1

Electrothermal Coordination in Cable Based

Transmission Grids
R. Olsen, Student Member IEEE, J. Holboell, Senior Member IEEE, and U. S. Gudmundsdóttir,
Member IEEE

Abstract - Electrothermal coordination (ETC) is introduced methodologies are executed by specifying a number of
for cable based transmission grids. ETC is the term covering loadability levels which the operator can relate to during
operation and planning of transmission systems based on emergency and faulted operation.
temperature, instead of current. ETC consists of one part The present paper introduces a method enabling operation
covering the load conditions of the system and one covering the of individual transmission cables based on temperature
thermal behaviour of the components. The dynamic temperature estimations (real DTR). The models developed for this DTR is
calculations of power cables are suggested to be based on
then combined with load flow simulations, such that ETC
thermoelectric equivalents (TEEs). It is shown that the thermal
behaviour can be built into widely used load flow software, becomes possible.
creating a strong ETC tool. ETC is, through two case scenarios,
proven to be beneficial for both operator and system planner. It Distributed temperature sensing (DTS) technology allows
is shown how the thermal behaviour can be monitored in real- for monitoring of the real time temperature in or around the
time during normal dynamic load and during emergencies. In cable. The DTS optic fibre can be implemented in the cable
that way, ETC enables cables to be loaded above their normal screen or, as it is done in Denmark, in a pipe laid close to the
rating, while maintaining high reliability of the system, which centre phase. However as the temperature on the conductor is
potentially can result in large economic benefits. the real limitation of the cable loadability, it is not sufficient
Index Terms--Cables, finite element methods, load flow
with such external measurements when controlling
analysis, power cable thermal factors, power system simulation,
prediction methods, temperature control, transmission lines transmission systems in real time. Modelling is thus necessary
in order to obtain the real time conductor temperature.
I. INTRODUCTION Furthermore, temperature measurements cannot directly be
used to make predictions about the future evolution of the
T his paper introduces electrothermal coordination (ETC) in
cable based transmission grids. ETC was introduced for
overhead lines in [1] and covers temperature based
cable temperature and accurate thermal models are thus
necessary in order to perform DTR and ETC.
Ref. [6] shows how DTS measurements can be integrated
dynamic operation of transmission systems and thermally into the thermal models in order to increase accuracy; however
based planning of such systems. Because of the short thermal some older cable systems are without temperature monitoring
time constant, it is in [1] argued that one can benefit mostly by and the thermal model developed in [6] thus take temperature
introducing ETC for overhead lines. However looking into the measurements as an optional input for increased accuracy.
future of e.g. the Danish transmission system, almost the The thermal models are developed with the scope of
entire grid will be undergrounded [2, 3], and the Danish implementation in the Danish transmission system, where
Transmission System Operator (TSO), Energinet.dk, thus direct burial in flat formation is the preferred installation
expects that much transmission capacity can be gained by method for transmission cables. Since most cables are
utilising the ETC concept on cable based systems as well. terminated above ground at a tower, in gas or in open air
The issue of dynamic operation of transmission lines has insulated substations, a discussion about the necessary thermal
been investigated by different researchers (e.g. [4, 5]), but models for cables in air is also included in this study.
many of these projects have focussed on the component level The installed cables are XLPE insulated and they are
(for most overhead lines) with limited considerations about the equipped with copper wire/aluminium foil metallic screens.
influence on an entire system. However the thermal models will, with smaller adaptions, be
On the component level, dynamic operation is discussed by applicable to other transmission systems as well.
using the term Dynamic Thermal Rating (DTR). DTR is a
calculation of the temperature in individual components and As it is seen from the above description of ETC (see also
based on this temperature the operator decides on how to act. [1]), a number of different models and principles are necessary
Some of the research in DTR has in reality ended in pseudo for ETC to become operational. In order to provide a clear
DTR, meaning that even though the temperature is somehow overview of the ETC methodology, the present paper is
estimated; network operation is still based on current instead therefore divided into several sections, each covering a
of temperature as the limiting parameter. Such operation necessary part of ETC in cable based transmission systems.
R. Olsen is with Section of Transmission Lines, Energinet.dk, Fredericia
7000, Denmark (e-mail: [email protected]).
II. THERMAL MODELLING OF TRANSMISSION CABLES
J. Holbøll is with the Dept. of Electrical Engineering, Technical Uni- A. Choice of Thermal Model
versity of Denmark, Kgs. Lyngby 2800, Denmark (e-mail: [email protected]).
U. S. Guðmundsdóttir is with Section of Transmission Lines, The thermal models of [1] are, as stated, concerned with
Energinet.dk, Fredericia 7000, Denmark (e-mail: [email protected]). overhead lines; but as the thermal conditions of OHL and
 2

