‘netting’ of aFRR activation requests of the TSOs of the CoBa. Hence
simultaneous activation of upward and downward aFRR control target is possible.
45 | P a g e
October 17, 2014

[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Clearing mechanism:
o
All TSOs send their aFRR control target (simultaneously) to the Activation
Optimisation Function. This function sums the aFRR control targets of the TSOs of
the Coordinated Balancing Area and activates the resulting aFRR control target on
the Common Merit Order List.
o
There is a ‘netting’ of aFRR control targets. There is no simultaneous activation of
upward and downward aFRR control target.
The continuous nature of the aFRR activation (the activation at time t depends of the activation at
time t-1) and the working frequency of the AGC controller (4-10s) seem to be incompatible with a
FCFS scheme or would result in a very complex mechanism. Therefore the working assumption
for the aFRR Activation Optimisation Function is a clearing mechanism as it is less
complex and ensures more optimal aFRR activation due to netting.
In a first stage the TSOs of the Coordinated Balancing Area forward their aFRR control targets to
the Activation Optimisation Function. Secondly the Activation Optimisation Function sums the
aFRR control targets of the involved TSOs and activates the resulting aFRR control target on the
CMOL. This is shown in Figure 21.
Figure 21: determination of aFRR control target by the AOF on the CMOL
As explained in section 4.2 the aFRR control request per bid (required physical activation) can be
calculated once the AOF determined the aFRR control target per aFRR bid.
The common optimisation of the aFRR activation leads to a situation where a single aFRR bid can
deliver aFRR balancing energy for >1 TSO at a time. This has to be considered in the settlement
of aFRR balancing energy. It requires a calculation of the amount of activated aFRR energy to be
settled by each TSO during a PTU. Solutions where an aFRR bid can only be activated for one
TSO would artificially increase aFRR energy balancing costs and are highly inefficient. Therefore
this option is not considered.
NC EB [Art. 40.11] states that the aFRR amount that a TSO can activate on the Common Merit
Order List is limited to the volume of bids that he submitted to the CMOL unless the TSOs of the
CoBa agree otherwise. Therefore there are two options for this pilot project:
46 | P a g e
October 17, 2014

[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
A TSO cannot activate more aFRR balancing energy as it made available on the Common
Merit Order List (cross-border exchanged aFRR capacity is also taken into account); or

A TSO can activate more aFRR capacity as it made available on the Common Merit Order
List but each TSO keeps prior access to its own aFRR balancing energy bids (contracted
bids and non-contracted bids of its own LFC Area).
The working assumption for this pilot project is the second option that allows a TSO in
question to activate more aFRR balancing energy in the case this aFRR balancing energy is not
needed by the other TSOs. This is illustrated in Figure 22.
Figure 22: unlimited activation of aFRR balancing energy bids on the CMOL with prior access to
Connecting TSO
The second option:

Optimizes the business case of the BSPs as they activate more aFRR balancing energy;

Allow the TSO in question to perform better LFC; and

Reduces unnecessary activation of mFRR in case aFRR balancing energy is still available
in the system.
The first option is sub-optimal as it forbids the activation of aFRR balancing energy bids while the
Connecting TSO has no need to activate them for own purposes. The initial reason for this was to
ensure that each TSO always has access to at least the volume of bids it submitted to the CMOL.
The second option resolves this by granting the Connecting TSO prior access to at least the
volume of bids it submitted to the CMOL once he needs their activation.
Exchange process
The Activation Optimization Function for aFRR balancing energy in the CoBa determines the
aFRR control target for each aFRR bid to fulfil the sum of the aFRR control targets of all TSOs in
the CoBa. This section describes how the exchange of aFRR balancing energy takes place on the
basis of the outcome of the AOF.
As explained in section 4.2 the aFRR control target is the desired activation of aFRR balancing
energy without considering the ramp rate limitations of the aFRR bids. The aFRR control request
is determined on the basis of the aFRR control target and respects the ramp rate limitations of the
aFRR bid. Hence only the aFRR control request can be physically delivered by the BSPs.
47 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
This leads to two different implementation schemes for the exchange of aFRR balancing energy
between TSOs of the CoBa:
 Model 1 which doesn’t consider the finite ramp rate of aFRR bids for the exchange of
aFRR balancing energy. This model is called the “exchange of aFRR control target”.

Model 2 which considers the finite ramp rates of the underlying aFRR bids for the
exchange of aFRR balancing energy. This model is called the “exchange of aFRR
control request”.
It must be highlighted that in both cases the TSOs exchange a firm amount of energy over the
border. As such the Connecting TSO bears the consequences of any mismatch between the
exchanged volume and the locally activated aFRR balancing energy. Such mismatch can be the
result of:

Imperfect activation of aFRR balancing energy by local BSPs; or

The exchange of an aFRR control target over the border which, due to ramp rate
limitations, cannot be physically delivered by the local BSPs.
Hence both models represent an exchange of responsibility from the Receiving TSO
towards the Connecting TSO (as required by NC EB).
The following sections provide a description of the technical functioning of both models:
Model 1: Exchange of aFRR control target
Under this option the TSOs exchange the difference between their initial aFRR control target and
the total aFRR control target allocated by the Activation Optimization Function to their local aFRR
bids over a Virtual Tie Line. This is illustrated in Figure 23. The numerical calculations for this
example can be found in Annex 4.2.
Figure 23: exchange of aFRR control target
48 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
The ‘exchanged aFRR control target’ is added to (deducted from) the imbalance (ACE Open
Loop) of the LFC Block of the Connecting (Receiving) TSO by means of a Virtual Tie Line. For this
example the difference between the initial aFRR control target of TSO A (40 MW) and the final
aFRR control target for aFRR bids in the area of TSO A (50 MW) is 10 MW, which is exchanged.
The numerical example of Annex 4.2 shows that for the above case in the previous AGC cycle –
before any exchange of aFRR- the aFRR control request of TSO A and B were resp. 16,4 MW
and 7,4 MW. Hence it is clear that the 10 MW that is exchanged between TSO A and TSO B is not
physically delivered at the same time in the area of TSO A. This will deteriorate the FRCE of
TSO A and improve the FRCE of TSO B compared to a situation without the exchange of
aFRR capacity.
Basically this model re-distributes the “problem” (being the imbalance of the TSO) by Virtual Tie
Line in such a way that the “problem” is placed at the TSO having the cheapest aFRR bids to
resolve it. Each TSO then activates locally aFRR bids on its local merit order list to resolve the
“problem” that was allocated to him.
Imbalance netting (iGCC) and the exchange of aFRR balancing energy within Germany (GCC) are
implemented according to this model. The main advantages of this model are its robustness and
fairly simple technical implementation.
Figure 24: exchange of aFRR control request
49 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Model 2: Exchange of aFRR control request
Under this model the Activation Optimization Function itself calculates the aFRR control request
for the aFRR bids of TSO A and TSO B. The Activation Optimization Function determines
afterwards how much of the aFRR control request needs to be exchanged between TSO A and
TSO B. This is shown –under identical conditions as the example for model 1- in Figure 24. The
numerical example for this case can be found in Annex 4.3.
The ‘exchanged aFRR control request’ is added to (deducted from) the imbalance (ACE Open
Loop) of the Connecting (Receiving) TSO. For this example there is an exchange of 1,2 MW of
aFRR control request over the Virtual Tie-Line
The exchange of aFRR control request implies that the exchanged energy over the border is
physically delivered at TSO A. In this case there is no impact on the FRCE (ACE) of TSO A and B
compared to a situation without the exchange of aFRR balancing energy (unless in case of
imperfect activation of the local aFRR bids).
This model ensures the activation of the cheapest aFRR bids to resolve the imbalance of the
involved TSOs. It is new and isn’t implemented yet in Europe. The main advantages of this model
are fully coordinated activation of aFRR in the CoBa and easier settlement. The technical
implementation might be more complex than model 1.
The above examples for model 1 and model 2 are based on exactly the same conditions. It
is clear however that there is a large difference in the exchanged power over the Virtual Tie
Line (10 MW for model 1 and 1,2 MW for model 2). The next sections will show that the
difference between the two models is substantial and has far reaching consequences.
Fundamental differences between model 1 and model 2
A. Mutual impact on FRCE regulation quality, TSO responsibility and aFRR
dimensioning
As already indicated above the exchange of aFRR control target impacts the FRCE (ACE)
regulation quality of the involved TSOs. This is not the case for the exchange of aFRR control
request. A simplified example for this is shown in the scheme below.
Section 1.2 explains that the TSOs of the LFC Block are responsible for the FRCE (ACE)
regulation quality of their LFC Block. Figure 25 shows that in case of the exchange of aFRR
control target there is a shift of FRCE (ACE) of TSO A to TSO B compared to the situation without
exchange of aFRR balancing energy. Hence the exchange of aFRR control target conflicts with
this TSO responsibility.
In practice this model might lead to a significant deterioration of the FRCE regulation quality of the
LFC Block of the TSO with the cheapest aFRR energy bids. The reason for this is that practically
all imbalances will be shifted to this TSO to be resolved by his aFRR bids. This TSO might not be
able anymore to respect its legal obligation to satisfy the FRCE Quality Target Parameters.
50 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
The exchange of aFRR control target is feasible between TSOs within an LFC Block (e.g. GCC
scheme in Germany) due to the shared FRCE regulation quality target parameters. However this
scheme cannot be applied for the exchange of aFRR between LFC Blocks as it conflicts with basic
TSO responsibilities.
Figure 25: link between exchange of aFRR control target / request and the FRCE (ACE) of Connecting
and Receiving TSO (simplified example).
In addition to the impact on the FRCE regulation quality, the exchange of aFRR control target
between LFC Blocks also risks to impact the (a)FRR dimensioning. The reasons for this are that:

this model shifts imbalance (ACE Open Loop) from one TSO to another. Section 1.2
explains that the dimensioning of reserves is based on ACE Open Loop of the LFC Block;

the FRCE regulation quality of the Connecting TSO (TSO with cheapest aFRR energy
bids) might deteriorate. As a result this TSO might need to procure more or faster aFRR
capacity.
It is important to note that any impact on the aFRR capacity volumes has a direct impact on the
access tariffs of the TSOs.
It can be concluded that the exchange of aFRR control request therefore is the preferred
model for the exchange of aFRR between LFC Blocks. The above graph shows that under this
model there is no shift of FRCE (ACE) from the Receiving to the Connecting TSO (unless in case
of imperfect activation of aFRR by the local BSPs).
B. Coordinated activation of aFRR
For the exchange of aFRR control target, the Activation Optimisation Function only decides upon
the re-distribution of imbalance (ACE Open Loop) from the Receiving to the Connecting TSO. The
51 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
actual activation of aFRR bids is still performed locally by both TSOs on their local aFRR merit
order lists. Hence there is no fully coordinated activation of aFRR bids within the Coordinated
Balancing Area.
For the exchange of aFRR control request, the Activation Optimisation Function fully determines
the activation of aFRR bids. Hence the activation of aFRR within the Coordinated Balancing Area
is fully harmonized.
In case of the exchange of aFRR control target the local TSOs decide upon the final activation of
aFRR. This requires full harmonization of the secondary controller and FRR activation strategies.
Indeed given the significant difference in functioning of local secondary controllers and FRR
activation strategies, a model where only the “problem” (imbalance) is exchanged will lead to
different aFRR and mFRR activations at the different TSOs of the CoBa. This might raise
questions on settlement and reciprocity.
Furthermore a fully coordinated activation of aFRR bids within the Coordinated Balancing Area
(exchange of aFRR control request) leads to more efficient activation and facilitates a clean
settlement.
C. Firmness of exchange, financial neutrality and unintended deviations
In case of the exchange of aFRR control target the impact of the exchange (transaction cost) is not
known at the time of activation. This is due to the following:

An “imbalance” is shifted from the Receiving to the Connecting TSO. As shown above this
will lead to an increase of activated aFRR energy and an increase in FRCE (ACE) at the
Connecting TSO. Hence the volume of activated aFRR energy at the Connecting TSO
doesn’t equal the volume of imbalance that was shifted between the TSOs. Furthermore it
will be really difficult to determine ex-post which aFRR bids and aFRR energy was
activated for the purpose of exchange of aFRR control target and for own purposes. This
will complicate settlement.

An “imbalance” is shifted from the Receiving to the Connecting TSO. However it is
unknown ex-ante which aFRR bids will be activated by the local secondary controller of
the Connecting TSO (no coordinated activation of aFRR energy bids).

Whereas without the exchange of aFRR balancing energy there would be a resulting
FRCE (ACE) at the requesting TSO, the FRCE (ACE) with exchange of aFRR control
target the FRCE (ACE) is now at the Receiving TSO. This will increase the unintended
deviations (ACE) of the Connecting TSO. NC EB requires financial settlement of
unintended deviations (ACE). The cost for these unintended deviations (volume and
price), due to the exchange of aFRR control target are unknown at the time of activation.
Also in practice it will be very hard to determine ex-post the exact cost of this (which
unintended deviations are due to exchange of aFRR control target and which ones are
due to other reasons).
The above allows to conclude that there is no “firmness of the bid” that is exchanged over the
border in case of the exchange of aFRR control target. Indeed it is unknown which aFRR bids will
be activated as a result of the exchange (no firmness of price) as well as which volume of aFRR
52 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
energy will be activated by the Connecting TSO (no firmness of volume). This is due to the fact
that aFRR activation itself remains a local responsibility of the TSOs and is performed by the local
secondary controllers on the local aFRR merit order list. This conflicts with the requirements of the
NC on Electricity Balancing for the exchange of balancing energy.
Furthermore there is a potential issue regarding financial neutrality. Financial neutrality means that
the exchange of aFRR balancing energy must be financially neutral for both the Connecting TSO
and the Receiving TSO. Indeed nor the Connecting or Receiving TSO should bear profit or loss as
a result to the exchange of aFRR balancing energy. As already shown above the transaction cost
of the exchange of aFRR control target is not known at the time of activation and will be very
difficult (if not impossible) to determine ex-post:

Which aFRR bids and which aFRR energy volume was activated for the exchange of
aFRR control target and which bids and energy was activated for own purposes?

