Integrated zonal-exchange and nodal-flow clearing

model in multi-zonal spot electricity markets

Andreas Vlachos Pandelis Biskas
Regulatory Authority for Energy (RAE) Department of Electrical & Computer Engineering
Aristotle University
Athens, Greece Thessaloniki, Greece
[email protected] [email protected]

Abstract— This paper presents a model for the solution of the C. Main Variables
flow-based model in multi-zonal European spot markets with
ࢄ࢙࢔ Supply volume in node ݊, in MWh
zonal-based exchanges. The model involves a straightforward
ࢄࢊ࢔ Demand volume in node ݊, in MWh
approach for solving inter-zonal exchanges and nodal injections ࡱࢄࢋ Volume of exchange e, in MWh
that retain branch physical flow limits, notably mixing branch ࡾ࢙࢔, ࡾࢊ࢔ Complement variable of supply and demand maximum
physical flows and inter-zonal exchanges in a single multi-clearing volume constraints in node ݊, respectively, in €/MWh
model. Its goal is to attain optimal supply, demand and market ࢛࢖
ࡾࢌ࢒࢒ , ࡾࢌ࢒࢒࢕
࢒
Complement variables of line l upper and lower flow limit
exchanges, endogenously bounded by grid physical constraints, constraints, respectively, in €/MW
࢛࢖
e.g. branch physical limits. The model attains simultaneously (a) ࡾࢋ࢞ࢋ , ࡾࢋ࢞࢒࢕
ࢋ Complement variables of exchange e upper and lower limit
intuitive exchanges between interconnected bidding zones, (b) constraints, respectively, in €/MWh.
accurate computation of flows at the critical intra-zonal and II. INTRODUCTION
inter-zonal branches, and (c) correct pricing at bidding zone level.
The model is compared in terms of social welfare and cleared According to the Commission Regulation 2015/1222
exchange volumes with the “intuitive patch” currently applied by establishing a guideline on capacity allocation and congestion
Euphemia solver followed by a “tuning” model for the simulation management (CACM Regulation) [1], there are two congestion
of redispatching to avoid congestion in physical branches. The management models, (a) the “Coordinated Net Transmission
comparison proves the model’s superiority in attaining overall Capacity” (CNTC) approach, and (b) the “flow-based” (FB)
optimal intuitive solutions avoiding redispatching costs caused by approach. The origins and computational framework of these
intrinsic inefficiencies in its design. models for managing congestion are different, although
stemming from the same concept (namely, security of supply
Index Terms— flow-based model, intuitive exchanges, mixed considering N-1 criterion constraints), leading to different
complementarity problem, complementarity conditions, nodal exploitation of the system inter-zonal and intra-zonal branches.
network configuration, zonal exchanges According to studies performed in several European regions
[2]-[3], the flow-based approach exhibits significant advantages
I. NOMENCLATURE in that (a) it maximizes the use of the transmission grid, offering
A. Indices and Sets enhanced trading opportunities with at least the same level of
࢔ Index of system nodes security of supply and thus (b) it maximizes the social welfare.
࢔ሺࢠሻ Index of system nodes within bidding zone ‫ݖ‬
Nevertheless, the application of flow-based market
ࢠ Index of bidding zones
ࢋ Index of exchanges between interconnected bidding zones coupling induces a significant “flaw” in the clearing of the
ࢋሺࢠǡ ࢠԢሻ Exchange from bidding zone ‫ ݖ‬to bidding zone ‫ݖ‬Ԣ commercial exchanges, namely the well-known “non-intuitive
࢒ Index of transmission lines (branches) exchanges” identified in several Central Western European
࢒ሺ࢔ǡ ࢔Ԣሻ Transmission line connecting nodes n and n’ (CWE) region studies [4]-[6]. Even though non-intuitive
B. Parameters exchanges relieve efficiently critical branch saturations leading
ࡽ࢙࢔ , ࡼ࢙࢔ Supply offer volume and price in node n, in MWh and to the maximum social welfare, such results are deemed
€/MWh, respectively undesirable for market designers and participants [7] since they
ࡽࢊ࢔ , ࡼࢊ࢔ Demand bid volume and price in node n, in MWh and €/MWh, are perceived as anti-competitive behavior. To surpass such
respectively flaw, a heuristic iterative algorithm has been developed in
࢛࢖
ࡲࡸ࢒ , ࡲࡸ࢒࢕
࢒
Remaining Available Margin (RAM), namely upper (positive CWE, called “intuitive patch” [4]. This process provides no
value) and lower (negative value) remaining physical flow
limits of line ݈, in MW, respectively
guarantees on the quality of the result, since it does not
ࢆࡵࡹࢋǡࢠ Exchange/zone incident matrix, denoting a positive (=1) / necessarily converge to the optimum, and it does not give an
negative (= -1) exchange ݁ transacted from / to bidding zone estimate of the error made [1], [4].
‫ݖ‬, respectively
ࡼࢀࡰࡲ࢒ǡ࢔ Power Transfer Distribution Factor of flow in branch ݈ with The above “flaw” has been resolved by the authors in [8],
respect to a nodal injection in node n by providing a proper formulation of complementarity
࢛࢖
࡭ࢀ࡯ࢋ ǡ upper and lower zonal exchange limits (Available Transfer conditions, that integrate the intuitive exchange and flow-based
࡭ࢀ࡯࢒࢕
ࢋ
Capacities) for exchange ݁, in MW, respectively solution. Nevertheless, the presented model in [8] exhibits a
ࡳࡿࡷࢠǡ࢔ Generation Shift Key of node n with respect to bidding zone nodal configuration in both the physical layer (congestion
‫ݖ‬
management in the network branches) and in the commercial
layer (commercial exchanges). Namely, the approach in [8]