underground cables are very different, another thermal model zones is shown in Fig. 2 where the insulation for illustrative
is necessary for this study. purposes is divided into three.
Different methodologies for temperature calculations on
cables have already been developed, where the most common
are: Finite Element Modelling (FEM, see e.g. [7]), Finite
Difference Modelling (FDM, see e.g. [8]), Thermoelectric
Equivalents (TEE, see e.g. [9]) and the Step Response Method
(e.g. [10]).
It is clear from the ETC-application, that the thermal model
to be selected must be fast, since the temperature has to be
calculated for a large number of cables, and accurate, because
the reliability of the entire transmission system depends on
accurate rating calculations. Based on these two requirements,
the TEE method was chosen for thermal modelling.

Compared to 2D and, in particular, 3D FEM simulations,

performed with commercially available FEM software, TEEs
can model the dynamic thermal behaviour of transmission Fig. 2: Cable design used for thermal modelling. r1 and r2 illustrate the outer
cables both with a sufficient accuracy and fast, [11]. The radius of Zone 1 and 2 respectively.
accuracy of TEEs was proven in [6], where the modelled
temperature of cables in a large scale experimental setup was Throughout this paper, the assumption is made that all
compared to measurements. metal parts of the thermal network can be considered as being
B. Design of Thermal Model isotherms. This is a valid assumption due to the high thermal
conductivity of metals as compared to dielectric materials.
The following briefly summarises dynamic temperature The thermal resistances used for modelling the different
calculations by TEE. subcomponents are calculated by using adapted versions of the
1) Calculating the Thermal Parameters guidelines given by IEC standard 60287-1 [12], as shown in
TEEs utilise the resemblance between heat flowing in a the following.
thermal system and current flowing in an electric system. Thus
thermal resistances of materials can be modelled by electric The thermal resistance of zone ‘x’ in the insulation is
resistances, as well as thermal capacitances can be modelled calculated as given in (1) and the thermal resistance of zone
by electric capacitances. The heat generated in conductor, ‘y’ in the jacket is calculated as given in (2).
dielectric, screen, etc. is resembled by current sources in the
electric analogy. therm,d r
T1, x ln x , x 1,2,3,..., s (1)
When modelling the heat flowing through a resistance, with 2 rx 1
a current, the voltage drop will be equal to the temperature where ρtherm,d is the thermal resistivity of the dielectric
difference on the two sides of the thermal resistance. Fig. 1 material, rx is the radius of zone x, r0 is the conductor radius
shows the simplest case of a TEE for a directly buried cable. and s is the total number of zones in the insulation.
therm, j ry
T3, y ln
, y 1,2,3,..., t (2)
2 ry 1
where ρtherm,j is the thermal resistivity of the jacket material,
ry is the radius of zone y, r0 is the screen outer radius and t is
the total number of zones in the jacket.
Fig. 1: Thermoelectric Equivalent, TEE, of a single phased cable without
armour. Wc are joule losses in the conductor, the sum of Wd1 and Wd2 are the
The thermal resistance, T4,z, of zone ‘z’ in the surrounding
dielectric losses and Ws are the screen losses. The C's are the thermal soil is calculated as given in (3).
capacitances and the T’s are the thermal resistances of the respective n d ' pk
subcomponents. r'z 1 r'z ln d pk
T4, z therm, sur
ln ln k 1 (3)
2 rz 1 rz u
In the figure, the thermal parameters of each subcomponent
(conductor, insulation, metal screen, jacket and surrounding z 1,2,3,..., u
soil) are lumped together into few components. As explained where ρtherm,sur is the thermal resistivity of the surroundings, L
in [6] and [11], such a coarse division of the thermal the burial depth, rz the radius of zone z (r0 is the outer radius
parameters, as seen in Fig. 1, will result in deviations between of the cable itself), r’z the distance from zone z to the image of
the modelled dynamic temperature and the temperature of a the cable itself above ground, dpk is the distance from the cable
cable in operation where the thermal parameters are under investigation, p, to cable k, d’pk is the distance from
continuously spread over the entire material. It was therefore cable p to the image of cable k above ground, n is the number
suggested in [6] that the different subcomponents should be of cables in vicinity of cable p and u is the total number of
divided into multiple thermal zones. Such a division into zones in the surroundings.
 3