Which part of unintended deviations was due to the exchange of aFRR control target and
which part not?
As a result financial neutrality will be hard to reach in case of the exchange of aFRR control target
and will require ex-post settlement and/or full cost sharing. Under the assumption that the
imbalance settlement must cover at least all balancing costs (see section 6.3) this will lead to
additional problems for systems in which the imbalance price is published near to real-time.
Indeed, as the balancing cost isn’t known in real-time, it is impossible to ensure in real-time
imbalance prices that cover costs for balancing energy and unintended deviations.
The exchange of aFRR control request ensures firmness of the bid over the border. The activation
of aFRR energy is fully determined by the Activation Optimisation Function (coordinated
activation). The exchanged energy therefore is only aFRR balancing energy. Hence such scheme
also allows for financial neutrality for the exchange of aFRR balancing energy.
Working assumption for this pilot project
The working assumption for this pilot project is the exchange of aFRR control request between Elia
and TenneT. Although the implementation of this scheme might be more complex than the
exchange of aFRR control target, this model seems to be the only one that allows for financial
neutrality and respects the local TSO responsibilities of TSOs of an LFC Block.
The exchange of aFRR control target between LFC Blocks can only be implemented in case a
solution is found for the different issues identified (TSO responsibility, financial neutrality,...). Such
option however can easily be implemented for the exchange of aFRR within an LFC Block. It’s
main advantages are its robustness and the technically less complex implementation.
Settlement process
Ex-post settlement is reasonably straightforward: the connecting TSO settles the balancing energy
with its BSPs (and, in the process, corrects their imbalances as appropriate). The cross-border
volumes are then settled as part of the separate TSO-TSO settlement.
53 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
CHAPTER 5: mFRR exchange
5.1 Overview and scope
The Frequency Restoration process as performed by each TSO allows for Balancing Service
Providers to provide explicitly, non-procured manually activated FRR to their connecting TSO on a
local Merit Order List, enlarging Internal Balancing Markets by product design. These products can
and will be used by the connecting TSO in local system balancing. Activated bids are considered
in determination of the imbalance price(s) (see Chapters 6, 7 and 8).
As explained in the Balancing Comparison Elia currently uses a direct activated mFRR product
with unit based bidding whereas TenneT uses a schedule activated mFRR product with portfolio
bidding. As working assumption for the second phase on exchange of mFRR it is assumed that
Elia implements a schedule activated mFRR product, similar to the present product definition in
NL.
Exceptionally Elia might locally activate such product in a direct activated way in order to ensure
faster mFRR response.
Products not on the local Merit Order Lists (directly activated, pre-contracted mFRR) are out of
scope for this second phase of this pilot project on X-Zonal balancing cooperation between Elia
and TenneT NL as cooperation already has been established via reserve sharing, offering
compliancy to the dimensioning obligations of each TSO at lower cost, with the justification that
they actually are activated, and exchanged only rarely.
The savings performed by reserves sharing far exceed potential savings from activation
optimization.
Moreover contracting less flexibility from the market leaves more flexibility within the market.
A possible cross-border exchange of mFRR balancing capacity is out of scope of the present Pilot
Project; as is exchange of balancing energy, direct activated from contracted mFRR. The rationale
for this is set out in the Balancing Comparison (see sub-section 5.6 in particular).
The principal impediment to the cross-border exchange of mFRR balancing capacity is the lack of
(firmly) available cross-border transmission capacity. Reservations of transmission capacity for this
purpose are possible only under rather strict conditions.
54 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
5.2 Working assumption schedule activated mFRR product
The product is a bid of manually activated Frequency Restoration Reserve, schedule activated, an
4
(electronic) message with an accountable character for imbalance adjustment and balancing
energy settlement, from a Balancing Service Provider to its connecting TSO (see Figure 26).
The Balancing Service Provider bids to sell (or buy) a fixed volume of energy to the TSO,
corresponding to bid volume * ISP, in a certain Imbalance Settlement Period (ISP of Delivery), at
a certain price (pay as bid), or at least that price (pay as cleared) settlement of balancing energy.
Figure 26: basic features of scheduled mFRR
The bids become firm in volume and price after Balancing Energy Gate Closure Time. It differs
from a Day Ahead and Intra Day markets in that the Balancing Market is a single buyer market,
with only the connecting TSO as counterpart to all Balancing Service Providers:

Whereas trades on Day Ahead Markets and Intra Day markets result in changes in
Internal or External Trade Schedules of a Balance Responsible Party, trades with the TSO
will result in an adjustment of the Imbalance of the Balance Responsible Party associated
with this bid from the Balance Service Provider .

Whereas trades on the Day Ahead and Intra Day Markets are not taken into account in the
determination of the Imbalance Price, trades on the Balancing Market are taken into
account (see Chapters 6, 7 and 8).

Whereas trades on the Day Ahead Markets and Intra Day markets create and redistribute
value (or social welfare) between market participants, the Balancing Market redistributes
value (or social welfare) between BSPs, BRPs, and the TSO (see Chapters 6, 7 and 8).
The connecting TSO is responsible for complying to the bid parameters as given by the BSP, and
for compliance to pricing and settlement rules: the volume requested by the TSO will be settled
with the Balancing Service Provider at the price of Balancing Energy in the requested direction
(see Chapters 6, 7 and 8).
Compliancy to delivery is enforced by applying an imbalance adjustment to the Balance
Responsible Party appointed by the Balance Service Provider for this product, and the local
4
Article 27 of NC EB requires that “each Balancing Energy bid from a Balancing Service Provider is assigned to one or more Balance
Responsible Parties to enable the calculation of an Imbalance Adjustment pursuant to Article 57”
55 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Imbalance Pricing system: over- respectively under delivery will be settled against the local
Imbalance Price applicable to the Imbalance direction (surplus or shortage).
The BSP is responsible for the bid parameters and acceptance of the financial result:

Any actual imbalances within the ISP of activation, ISP of delivery, and ISP after delivery,
due to activation, and financial imbalance risks associated can be priced into the bid price
by BSPs.
The following rationale applies for schedule activated mFRR product:

provides easy access product to Market, enlarging Internal Balancing Markets by product
design (see Chapter 2), to be traded on a, residual, Balancing Market for non procured
bids,

similarity to Day Ahead and Intra Day Market products,

similarity to supportive Imbalance, but with a reduced settlement price risk,

few prequalification/verification requirements, compared to mFRR directly activated and
aFRR [ch aFRR status Quo].

Impact of activation of mFRR balancing energy bid on the Imbalance price of one ISP
only;

Allows netting of the activation requests of different TSOs to avoid bi-directional activation
of mFRR energy
A late Balancing Energy Gate Closure Times suits for RES and demand resources: All flexibility is
allowed to participate to bid in flexibility as long as offered energy is delivered within ISP of
delivery. The local Merit Order List provides visibility to Market and TSO on actual flexibility
available in the system, in volume and prices.
5.3 Activation philosophies
The two TSOs practice the same activation philosophies for the use of mFRR. Neither TSO
activates aFRR resp. mFRR for the purpose of economic optimisation; in other words, a possible
price difference between aFRR and mFRR does not lead to changes in how (and how much) the
two products are activated. Both TSOs make use of mFRR in order to free / reconstitute their
automatic FRR; i.e., particularly in case of larger imbalances.
However Elia activates mFRR in function of the ACE open loop where TenneT activates mFRR in
function of the ACE open loop after iGCC correction. Liquidity in the Belgian market for aFRR is
low, so Elia procures only a relatively small volume of this product (140 MW in contrast with 300
MW of aFRR procured by TenneT). In order to achieve a satisfactory balancing quality
nevertheless, Elia - in contrast to TenneT - needs to make extensive use of mFRR. Finally, given
Elia's current pro-rata activating of aFRR with fixed prices, activating mFRR in addition to aFRR
has the side effect of creating some variability in the imbalance price. However, the latter effect will
become less relevant if and when Elia activates aFRR on a merit-order basis.
For TenneT, one particular concern in activating mFRR (the "free bids" of which are already
schedule-activated in the Netherlands) is the risk of counteractivation; specifically, the risk of
56 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
5
activating mFRR in one direction and then having to activate aFRR in the opposite direction .
Given the highly unpredictable nature of the sign and magnitude of the ACE open loop even within
an ISP, that risk is high indeed.
Counteractivation of mFRR and aFRR is technically inefficient and, under TenneT's and Elia's
current system of imbalance pricing and financial neutralisation have different distributional effects
between BRPs and tariff payers. Annex 5.1 provides some quantitative evidence on the incidence
of counteractivation during the second half of 2013.
5.4 Introduction to exchange of mFRR
6
7
The FGEB and NC EB building on the FGEB envisage a regional resp. European target model
for the integration of balancing markets (NC EB; Articles 14 and 15). These models are to be
based upon a multilateral TSO-TSO Model with Common Merit Order Lists (CMOLs) and so-called
Standard Products (NC EB; Article 28). Since the present Pilot Project is explicitly meant to move
forward the development of these concepts, the cross-border exchange of scheduled activated
mFRR balancing energy between Elia and TenneT is proposed to follow the TSO-TSO model. The
two TSOs have also developed, as a working assumption, a harmonized, scheduled activated
mFRR balancing energy product which they propose to exchange cross-zonally and which can
serve as a precursor to a future European Standard Product.
5.5 Common merit order list product
As a working assumption a common, harmonized, scheduled activated product, non-procured
balancing energy product is available to both TSOs [See 5.2]. The product on the Common Merit
Order List will be a schedule-activated energy product. Delivery during Imbalance Settlement
Period (ISP) (T) can be demanded at any point during ISP (T-1); even very close to the end of ISP
(T-1). The activation requests of the TSOs can be matched or not (see later). However, during ISP
(T-1) activation for ISPs later than (T) (for example, (T+1), (T+2), etc.) is not possible.
Activation is for one ISP only and deactivation is automatic (unless, in case of larger resp. more
persistent system imbalances, the same bid is activated a second time during ISP (T) for delivery
during ISP (T+1)).The mFRR schedule activated product is an energy product, similar to the ISP
length tradable product beween BRPs on the hub.
5
Abstracting from cross-zonal counteractivation, there are various types of counteractivation that may
occur even within an LFC Area resp. LFC Block: simultaneous vs. sequential (yet within the same ISP); aFRR
vs. aFRR (this phenomenon can arise even simultaneously due to so-called "dummy energy"; e.g., positive
aFRR bids still being ramped down while negative aFRR bids are already being activated; however, it is
more straightforward to think of this type of counteractivation as occuring sequentially); aFRR vs. mFRR; in
an extreme case theoretically even mFRR vs. mFRR. Counteractivation might also be understood to be
counteractivation of an FRR product and the IGCC contribution to balancing the system.
6
ACER, Framework Guidelines on Electricity Balancing, 18 September 2012; available at
http://www.acer.europa.eu/Official_documents/Acts_of_the_Agency/Framework_Guidelines/Framework
%20Guidelines/Framework%20Guidelines%20on%20Electricity%20Balancing.pdf
7
Available at
https://www.entsoe.eu/fileadmin/user_upload/_library/resources/BAL/131223_NC_EB_FINAL.PDF
57 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
The characteristics of the harmonised local product can most easily be summarised in Figure 27.
Figure 27: Characteristics harmonized scheduled activated mFRR
The market is considered to be able to incorporate potential Imbalance risks in their bid price and
dispatch behaviour. The risks for both TSO and BSP differ as a function of the actual activation
moment. Since in the current reactive market design and activation strategy by TenneT the BRPs
are likely to already support system balance in Imbalance, the actual activation will merely result in
a change of volumes to be settled in Imbalance, respectively Balancing Energy.
Furthermore the following applies:

volume settled is volume requested

imbalance adjustment for the volume settled

pricing arrangements (pay-as-bid, pay marginal, relation to imbalance pricing)

no physical verification / monitoring remains a local TSO responsibility
5.6 Bidding process
Exchange of scheduled activated mFRR balancing energy between Belgium and the Netherlands
will follow the TSO-TSO Model. Each of the two TSOs will continue to be responsible for all
communication with the BSPs for whom it is Connecting TSO; the other TSO will not be in direct
contact with these. BSPs will submit their bids to their Connecting TSO from whom they will
receive activation instructions and who will ensure settlement (of both balancing energy and
imbalances, and TSO-TSO exchanges).
58 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 28: Common Merit Order List in a TSO-TSO model
Both TSOs will include all or a subset of the bids in their local merit order list in the Common Merit
Order List and will ensure that this list is continuously up-to-date. This applies in particular if a bid
becomes unavailable due to congestion, an outage, or a similar problem. The Common Merit
Order List for mFRR balancing energy is thus based upon a Standard Product with a harmonised
product definition, as detailed in Annex 5.2. BSPs submit their bids to their respective TSO's local
merit order list. Their TSO will forward them to the CMOL.
If bids become unavailable due to congestion or an outage, it is the Connecting TSO's
responsibility to update the CMOL accordingly. BSPs will only ever have to communicate with their
local (connecting) TSO. Note that "local" product characteristics (such as the "divisibility" option in
Belgium) may be retained on the CMOL. However, it is the Connecting TSO's responsibility to
ensure that translation into CMOL bids is correct and immediate. After the balancing energy gate
closure time (which both TSOs would eventually like to move towards real-time as much as
possible and which they aim to set at the earliest to (H-1) in the first instance), bids become firm
and may no longer be modified.
5.7 Activation
Both TSOs determine their respective need for mFRR and, in principle, forward this request to the
Activation Optimisation Function (AOF) which administers the CMOL. Having said that, both TSOs
agree that it will remain possible for the Connecting TSO to activate mFRR from the local MOL
according to local rules. However, in that case the Connecting TSO will have to mark the CMOL
bids thus becoming unavailable accordingly. The default scenario, however, will be for all
activation requests to be sent to the Activation Optimisation Function.
The Activation Optimisation Function (AOF) administers the CMOL and has access to all bids
included in the CMOL. No restrictions are envisaged with respect to the volume of mFRR that a
TSO is allowed to request. Specifically, the TSOs will not be limited to activating the volume which
they themselves have made available to the CMOL. However, each TSO will have priority access
to those bids that it submitted to the CMOL. If TSO A activates a cross-border bid during ISP (T),
then a request by TSO B will result in this bid becoming available to TSO B from ISP (T+1)
onwards (if otherwise the remaining volume available for B were insufficient).
59 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
One important question relates to whether offsetting requests by the two TSOs are matched or not.
Clearly, if requests sent to the AOF are forwarded to BSPs immediately, matching a later
(offsetting) request by the other TSO is impossible. However, waiting for possible matching
opportunities requires activating bids for delivery during ISP (T) late in ISP (T-1). This leads to
delayed physical activation and increased discrepancies between physical delivery and financial
settlement (and thus increased risk resp. costs to the BSP and a deterioration in balancing quality
for the Connecting TSO). On the other hand, cross-zonal counteractivation should be avoided just
like intra-zonal counteractivation, as any kind of counteractivation makes it more difficult for TSOs
to achieve financial neutrality. In light of these considerations, the TSOs have decided that the
default solution will be make use of netting opportunities to the extent possible and, therefore, to
activate bids late and match resp. net them if possible.
In case of cross-border activation, transmission capacity needs to be available. The AOF therefore
needs to have a (two-way) link to all relevant capacity management modules (incl. the IGCC one)
and takes availability of capacity into account when selecting bids for activation. Once the AOF
has selected bids for activation, the actual activation requests are sent by the Connecting TSO
who will also remain responsible for correct delivery by its BSPs.
The mechanics of the actual cross-border activation are the subject of the following section.
However, before moving on the subject of (intra-zonal) congestion management should be
addressed. Congestion is relevant in at least two respects:
1. Internal Congestion might, in principle, make bids unavailable (i.e., they cannot be
activated without violating security constraints). This problem is taken care of by the
connecting TSO's responsibility for keeping the CMOL updated and marking bids as
unavailable if and when necessary.
2. 2. A TSO might want to use mFRR bids submitted to the CMOL for the purpose of
curative congestion management rather than balancing (within the restrictions set by
the NC EB; see in particular Article 39). This problem is addressed via the abovementioned convention of allowing the Connecting TSO to activate mFRR from the
local MOL according to local rules (and then having to mark the CMOL bids thus
becoming unavailable accordingly).
5.8 Cross-border exchange
The actual cross-border transfer of the mFRR balancing energy activated will be implemented via
a firm, predetermined virtual tie-line block profile as shown in Figure 29.
60 | P a g e
October 17, 2014
Early:
Connecting TSO:
reduces control
inaccuracies
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Late:
Requesting TSO:
facilitates efficient
activation
Firm Virtual Tie-Line
Block Profile
ISP of Activation
ISP of Delivery
Figure 29: Cross-border exchange of schedule activated mFRR
During the ISP of delivery, the ACE of the requesting TSO and the ACE of the connecting TSO will
be adjusted accordingly; i.e., this adjustment will be based upon the block profile corresponding to
the mFRR product. Power differences between the locally activated mFRR schedule activated
energy product, and the X-Zonal exchanged firm power profile will increase the connecting TSO's
ACE and thus may lead to a deterioration of its balancing quality. The latter effect is explicitly
accepted by both TSOs.
In the case of mFRR balancing energy, the nature of the harmonised product, an energy product,
means that a distinction between control request and control target does not apply, as it does in
case of aFRR. Control target and control request are one and the same.
5.9 Settlement
Ex-post settlement is reasonably straightforward: the connecting TSO settles the balancing energy
with its BSPs (and, in the process, corrects their imbalances as appropriate). The cross-border
volumes are then settled as part of the separate TSO-TSO settlement.
61 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
CHAPTER 6: Settlement and pricing – considerations regarding
harmonization
6.1 Requirements by NC EB
The guiding principles of NC EB are for integration, coordination and harmonisation of the
Balancing regimes in order to facilitate electricity trade within the EU in compliance with the
Electricity Regulation (EC) 714/2009 and Directive 2009/72/EC. More specific the FGEB request
the harmonisation of the following aspects regarding settlement:

All design elements required for European balancing market integration

The pricing method for settlement of balancing energy products

The imbalance settlement period

Principles for calculating imbalances

Main features of the imbalance settlement
When discussing harmonisation of the balancing markets it is important to know that the FGEB
and as a consequence also NC EB do not always have strict requirements regarding settlement.
The FGEB only request a harmonisation of the principles for the calculation of the imbalance
volumes and a harmonisation of the main features of the imbalance settlement. Although NC EB
has a preference for marginal pricing, it is still possible for TSOs to apply a pay-as-bid
pricing in case they can provide “a detailed analysis demonstrating that a different pricing method
is more efficient for EU-wide implementation”.
It is also relevant to consider that currently in the Dutch and Belgian balancing market the TSOs try
to promote that Balance Responsible Parties are helping to restore the system imbalance. In both
countries this is done via an imbalance settlement based on the principle of ‘single marginal
pricing’ and by publishing close-to-real time balancing information. Closer analysis reveals
nevertheless some differences between both countries. Based on our assessment we believe
however that that the main features for settlement and principles for calculation of
imbalances are already harmonised between both countries. Nevertheless in this report we
shall give an overview about the remaining differences and assess in a qualitative way the
possibilities for harmonisation.
Currently the imbalance settlement in the Netherlands and Belgium is done per 15 minutes. Hence
the requirement to harmonise the imbalance settlement period is not applicable for this study.
6.2 Harmonisation for market functioning
When establishing a cross border balancing market it is important to create a level playing field
between the Balancing Service Providers and the Balance Responsible Parties of the involved
countries. It is important to consider that such a well-functioning balancing market might already be
achieved by a harmonisation a set of design elements of the balancing market. Hence a full
harmonisation of the balancing market might not be required to create such a level playing field.
On the other hand some market design elements have to be harmonised in order to enable the
exchange of Balancing Energy and Balancing Capacity between different countries. Indeed some
62 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
aspects of the balancing market need to be harmonised in order to allow the creation of the
common merit order list of balancing energy products which then can be activated by the involved
TSOs.
In current step of the pilot project (Step 2) we are focusing on those aspects of the
balancing market which are required to enable a cross-border exchange of balancing
energy. We don’t exclude to investigate –if required– the added value of harmonising other design
elements which are not required to enable the cross border exchange of balancing energy in the
cost benefit analysis [Step 3].
6.3 Financial impact –BSP vs. BRP Settlement
An important aspect regarding harmonisation is the interaction between the settlement of
balancing energy and the imbalance settlement with respect to financial neutrality. In principle the
cost of balancing should be covered by the imbalance settlement. However this is not guaranteed
in particular harmonisation scenarios.
Existence of counteractivation costs
In theory if the volume (energy) and prices used in the imbalance settlement and settlement of balancing energy are
equal, the financial neutrality should be guaranteed. However counteractivations - in case within the same ISP upwards
and downwards reserves are activated - might jeopardise this principle.
The reason for this is that in case a TSO performs counteractivations the gross volume he is settling with BSPs might
be bigger than the quarter-hourly netted imbalance volumes he settles with the BRPs. As a consequence it is possible
that income out of imbalance settlement is smaller than the balancing costs. Therefore, for financial neutrality reasons it
is important for a TSO to keep the counteractivations costs as low as possible. Note that in a balancing market it is
inevitable to have counteractivations (aFRR vs iGCC, aFRR vs. mFRR, mFRR vs mFRR, aFRR vs. aFRR).
Example
Balancing energy (BSP)
Imbalance settlement (BRP)
100 MWh upwards mFRR @ 100 €/MWh
=> 10.000 € paid by TSO
Short BRPs 250MWh pay 25.000 € (250MWh*100€/MWh)
40 MWh downwards aFRR @ 20 €/MWh
=> 800 € received by TSO
Long BRPs 190 MWh receive 19.000 € (190 MWh*100€/MWh)
Result
=> 9.200 € paid by TSO
Result

=>6.000 € received by TSO
Net Result for TSO -3200 €
Another important feature is to which extend the imbalance settlement should cover other cost
than those of the activation of balancing energy. Indeed the imbalance settlement might cover also
other costs related to balancing such as procurement costs of balancing capacity.
As it possible to have a negative or positive outcome out of the settlement of balancing energy and
imbalance settlement there’s an indirect link to the local access tariffs. A surplus might be used to
cover costs which would have been otherwise covered by access tariffs. A deficit might lead to an
increase of local access tariffs.
For the further analysis in this study we shall assume that imbalance settlement should at least
cover all costs of balancing energy.
63 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
6.4 Financial Impact – Local vs. Cross Border Settlement
When designing a balancing market is important for TSOs that the cross border exchange of
balancing energy is not creating a deficit in the settlement for the reserve connecting TSO. More
specific a difference between the energy volume or price used for the settlement at the border
between both TSOs and the energy or price used used locally between the reserve connecting
TSO and its BSP may result in distributional effects between all parties affected.
Price difference
According to NC EB the Balancing Energy Bids offered on a common merit order list should be
firm. In other words after the Balancing Energy Gate Closure Time it is not possible anymore to
modify the price of bid which was transferred to the common merit order list.
Moreover NC EB also stipulates that Connecting TSOs need to submit the bids received from the
BSPs to the common merit order list.
Hence it is not possible that there’s a difference between the price used for the settlement
between the BSP and the connecting TSO and the price used for the settlement between
the connecting and requesting TSO. If a different price would have been used, the reserve
connecting TSO would be exposed to a potential deficit or surplus.
Volume difference
The firmness principle of NC EB is also applicable for volumes of bids offered on the common
merit order list. If the volume exchanged over the border is not equal to the volume that is
settled with the BSP there will be a financial delta for the connecting TSO.
It is important to consider these ‘equality’- principles when designing the operational processes for
exchanging aFRR and mFRR. Indeed some exchange processes would imply to have differences
in the settlement of volumes or would require that the initial bid price is not guaranteed any more.
In such a case – for reasons of financial neutrality- more complex solutions might be required
which might affect the transparency.
6.5 Local TSO responsibility
The balancing market is often mistakenly compared with day ahead markets where it is possible to
have one clearing price for the uncongested region (composed of multiple control areas). As it is
possible to have one clearing price for the uncongested region on day markets, the reasoning is
that it should be also possible to have the same imbalance price for the uncongested region.
However, when considering this, the following should be taken into account:
1. BRPs should not be allowed to balance themselves out over the uncongested region as
this will lead to non-compliant use of interconnection capacity.
2. In contrast to day ahead markets the settlement price in the balancing market is not only
used to settled the exchanged energy but also to determine the imbalance price which is
giving locally financially incentives for BRPs to be balanced and/or to restore the system
64 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
imbalance per LFC Area/Block. Hence there is a direct link with the local TSOs
responsibilities as explained in section 1.2.
1: Non-compliant use of interconnection capacity
As explained in Chapter 9, an administrative amount of cross zonal capacity is made available to
the market. To make use of this capacity, market parties need to nominate it in one of the
timeframes (forwards, D-1, or intraday).
In case both LFC blocks carry the same single imbalance price, market parties that are active on
both sides of the border can adjust their supply or demand of energy on both sides in opposite
directions without it affecting their imbalance payments. In this way they can create a flow of
energy across the border which does not correspond to their nominations. This can be seen as a
non-compliant use of interconnection capacity.
Should this occur, due to the distinction between the administrative capacity available for cross
zonal balancing and the physical capacity of the cross-border connections, there is no capacity
based mechanism that will allow imbalance prices to diverge when the border becomes physically
congested as a result of market parties balancing themselves out across bidding zones.
Example:
Assume that both LFC Blocks are short and the marginal price is currently 120 €/MWh and the available cross border
capacity is almost completely used. In such a case the imbalance prices would be:
Assume a BRP who is having a licence in both countries and a flexible unit with marginal cost 100 €/MWh fully activated in
the control area of Elia and another flexible unit, partially activated -with a cost of 60 €/MWh in the control area of TenneT.
This BRP would have the interest of performing the following arbitrage: first lower the production (ex. 100MWh) in Elia and
increase the production in the control area of TenneT (ex. 50MWh). This would lead to the following benefit
-(100 MWh*120€/MWh)IMB ELIA + (100 MWh*110€/MWh) COST↓ + (100 MWh*120€/MWh)IMB TenneT – (100
MWh*60€/MWh) COST↑ = 6000 €
This behaviour is in fact a non-compliant use of XB capacity as the BRP is transferring energy without nomination of the
use. Moreover as the use is not known by both TSOs cross-border capacity will be not considered as fully used and
balancing market will not split, i.e. imbalance prices will remain equal between both TSOs.
2: Local TSO Balancing responsibility
NC LCFR stipulates that TSOs are responsible for the balance of their own LFC Blocks. Quality
targets are defined and TSOs have to assure that sufficient reserves are available in order to fulfil
these obligations.
Imbalance prices are a key element for these local TSO balancing responsibilities. They should
always give incentives to BRPs to balance themselves and/or help to restore the system
imbalance per LFC Area/Block. It should be avoided that imbalance prices incentivise market
65 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
parties to balance themselves out over different LFC Area/Block (see example previous section)
because this is not compatible with the local TSO responsibilities as defined in NC LFCR.
Example:
BRPs react (passive contribution of +100MW) on imbalance price exceeding +150 €/MWh
TSO A has an imbalance of 20 MW and activates 20MW of bids
TSO B has an imbalance of 500 MW and activates 400 MW of bids
The example below is showing that a large imbalance in one LFC Block B might lead to increase of the imbalance price of
another LFC Block. This increase in imbalance prices is leading to a reaction by market parties and resulting into an
imbalance for LFC area B.
Figure 30: issue of cross-border single pricing
Erroneously it is often assumed that imbalance netting will in all cases resolve this issue. The
example below shows the effect of having a region for imbalance netting which differs from the
region with the single imbalance price, in case there is no subnetting in the latter region.
Example:
The international group X is having a BRP licence in the control areas of TSO A & TSO B.
TSO A, TSO B, TSO C & TSO D are performing imbalance netting.
Assume that the initial system imbalances of the TSOs are
TSO A: 10 MW;
TSO B: 50 MW;
TSO C: -50 MW;
TSO D: 400 MW
TSO C: 0 MW;
TSO D: 356,5 MW
After proportional netting between TSOs the situation would be
TSO A: 8,9 MW;
TSO B: 44,5 MW;
For economic reasons group X decides to create a long imbalance in TSO A (+200 MW) and short imbalance in TSO B (200 MW); total imbalance position in the 2 LFC Blocks is zero. This can be done because single marginal pricing is applied
in the LFC Blocks of TSO A and TSO B (see previous example)
The imbalances of the TSOs are now the following
TSO A: 110 MW;
TSO B: -50 MW;
TSO C: -50 MW;
TSO D: 400 MW
After proportional netting between TSOs the situation would be

TSO A: 86 MW;
TSO B: 0 MW;
TSO C: 0 MW;
TSO D:313 MW
Conclusion; when BRPs are balancing themselves out over an uncongested area imbalance netting is
not a solution to resolve the issue of local TSO responsibility
66 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
A potential solution for the problems described in this section is to apply double imbalance pricing
instead of single marginal pricing. This solution is considered by both TSOs as unacceptable as
experience shows that single marginal imbalance pricing is the best solution for imbalance
settlement.
Another solution would be to follow up in real time the physical reality of the interconnections and
to adjust the imbalance prices if this is required, causing a price divergence when the border
becomes congested. This might be indeed a solution however further analysis is required.
Experience with flow based implementation for day ahead market also shows that such solutions
are highly complex and may require a lot of time to be implemented.
Currently both TSOs accept the working assumption that the issues mentioned in this section
might be resolved by applying a local pricing scheme. This shall be explained later on in Chapter
8.
67 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
CHAPTER 7: Pricing and settlement - Imbalance pricing
7.1 Similarities between Netherlands and Belgium
In this section we are describing the similarities in imbalance settlement between Belgium and the
Netherlands. As explained previously the main features for settlement and principles for
calculation are already harmonised between both countries.
Imbalance price calculation
Imbalance prices are used in both LFC Blocks as important local tool to trigger correct real-time
reactions of balancing responsible parties. As a consequence the imbalance tariffs in the
Netherlands and Belgium are showing a lot of similarities which will be described in this section. In
the next sections we will describe the differences.
In both countries the imbalance prices are derived from the activated bids which were used for the
balance of the LFC Block. For each quarter hour the marginal price for upwards and/or downwards
regulation is applied dependant the regulation state during a quarter hour.
Both TSOs are applying a “single marginal pricing” mechanism for the determination of the
imbalance prices, in other words the imbalance price for a negative and a positive imbalance are
almost equal for each quarter-hour and equal to the marginal price of activated Balancing Energy.
The main feature of such a mechanism is that it enforces the balance responsibility and it
promotes self-balancing. Please note that in the Netherlands – as explained in the next section- it
is also possible to have a double imbalance price, e.g. the imbalance price for a negative and
positive imbalance are different for specific quarter- hours. As in practice the occurrence of double
pricing is very low, we believe it is fair to state that in practice the Netherlands is also applying
single marginal imbalance pricing.
A detailed analysis of the imbalance pricing schemes in both countries is revealing 4 differences:

Regulation states

Additional components

Use of iGCC

Price determination
Those 4 differences shall be explained into detail in the next sections. We shall focus whether
harmonisation is required to enable cross-border exchange of balancing energy.
Imbalance volume calculation
In both countries the imbalance volumes of a BRP are calculated per Imbalance Settlement period
of a quarter hour.
Both countries use the same components:

The net trade position of the BRP, as indicated by X-Zonal import/exports, and sales and
purchase to other BRPs, as notified by the BRP to the TSO.
68 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]

The net allocated volume of injections and withdrawals at injection and offtake points
that determine the balance perimeter of the BRP, based on metered or profiled data.