‹,(((

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on July 30,2020 at 08:58:32 UTC from IEEE Xplore. Restrictions apply.
 considers commercial exchanges in single physical lines, which B. Generalized LP formulation of SEB model:
is not the standard practice in European markets, where The SEB generalized model is formulated by (1)-(3) and the
commercial exchanges are regarded at inter-zonal level following equations:
(between interconnected bidding zones).
a) Inter-zonal exchange upper and lower limits:
In this paper a solution that achieves a remedy to the ௨௣ ௨௣
aforementioned drawbacks/problems is presented. In this ‫ܺܧ‬௘ ൑ ‫ܥܶܣ‬௘ ‫ݔܴ݁ ׷‬௘ ൒ Ͳǡ ‫݁׊‬ (8)
respect, this paper presents a straightforward approach for ‫ܺܧ‬௘ ൒ ‫ܥܶܣ‬௘௟௢ ‫ݔܴ݁ ׷‬௘௟௢ ൒ Ͳǡ ‫݁׊‬ (9)
solving inter-zonal exchanges and nodal injections that retain b) Zonal power balance:
physical flow limits, notably mixing flows and exchanges in a
single multi-clearing model. The model attains simultaneously ܼ‫ܫ‬௭ െ σ௘൫ܼ‫ܯܫ‬௘ǡ௭ ή ‫ܺܧ‬௘ ൯ ൌ Ͳ ‫ܲܯ ׷‬௭ ǡ ‫ݖ׊‬ (10)
(a) intuitive exchanges between interconnected bidding zones,
and (b) accurate computation of flows at the critical intra-zonal where zonal injection ܼ‫ܫ‬௭ is defined as
and inter-zonal branches, that may reach their physical limits. ܼ‫ܫ‬௭ ൌ σ௡ሺ௭ሻሺܺ‫ݏ‬௡ െ ܺ݀௡ ሻ ൌ σ௡ሺ௭ሻሺܰ‫ܫ‬௡ ሻ (11)
The model exhibits a “hybrid” network configuration: (a) a
nodal-based network for the congestion management, defining IV. CURRENT APPROACHES IN SOLVING NODAL-BASED
a “physical layer” and (b) a zonal configuration for the FLOW AND ZONAL-EXCHANGES MODEL
commercial exchanges between the market zones, defining a A simplified combination of SFB and SEB model has been
“commercial layer”. This hybrid configuration conforms analyzed in [8], where the theoretical ground has been presented
perfectly to the currently applied scheme in Europe, where the to justify the endogenous presence of non-intuitive exchanges,
“flow-based” approach with a hybrid nodal/zonal configuration namely to prove that the conditions of the combined model do
is the preferred congestion management modeling approach for not guarantee an intuitive solution for the dependent exchange
the day-ahead and intra-day electricity markets [1]. The model variables. Besides paper [8], where a nodal configuration has
is compared in terms of social welfare and cleared exchange been adopted for commercial exchanges, there is no published
volumes with the “intuitive patch” currently applied by scientific literature bearing a nodal/zonal configuration, except
Euphemia solver plus a “tuning” model for the simulation of from some reports of CWE [4]-[6] documenting the progress of
redispatching to avoid congestion in physical branches. internal research towards the implementation of the intuitive
flow-based modeling in the Day-Ahead Market of CWE. The
III. BASIC CONGESTION MANAGEMENT MODELS CWE approach is based on the use of Generation Shift Keys
Basic congestion management is set out in two basic (GSKs) designated as ‫ܭܵܩ‬௭ǡ௡ that denote the distribution factor
models: (a) the Standard Flow-Based (SFB) Model, which of zonal injection in its internal nodes, nodal injections:
resembles the FB approach in CACM Regulation [1], and (b)
the Standard Exchange-Based (SEB), which is the CNTC ܰ‫ܫ‬௡ ൌ ‫ܭܵܩ‬௭ǡ௡ ή ܼ‫ܫ‬௭  (12)
approach in CACM Regulation [1]. An analytical description of Following the standard definitions of the flow-based model
the structure, design and constituent constraints of these models (1)–(6), combined with (10)-(11) and equation (12), a
has been presented in [8]. For reference reasons, the generalized Combined Exchange Flow-Based CEFB model is formulated.
mathematical formulation of these models is provided below The definition of power flow on lines is transformed to:
for a single trading-period.
σ௡൫ܲܶ‫ܨܦ‬௟ǡ௡ ‫ܭܵܩ ڄ‬௭ǡ௡ ή ߄‫ܫ‬௭ ൯ ൌ σ௭൫ܼܲܶ‫ܨܦ‬௟ǡ୸ ‫ܫܼ ڄ‬௭ ൯ (13)
A. Generalized LP formulation of SFB model: The term σ௡‫א‬௭൫ܲܶ‫ܨܦ‬௟ǡ௡ ‫ܭܵܩ ڄ‬௭ǡ௡ ൯ practically represents
‹ σ௡ሺܲ‫ݏ‬௡ ή ܺ‫ݏ‬௡ െ ܲ݀௡ ή ܺ݀௡ ሻ (1) the power transfer distribution factor of a zonal net injection
ܼ‫ܫ‬௭  on line flow ‫ܮܨ‬௟ , denoted hereafter as ܼܲܶ‫ܨܦ‬௟ǡ௭ . By using
subject to the following constraints:
(13) in (4) and (5), along with (1) – (3) and (10) – (11), the SFB
a) Supply volume capacity constraint: model is combined with the SEB model, embedding ‫ܺܧ‬௘ as
primal decision variables. The substitution of (10) in equation
ܺ‫ݏ‬௡ ൑ ܳ‫ݏ‬௡ ǣ ܴ‫ݏ‬௡ ൒ Ͳǡ ‫݊׊‬ (2) (13) – and subsequently in (4) and (5) – resolves to an
b) Demand volume capacity constraint: exchange-based representation of branch nodal flows:
௨௣
ܺ݀௡ ൑ ܳ݀௡ ǣ ܴ݀௡ ൒ Ͳǡ ‫݊׊‬ (3) ‫ܮܨ‬௟ ൌ σ௡ሺߜܼܲܶ‫݈ܨܦ‬ǡ݁ ‫ܺܧ ڄ‬௘ ሻ  ൑ ‫ܮܨ‬௟ (14)
‫ܮܨ‬௟ ൌ σ௡ሺߜܼܲܶ‫݈ܨܦ‬ǡ݁ ‫ܺܧ ڄ‬௘ ሻ  ൒ ‫ܮܨ‬௟௢