It is seen that the basic concept of Kennelly’s hypothesis, c1 e (t 0 ) e (t )

stating that the ground surface is isothermal, is maintained in v1 v2 v3 c2 s (t 0 ) s (t ) (8)
the calculation, as well as the thermal influence of the cables c3 c (t 0 ) (t )
in the vicinity is seen also to be included in the calculation. c

Note that by summing up T1,x over all x, T3,y over all y and Denoting the right hand side of (8) by θ(t0)-θ(t∞), the
T4,z over all z, one obtains the standardised equations of [12]. constants c1, c2 and c3 are found by performing Gauss-Jordan
Elimination on the matrix:
The thermal capacitances are calculated based on the cross v1 v2 v 3 (t 0 ) (t ) (9)
section of the different zones (4). It is assumed that the cross
section of the zones in the surroundings can be approximated The total cable temperature is then found by adding the
by means of concentric circles. results of (7) to the measured (or modelled, [14]) undisturbed
C1, x ctherm,d
2
rx rx 1
2 soil temperature at cable depth.
As suggested in [6], this study utilises a division of the
C 3, y ctherm, j ry
2
ry
2 (4) surroundings into 100 zones, the insulation into 10 zones and
1
2 2 the jacket into 3 zones. The system matrix of (5) therefore
C4, z ctherm,sur rz rz 1 becomes a 113 x 113 matrix, as well as the vectors of (5) and
Here ctherm,d, ctherm,j and ctherm,sur are the specific heat of the (7) will contain 113 elements. This will lead to 113 individual
dielectric, jacket and surroundings respectively, x=1,2,…,s, eigenvalues and eigenvectors. As there is no general analytical
y=1,2,…,t and z=1,2,…,u is the number of the zone in the method for solving a degree 113 polynomial equation, the
insulation, jacket and surroundings, respectively. eigenvalues and vectors, have been determined numerically.
Subcomponents such as semiconductive layers, thermal It should be noted that the electric resistivity, the thermal
backfill, etc. have been left out of the model. This is not a resistivity and the specific heat are temperature dependent
limitation in the methodology, but merely an attempt to properties; however in the present study the thermal
simplify the description. parameters of the model have been treated as constant. This is
not a limitation in the methodology, [6], but an assumption
The joule losses in the conductor, dielectric losses in the which is made for the purpose of simplification only.
insulation and induced joule losses in the screen have been It should be acknowledged that thermal modelling must be
calculated as described in IEC 60287 [12]. performed on hotspot locations as the hotspots will be the
limiting operating points for the cable system. However the
2) Calculating the Temperature
thermal model described above is well suited to deal with
In the TEE model, which in its nature is discrete in space,
hotspots, as studies such as [15] proves that hotspots wider
the temperature increase above ambient at each node is
than 3 meters can be treated without considering axial heat
calculated analytically by solving the differential equation
flow. The assumption that hotspots are wider than 3 meters is
system shown in (5). The specific equation in (5) relates to the
valid, at least in the Danish transmission system, as hotspots
simplest TEE as it is shown in Fig. 1.
are generally located when cables cross under roads, etc.
e
1
C 4T3
1
C 4T4
1
C 4T3
0 e
0 3) Cables in Air and other Special Conditions
1 1 1 1 1 Wd 2 Ws (5) As mentioned, underground cables are in most cases
s C3T3 C3T1 C3T3 C3T1 s C3
0 1 1 terminated above ground. The part of the cable in air will be
1
subjected to different thermal conditions than the
c C1T1 C1T1 c
C1 Wc Wd 1