The net adjustment due to the activated balancing energy.
There’s a small difference in Belgium where the allocated volume includes an additional offtake
attributed to grid losses. Considering the fact the compensation in nature by BRPs in Belgium does
not affect the well-functioning of a cross-border balancing market, these topics shall not be treated
in the next sections when the design of a cross-border balancing market is discussed.
7.2 Differences between Netherlands and Belgium: regulation state
Difference
In Belgium the regulation states are determined based on the Net Regulation Volume. This
volume is calculated for each quarter-hour by using the difference between the sum of the energy
of all upward regulations and the sum of the volumes of all downward regulations requested by
Elia.

In case of the net regulation volume was downwards the price of the marginal downwards
activated balancing energy is used to determine the imbalance price for positive and
negative imbalances

In case of the net regulation volume was upwards the price of the marginal upwards
activated balancing energy is used to determine the imbalance price for positive and
negative imbalances.
Figure 31: Regulation states and imbalance pricing in Belgium
There has been no specific reasons way there are only 2 regulations states in Belgium. This has
always been like this and this approach is similar to other countries. Nevertheless, as explained
further on, after analysis it appears that applying three different regulations states would affect the
balancing market design philosophy of Elia in a fundamental way.
The Netherlands have three different regulations states which are based on the gross regulation
volume and the direction of the requested regulation:

When continuous going from upward regulating to less upward regulating or to downward
regulating the regulating state will be downward regulating;

When continuous going from downward regulation to less downwards regulation or to
upwards regulation the regulation state will be upwards regulation;

When this cannot be unambiguously established dual pricing ensues, using the marginal
prices of Balancing in either direction.
69 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 32: regulation states and imbalance pricing in the Netherlands
The principal reason for applying a third regulation state in the imbalance pricing structure is to
prevent overshooting of BRPs. Large incentives combined with small imbalances might lead to a
too strong reaction from market parties.
Harmonisation assessment
The report of step 1 of our pilot project concluded that the exchange of mFRR energy might lead to
more frequent dual pricing in the Netherlands even in case large imbalances are occurring. This
evolution should be avoided as dual imbalance pricing is not compatible with the balancing
philosophy that Balancing Responsible Parties should help to restore the system imbalance.
Therefore a move to 2 regulations states for imbalance settlement in the Netherlands should be
investigated.
Note that such a modification will have an impact on the financial outcome of the settlement
process and hence an indirect impact on the access tariffs.
The determination as such of the 2 regulation states is not a requirement to enable the exchange
of balancing energy. Although a harmonisation for the determination of the regulation might have a
positive impact for the functioning of the markets, TenneT and Elia have decided not to harmonise
this when setting up a cross border balancing market between Belgium and the Netherlands. The
reasons are twofold:

At first view the added value of a full harmonisation seems to be limited. Indeed the largest
share of welfare triggered in a cross border balancing is coming from the exchange of
balancing energy and reserve capacity and not from the fact of having 100% equal rules.

Taking a decision on a harmonisation between Belgium and the Netherlands in the current
stage is not efficient as on short term the risk of a revision of this decision and potential
implementation is significant. Indeed a first analysis is showing that the determination of
the regulation states in Germany and France is also different. Hence in case of extension
of the Belgium-Dutch balancing market to a third country, the harmonisation of the
regulation stated need to be re-discussed. Moreover, NC EB foresees a harmonisation of
the main features of imbalance settlement two years after the Network Code enters into
force. Potentially an EU wide decision on the determination of the regulation state shall be
taken then.
Working assumptions are:
- To have 2 regulation states (columns) for the imbalance settlement in both countries.
- The determination of the regulation state itself will not be harmonised.
70 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
7.3 Differences between Netherlands and Belgium: additional
components
Current Elia and TenneT are applying additional incentive components when calculating the
imbalance price. However these components are different.
Difference
In Belgium there are 4 different incentive components in the imbalance pricing structure, however
the components β1 and β2 are currently not active (set to zero).
Figure 33: additional components in imbalance pricing in Belgium
Only the 2 components α1 and α2, which are applicable to BRP imbalances which are aggravating
the system imbalance, are getting a value different to zero once the system imbalance is larger
than 140 MW. This is done via a quadratic formula which is considering the imbalances of the last
8 quarter hours.
With formula for α:
If ABS(System imbalance) < 140 MW
α1 = α2 = 0
If ABS(System imbalance) > 140 MW
α1 = α2 = AVG8qh (System Imbalance)²/15.000 (€/MWh)
System imbalance = ACE – NRV
The principal reason for applying this component is giving additional incentives to Balance
Responsible Parties to maintain their balance.
A difference with Netherland is that this component is only applied for the imbalance price
applicable to imbalances which are aggravating the system imbalance.
In the Netherlands a component is added to the marginal regulation price in case the quality of
the system balance has been unsatisfactory in the preceding week.
TenneT developed some criteria which are evaluated each week in order to determine whether
this K-factor needs to increase or not. Unless there is an increase compared to the previous week
the incentive component will return to 0 EUR/MWh immediately. The principal reasons for applying
this component are:
 The K component is giving additional incentives to Balance Responsible Parties to be
balanced; as such not very significant, given its small magnitude, compared to the actual
range of Imbalance Prices.
71 | P a g e
October 17, 2014

[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
When the K-component is higher than zero, the requested balancing energy better is
rewarded than energy which is not offered to TenneT but used to react on the imbalance
prices. Hence the K-component is creating an incentive for market parties to bid in
balancing energy bids; as such not very significant, given its small magnitude, compared
to the actual range of Imbalance Prices.
Harmonisation assessment
Elia and TenneT have decided not to harmonise the use of additional components in imbalance
pricing. There are 2 reasons for this:

Currently the bidding behaviour of market parties in the Netherlands and Belgium is very
different. Market Parties tend to bid more extreme prices in the Netherlands compared to
Belgium. Therefore for the Netherlands it might be acceptable to remove any additional
component for the determination of the imbalance prices. In Belgium however the price
curve of the bids is less steep. Hence if no cross zonal interconnection capacity is
available for the exchange of balancing energy between both countries the imbalance
prices in Belgium shall be only determined from the local activated bids. The additional
component ensures stronger incentives to BRPs.

As explained in chapter 8 it is possible that in particular scenarios not all balancing cost
are covered by imbalance prices. This abolition of the double imbalance pricing column in
the Netherlands (see previous section) might lead to a decrease of income out of
imbalance settlement with potential risk to be insufficient to cover all balancing costs. For
Belgium changes in the pricing schemes for the settlement of balancing energy might lead
to a decrease of the income out of settlement with as result that counteractivation costs
aren’t covered anymore by imbalance settlement.

At first view the added value of a full harmonisation seems to be limited. Indeed the largest
share of welfare triggered in a cross border balancing is coming from the exchange of
balancing energy and reserve capacity and not from the fact of having 100% equal rules.
NC EB shall allow that other costs related to balancing are covered by imbalance
settlement. For instance the local regulatory decision to include local costs for reserves in
imbalance prices can only be achieved via local additional components.
Working assumption is not to harmonise the use of additional components in the imbalance
pricing.
7.4 Differences between Netherlands and Belgium: iGCC
TenneT and Elia joined the International Grid Control Cooperation (iGCC) agreement as of 1
February 2012 and 1 October 2012, respectively. In this cooperation the imbalance of the
individual LFC Areas/Blocks are first are netted, which results in a reduced activation of balancing
energy whenever the individual system imbalances are having different signs.
Currently in both countries iGCC is considered in a different way for the determination of the
imbalance prices.
72 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Difference
In the Netherlands iGCC is currently kept outside the balancing market. More specific:

an exchange of imbalances (a reduction of control targets for aFRR) via iGCC isn’t
considered as an element of the activated Balancing Energy.

as iGCC isn’t considered in the regulation volume logically the marginal imbalance price
isn’t affected by the settlement price of iGCC.

In case no bids of Balancing Energy have been activated, the marginal price to be
considered in the Imbalance piece determination is established as the average price of the
first bids in either direction on the merit order list (Regulation state 0); this is also a single
price determination.
In Belgium however the exchanged iGCC volumes are considered in the calculation of the
imbalance prices;

The netted volumes of imbalances of iGCC are considered in the determination of the
regulation volumes;

As the iGCC volumes are consider in the regulation volume they should also have a price
to determine the marginal imbalance price. It was decided to take the weighted average
price of aFRR because:
o
The real settlement price of iGCC is only known multiple weeks after the relevant
15 min period. Currently Elia is publishing the imbalance prices 2 minutes after
each quarter hour. Another price is required if this publication is maintained;
o
The iGCC settlement price is based on opportunity costs of other TSOs; hence the
iGCC price doesn’t necessarily reflect the imbalance situation in Belgium.
Reason for the Belgian approach is that not considering the iGCC volumes in the determination of
the regulation state might give wrong incentives to market parties, as explained in the example
below.
Example: (during the same quarter hour)
120 MWh export via iGCC
10 MWh upwards aFRR
Without considering iGCC exports imbalances of BRPs would be settled at “a marginal
upwards price” and hence give a wrong signal that the zone is short.
When considering the iGCC exports imbalance of BRPs are correctly settled at “a marginal
downwards price” and hence give a correct signal that the zone is long.
Harmonisation assessment
Elia and TenneT have decided that currently it is not possible to take a common position on how
iGCC should be considered in the imbalance pricing.

Currently the maximum volume of imbalances which can be netted off with other LFC
Areas/Blocks is limited to the available volumes of aFRR. Hence in case the netted iGCC
volume suddenly is unavailable (due to cross zonal capacity or a change in imbalances in
other LFC Areas/Blocks) there is sufficient aFRR available to restore the imbalance of the
73 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
LFC Area/Block. Nevertheless due to ramping restrictions the local control area shall face
a sudden increase of the ACE.

It is foreseen as a next step to abandon the current limits and to start netting volumes
beyond the available volumes of aFRR. This would mean that if suddenly iGCC netting
stops, the local TSO will need to resolve the sudden increase of imbalances with aFRR
and slower mFRR. Hence the local control shall have for a longer period an increase of
ACE.

Another foreseen step is to start with aFRR assistance. In this solution other TSOs shall
assist TSOs who’s aFRR volumes are fully activated
Currently the operational and financial modalities of these futures changes are unknown. It is not
possible for Elia and TenneT to take any position without these elements.
There’s also a strong link to the local TSOs responsibilities as described in section 1.2. In order to
avoid a decrease of quality of the ACE Regulation some TSOs prefer that a BRP is resolving each
imbalance, even if this imbalance can be netted with other TSOs. In such a case the possibility to
have a sudden increase of the ACE when iGCC become unavailable is limited.
Working assumption will be not to harmonise the way of iGCC is considered in the
determination of the imbalance prices.
7.5 Differences between Netherlands and Belgium: price
determination
TenneT and Elia are applying currently a marginal pricing principle for the determination of the
imbalances. However also here there some differences.
Currently the Netherlands applies a cross-product marginal pricing mechanism for the settlement
of balancing energy. In other words for each 15 min period positive bids are settled at the price of
the most expensive upward deployed or dispatched bid during that 15 min period (i.e. the price for
upward adjustment). Negative bids are settled at the price of the marginal downward deployed bid
8
during that 15 min period (i.e. the price for downward adjustment) . This settlement price is used
for the determination of the imbalance tariffs.
In Belgium however – as explained in the next section – a pay-as-bid pricing mechanism is
applied for settlement of balancing energy. The marginal imbalance price is calculated by taking:

The highest (lowest) activation price used for the settlement between Elia and the BSP for
upwards (downwards) regulation bids of manual FRR (R3)

The weighted average price of activate automatic FRR (R2). The underlying reason is that
in a pro-rata mechanism the pre-selected bid with the highest price is always activated
independent of the magnitude of the total requested volume of the TSO. Hence in such a
8
If no price for upward or downward adjustment is available, the volume to be allocated to suppliers per
15 Min per direction is settled at the upward or downward adjustment price of the previous 15 Min .
74 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
case, even with a small imbalance, the bid with the highest price might set the imbalance
price all the time. In order to avoid the possibility of market abuse by one provider setting
the (high) imbalance price for all BRPs, the weighted average price of activated aFRR bids
is used when the marginal imbalance price is determined. Obviously if Elia would start with
a merit order activation of aFRR this exemption would not be applicable anymore.
Current study takes as working assumption that in Belgium aFRR bids should be activated in a
merit order way. Once implemented this would allow Belgium to calculate the imbalance price in
the same way as the Netherlands. The highest (lowest) activation price for the settlement between
Elia and the BSP for upwards (downwards) regulation bids for manual FRR and aFRR shall be
used for the determination of the marginal price applicable in the imbalance tariffs.
75 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
CHAPTER 8: Pricing and settlement - Settlement of balancing energy
8.1 Description current pricing mechanism for settlement
Currently in Belgium a pay-as-bid pricing mechanism is applied for the settlement of balancing
energy being activated upon demand of Elia. A provider is able to send in a different bid prices per
15 min; hence energy settlement is occurring per 15 min.
In the Netherlands however a cross-product marginal pricing mechanism is applied for the
settlement of balancing energy. In other words for each 15 min period positive bids are settled at
the price of the highest bid deployed or dispatched during that 15 min period (i.e. the price for
upward adjustment). Negative bids are settled at the price of the lowest bid deployed during that
9
15 min period (i.e. the price for downward adjustment) .
An important aspect when discussing the marginal pricing mechanism in the Netherlands is the
use of the manual FRR and automatic FRR. TenneT selects automatic FRR in accordance with
the merit order and activates pre-selected bids in parallel. Manual FRR is only used in case of
insufficient automatic FRR. Hence the balancing market is dominated by automatic FRR setting
almost all the time the marginal price. In practice this means that very short and temporary
deviations within a 15min period may reach out far on either side of the MOL for FRR, and hence
in a corresponding increase in the marginal price for balancing energy.
There are several reasons why TenneT applies this pricing mechanism:
 Simplicity: Current pricing mechanism is simple to apply; there are maximum 2
settlement prices for all activating balancing energy per 15 min period.
 Market functioning: The economic theory states that marginal pricing will encourage
balancing service provider to bid in at marginal cost.
In the next section we shall assess the different possibilities for harmonisation of the pricing
mechanism for the settlement of balancing energy, considering the different aspects described in
chapter 8.
Conclusions
Both TSOs perform a settlement per 15min period however there’s a fundamental
difference in the pricing mechanism used for the settlement of balancing energy
9
If no price for upward or downward adjustment is available, the volume to be allocated to suppliers per
15 Min per direction is settled at the upward or downward adjustment price of the previous 15 Min .
76 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
8.2 Options for the settlement of balancing energy
Close analysis shows that there are multiple ways to implement pricing schemes for balancing
energy. In general 3 different questions need to be answered for a final decision.
A first basic question is whether the settlement should be performed based on a pay-as-bid
scheme or a marginal pricing (pay-as-cleared) scheme. In the economic theory it is often
argued that in a competitive market with uniform product & perfect information marginal pricing is
the best solution as providers tend to offer their services at the marginal cost. However it is unclear
whether these conditions are met for all balancing markets. For balancing markets where market
power is unevenly distributed between market parties it is unclear whether it is feasible to
implement such a pricing mechanism.. The development of a cross-border market might improve
the competition in a local concentrated market, however there’s no guarantee that sufficient cross
zonal capacity shall be available at all time.
Ownership structure of Dutch generation park
Ownership structure of Belgian generation park
400
420
420
4.040
ELECTRABEL SA
1.038
12.769
SPE-LUMINUS
NUON NV
7.022
ELECTRABEL NEDERLAND
1.050
ENERGIELEVERANCIER E.ON
ENERGIELEVERANCIER E.ON
1.274
ESSENT NV
ZANDVLIET POWER NV
758
NV EPZ
MARCINELLE ENERGIE
AKZO NOBEL
T-POWER
920
Others
5.719
3.892
Others
1.920
Figure 34: ownership structure of Belgian and Dutch generation park
NC EB has a clear preference for marginal pricing, however an opening is still left for other
solutions: “No later than one year after the entry into force of this Network Code, all TSOs shall
develop a proposal for the pricing methods of each Standard Product for Balancing Energy. The
pricing methods shall be based on marginal pricing (pay-as-cleared), unless TSOs complement the
proposal with a detailed analysis demonstrating that a different pricing method is more efficient
for European-wide implementation pursuing the general objectives defined in Article 10. “
Therefore we can conclude that NC EB doesn’t provide a final answer to our first design question,
although it points in the direction of marginal pricing.
A second important question is how the cross-border exchange of balancing energy affects
the settlement price. One option could be to apply a cross-zonal pricing scheme. In such
mechanism the settlement price in one LFC Block might be set by the price of a bid activated in
another LFC Block. An alternative solution might be to apply a local pricing scheme. In this option
the settlement price in one control area might not be influenced by the prices of bids activated for
other LFC Blocks.
In the balancing market different kind of balancing product are used for different balancing
purposes and with different activation methods. A third important question is whether different
products (aFRR, mFRR) might influence each other settlement price. One option could be to
apply a cross product pricing scheme (as currently in the Netherlands) where the settlement price
of one type of product might be set by the price of a bid of another type of product. An alternative
solution would be to apply a pricing scheme per product where one type of product doesn’t
influence the settlement price of another product.
As a conclusion the possible combinations of design choices are indicated in the table below:
77 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Table 10: options for balancing energy prices in cross-border configuration
Local
Cross Zonal
Pay-as-bid
#N/A#
Marginal pricing per product
Local marginal pricing per
product
Cross-border marginal pricing per
product
Marginal pricing across
Local marginal pricing across
products
Cross-border marginal pricing across
products
Pay-as-bid pricing
products
8.3 Local pricing principle
As previously explained in a local pricing scheme the settlement price in one LFC Block is not
influenced by balancing actions taken by another LFC Block. In this section we shall explain the
functioning of such a pricing mechanism.
A first important point to consider is that the principle of local pricing has no impact on the real
time balancing process. It doesn’t affect the process of sending bids, creating merit order lists
and activation of cheapest bids. Indeed local pricing is in fact an ex-post (close to real-time)
settlement process in which activated balancing bids are allocated to the different involved TSOs
according to a predefined set of rules. Once all activated balancing energy is allocated to a TSO,
each TSO can determine the settlement price.
Figure 35: cross-border pricing and local pricing of balancing energy
The rules for the allocation of activated balancing energy are as follows:

All activated balancing is allocated to the involved TSOs.

A balancing bid can be activated by different TSOs. The Activation Optimisation Function
of the Coordinated Balancing Area shall first sum up all control targets of all TSOs and
then activate the cheapest bids in function of total control target. Hence it is possible that a
cheap indivisible bid is activated by 2 different TSOs.

The local pricing method ensures that, for settlement purposes, the local cheap balancing
bids are first allocated to the reserve connecting TSOs. This is very important in a first
come first serve exchange model for mFRR where some TSOs –due to differences in
78 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
mFRR activation strategy- might always activate the cheaper mFRR bids on the CMOL
before other TSOs (how decide upon activation of mFRR closer to real time). This might
systematically increase or decrease the imbalance settlement prices of the concerned
TSOs. The exact way how balancing resources are activated cannot be harmonised due
to the differences in system dynamics and the local TSOs responsibility.

If more bids in a LFC block were activated than requested by the reserve connecting TSO,
only the cheapest bids are allocated to the reserve connecting TSOs. The remaining more
expensive activated bids are allocated to other TSOs of the Coordinated Balancing Area in
proportional way.

The settlement price shall be determined based on the balancing energy bids which were
only used for own balancing purposes.
Note that such a pricing mechanism is not creating an economic disadvantage for BSPs. Indeed
BSPs will always receive at least the price which they would have received if their country would
not be a member of a Coordinated Balancing Area for the exchange of Balancing Energy.
8.4 Considerations on cross product & cross border pricing
There are several considerations regarding cross border pricing that might be solved by applying a
local pricing mechanism instead. A cross border pricing mechanism has a fundamental impact on
the required harmonisation of balancing markets, the future configuration of cross-border
balancing markets, the setting of the Gate Closure Times of the different Standard Products used
for different balancing processes, the control strategies of TSOs and hence indirectly the
imbalance settlement period, the balancing incentives for BRPs to be balanced and many other
aspects. Some considerations are detailed further below. A local pricing mechanism avoids difficult
questions of harmonisation and facilitates a faster implementation of cross zonal balancing
markets.
Working assumption will be local pricing for the settlement of balancing energy
Firmness of Common Merit Order lists
For balancing their control area TSOs use different kinds of products for different balancing
processes. In general we distinguish:

Automatic Frequency Restoration reserves: a power reserve which is activated
automatically by a local controller.

Manual Frequency Restoration reserves: a power reserves which is activated on request
of an operator at the control room of a TSOs.

Replacement Reserves; this is a reserve is not used by Elia and TenneT. However the
issues regarding interactions between aFRR and mFRR are also applicable for RR.
The decision for activation these reserves are taken at a different moment in time. Automatic FRR
is activated in real time for short periods to deal with sudden variations of the system imbalance
whereas manual FRR is activated for a longer period (15 minutes) to deal with longer system
imbalances.
79 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Currently the FGEB foresees a stepwise implementation of the cross border balancing market.
According to the FGEB the exchange of Replacement Reserves, the exchange of manual
Frequency Restoration reserves and the exchange of automatic Frequency Restoration Reserve
should respectively be implemented 2 years, 4 years and 6 years after NC EB enters into force.
The FGEB also allows different geographical configurations for the exchange of the different
balancing energy products.
The example below is showing the problems which will occur when activating bids from different
CMOs
Example:
Assume that Belgium and the Netherlands are having a Coordinated Balancing Area for the exchange of aFRR.
On the same time both TSOs and Germany are member of a Coordinated Balancing Area for the exchange of mFRR.
The Blue merit order list represents the merit order list of German (TSO A) mFRR bids
The Green merit order list represents the merit order list of the Belgian (TSO B) and Dutch (TSO C) mFRR bids.
Both merit order lists shall be merged into one list (3th merit order list from the left).
Figure 36: issue of cross-product pricing and firmness of CMOL
If TSO A (Germany) wants to activate mFRR bids the Activation Optimisation Function shall select for him 1 German and 3
non-German bids.
However in real time Belgium and the Netherlands are activation more expensive aFRR bids in their own Coordinated
Balancing Area. Due to cross product pricing the settlement price of the mFRR bids activated for Germany is increasing to
the price level of the activated aFRR bids in Belgium and the Netherlands.
This means that ex-post the merit order list of mFRR is looking differently. Due to the price increase of the mFRR bids it
would have been more economic efficient for Germany to activate only local bids.
Based on the example we conclude the cross product pricing can only be implemented in
combination with local pricing in case the coordinated balancing areas for mFRR and aFRR are
different. If not the firmness of the merit order list and efficient economic dispatch cannot be
guaranteed.
Different control strategies
As previously explained in this document automatic FRR reserves are activated by the local
controller per LFC Block. All TSOs are having the local TSO responsibility to resolve the imbalance
in such a way that the ACE quality targets are met. Because of different dynamics in their LFC
80 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Block the parameters of the local controller are set by the TSOs in a different way in order to react
the most optimal to a sudden variation of system imbalances.
Moreover some TSOs – like TenneT – might even activate more bids in parallel in order to have
faster reaction in case strong imbalances are suddenly occurring.
The example below illustrates potential problems in case different control strategies are applied in
combination with cross border pricing. Due to differences in control strategies it is possible that
some TSOs (artificially) increase the settlement prices of other TSOs.
Example:
Assume that TSO A activates bids in parallel to deal with sudden imbalances which are frequently occurring in his system.
Assume that TSO B doesn’t require a parallel activation because of a stable system
In case of cross border pricing, TSO A shall systematically drive up the settlement price for TSO B as he is activating more
often more expensive bids
Figure 37: impact of local TSO activation strategy on balancing energy settlement
The different dynamics of LFC Block might require TSOs to apply different control strategies for
using the same balancing energy in order to be compliant with the ACE quality targets. A
harmonisation of control strategies is not (or less) required in case a local pricing mechanism is
implemented.
Wrong local incentives to BRPs
TenneT and Elia apply single marginal imbalance pricing in combination with close-to-real time
information in order to encourage Balance Responsible Parties to support the system imbalance.
Chapter 2 showed that the shift in Belgium from a double to a single imbalance pricing scheme led
to a significant decrease of system imbalances and hence a better regulation quality at a lower
cost. This means that the imbalance settlement in both counties, in addition to recovering cost of
balancing, is also a ‘control tool’ for a TSO to balance its own system.
In case of cross-border pricing the imbalance price no longer reflects the imbalance of the LFC
Block the BRP belongs to. Hence there’s a risk that local TSO responsibility and ACE quality are
affected in a negative way.
81 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Example:
Assume that TSO A & TSO B are encouraging BRPs to help to restore the system imbalance. BRPs start to react on the
incentives given by imbalance prices once the imbalance price is larger than 150 €/MWh with +/- 100MW.
Assume that TSO A is having an imbalance of 20 MW and is activating 20 MW of bids to resolve his imbalance.
Assume that TSO B is having an imbalance of 500 MW and is activating 400 MW of bids to resolve his imbalance. Based
on the merit order list he anticipates on a reaction of 100 MW by BRPs.
Figure 38: cross-border single marginal pricing and local BRP incentives
Because TSO A did know upfront that TSO B was going to have a large imbalance. If cross border pricing is applied his
imbalance will increase above 150 €/MWh and market parties will react with 100MW although there’s locally only an
imbalance of -20 MW.
Our conclusion is that single marginal imbalance pricing with the publication of close-to-real time
information is only possible in combination with a local pricing scheme for the settlement of
balancing energy.
Extendibility of the pilot project
Local pricing ensures that the extendibility of this pilot project can also be extended to TSOs that
perform economic optimisation between merit order lists of resp. aFRR, mFRR and RR activation.
These TSOs do not activate products on the basis of technical needs, but on least cost basis.
This is not a concern for the cooperation between TenneT and Elia, as neither TSO performs any
economic optimisation between different products. The activation trigger for both products is based
on technical considerations of underlying price of products.
Nevertheless it is not excluded that one day TenneT & Elia will collaborate with a TSO performing
an economic optimisation between RR & mFRR and between aFRR & mFRR. Moreover one year
after the Network Code enters into force all TSOs should propose an unique pricing mechanism for
the settlement of balancing energy. Obviously this proposal should consider the fact that some
TSOs perform economic optimisation between different types of reserves.
The below example demonstrates that cross border pricing requires harmonisation of balancing
processes. Likely it will be impossible to perform an economic optimisation between mFRR and
82 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Replacement Reserves and between mFRR and aFRR in case of a Cross Border Pricing
mechanism.
Example:
Assume that TSO A avoids the activation of the most expensive mFRR bids
With an imbalance of 80 MW TSO A shall limit the activation of mFRR to 20 MW in order to avoid a high
settlement price. The remaining 60 MW are activated via RR
However a few seconds later TSO B faces a sudden imbalance of 80 MW which he needs to resolve directly via mFRR.
Figure 39: issue of cross-border pricing and economic optimisation of balancing energy activation
The example above shows that the economic optimisation of TSO A is not efficient due to the activation of balancing
energy of another TSO
Our conclusion is that a future collaboration with TSOs performing economic optimisation
between different kinds of reserve products is only possible if a local pricing scheme is
applied.
83 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
CHAPTER 9: Use of Cross Zonal Capacity
9.1 Availability of Cross Zonal Capacity
The possibilities for cross border balancing between Belgium and the Netherlands are limited by
the available capacity between the two bidding zones. In order to have a functioning system for
cross border balancing, it needs to be determined how the available capacity between bidding
zones, like Belgium and the Netherlands, is distributed between the different processes influencing
actual cross border flows. It is interesting to note that this is one of the ways in which our pilot
project differs from the situation in Germany, as the exchange of balancing energy between the
LFC blocks there all takes place within the same bidding zone and is not limited by capacity
considerations outside of internal congestion management.
There are three balancing processes that are in scope for this chapter, as they will peruse cross
zonal capacity, and will influence or be influenced by each other. These three processes are:

the intended exchange of energy as a result of mFRR;

the intended exchange of energy as a result of aFRR; and

the intended exchange of energy as a result of operating the imbalance netting process.
The cross zonal capacity that remains after the intraday
cross zonal gate closure time, calculated in line with NC
CACM, is available to be used for the cross zonal
balancing purposes. The influence between intraday and
balancing capacity is one-sided; there is no influence of
cross border balancing on the capacity available for the
intraday commodity market. The current scope of the
pilot project doesn’t consider reservation of cross zonal
capacity for the exchange of FRR balancing energy.
After the intraday cross zonal gate closure time, there is
capacity available for cross zonal balancing processes in
each direction in an interval [0,X] MW, where X is positive
or
zero.
X
is
determined
by
X=ATC-∑(External
Commercial Trade Schedules), or an equivalent when Figure 40: Available Capacity X
flow-based
capacity
calculation
is
used.
This
is
schematically shown in Figure 40, and means that the combined intended exchange between
TSOs in a direction, meaning the sum of the calculated capacity need of imbalance netting, and
exchange of balancing energy from aFRR and mFRR, ∑(mFRR, aFRR, Imbalance Netting), is
limited to X MW. In case X is small in one of the directions, there may not be sufficient capacity for
the TSOs on each side of the border to use the three processes as they would like.
This chapter will determine how to deal with limited capacity in such a situation and how the
capacity should be allocated to the different processes. It will describe the capacity needs of the
different processes based on their specific attributes:
84 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]