௟ (15)
c) Line power flow capacity limits:
௨௣ ௨௣ where ߜܼܲܶ‫ܨܦ‬௟ǡ௘ ൌ ܼܲܶ‫ܨܦ‬௟ǡ௭ െ ܼܲܶ‫ܨܦ‬௟ǡ௭ʅ
‫ܮܨ‬௟ ൌ σ௡ ܲܶ‫ܨܦ‬௟ǡ௡ ‫ܫܰ ڄ‬௡  ൑ ‫ܮܨ‬௟ ǣܴ݂݈௟ ൒ Ͳǡ ‫݈׊‬ (4) In the CWE approach, the GSK factor are exogenous static
‫ܮܨ‬௟ ൌ σ௡ ܲܶ‫ܨܦ‬௟ǡ௡ ‫ܫܰ ڄ‬௡ ൒ ‫ܮܨ‬௟௢ ௟௢
௟ ‫݈݂ܴ ׷‬௟ ൒ Ͳǡ ‫݈׊‬ (5) parameters, i.e. they represent a user-defined, static distribution
d) Nodal net injection and system energy balance: of zonal to nodal injection. To this extent, the approach resolves
theoretically to a suboptimal solution, since the solution space
σ௡ሺܰ‫ܫ‬௡ ሻ ൌ Ͳ ‫ܮ  ׷‬௡ ǡ ‫݊׊‬ (6) is defined by the values of the static, user-defined GSK factors.
where the net injection in node ݊, ܰ‫ܫ‬௡ , is defined as A plain combination of SFB and SEB models may result to
ܰ‫ܫ‬௡ ൌ ܺ‫ݏ‬௡ െ ܺ݀௡ (7) non-intuitive directional exchanges, i.e. exchanges from a
higher price bidding zone to a lower price bidding zone. To
surpass the “flaw” of non-intuitive exchanges, a heuristic