Equation (5) can in short be given as (6). undergrounded part, and a different model must therefore be
utilised. The thermal conditions for the internal parts of the
A W (6) cable are identical for the air part and underground part, but
where A is the system matrix and ⃗⃗⃗ is related to the applied the surface of a cable in air can be subjected to solar radiation
external influences, in this case the different losses. and the heat losses to the surroundings are highly nonlinear,
i.e. most research within power cables assumes that the losses
The general solution to the differential equation system of may be evaluated as given in (10), [9].
(5) is as given in (7), [13]. 4 4
Wrad hrad e air
(t ) (t ) (10)
e e
3t
Wconv hconv e air
s (t ) c1v1 e 1t c2 v 2 e 2t
c3v3 e s (t ) (7) where Wrad and Wconv are the losses caused by radiation and
c (t ) c (t ) convection respectively, hrad and hconv are the radiation and
convection constants respectively (each dependent on the
where v i are eigenvectors of the system matrix A.
materials and cable geometry) and θair is the air temperature.
Similarly λi are the eigenvalues which are all non-positive,
Because of the nonlinearity in the heat balance equations,
since the TEE is a stable system. The constants ci are
they must be solved with numerical solution methods or by
calculated based on the end point constraints, as given in (8).
linearizing the losses such as done in the step response
The end point constraints are the initial (θ(t0)) and infinite
method, [9].
(θ(t∞)) steady state temperatures, which are simply calculated
In the remaining part of this study it is assumed that the
by removing the capacitances from the TEE and calculating
bottlenecks of the cables are on the underground part, and that
the voltage drops across the thermal resistances.
the temperature thus can be calculated by using (7), but it
 4

should be pointed out that measures to mitigate possible which makes the algorithm both flexible and efficient. All grid
hotspots above ground might be necessary. quantities such as bus voltages, etc. of course have to be
maintained within relevant regulations.
C. Load Flow Calculations
The AC-balanced load flow (positive sequence) calculations
The novelty of ETC lies in the close connection between are performed by DSPF, which solves the classical power
temperature calculations and load flow simulations within the equations by using the Newton-Raphson method.
transmission system. The connection between the load flow calculations and the
On the basis of the predicted future load flow scenario, it is thermal calculations is, as mentioned, primarily the conductor
possible to forecast the evolution of the temperature in all current. The actual load current of each cable in the grid is
transmission lines in the system. therefore written to a text file, which then is loaded by Matlab
Relatively fast and accurate methods are available for load for the further thermal analysis.
flow calculations and the present study uses the Newton- The feedbacks from Matlab to DSPF are the calculated
Raphson methodology. temperatures plus the resistivity of the individual cables in the
D. Connecting Load Flow and Thermal Calculations grid, the latter is to be used in the next load flow calculation.
The implementation of this 'DSPF-Matlab ETC Package' is
The thermal calculations and the load flow calculations are visualised in Fig. 3 and the individual parts are explained in
interdependent in several ways. the following.
The temperature is via the joule heating directly related to
The accuracy of the, dynamically evolving, calculated
the current passed through the conductor. However, the
temperature is highly dependent on the losses, which again are
current based joule heating is also connected to the
highly dependent on the time step. The time step should
temperature because the electric resistance of the conductor is
therefore be carefully chosen as a compromise between
temperature dependent. Furthermore, this resistance is a part computational time and accuracy.
of the impedance of the transmission line, which again is an
essential parameter in the load flow studies.
i. To adapt component parameters via DPL it is necessary to
In the present study these interdependencies have been
initialise the grid by running a load flow calculation.
implemented as follows.
ii. A description of the transmission grid is written to txt-files.
The temperature dependence of the electric resistance is This includes load-names, line-names, etc.
included by adapting the electric resistance to the temperature iii. All parameters in the system can be changed via DPL-
calculated at the last time step. This approximation is assumed
scripts. Changes will most often be the size of loads,
valid as long as a low time step is used compared to the time
lengths of cables, sizes of cables, sizes of generators, etc. It
constant of the conductor temperature, screen temperature, etc.
is assumed that all parameters of the grid have been static
An iterative process for calculation of the resistance would be
for a long time, such that the temperatures are in steady
more correct, the gained accuracy hereof however is, due to state before the dynamic simulation begins.
the high thermal time constants involved, expected to be low All cables are compensated by half at each end and the
compared to the computational time required.
necessary compensation per phase is as given in (11).
III. IMPLEMENTATION IN LOAD FLOW SIMULATIONS Q 2 f C ' l U 02 (11)
The thermal models of section II. have been implemented Here Q is the reactive power, f is the system frequency, C'
in the commercially available system analysis software is the capacitance per length of the cable, l is the length
DigSilent Power Factory (DSPF). and U0 is the line-to-screen voltage.
This software is widely used for load flow calculations by iv. The eigenvalues and eigenvectors of (7) for all cables in
utility companies, transmission system operators, as well as the system are evaluated and the initial temperatures are
many universities and other research institutions. The calculated.
implementation of ETC in DSPF will therefore address a large v. The temperatures are loaded into DSPF from a txt-file.
part of the power industry. vi. The cable resistance is redefined.
One of the strengths of DSPF is the graphical interface, vii. Having performed the load flow calculations, the current
which enables the user to get an easy overview of the grid, running in all cables is extracted and exported to a txt-file.
including electrical variables, etc. In addition to the graphical viii. The currents are loaded in Matlab where the conductor
interface, it is possible to change the grid parameters in two losses (Wc) and steady state temperatures are calculated for
different scripting environments, DigSilent Simulation all lines.
Language (DSL) and DigSilent Programming Language ix. The temperatures are stored in a mat-file for later analysis
(DPL), which makes it a strong tool for a large variety of and the temperatures which are important for DSPF are
simulation purposes. The grid calculations are obviously exported to txt-files.
inherent parts of the DSPF software, however the thermal x. Check whether the results have changed from the previous
calculations are not. Therefore the major part of the algorithm iteration to the present. If large differences exist in either
to be developed in the following clauses lies in this area. load flow or temperatures, it is necessary to run the
Since ETC here is treated with no concern about stability calculations again until they have stabilised. Stability here
(neither voltage nor frequency) related issues, the only grid is defined as temperature differences below 0.01 K and
calculations needed are load flow simulations. It is therefore load differences below 1 A.
possible to use the DPL instead of the DSL environment,
 5