mFRR: scheduled, predetermined allocation, no legacy

aFRR: real time allocation, legacy

imbalance netting: real time allocation, no legacy
Here, the legacy of aFRR means that exchange of balancing energy from aFRR may result in the
exchange of a signal over a virtual tie-line in the imbalance settlement period following the
imbalance settlement period of activation. Pricing and settlement arrangements in both countries
(see chapters 6, 7 and 8) will determine the potential value of these processes and hence the
allocation of capacity to these processes.
9.2 What is congestion of the border?
Cross border congestion management is aimed at not violating ex-ante the n-1 security criterion by
making a comparison of available cross zonal capacity and administrative use of this capacity. The
capacity criterion is calculated on the bidding zone border using either an ATC of a flow-based
methodology. The capacity is determined in advance for each market period, and is therefore
known for each imbalance settlement period. Only the remaining capacity after the intraday gate
closure is important for our purposes, at which point the market has had full opportunity to use the
capacity for their own purposes.
It is important to note that the ATC value for the NL-BE border is an administrative value that does
not relate directly to the physical flow over this border; in a meshed grid, there are for example
physical unscheduled flows such as loop flows. Continuous intraday flow-based capacity
calculations will improve visibility of the congestion.
9.3 Capacity needs of different processes
There are three processes of which the cross zonal capacity requirements will be detailed in this
chapter: exchange of mFRR, exchange of aFRR and imbalance netting. Each of these three
processes carries its own specific timelines, which makes allocating cross zonal capacity to these
processes
more
than
a
trivial
matter.
Capacity requirements of mFRR
When exchanging balancing energy
from mFRR, the cross zonal mFRR
product is a fixed virtual tie-line
profile corresponding to a cross
zonal schedule change for a single
imbalance
settlement
period
of
delivery, as is explained in chapter
5. This firm power profile (positive Figure 41: mFRR capacity use
or negative) will be deducted from
the ACE_OL of the reserve requesting TSO and added to the ACE_OL of the reserve connecting
TSO. It can only be activated in the imbalance settlement period prior to the imbalance settlement
period of delivery.
85 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Therefore, when using cross zonal capacity for the exchange of mFRR, this capacity will be
required to be available for the full duration of the next imbalance settlement period, as shown
schematically in Figure 41. That means that all activated cross zonal mFRR will offset the available
capacity during the next imbalance settlement period by exactly the activated amount for exactly
the duration of the imbalance settlement period. This is irrevocable, and can only be nullified either
by counter-activation within the connecting LFC Block, or activation of a bid in the same regulation
direction in the receiving LFC Block. When insufficient capacity is available for full activation of a
bid, divisible bids could still be partially activated.
When balancing energy from mFRR is exchanged, this will alter the available capacity for the full
duration of the next settlement period of delivery, increasing it in one direction, and limiting it in the
other. Exchange of balancing energy from mFRR therefore affects the exchange of balancing
energy from aFRR and imbalance netting.
Capacity requirements of aFRR
The product exchanged for aFRR is detailed in chapter 4. When exchanging balancing energy
from aFRR across the border, this is done through a continuous signal over a virtual tie-line: the
exchange of a control request. This signal takes into account ramp rates of bids, both in the
imbalance settlement period in which a bid is active, and in imbalance settlement periods in which
there is a legacy of a previously active bid ramping to 0 MW again; dummy energy is exchanged
as a power profile over the virtual tie-line. During the imbalance settlement period of activation, the
resultant virtual tie-line signal, corresponding to the control request sent to the balancing service
providers will have to remain within the capacity limits, as is shown schematically in Figure 42 for a
single aFRR bid.
Imbalance settlement period
aFRR capacity use
Figure 42: aFRR capacity use
Dummy energy
The legacy of aFRR in the next imbalance settlement period is the dummy energy, corresponding
to the area underneath the red line in Figure 42: aFRR capacity useFigure 42, is an aspect which
is unique to aFRR. mFRR is a block product corresponding to a single imbalance settlement
86 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
period. Imbalance netting is instantaneous. The capacity requirements of aFRR depend on the
way we deal with the dummy energy.
There are several options:
1. Cross border aFRR bids will be ramped down prematurely in order to take into account
limited capacity in the next imbalance settlement period.
2. Cross border aFRR bids will not be activated unless there is sufficient capacity for full
activation in the next imbalance settlement period.
3. Limited capacity in the next imbalance settlement period will not be taken into account;
dummy energy is not required to stay within capacity limits.
The chosen answer can have significant effects on the usefulness of aFRR exchange, because
current ramp rates are at least half an imbalance settlement period for full activation or
deactivation. Both the first and second options may significantly limit the benefits of cross border
exchange of balancing energy from aFRR, because fewer opportunities for the exchange may
present themselves. This is not the case for the third option, because if limited capacity in the next
imbalance settlement period will not be taken into account, it cannot limit the exchange of
balancing energy in the current. Therefore the third option presents the largest amount of
opportunities for cross border exchange of balancing energy.
Aside from this advantage, the third option has the added benefit of being the simplest option to
implement. The main disadvantage of this option over the others is that energy is exchanged over
a virtual tie-line for which capacity availability is ignored. This can be justified, because, though
explicit, this exchange is directly comparable to the implicit exchange of energy within the control
program: there, a ramping period is used to prevent large jumps at the transition point between two
imbalance settlement periods from unnecessarily activating local reserves, and this implicit
exchange does not take into account available capacity. The approximation of the physical reality
by virtual tie-line for the exchange of mFRR carries similar uncertainties. Furthermore, partial
activation of multiple bids is likely, which will be ramped down a lot quicker, even if done exactly in
accordance with the signal, than fully activated bids.
Based on these considerations, the preferred option for dealing with limited capacity in the next
imbalance settlement period is not to require capacity to be available for the exchange of dummy
energy over the virtual tie-line. This means that when allocating capacity to the different processes,
dummy energy is not taken into account, but is treated as any other uncertainty.
Control targets and control requests
Signals for aFRR activation are sent every four to ten seconds, based on the data that is available
at that time. When activating aFRR across the bidding zone border, the available capacity should
be part of that data, ensuring that the control request – the signal sent for activation across the
border – remains within the capacity constraints.
When exchanging aFRR across the border, the capacity requirements are dependent on the time
of activation, the ramping rate of the aFRR bid(s) to be activated, the volume of the aFRR bids to
be activated, the time of deactivation and the time at which the last ISP of activation ends.
87 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
The capacity requirement for a single aFRR bid is then represented by the shaded area in Figure
42.
In practice, aFRR activation is a continuous process and the capacity requirements will not follow
the simplified example of a single bid. The following holds:

ACE
= sum(trade schedules)-sum(tie-lines)+kdF
=sum(trade schedules)-sum(virtual+metered tie-lines)+kdF
While the schedules are administrative reference values, the inputs from the virtual tie-lines, the
metered values, and the frequency are dynamic. The control target for aFRR is calculated each
interval based on the FRCE, and will therefore be dynamic as well. Exchange of aFRR will lead to
an additional virtual tie line value representing the exchanged control request., It suffices to focus
on the control request of aFRR exchanged to recalculate the remaining cross zonal capacity
available for imbalance netting.
A straightforward method of checking whether aFRR bids can be activated is to check solely
whether the control target of all activated bids remains within the capacity limits..
Advantages of checking the control request against capacity constraints:

In practice, multiple bids will be activated simultaneously and bids may be activated
partially. Using the control request is in line with actual activation.

When control requests are used, it reduces the risk of bypassing bids on the CMOL.
Advantages of checking the control target against capacity constraints:

Technical implementation is easier, because capacity only needs to be checked at
activation of a bid, rather than adding another step in the calculation every AGC cycle;

This option adheres more closely to the standard CMOL approach, and is therefore more
likely to avoid difficulties for pricing and settlement.
When technical implementation is possible, and when issues with pricing and settlement can be
solved, the option of checking the control request against the capacity constraints is preferred as
the optimal solution. A consequence of using the control request to assess available capacity is
partial activation of cross-border aFRR bids due to capacity constraints. The remaining part of the
control target will have to be rerouted to the local MOL.
Since during a single imbalance settlement period the available capacity after intraday and mFRR
activation is represented by a constant, and the power profile resulting from dummy energy is not
required to stay within capacity limits during the next ISP, abiding by capacity requirements during
one interval, means it is possible to abide by them during the next.
During the imbalance settlement period of activation, aFRR bids affect remaining capacity
available for imbalance netting by a decrease in one, and increase in the other direction. The
remaining capacity available for imbalance netting will be based on the control request signal
88 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
exchanged over the virtual tie-line, excluding the dummy energy, even if the control target is used
for determining whether capacity is available for aFRR.
Capacity requirements of imbalance netting
Imbalance netting is unique in relation to Cross Zonal Capacity in that it can be fully activated and
deactivated at any point in real-time, de facto every four seconds., Available capacity will
automatically limit imbalance netting; all available capacity can be used. The actual imbalance
netting can never be an obstacle for subsequent cross-border activation of aFRR or mFRR.
9.4 Allocating capacity to different processes
All three processes make use of the same available cross zonal capacity in both directions. The
sum of the external commercial trade schedules and virtual tie-line values, excepting dummy
energy as explained above, should remain within the interval between the congestion limits in
either direction.
The virtual tie-line values for cross-border balancing are set by:
1. imbalance netting: real-time variable value, no ramping limitations
2. cross-border activation of aFRR: real-time variable value, ramping limitations in exchange
signal, existence of dummy energy
3. cross-border activation of mFRR: predefined fixed profile
When capacity is limited, different processes compete for the available capacity. Therefore, the
available capacity needs to be optimally allocated to the different processes, and updated. The
objectives are:
1. compliance with the frequency restoration target parameters;
2. cost-effectiveness; and
3. preventing counteractivation within and across each LFC area/block, in order to avoid
inefficient use of flexibility.
Exchange of imbalance netting is a direct cost saving balancing solution. Therefore, in theory,
available capacity should be allocated to imbalance netting before being used for the other
processes.
Unlike
the
other
processes,
imbalance
netting
reduces
counteractivations.
Furthermore, exchange of aFRR is less likely to lead to counteractivation than exchange of mFRR.
Therefore, in theory, available cross zonal capacity should be allocated to exchange of aFRR
before being used for the exchange of balancing energy from mFRR.
The effects of cross border activation of mFRR on both cross border activation of aFRR and
imbalance netting are unclear. On the one hand, it could limit capacity available for the other
processes, thereby preventing processes for which there is a technical preference. On the other
hand, in practice it could also lead to more capacity being available for those processes, allowing
them to take place where otherwise they could not. Similarly, cross border activation of aFRR
could be both detrimental or beneficial for the availability of capacity for imbalance netting.
89 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
In case cross zonal activation of mFRR (or aFRR) limits subsequent capacity available for aFRR
or imbalance netting, this could lead to counteractivation and higher costs. This priority of capacity
use suggests that some cross zonal capacity should be withheld from mFRR to be used for aFRR
and imbalance netting, and similarly, some cross zonal capacity should be reserved specifically for
imbalance netting.
However, depending on pricing arrangements in both countries the inability to activate mFRR
cross border because of limited capacity may lead to higher costs, especially when it turns out the
capacity was not needed for the other processes. The same holds for withholding capacity from
aFRR for possible imbalance netting. Depending on the chosen amount of capacity to be reserved
for the different processes, this could mean reserving capacity turns out to be more expensive than
not reserving capacity.
When allocating capacity between the three processes, it is therefore safe to conclude that a
straightforward choice of allocating capacity would lead to imbalance netting being preferred above
aFRR, and aFRR being preferred above mFRR. However, a straightforward choice is not possible,
and further investigation is necessary to know if, in practice, this requires capacity to be withheld
from mFRR for subsequent activation of aFRR or imbalance netting, or from aFRR for subsequent
activation of imbalance netting in order to obtain the most optimal solution.
9.5 Cross zonal capacity reservations
As explained above, there is a preferred ordering of allotting capacity, where if a choice would be
possible, imbalance netting would be preferred above activation of aFRR, and the latter would be
preferred above activation of mFRR, and in practice it may turn out that cross-border exchange of
mFRR is unnecessarily limiting towards the other cross border processes. For this reason, it is
prudent to include the option of reserving capacity for both aFRR and for imbalance netting when
implementing the cross border exchange of balancing energy from frequency restoration reserves.
Here it is assumed that the coordinated balancing area only consists of Belgium and the
Netherlands, and we are only discussing capacity reservations on the Dutch-Belgian border for the
purposes of exchanging balancing energy between the Elia and TenneT TSO BV. When cross
zonal capacity will become necessary for exchanges of balancing energy within a different
coordinated balancing area, this requires its own analysis. Such an option to reserve capacity
should be done with the understanding that the amount reserved for both aFRR and imbalance
netting will, at least at first, be set to zero. This is a pragmatic approach to have the option to
reserve capacity in place, but to only make use of it if and when it should become necessary. A
mechanism to change the amount of reserved capacity is required as well.
In order to determine whether or not capacity shall be reserved in the future, the effects of the
exchange of mFRR on the availability of capacity for aFRR and imbalance netting, and the effects
of the exchange of aFRR on the availability of capacity for imbalance netting shall be monitored.
90 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
9.6 Cross-border balancing processes in practice
The practical aspects of activating, and deactivating cross-border bids in relation to subsequent
capacity use, can be summarised per product as follows:
mFRR

active decision by requesting TSO, activation optimisation module

cannot be deactivated

firm virtual tie-line block signal corresponding to ISP, no legacy across ISP boundary

available capacity for the entire ISP needs to be assessed at activation, singular decision

partial activation of divisible bids due to capacity limits allowed
aFRR

active decision by requesting TSO, activation optimisation module

can be deactivated but with ramping down signal, energy legacy converted to power
legacy

the virtual tie-line signal based on the control request, excluding dummy power from
deactivated bids, will need to stay within capacity limits during the entire ISP of activation

dummy power will be calculated based on the control request sent to BSPs for deactivated
bids, which is based on the ramp rates corresponding to the bids. It could be exchanged
over a separate virtual tie-line not taken into account when checking capacity limits, or
included in the capacity calculation as a separate input to offset the virtual tie-line value.

possibility of checking either control target or control request against capacity limits.