,(((0LODQ3RZHU7HFK

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on July 30,2020 at 08:58:32 UTC from IEEE Xplore. Restrictions apply.
 iterative workaround technique has been applied for CWE, conditions. Moreover, no PTDF modification (elimination) is
called “intuitive patch”, in each iteration of which the counter- assumed with respect to power flows, thus the solution space is
exchanges associated with negative ߜܼܲܶ‫ܨܦ‬௟ǡ௘ (ߜܼܲܶ‫ܨܦ‬௟ǡ௘ ൌ not reduced (as in the aforementioned intuitive patch). Thus,
ܼܲܶ‫ܨܦ‬௟ǡ௭ െ ܼܲܶ‫ܨܦ‬௟ǡ௭ʅ , where exchange e refers to an energy physical flows are allowed to reach their binding limits, since
transfer from zone z to zone zǯ) are ignored or eliminated in the no artificial tighter limits are imposed to the critical branches
computation of the (prevailing) flow on the critical/congested through the said elimination of specific PTDFs.
branches [4], [9]. Despite the introduction of exchange variables in the model
In a flow-based market setup with portfolio-based bidding, (1)-(5), (7), (10)-(11), its solution, as a standard LP
after the Day-Ahead Market clearing the market participants optimization problem with objective (1), still does not
define their production and consumption schedules – self- guarantee intuitive exchanges; moreover, it may yield
scheduling process – according to the traded (bought, sold) paradoxically accepted or rejected supply/demand bids with
energy in each bidding zone. These self-schedules define the respect to the zonal clearing price ‫ܲܯ‬௭ (shadow price of
actual scheduled GSKs, which are in general different from the constraint (10)). These facts are theoretically expected and
forecasted GSKs defined at day-ahead level by the TSOs. The explained by further analyzing the KKT conditions of the
mismatch between the static pre-defined GSKs and the actual formulated problem:
GSKs (based on scheduled production/consumption at each
node) leads to erroneous calculated flows in network branches a) exchange variables are dependent variables, and
at day-ahead level (with respect to actual flows), and thus leads b) KKT optimality conditions implicitly define nodal
to situations where: prices that prevail over the zonal prices and drive the optimal
a) some line flow constraints are not binding (active) in solution of ܺ‫ݏ‬௡ and ܺ݀௡ . These nodal prices are implicitly
௨௣
the market clearing results, but they may be violated in actual formed as ‫ܲܯ‬௭ െ σ௟ ቀܲܶ‫ܨܦ‬௟ǡ௡ ‫ ڄ‬൫ܴ݂݈௟ െ ܴ݂݈௟௟௢ ൯ቁ . The term
operation considering the self-schedules,
σ௟ ቀܲܶ‫ܨܦ‬௟ǡ௡ ‫ ڄ‬൫ܴ݂݈௟௨௣ െ ܴ݂݈௟௟௢ ൯ቁ shall be hereafter referred as
b) whereas some line flow constraints are binding in the
market clearing results, but they may not be binding in actual ܰ‫ܸܫ‬௡ . This term expresses the “nodal injection congestion
operation considering the self-schedules. value” referring to the congestion of all flow constraints.
Case (a) above necessitates the use of ex-post Remedial Thus, in the classical LP approach, it may resolve that
Actions in real-time (mainly redispatching) to resolve the line supply volumes may be cleared at a zonal price lower than their