partial activation due to capacity limits allowed when comparing control request rather than
control target to capacity requirements.
imbalance netting

passive

automatically limited by remaining capacity

instantaneous virtual tie-line value, no legacy

no decision on activation necessary, instantaneous deactivation possible
After possible activation of mFRR and aFRR, the remaining capacity will automatically limit
imbalance netting. To determine the available capacity for iGCC, the control request for aFRR will
be used, excluding dummy energy, even if it is decided to use the control target to determine
whether capacity is available for aFRR itself.
This means that aFRR dummy energy is not taken into account when determining available
capacity for aFRR, and also not when determining available capacity for imbalance netting.
9.7 Pricing and Imbalance Settlement in situations with congestion
In the above considerations on capacity use, it has implicitly been assumed that whenever
capacity is available, it can be used for exchange of balancing energy, even if its unavailability has
prevented cross-border activation of bids earlier for the same imbalance settlement period. For
example, cross-border activation by Elia of an upwards mFRR bid located in the Netherlands could
91 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
not be possible for a given ISP, because capacity is limited, but subsequently they are able to
activate an upwards aFRR bid located in the Netherlands for that same ISP that requires less
capacity or for which capacity has been reserved. This assures an optimal use of cross-border
capacity and gives the most freedom to BSPs in selling their products.
A consequence of this assumption is that in certain situations a TSO may activate a local bid at a
higher price early in the imbalance settlement period, but activate a cross-border bid on the CMOL
at a lower price later in the same imbalance settlement period. Furthermore, while capacity
considerations avoid the TSOs from activating cross-border bids in one direction, they are at the
same time still able to activate a cross-border bid in the other direction. Pricing and settlement
arrangements in both countries need to be such that they allow this practice of activation.
92 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Conclusions
In this second step of the pilot project between Elia and TenneT TSO BV it has been analysed
what is necessary in order to make cross-border balancing possible between two separate LFC
blocks and bidding zones applying a reactive balancing concept. The differences between the
Dutch and Belgian system have also been taken into consideration. Integrating the two balancing
markets remains a complex affair. A number of working assumptions have been formulated which
will be examined further during the third step of the pilot project: the cost-benefit analysis.
The working assumptions include, first of all, details on the balancing energy products to be
exchanged, and on how to exchange them. Virtual tie-lines will be used for the exchange of
balancing energy from FRR, bids will be transferred by TSOs to the CMOL in a firm way, and the
requested volumes of activation may exceed the volumes transferred by individual TSOs. In case
of mFRR, the product will be a block energy product with the duration of one ISP exchanged over
a virtual tie-line as a constant power profile. Only non-contracted mFRR will be exchanged. In case
of aFRR, the working assumption is to harmonise the ramp rates to a common value of 7.5
minutes. The exchange of balancing energy from aFRR shall be based on the exchange of the
control signal, which contrasts with the solution chosen within the German LFC Block, where the
control target is used. Such a solution is not feasible for exchange between LFC Blocks due to
local TSO responsibilities.
Another important working assumption is related to the activation strategy of mFRR. It has been
agreed that activation of mFRR shall be based on the degree of saturation of activation of aFRR,
which means that there shall only be technical and no economic considerations for the choice
between activation of aFRR and mFRR.
In order to exchange balancing energy, sufficient cross zonal capacity needs to be available. The
working assumption is that only the capacity remaining after the intraday market will be used. This
means that the market will be able to make full use of the cross zonal capacity for their own
purposes and will not be limited by the exchange of balancing energy by the TSO. The remaining
cross zonal capacity will need to be efficiently allocated between different processes: exchange of
aFRR and mFRR, and imbalance netting.
Within a reactive balancing market, there is a lot of emphasis on pricing and settlement
arrangements. Within this pilot project, the assumption is that both TSOs will apply a single pricing
mechanism with two regulation states. For settlement of balancing energy, local pricing
arrangements are assumed.
All working assumptions will serve as the basis for the cost-benefit analysis in step 3. However,
alternative solutions remain under consideration, dependent on the results of the cost-benefit
analysis.
93 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
References
1. Qualitative Analysis of Cross-Border Exchange of Balancing Energy and Operational
Reserves between Netherlands and Belgium
(http://www.tennet.eu/nl/fileadmin/downloads/About_Tennet/Publications/Technical_Public
ations/CF13VKH945.pdf)
2. E-Bridge consulting and UMS group, White paper on a sustainable design of the electricity
market, Challenges from Renewable Energies and Guidelines for a Sustainable Market
Design
(http://www.tennet.eu/nl/fileadmin/downloads/News/White_Paper_on_a_Sustainable_Mark
et_Design__1_.pdf)
3. ACER, Framework Guidelines on Electricity Balancing (FGEB)
(http://www.acer.europa.eu/Official_documents/Acts_of_the_Agency/Framework_Guidelin
es/Framework%20Guidelines/Framework%20Guidelines%20on%20Electricity%20Balanci
ng.pdf)
4. ENTSO-E, Network Code on Electricity Balancing (NC EB, version 3.0 of 6 August 2014)
(https://www.entsoe.eu/Documents/Network%20codes%20documents/NC%20EB/140806
_NCEB_Resubmission_to_ACER_v.03.PDF)
5. ENTSO-E, Network Code on Load Frequency Control and Reserves (NC LFCR, version of
28 June 2013)
(https://www.entsoe.eu/fileadmin/user_upload/_library/resources/LCFR/130628NC_LFCR-Issue1.pdf)
94 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
List of abbreviations
ACE
Balancing Comparison
BRP
BSP
(C)MOL
CoBA
FGEB
FRCE
FRR
ISP
LFC
NC EB
NC LFCR
RR
Area Control Error
Report on the first step study of the pilot project, "Qualitative Analysis of
Cross-Border Exchange of Balancing Energy and Operational Reserves
between Netherlands and Belgium"
Balance Responsible Party
Balance Service Provider
(Common) Merit Order List
Coordinated Balancing Area
Framework Guidelines on Electricity Balancing
Frequency Restoration Control Error
Frequency Restoration Reserves
Imbalance Settlement Period
Load-Frequency Control
Network Code on Electricity Balancing
Network Code on Load-Frequency Control and Reserves
Replacement Reserves
95 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annexes
Annex 1: FRR dimensioning rules
The FRR dimensioning rules (applicable on the level of the LFC Block) [NC LFCR art 46] stipulate
that:

The FRR Capacity must cover the Dimensioning Incident (positive and negative);

The FRR Capacity shall be sufficient to cover at least 99% of the respective positive and
negative expected LFC Block Imbalances (analysis based on historical LFC Block
Imbalances, anticipated changes in LFC Block Imbalances and availability of FRR
Capacity); and

The TSOs of an LFC Block shall define the split between aFRR and mFRR Capacity, as
well as the aFRR and mFRR Full Activation Time, with the goal of respecting the
Frequency Restoration Control Target Parameters defined in NC LFCR.
96 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 2: IGCC
IGCC settlement
All energy exchanged due to IGCC is settled against one price for both directions, the volume
weighted average of all volumes exchanged and valued by each TSO at opportunity cost:
REGULAR REPORT ON SOCIAL WELFARE
PAGE 4
05/10/13
IGCC-Settlement – Basic Principle
Opportunity Price
for Imbalance
Netting
without IGCC
with IGCC
SCEbefore IGCC [MWh]
x
SCE pricebefore IGCC
[€/MWh]
IGCC exchange
–
Opportunity Price =
Opportunity Value/IGCC Volume
SCEafter IGCC [MWh]
x
SCE priceafter IGCC [€/MWh]
IGCC Settlement Price (
[(SCEbefore IGCC * SCE pricebefore IGCC) (SCEafter IGCC * SCE priceafter IGCC) ] / IGCC
exchange
): Energy weighted (
average of the opportunity prices (
IGCC Settlement
Price
–
–
Value of Avoided
Activations
and
and
)
)
Single price for all IGCC exchanges
Value of avoided activations for a participant is driven by the
spread between the opportunity price and the IGCC settlement
price
Figure 43: basic principles of iGCC settlement
Improved ACE quality
IGCC has improved ACE quality in the Netherlands and Belgium.
100
Impact IGGC on Control Quality TenneT-NL
Net Market Error
80
Balancing Energy + Exchanged Control Error
|GWh|/month
Actual remaining Error ENTSO-E CE
60
Theoretical remaining Error ENTSO-E CE
40
20
0
Figure 44: iGCC improved the ACE quality in both NL and BE
97 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 3.1: Changes to Table 3 of Balancing Comparison
Table 3 of the Balancing Comparison must be changed to:
Table 11: update of Table 3 of Balancing Comparison
Belgium
Procurement of capacity
Annual tender (part of volume)
Monthly tender (part of volume)
The Netherlands
Annual tender
Availability requirement
100%
Products
Volumes differentiated by time Constant volume over
(peak, off-peak)
to be bid as R2 on MOL
Provision
Portfolio-based
Remuneration
Pay-as-bid for reserve ( MW)
time,
Source: DNV KEMA
98 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 3.2 Functioning of a secondary controller
This annex provides a more detailed overview of the functioning of a secondary controller.
Figure 45: overview of modules of secondary controller
First of all there is the input module gathering all inputs required to calculate the Frequency
Restoration Control Error (FRCE), also known as the Area Control Error (ACE) of the LFC Block.
Figure 46: input module of secondary controller
The main inputs for this module are:

Frequency set-point (standard 50 Hz), frequency measurement and K-factor of the LFC
Area for the calculation of the reaction of the LFC Area on frequency deviations in the
Synchronous Area;

Measured tie-line flows, scheduled exchange with neighbouring LFC Areas and Virtual
Tie-Line flows (e.g. due to imbalance netting) for the non-scheduled agreed exchanges
with other TSOs to calculate the difference between the intended exchange and actual
realized exchange with neighbouring TSOs.
On the basis of this data the FRCE (ACE), which acts as input for the secondary controller, is
calculated.
In a next stage the FRCE is processed by the AGC controller itself to calculate the aFRR control
target. The aFRR control target represents the required amount of aFRR balancing energy
activation without taking the limited ramp rate of the aFRR bids into account.
99 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 47: AGC controller
The AGC controller is a proportional-integral controller. The PI-controller is often complemented
with additional logic such as:

Filters: most TSOs apply a filter on the FRCE at the input side of the PI-controller.
Common examples are:
o Low-pass filter: the filter reduces the noise and tempers very fast deviations of
the FRCE to reduce unnecessary fluctuations in the required amount of aFRR
o
activation;
Quadratic filters: the filter avoids unnecessary activation of aFRR balancing
energy in the case of small FRCE (mostly applied in merit order aFRR schemes);

Additional logic:
o
The value of the integrator part is often limited (anti-windup logic) to the available
aFRR capacity. This avoids that, in case of a persisting FRCE, the integrator value
becomes very large thereby creating a large overshoot if the FRCE reduces or
changes sign;
o
Manipulation of the integrator value in case the FRCE changes sign (e.g. reset or
restriction of the value) to reduce the activation of aFRR balancing energy in the
wrong direction;
o
Reduce the volume of counter-activation between mFRR and aFRR balancing
energy;
o
…
The output of the PI-controller defines the aFRR control target.
The aFRR control target doesn’t consider the finite ramp rate of the aFRR bids. The bid activation
module determines the actual activation request for the aFRR bids while taking into account their
limited ramp rates. This is called the aFRR control request. The aFRR control request therefore
represents the physical aFRR activation request for an aFRR bid.
100 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 48: bid activation module
This requires a translation of the aFRR control target to the aFRR control request (= aFRR
activation/setpoint that has to be delivered by each individual bid and/or BSP). This requires the
following inputs:

aFRR submitted bids: capacity [MW], ramp rate [MW or % /min], aFRR energy price per
bid [€/MWh];

aFRR control request in previous AGC cycle per bid or per BSP [MW].
The bid activation module either functions in a pro-rata way or a merit order way to determine the
aFRR control request.
Finally the communication module sends the aFRR activation request to the BSP.
Figure 49: communication module
This module is highly specific per TSO:

Separate aFRR control request for each bid or aggregated control request for all the bids
of a BSP;

Separate aFRR control request for upward and downward aFRR activation or not;

Relative control request [0 to +/-100%] or absolute control request [MW];

…
101 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 3.3: Examples pro-rata and merit order activation
This annex provides a numerical example for a pro-rata and merit order aFRR activation scheme.
For the pro-rata scheme:
Table 12: pro-rata scheme numerical example
Bid
nr.
BSP
1 BSP A
2 BSP B
3 BSP A
4 BSP A
5 BSP A
6 BSP B
7 BSP B
8 BSP A
9 BSP B
10 BSP A
Max. ramping
aFRR control
Bid size Bid price per AGC cycle aFRR control
request(t-1)
[MW] [€/MWh] [MW]
taget(t) [MW] [MW]
5
50
0,7
2,7
10
51
1,3
5,5
9
53
1,2
4,9
8
54
1,1
4,4
7
60
0,9
3,8
12
61
1,6
6,6
5
63
0,7
2,7
20
65
2,7
10,9
13
70
1,7
7,1
8
75
1,1
4,4
Total
53
aFRR control
request(t) [MW]
4
6
3
2,5
2
2,3
2,1
1,9
0
0
3,3
4,7
4,2
3,6
2,9
3,9
2,7
4,6
1,7
1,1
32,7
For the merit order scheme:
Table 13: merit order scheme numerical example
Max. ramping
aFRR control
Bid
Bid size Bid price per AGC cycle aFRR control
request(t-1)
aFRR control
nr.
BSP
[MW] [€/MWh] [MW]
taget(t) [MW] [MW]
request(t) [MW]
1 BSP A
5
50
0,7
5
4
4,7
2 BSP B
10
51
1,3
10
6
7,3
3 BSP A
9
53
1,2
9
3
4,2
4 BSP A
8
54
1,1
8
2,5
3,6
5 BSP A
7
60
0,9
7
2
2,9
6 BSP B
12
61
1,6
12
2,3
3,9
7 BSP B
5
63
0,7
2
2,1
1,4
8 BSP A
20
65
2,7
0
1,9
0,0
9 BSP B
13
70
1,7
0
0
0,0
10 BSP A
8
75
1,1
0
0
0,0
Total
100
53
23,8
28,0
It can be concluded that in this case the ramp rate of the pro-rata scheme outperforms the ramp
rate of the merit order scheme. In both schemes the aFRR control target is 53 MW. The pro-rata
scheme effectively results in the activation of 32,7 MW whereas the merit order scheme only
results in the activation of 28 MW.
This shows that a pro-rata scheme outperforms a merit order scheme in terms of ramp rate for
cases where not all bids are activated in the merit order scheme (under the assumption of identical
aFRR energy bids in both systems).
102 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 4.1: aFRR Full Activation Time, aFRR dimensioning, aFRR
activation scheme and FRCE (ACE) regulation quality
A. Ramp rates in aFRR pro-rata and merit order schemes
In a pro-rata scheme all participating aFRR bids are always activated in parallel, regardless of the
amount of required aFRR activation. As a result the maximum ramp rate of a pro-rata aFRR
scheme is constant over the entire aFRR activation range and equals:
𝑎𝐹
𝑟𝑎𝑚𝑝 𝑟𝑎𝑡𝑒
𝑇𝑜𝑡𝑎𝑙 𝑎𝐹
𝑣𝑜𝑙𝑢𝑚𝑒
𝐹𝑢𝑙𝑙 𝑎𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒
Where:

Total aFRR volume is the volume of all positive or negative participating aFRR energy
bids; and

Full Activation Time is the maximum time in which the bid must be able to be fully
activated (imposed by the TSO).
It can be concluded that, for a pro-rata scheme, the maximum ramp rate can be increased by
either increasing the aFRR volume or by decreasing the Full activation time of the aFRR bids.
In a merit order scheme the number of activated bids, and hence the aFRR ramp rate, depends on
the required aFRR activation. In case of a small activation request for aFRR energy only a limited
number of bids (and capacity [MW]) are activated. This results in a limited ramp rate. In case of a
large activation request of aFRR capacity more aFRR bids are activated, resulting in a higher ramp
rate. The aFRR ramp rate for a merit order aFRR scheme therefore is:
𝑎𝐹
𝑟𝑎𝑚𝑝 𝑟𝑎𝑡𝑒
𝐴𝑐𝑡𝑖𝑣𝑎𝑡𝑒𝑑 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑜𝑓 𝑎𝐹
𝑏𝑖𝑑𝑠 [𝑀𝑊]
𝐹𝑢𝑙𝑙 𝑎𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒
Where:

Activated capacity of aFRR bids [MW] is the total volume of the bids that are activated by
the merit order scheme (~aFRR activation request); and

Full activation time is the maximum time in which the bid must be able to be fully activated
(imposed by the TSO).
The aFRR ramp rate of a merit order scheme can therefore be increased by:

For small imbalances: decreasing the Full activation time or increasing the activated
capacity of aFRR bids [MW] for small imbalances. In such case the merit order scheme is
operated more than a pro-rata scheme. It can be concluded that in such case a trade-off is
made between economic efficiency and regulation quality;

For (very) large imbalances: decreasing the Full activation time of the bids.
The graph below gives a qualitative explanation of the difference in ramp rates between a pro-rata
and merit order scheme (for the same total aFRR volume and Full activation time):
103 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Figure 50: aFRR activation ramp rate for pro-rata and merit order aFRR schemes
It must be noted however that the above graph is not fully correct for the merit order scheme as:

in fact the aFRR ramp rate in such a scheme depends on the volume of the activated
aFRR bids that aren’t fully delivering their required energy (saturated) yet;

The activated capacity of aFRR bids [MW] is not fully linear with the activation request
(BSPs are free to define the size of the aFRR bids they submit).
Some TSOs operating a merit order aFRR scheme included additional logic to allow for faster
activation of aFRR bids. TenneT e.g. switches to parallel activation of all bids in case of a very
steep increase of the ACE_ol over several consecutive AGC cycles is observed.
B. aFRR pro-rata and merit order schemes and FRCE (ACE) regulation quality
The difference in ramp rate performance between a pro-rata and merit order scheme has also an
effect on the. The graphs below [source: DNV KEMA] show the difference in FRCE regulation
quality between pro-rata and merit order schemes. They represent the distribution of the FRCE
(red: merit order scheme; black: pro-rata scheme) in case of the same aFRR Capacity and aFRR
Full Activation Time. It is clear that under such circumstances the regulation quality of the pro-rata
scheme is better.
Figure 51: FRCE (ACE) regulation quality for pro-rata and merit order schemes [source: DNV KEMA]
The FRCE Quality Target Parameters -to be respected by the TSOs of the LFC Block- of the NC
LFCR are:
104 | P a g e
October 17, 2014

[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Level 1 Range: the TSOs must ensure that the absolute value of the ACE per PTU of 15
minutes doesn’t exceed this value for more than 30% of time. In fact this target focuses
more or less on a good regulation quality for relatively small imbalances;

Level 2 Range: the TSOs must ensure that the absolute value of the ACE per PTU of 15
minutes doesn’t exceed this value for more than 5% of time. In fact this target focuses
more or less on a good regulation quality for larger imbalances.
This pilot project considers the exchange of aFRR between merit order aFRR schemes. As
concluded above the regulation quality for small imbalances in such case can only be increased by
either decreasing the full activation time of the aFRR bids or activating more bids in parallel for
small imbalances. For a pro-rata scheme the regulation quality for both small and large imbalances
increases in case of larger volumes.
The above conclusions were confirmed by simulations performed by Elia to estimate the impact on
the FRCE Quality Target Parameters for the case of a transition of a pro-rata towards a merit order
aFRR scheme. This is shown in the table below:
Table 14: simulated FRCE (ACE) regulation quality for the Belgian system for pro-rata and merit order
aFRR schemes
Pro-rata
Merit order
aFRR volume
140
140
140
240
aFRR full activation time
13,3%
20%
13,3%
6,7%
Level 1 range
100%
93%
113%
135%
92%
109%
109%
(baseline)
Level 2 range
100%
(baseline)
This shows that:

A transition from a pro-rata scheme towards a merit order system in case of 140 MW of
contracted aFRR capacity and a 13,3% per minute ramp rate decreases the FRCE
regulation quality;

By increasing the aFRR ramp rate requirement to 20% per minute the FRCE regulation
quality would improve compared to the pro-rata scheme;

In case of the decrease of the aFRR ramping requirement to 6,7% per minute and an
increase of aFRR capacity with 100 MW (+71%) the regulation quality would decrease.
There is a negative impact on the Level 1 range (small imbalances). The negative impact
on the Level 2 range was limited. It was even observed that the regulation quality for the
very large imbalances even improved in such case.
The above table shows that the transition towards a merit order scheme at Elia -without a drastic
increase of aFRR volumes to mitigate the impact on FRCE regulation quality and hence increase
of aFRR capacity procurement costs- can only be performed in case the ramp rate requirement for
aFRR bids is maintained (13,3% per minute) or is even increased.
105 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 4.2: exchange of aFRR control target
Below an example is given explaining more in detail for the exchange of aFRR control target.
Suppose that there are two TSOs in the CoBa for the exchange of aFRR balancing energy:

Both TSOs make 50 MW of aFRR bids available on the CMOL ( BSP A and BSP B);

The aFRR control target of both TSOs is 40 MW (80 MW of total aFRR control target);

The Activation Optimization Function activates 50 MW of aFRR bids from TSO A and 30
MW of aFRR bids of TSO B.
Table 15: exchange of aFRR control target
Bid
nr.
Bid size Bid price aFRR control
BSP
[MW] [€/MWh] taget(t) [MW]
1 BSP A
5
50
5
2 BSP A
10
51
10
3 BSP B
10
53
10
4 BSP A
8
54
8
5 BSP A
7
60
7
6 BSP B
12
61
12
7 BSP B
5
63
5
8 BSP A
20
65
20
9 BSP B
13
70
3
10 BSP B
10
75
0
Total
100
80
BSP A
50
50
BSP B
50
30
The difference between the initial aFRR control target of TSO A (40 MW) and the aFRR control
target allocated by the AOF to the bids of TSO A (50 MW) is 10 MW. Hence 10 MW is exchanged
from TSO A to TSO B over the Virtual Tie-Line.
aFRR bids have only limited ramping rate capabilities. Therefore the actual aFRR control request
per bid will differ from the aFRR control target per bid. The Activation Optimization Function
requires the activation of 50 MW and 30 MW of aFRR balancing energy from respectively TSO A
and TSO B. In reality the aFRR control requests in this example will result in 23,1 MW (12,7 MW)
of activated aFRR balancing energy at TSO A (TSO B). This is shown in the table below:
Table 16: exchange of aFRR control target
Bid
nr.
Max. ramping
aFRR control
Bid size Bid price per AGC cycle aFRR control
request(t-1)
aFRR control
BSP
[MW] [€/MWh] [MW]
taget(t) [MW] [MW]
request(t) [MW]
1 BSP A
5
50
0,7
5
4
4,7
2 BSP A
10
51
1,3
10
6
7,3
3 BSP B
10
53
1,3
10
3
4,3
4 BSP A
8
54
1,1
8
2,5
3,6
5 BSP A
7
60
0,9
7
2
2,9
6 BSP B
12
61
1,6
12
2,3
3,9
7 BSP B
5
63
0,7
5
2,1
2,8
8 BSP A
20
65
2,7
20
1,9
4,6
9 BSP B
13
70
1,7
3
0
1,7
10 BSP B
10
75
1,3
0
0
0,0
Total
100
80
23,8
35,8
BSP A
50
50
16,4
23,1
BSP B
50
30
7,4
12,7
106 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Given the fact that in this example in the previous AGC cycle (without exchange of aFRR
balancing energy) the aFRR control request at TSO A and TSO B are resp. 16,4 MW and 7,4 MW
it is clear that the 10 MW that is exchanged is not physically delivered at TSO A. Therefore there is
also a shift of FRCE (ACE) from TSO B to TSO A. This impacts the FRCE regulation quality of
both TSOs (improvement for TSO B, deterioration for TSO A).
107 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 4.3: Exchange of aFRR control request
Below a numerical example is provided for the exchange of the aFRR control request between two
TSOs. Section Y shows different ways to perform the exchange of aFRR control request.
Table 17: exchange of aFRR control request
Bid
nr.
Max. ramping
aFRR control
Bid size Bid price per AGC cycle aFRR control
request(t-1)
aFRR control
BSP
[MW] [€/MWh] [MW]
taget(t) [MW] [MW]
request(t) [MW]
1 BSP A
5
50
0,7
5
4
4,7
2 BSP A
10
51
1,3
10
6
7,3
3 BSP B
10
53
1,3
10
3
4,3
4 BSP A
8
54
1,1
8
2,5
3,6
5 BSP A
7
60
0,9
7
2
2,9
6 BSP B
12
61
1,6
12
2,3
3,9
7 BSP B
5
63
0,7
5
2,1
2,8
8 BSP A
20
65
2,7
20
1,9
4,6
9 BSP B
13
70
1,7
3
0
1,7
10 BSP B
10
75
1,3
0
0
0,0
Total
100
80
23,8
35,8
BSP A
50
50
16,4
23,1
BSP B
50
30
7,4
12,7
In (t-1) there was in totally 23,8 MW of aFRR activated. 17 MW of this energy was allocated to
TSO A and 6,8 MW was allocated to TSO B. We know that between (t-1) and (t) an additional 12
MW of extra aFRR is actually activated. Both TSOs request the activation of 40 MW of aFRR
capacity in (t). For TSO A this is an increase with 23 MW compared to the aFRR activated in (t-1),
whereas this is 33,2 MW for the case of TSO B. Therefore the 12 MW is divided proportionally
over both TSOs. The aFRR allocated to TSO A increases with 4,9 MW (21,9 MW) whereas the
one for TSO B increases with 7,1 MW (13,9).
Given the fact that in (t) 23,1 MW is activated at TSO A, it is clear that 1,2 MW is transferred by
VTL from TSO A to TSO B. This 1,2 MW is also actually produced in the area of TSO A. Therefore
the impact on the local FRCE is eliminated completely.
108 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 5.1: Counteractivation during 2013-H2
The table below quantifies the FRR counteractivations within the Belgian and Dutch systems
during the second half of 2013 (July – December 2013). In the Netherlands the IGCC volume is
not taken into account when assessing whether counteractivation (within an ISP) occurred; in
Belgium it is. The first part of the table investigates counteractivation between aFRR and mFRR.
The second part of the table shows (for the Belgian case only) the impact of the IGCC
contribution.
Table 18: balancing energy counteractivation at TenneT and Elia for the year 2013
2013
aFRR – aFRR infra-ISP counteractivation [GWh]
mFRR – mFRR infra-ISP counteractivation
[GWh]
‘Net’
aFRR
–
‘net’
mFRR
infra-ISP
counteractivation [GWh]
Total aFRR-mFRR counteractivation [GWh]
Total aFRR activation [GWh]
Total mFRR activation [GWh]
Total activation [GWh]
Counter-activation / total activation [%]
% of ISPs with counter-activation
% of time ‘double pricing’
% of time mFRR activation
% of time mFRR activation with infra-ISP
counteractivation
Additional counteractivation due to IGCC
contribution [GWh]
% of ISPs with counteractivation
% of time mFRR activation & counteractivation
TenneT
10.9
0.0
Elia
21.7
0.1
0.1
22.3
11.0
239.6
1.0
240.6
4.6%
26.0%
9.7%
0.3%
31.6%
44.1
316.3
109.6
425.9
10.3%
41.8%
# NA
19.3%
51.6%
Not considered
34,4
Not considered
Not considered
59.8%
64.7%
Conclusions:
In the Belgian system the volume of infra-ISP counteractivation is significantly larger than in the
Dutch system (larger by a factor of four). One factor in explaining this is the larger volume of
mFRR balancing energy activated in Belgium. The relative volume of counteractivations in Belgium
is higher than in the Netherlands by a factor of two only. 26 % of ISPs are characterized by some
kind of counteractivation (excl. counteractivation of FRR vs IGCC) in the Netherlands, whereas the
corresponding figure for Belgium is a little less than 42%.
The table also shows that if the IGCC contribution is considered to be part of regulation volume,
this has a non-negligible effect on the incidence of counteractivation in the Belgian system. mFRR
activation has a large impact on the likelihood of counteractivation occurring (almost 52 % of ISPs
in which mFRR is activated are characterised by counteractivation; that share increases to almost
65% if counteractivation against the IGCC contribution is included).
109 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
Annex 5.2: Product definition schedule activated mFRR product
The below product definition is identical for the different settlement interfaces:
-
TSO-BSP;
-
TSO-BRP; and
-
Connecting TSO- Receiving TSO.
Table 19: mFRR product design characteristics for this pilot project
Parameter; Definition
(a) Preparation Period
NC EB; 2013-12-09; Article 2
“Preparation Period means the time
duration between the request by the
TSO and start of the energy delivery”
(b) Ramping Period
NC LFCR; 2013-06-28; Article 2
“Ramping Period means a period of
time defined by a fixed starting point
and a length of time during which the
input and/or output of Active Power will
be increased or decreased.”
(c) Full Activation Time
NC EB; 2013-12-09; Article 2
“Full Activation Time means the time
period between the activation request
by TSO and the corresponding full
activation of the
concerned product"
(d) minimum and maximum quantity
No definition in the NCs
This parameter refers to possible
restrictions with respect to minimum and
maximum bid size
(e) Deactivation Period
NC EB; 2013-12-09; Article 2
"Deactivation Period means the time
period for ramping, from full delivery or
withdrawal back to a set point”
(f) price of the bid
No definition in the NCs
Value
Not defined explicitly
Not defined explicitly
Full Activation Time is defined as ISP-1 for the settled
reaction.
Technical pre-qualification
responsibility.
remains
local
TSO
Minimum of 1 MW
1 MW increments
Maximum of 999 MW
Deactivation is not applicable for a scheduled product
Unit to be used is EUR / MWh with two decimals
To be defined by the BSP.
110 | P a g e
October 17, 2014
[FINAL REPORT OF STEP 2 OF XB BALANCING PILOT PROJECT BE-NL]
(g) Divisibility
NC EB; 2013-12-09; Article 2
“Divisibility means the possibility for the
TSO to use only part of the Balancing
Energy bids or Balancing Capacity bids
offered by the Balancing Service
Provider, either in terms of power
activation or time duration”.
(h) minimum and maximum duration of
Delivery Period
NC EB; 2013-12-09; Article 2
“Delivery period means a time period of
delivery during which the Balancing
Service Provider delivers the full
requested change of power in-feed or
withdrawals to the system”
(i) location
No definition in NCs
(j) Validity Period
NC EB; 2013-12-09; Article 2
“Validity Period means the time period
when the Balancing Energy bid offered
by the Balancing Service Provider can
be activated,
whereas all the characteristics of the
product are respected. The Validity
Period is defined by a beginning time
and an ending time”
(k) Mode of Activation
NC EB, 2013-12-09; Article 2
“Mode of Activation means the
implementation
of
activation
of
Balancing Energy bids, manual or
automatic, depending on whether
Balancing Energy is triggered manually
by an operator or automatically by
means of a closed-loop regulator”
(l) minimum duration between the end
of Deactivation Period and the following
activation
Divisibility is an optional product characterstic on the
CMOL. Only in Belgium the BSPs have the right to
indicate mFRR bids as divisible.
Delivery Period is exactly one ISP; i.e., minimum
Delivery Period = maximum Delivery Period = one ISP
Location information indicates bidding zone on the
CMOL.
Bids can be marked unavailable for activation by the
connecting TSO according to local rules approved by
the NRA.
Bids only become firm after the Balancing Energy Gate
Closure Time at the most one full clock hour ahead of
the delivery ISP (i.e., a bid relating to the ISP from
15:15h to 15:30h becomes firm at 14:00h).
The connecting TSO forwards the bids to the CMOL
after the Balancing Energy Gate Closure time.
Validity period of a bid is always one ISP.
By definition, mFRR is manual
Locally managed by connecting TSO.
111 | P a g e

A Method for Load Balancing based on Software

October, 2014 News: Happy 2 be Home as had some great PR in

4M Market Coupling

COJ 20X6_17.indd

PowerPoint - David Haimes Oracle Intercompany Financials Blog

MIRA 304 TY 2013, English, v13.2

1 CORRIGENDUM FOR E-TENDER REF NO. ROJ/RKVY

Procurement of A Turnkey Solution for Connectivity in Secondary

Language in children with early unilateral brain

BAL-CC1101-01D3