flow violations. Such Remedial Actions lead to an overall respective bid, or equivalently, demand volumes may be
increased cost (day-ahead cost, plus redispatching costs), or cleared at a zonal price higher than their respective bid; these
equivalently lower welfare to the society. cases are collectively called paradoxically accepted orders. In
fact, the LP solution of (1)-(5), (7), (10)-(11), is identical to the
Based on the above, the calculation of GSKs is critical for solution of the SFB model. In order to attain a solution with
real-time congestion management. In [9] the way GSKs are intuitive exchanges and non-paradoxically cleared supply/
computed in several European countries is presented. demand volumes, the following conditions must hold:
Moreover, in [3], [10] the nine possible options for the
computation of GSKs in the Nordic region are proposed. In any a) Supply and demand volumes must be cleared, only
case, the choice of GSKs influences significantly the market under a positive surplus of their (nodal) bid price with respect
[3], [10]. GSKs are actually the major source of inaccuracies of to the zonal clearing price, i.e.
the applied flow-based parameter (zonal PTDFs, i.e. ܼܲܶ‫ܨܦ‬௟ǡ୸ ) ܺ‫ݏ‬௡ ൐ Ͳǡ ‹ˆ‫ܲܯ‬௭ ൐ ܲ‫ݏ‬௡ ܺ݀௡ ൐ Ͳǡ ‹ˆ‫ܲܯ‬௭ ൏ ܲ݀௡
calculation and subsequently of the market clearing results. The
ܺ‫ݏ‬௡ ൌ Ͳǡ ‹ˆ‫ܲܯ‬௭ ൏ ܲ‫ݏ‬௡ ƒ†ܺ݀௡ ൌ Ͳǡ ‹ˆ‫ܲܯ‬௭ ൐ ܲ݀௡ (16)
incorrect GSKs lead to inaccurate zonal clearing prices -derived
from the market clearing-, which lead to (a) wrong economic ܺ‫ݏ‬௡ ൒ Ͳǡ ‹ˆ‫ܲܯ‬௭ ൌ ܲ‫ݏ‬௡ ܺ݀௡ ൒ Ͳǡ ‹ˆ‫ܲܯ‬௭ ൌ ܲ݀௡
signals to the market participants, and (b) inaccurate congestion b) Considering that ‫ܺܧ‬௘ can be split in two directional
income, not corresponding to the actual congestion of the ା ି
non-negative variables ‫ܺܧ‬௘൫௭ǡ௭ ᇲ ൯ ൌ ‫ܺܧ‬௘൫௭ǡ௭ ᇲ ൯ െ  ‫ܺܧ‬௘൫௭ǡ௭ ᇲ ൯ ,
electricity grid.
these directional exchange volumes must be cleared only under
V. INTEGRATING ZONAL EXCGANGES WITH NODAL a positive surplus with respect to the zonal clearing price
FLOWS IN MULTI-ZONAL CLEARING difference:
A. Method outline ‫ܺܧ‬௘ା ൐ Ͳǡ ‹ˆ‫ܲܯ‬௭ ᇲ ൐ ‫ܲܯ‬௭ ‫ܺܧ‬௘ି ൌ Ͳǡ ‹ˆ‫ܲܯ‬௭ ᇲ ൐ ‫ܲܯ‬௭
The proposed method considers a multi-zone electricity  ‫ܺܧ‬௘ ൌ Ͳǡ ‹ˆ‫ܲܯ‬௭ ᇲ ൏ ‫ܲܯ‬௭ ƒ† ‫ܺܧ‬௘ି ൐ Ͳǡ ‹ˆ‫ܲܯ‬௭ ᇲ ൏ ‫ܲܯ‬௭ (17)
ା
market with the underlying nodal physical transmission system. ‫ܺܧ‬௘ା ൒ Ͳǡ ‹ˆ‫ܲܯ‬௭ ᇲ ൌ ‫ܲܯ‬௭  ‫ܺܧ‬௘ି ൒ Ͳǡ ‹ˆ‫ܲܯ‬௭ ᇲ ൌ ‫ܲܯ‬௭ 
The physical flow constraints are modeled at nodal level, i.e.
equations (2)-(5), (7). Instead of system level power balance In the proposed model the above conditions (16)-(17) are
(6), zonal power balance equations (10)-(11) are used. In this formulated as complementarity equations, associated to the
way, the fundamental parameters affecting the power flows are respective energy volumes variables establishing the market
nodal net injections ܰ‫ܫ‬௡ ൌ ܺ‫ݏ‬௡ െ ܺ݀௡ . No GSK parameters clearing conditions, i.e. the market layer. The (1)-(5), (7), (10)-
are considered; thus, the nodal net injections are the primal (11) are formulated also as complementarity equations
drivers of the problem, and the distribution of zonal injections establishing system conditions, i.e. the physical layer.
to internal nodes is intrinsically resolved by optimality

,(((0LODQ3RZHU7HFK

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on July 30,2020 at 08:58:32 UTC from IEEE Xplore. Restrictions apply.
 B. Clearing conditions of supply bids difference. For each direction of exchanges, two auxiliary price
The clearing conditions of supply bids are defined at node difference variables οܲା ି
௘ and οܲ௘ are defined as:
level, with respect to the zonal market clearing price of the zone െ‫ܲܯ‬௭ ᇲ ൅ ‫ܲܯ‬௭ ൅ οܲା ା
(22)
௘ ൒ Ͳ ٣ οܲ௘ ൒ Ͳǡ ‫݁׊‬
where the node belongs to. Supply clearing should resolve to ൅‫ܲܯ‬௭ ᇲ െ ‫ܲܯ‬௭ ൅ οܲି ൒ Ͳ ٣ οܲି
(23)
௘ ௘ ൒ Ͳǡ ‫݁׊‬
positive surplus of cleared nodal volumes with respect to the
zonal market clearing price. In this aspect, the clearing Conditions (22) - (23) impose that only one of the οܲା ௘,
condition of nodal supply bids takes into account the shadow οܲି௘  would be positive, whereas the other would be zero.
(added) value of any constraint that defines an upper bound for Based on which direction renders the positive price difference,
the supply volume. Supply volumes are constrained explicitly the following conditions impose that only the directional
by their respective capacity limits (2), and implicitly by flow exchange with ߂ܲ ൒ Ͳ will be cleared, while, the counter
constraints (4), (5). More specifically, in the KKT (dual) directional exchange with ߂ܲ ൌ Ͳ would be zero.
condition of the SFB model, this bound is implicitly imposed
െ‫ܲܯ‬௭ ᇲ ൅ ‫ܲܯ‬௭ ൅ οܲା ା
௘ ൒ Ͳ ٣ ‫ܺܧ‬௘ ൒ Ͳǡ ‫݁׊‬ (24)
by the “nodal injection congestion value” ܰ‫ܸܫ‬௡ .
൅‫ܲܯ‬௭ ᇲ െ ‫ܲܯ‬௭ ൅ οܲ௘ ൒ Ͳ ٣ ‫ܺܧ‬௘ି ൒ Ͳǡ ‫݁׊‬
ି
(25)
ܲ‫ݏ‬௡ െ ‫ܲܯ‬௭ ൅ ܴ‫ݏ‬௡ ൅ ܰ‫ܸܫ‬௡ ൒ Ͳ ٣ ܺ‫ݏ‬௡ ൒ Ͳǡ ‫݊׊‬, (18)
Depending on the price difference ‫ܲܯ‬௭ ᇲ െ ‫ܲܯ‬௭ , conditions
In order to resolve to non-paradoxically cleared supply (24) - (25) impose (due to ൒ Ͳ constraint) that only one of the
volumes, (18) is modified, by eliminating negative values of ‫ܺܧ‬௘ା ǡ ‫ܺܧ‬௘ି would be positive, in the direction of a positive price
ܰ‫ܸܫ‬௡ , i.e. considering only the state where ܰ‫ܸܫ‬௡ ൐ Ͳ, that is difference ߂ܲ, while the counter variable would be zero. In
summary, conditions (22)-(25) guarantee that cleared exchange
ܲ‫ݏ‬௡ െ ‫ܲܯ‬௭ ൅ ܴ‫ݏ‬௡ ൅ ܰ‫ܸܫ‬௡ ȁவ଴ ൒ Ͳ ٣ ܺ‫ݏ‬௡ ൒ Ͳǡ ‫݊׊‬ (19)
volume between two bidding zones are intuitively coherent to
Condition (19) provides that the overall congestion defines the price difference of these zones.
an implicit upper bound to ܺ‫ݏ‬௡ and perceived nodal price
E. Complete model and results
‫ܲܯ‬௭ െ ܰ‫ܸܫ‬௡ is lower than the zonal price ‫ܲܯ‬௭ ; thus, only
supply volumes with a bid lower than the zonal price may be The complete formulation of the proposed method is set by
cleared. simultaneously solving conditions (2)-(5), (10)-(11), (19), (21)
and (22)-(25). The solution method presented herein can thus
C. Clearing conditions of demand bids, be formulated as a mathematical programming model, e.g. as a
The clearing conditions of demand bids are defined at node Mixed Complementarity Problem [11] that can be solved either
level, with respect to the zonal market clearing price of the zone by commercially available solvers [12].
where the node belongs to. Demand clearing should resolve to
VI. ILLUSTRATIVE IMPLEMENTATION
positive surplus of cleared nodal volumes with respect to the
zonal market clearing price. In this aspect, the clearing A. Description of the case system
condition of nodal demand considers the shadow (added) value The examined test system is based on a version of the
of any constraint that defines an upper bound for the demand continental European electricity full grid, which has been
volume. Demand volumes are constrained explicitly by their provided to the authors by ENTSO-E. The test system consists
respective capacity limits (3), and implicitly by flow constraints of about 3490 nodes grouped in 26 bidding zones and 5810
(4), (5). More specifically, in the KKT (dual) condition of the lines; 47 exchanges are considered between zones that are
SFB model, this bound is implicitly imposed by the term ܰ‫ܸܫ‬௡ : physically interconnected with AC branches. The generation –
െܲ݀௡ ൅ ‫ܲܯ‬௭ ൅ ܴ݀௡ െ ܰ‫ܸܫ‬௡ ൒ Ͳ ٣ ܺ݀௡ ൒ Ͳǡ ‫݊׊‬ (20) demand mix of the case comprise 800 nodal supply offers,
represented as step-wise linear blocks (3-10 steps) and 2270
In order to resolve to non-paradoxically cleared demand nodal demand bids depicted as one divisible block volume.
volumes, (20) is modified, by eliminating positive values of Price steps of generation offers are randomly generated, in a
ܰ‫ܸܫ‬௡ , i.e. considering only the state where ܰ‫ܸܫ‬௡ ൏ Ͳ, that is range of 20-149€/MWh. The price for all divisible demand
blocks block is set to 150€/MWh. Under this assumption
െܲ݀௡ ൅ ‫ܲܯ‬௭ ൅ ܴ݀௡ െ ܰ‫ܸܫ‬௡ ȁழ଴ ൒ Ͳ ٣ ܺ݀௡ ൒ Ͳǡ ‫݊׊‬ (21)
demand is practically treated as fixed, yet available for shedding
Condition (21) provides that the overall congestion defines at very high market prices; demand surplus is considered in the
an implicit upper bound to ܺ݀௡ and perceived nodal price total surplus of the system.
‫ܲܯ‬௭ െ ܰ‫ܸܫ‬௡ is higher than the zonal price ‫ܲܯ‬௭ ; thus, only
B. Solution of the Standard Flow Based (SFB) model
demand volumes with a bid higher than the zonal price may be
cleared. The 1st column of Table I summarizes the solution by the
SFB LP formulation. This model is not suitable for the scope of
D. Clearing conditions of intuitive exchanges resolving intuitive inter-zonal exchanges, in a mixed
By definition the clearing for intuitive exchanges, must nodal/zonal physical/market system; yet it’s solution may be
result to exchange volumes transacted from a bidding zone with considered as the best objective value. The standard LP solution
a lower price to a bidding zone with a higher price.An exchange practically provides nodal prices, which are endogenously
‫ܺܧ‬௘ is considered as the difference of two directional non- computed, as an overhead of constraints shadow prices on the
negative components ‫ܺܧ‬௘ ൌ ‫ܺܧ‬௘ା െ  ‫ܺܧ‬௘ି , only one of which zonal prices (zonal power balance equation shadow price). In
might be non-zero. Directional exchange volumes should be this sense, zonal exchanges are dependent variables, that are
cleared intuitively with respect to the zonal clearing price resolved just to fit the optimal nodal dispatch.

,(((0LODQ3RZHU7HFK

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on July 30,2020 at 08:58:32 UTC from IEEE Xplore. Restrictions apply.
 C. Solution of the proposed model TABLE I. SOLUTION SUMMARY OF ALL MODELS

In the 4th column of Table I a summary of the proposed SFB

“Intuitive “Patch & Proposed
solution is provided. The term “intuitive” provides the Patch” Tune” model
Supply 5,409,618 5,679,547 5,502,699 5,607,778
indication whether the exchange direction is coherent to the
Demand 18,574,802 17,344,076 17,091,582 17,842,674
price difference between the sending and receiving bidding Surplus (€) not
zones. Zero volume exchanges or exchanges between nodes 714,091 714,091 302,894
Exchanges applicable
with equal prices are indicated as “neutral”. All line flows are Total 23,984,420 23,737,714 23,308,371 23,753,347
within their physical limits; a small set of line flows result to Supply/
189,534 190,791 186,614 185,258
the respective binding limits. The latter verifies that line Volumes Demand
physical limits remain the critical/restrictive active boundary (MW) not
13,860 13,860 30,933
Exchanges applicable
conditions regarding the energy transactions. It should be
Over Limit 0 61 0 0
stressed that both flow and exchange volumes satisfy the zonal Count of
Binding 57 0 39 55
injections balance. Table II presents a solution summary of Flows
Under Limit 5,759 5,755 5,777 5,761
injections balance and surplus at zonal level. It should be noted Intuitive 13,860 13,860 30,933
Exchange not
that supply/demand surplus is computed with respect to the Non-Intuitive
applicable
0 0 0
Volumes
resulting zonal price; all cleared supply and demand bids are Neutral 0 0 0
consistent to zonal prices, i.e. cleared supply blocks have a Intuitive 18 18 47
Count of not
lower price and cleared demand has a higher price than the Non-Intuitive 0 0 0
Exchange applicable
Neutral 0 0 0
respective zonal price. In this way, none of the cleared bids
(either supply or demand) surplus renders negative value. This outcome is, in general, unavoidable, since fixed ex-
Accordingly, exchange surplus is computed with respect to ante computed GSK are used in flow computations. Insofar, the
exchange direction and price difference between respective “intuitive patch” technique applied on a zonal CEFB model, is
bidding zones. The intuitive solution guarantees that none of quite different, as opposed to the new solution model proposed
exchange surplus is negative. in this paper; practically, the solution of the two cannot be
directly compared, since the “intuitive patch” solution leads to
D. Solution of CEFB and the “Intuitive Patch” model a solution that is infeasible (in terms of overloaded lines) in
This sub-section provides the solution attained by applying actual operation.
the approach of CWE, utilizing a CEFB model and the
E. Adjustment of “Intuitive Patch”, “Patch and Tune” model
“intuitive patch” technique applied in CWE. This approach
assumes an input of GSK factors, i.e. a predefined distribution In order to obtain a physically feasible solution for the
of zonal net injections to the nodes of each respective zone. The comparison purposes, generation and demand schedules
combination of the GSK factors with nodal PTDFs provides a (acquired from the “intuitive patch” solution) should be
set of zonal PTDFs that are used in an LP formulation to solve adjusted. For such adjustment, the authors adopted a simple
for generation/demand dispatch at zone level and inter-zonal single-step fine-tuning of the solution provided by the “intuitive
exchanges, with respect to AC lines’ flow limits; the latter are patch” technique, as follows: the attained zonal market prices
expressed as linear equations of zonal net injections and zonal and inter-zonal exchange volumes are fixed, and the nodal
PTDFs. As explained in Section IV, the LP problem may result injections (i.e. cleared supply/demand bids) are re-optimized /
to non-intuitive exchanges, and the “intuitive patch” technique shifted to alleviate overloadings in all inter- and intra-zonal
is utilized to resolve for “intuitive” exchanges. For the purposes branches, considering the nodal (actual) PTDFs and the initial
of the illustration, the set of GSK values are computed using the supply / demand bids. The standard SFB model is used for this
distribution of zonal injection to internal nodes, as resulted by re-optimization (performed at-once for all bidding zones – i.e.
the standard LP model (Section VI.B). for the entire grid), adjusting only supply/demand volumes
whose respective offer/bid prices are consistent with the already
In the 2nd column of Table I a summary of the solution attained zonal market prices.
attained by the iterative application of the “intuitive patch” is
presented. As expected, the “intuitive patch” technique can The solution attained by the fine tuning of the “intuitive
provide a fully intuitive solution. The resulting surplus appears patch” solution is presented in the 3rd column of Table I. As
greater even than the solution of the standard SFB model. This expected, the overall “Patch and Tune” model leads to lower
is explained by the fact that the CEFB model is less restrictive social surplus, as compared to the new proposed solution
than the standard SFB: the formulation of line flows is method. This signifies the importance of the redispatching costs
simplified assuming a specific distribution of zonal injections in order to achieve a feasible solution in actual operation. The
in the respective internal nodes; the formulation neglects the “intuitive patch” solution itself leads to a similar welfare as
actual nodal injection, since generation and demand bids are compared to the proposed model, yet it exhibits a number of
aggregated at zonal level. A reverse ex-post computation of flows over limits (overloads). The tuned solution resolves to a
nodal injections may result to inconsistent cleared volumes per lower welfare, making the proposed model preferable by about
node, compared to the generation / demand bids cleared 2% (0.45 m€), for a single trading snapshot.
volumes. As a result, although line flow constraints are Also, the sold/bought quantities and the exchange volumes
perceived by the aggregated model as fully satisfied, actual are lower in the “Patch and Tune” model results, as well. Table
flows may render over limit flows; this constitutes a “fictitious” II presents the solution summary of injections balance and
feasible vs. an actual infeasible solution paradox. social surplus at zone level. Comparing the left columns of
Table II (proposed solution method results) with the right

,(((0LODQ3RZHU7HFK

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on July 30,2020 at 08:58:32 UTC from IEEE Xplore. Restrictions apply.
 TABLE II. ZONAL SUPPLY / DEMAND, PRICES AND WELFARE

Proposed model in this paper “Patch and Tune” Model

Clearing Net Supply Supply Demand Clearing Net Supply Demand
Zone Demand Supply Demand
Price Injection Volume welfare Volume Price Injection Volume Volume
ID welfare (€) welfare (€) welfare (€)
(€/MWh) (MWh) (MWh) (€) (MWh) (€/MWh) (MWh) (MWh) (MWh)
AL 90.48 -69 503 24,566 -572 34,043 128.83 -69 503 43,852 -572 12,112
AT 54.37 250 913 20,873 -663 63,415 24.91 324 987 4,698 -663 82,946
BA 81.22 88 1,191 40,858 -1,103 75,872 80.11 0 1,134 39,537 -1,134 79,253
BE 40.11 810 2,973 69,315 -2,163 237,712 18.83 1,254 3,417 6,258 -2,163 283,753
BG 85.63 334 2,116 119,570 -1,782 114,688 43.90 58 2,129 35,604 -2,071 219,744
CH 41.33 454 4,050 84,975 -3,596 390,750 23.23 156 3,752 14,527 -3,596 455,856
CZ 52.59 2,621 6,721 220,838 -4,100 399,379 24.54 3,727 7,827 33,975 -4,100 514,407
DE 54.40 4,641 23,791 834,570 -19,150 1,830,760 20.75 3,326 22,468 63,680 -19,142 2,474,174
DK 50.96 369 541 27,567 -172 17,055 0.0 0 172 0 -172 25,830
ES 44.12 -52 27,691 651,284 -27,743 2,937,464 39.13 -78 27,577 513,565 -27,654 3,066,028
FR 42.42 8,477 54,319 1,264,832 -45,842 4,931,521 22.28 2,079 47,939 178,660 -45,860 5,857,168
GB 36.61 1,768 1,768 37,010 0 0 12.27 0 0 0 0 0
GR 90.30 -695 7,725 398,422 -8,420 502,671 116.68 145 9,394 611,265 -9,249 308,167
HR 57.45 -126 567 8,931 -693 64,157 134.98 -273 873 56,001 -1,146 17,205
HU 78.45 -1,333 1,141 68,166 -2,474 177,021 46.24 -1,085 1,389 37,217 -2,474 256,700
IT 42.43 -7,456 22,676 467,468 -30,133 3,241,262 113.82 -1,847 28,264 2,280,417 -30,110 1,089,457
XX 71.96 -350 0 0 -350 27,316 150.00 -311 0 0 -311 0
ME 87.21 -8 456 21,510 -464 29,133 150.00 0 456 50,141 -456 0
MK 98.92 -26 0 0 -26 1,308 135.93 -356 176 678 -532 7,483
NL 79.66 -6,105 2,224 97,264 -8,329 585,869 88.78 -4,322 3,236 149,827 -7,558 462,654
PL 78.70 -937 11,622 622,255 -12,558 895,345 83.10 -1,377 10,809 643,606 -12,186 815,262
PT 44.25 -1,262 3,735 88,508 -4,997 528,413 114.33 0 4,997 389,777 -4,997 178,243
RO 73.50 133 4,618 225,933 -4,485 343,087 44.04 0 4,485 90,152 -4,485 475,191
RS 81.93 -173 2,731 163,741 -2,904 197,657 101.86 108 3,012 221,728 -2,904 139,796
SI 54.26 -309 29 39 -338 32,358 131.50 -324 553 24,292 -877 16,227
SK 63.81 -701 1,158 49,283 -1,859 160,233 32.39 -794 1,065 13,241 -1,859 218,630
UA 79.51 -343 0 0 -343 24,184 47.13 -343 0 0 -343 35,296
columns of the same Table (“Patch and Tune” model results), it [5] “CWE Flow-Based Market Coupling Forum”, EPEXSPOT, Oct. 2013.
is evident that the two solutions exhibit a large perturbation. In [Online]. Available:
https://www.epexspot.com/document/24695/CWE%20FB%20MC%20
Table II zone XX denotes the outer bounds of European region, Market%20Forum%2C%20October%2010th%2C%202013.pdf
namely the interconnections with non-EU countries. [6] “Position Paper of CWE NRAs on Flow-Based Market Coupling”,
CREG, Mar. 2015. [Online]. Available:
VII. CONCLUSIONS AND FURTHER RESEARCH http://www.creg.info/pdf/Opinions/2015/b1410/CWE_NRA_Position_P
This paper presents a model for the clearing of electricity aper.pdf
markets with a hybrid network configuration, namely nodal- [7] NRA public consultation on Flow Based Market Coupling, Synthesis of
respondents’ answers, CREG, July 2014. [Online]. Available at:
based configuration for congestion management in system http://www.cre.fr/en/documents/public-consultations/flow-based-
branches (physical layer), and an inter-zonal configuration for market-coupling-fb-mc-and-its-implementation-in-the-cwe-region/read-
the commercial exchanges between interconnected bidding the-synthesis-of-the-answers-to-the-public-consultation
zones (commercial layer).As compared with the traditional [8] A.G. Vlachos, P.N. Biskas, “An Integrated Intuitive Exchange- and Flow-
“intuitive patch” applied in CWE, the conditions of the Based Market Clearing Model”, IEEE Transactions on Power Systems,
vol. 31, no. 5, Sept. 2016.
presented model guarantee that all physical flows will be [9] Explanatory note on the day-ahead and intraday common capacity
retained to their limits (cf. to artificially tighter limits in the said calculation methodologies for the Core CCR, June 2018. Available at:
“intuitive patch”) and zonal injections will be distributed in https://docstore.entsoe.eu/Documents/Network%20codes%20documents
internal nodes dynamically (cf. statically in the intuitive patch /Implementation/ccr/methodologies/core/cacm-deliverables/da-and-id-
through static pre-defined GSKs). This constitutes the main ccms-art-20ff/20180604-core-tsos-explanatory-note-for-core-da-id-fb-
ccm-fv.pdf
enhancement of the proposed model with respect to the currently [10] All TSOs’ of the Nordic Capacity Calculation Region proposal for
applied “intuitive patch” used in Euphemia [13]. capacity calculation methodology in accordance with Article 20(2) of
Commission Regulation (EU) 2015/1222 of 24 July 2015 establishing a
REFERENCES guideline on capacity allocation and congestion management, May 2018.
[1] Commission Regulation (EU) 2015/1222. Available: https://energinet.dk/-/media/7191C3A0B2314DF7AABE419
[2] “CWE Enhanced Flow-Based MC feasibility report”, CWE REM, Oct. C2D2B34C2.pdf
2011. [Online]. Available: http://www.casc.eu/media/CWE%20FB%20 [11] M. C. Ferris, J. S. Pang “Engineering and Economic applications of
Publications/CWE_FB-MC_feasibility_report_2.0_1910 2011.pdf complementarity problems”, Society for Industrial and Applied
[3] “Stakeholder consultation document and Impact Assessment for the Mathematics, Vol. 39, No. 4, pp. 669-713, Dec. 1997.
Capacity Calculation Methodology Proposal for the Nordic CCR”, Nordic [12] General Algebraic Modeling System. [Online]. Available:
TSOs, Sep. 2017. [Online]. Available: http://d34718316.u65.surf- http://www.gams.com
town.net/CCM-documents/SH_archive/20170912/supp.pdf [13] “EUPHEMIA public description”, Dec. 2016. [Online]. Available:
[4] "CWE enhanced flow-based MC intuitiveness report", CWE REM, Oct. https://www.nordpoolspot.com/globalassets/download-
2013. [Online]. Available: https://www.apxgroup.com/wp-content/ center/pcr/euphemia-public-documentation.pdf
uploads/2-CWE_FB-MC_intuitiveness_report_October-2013.pdf

,(((0LODQ3RZHU7HFK

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on July 30,2020 at 08:58:32 UTC from IEEE Xplore. Restrictions apply.