xi. The dynamic simulation starts by loading the steady state

temperatures into DSPF and adjusting the grid parameters
(resistances) to these temperatures.
xii. Test if the temperature has exceeded the individual limits
of these components.
xiii. If the temperature limit has been exceeded for one or
more cables, the relevant lines are disconnected from the
grid.
xiv.Check if it is time for disturbances of system.
xv. Disturb the system. This can for example be an N-1 case (a
grid with a major single outage).
xvi. Knowing the temperatures, the cable impedances are
adapted as in vi.
xvii. DSPF runs a load flow calculation.
xviii. The currents obtained from the most recent load flow are
exported to txt-files.
xix. These currents are imported in Matlab and the thermal
response to the current is calculated on the basis of (7).
xx. Finally, test whether or not the end time has been reached.
If not, the algorithm runs the steps from xi to xix again.

It should be acknowledged that the described methodology

has been developed in close cooperation with all stakeholders
at the Danish TSO, and the software integration can thus be
seen as technically ready for implementation into the Danish
transmission system.

IV. CASE STUDY

In order to maintain a clear overview of the benefits of ETC
the basic case used in this study is a simple two-line system
with two generators and one load, Fig. 4, which is also found
to be the case studied in papers that have previously dealt with
ETC [1].

Fig. 4: Two line system used for case study.

The ETC software package has, with success, been tested

on significantly larger systems than this two line test case,
hereunder an adapted version of the IEEE 14-bus test network
containing 20 cables, however the described two line system
enables more comprehensible analysis, and discussing ETC as
a concept is thus easier.

The base voltage (Vbase) is 132kV. The cables are equipped

with metallic screens, which may cause additional heating.
The transmission lines have been modelled with cable
dimensions which are available online [16], and the most
important geometric parameters are summarised in Table I.
The cables are 100 km long [1]. The reactive power
generated by the cables is compensated by half at each end.
The cable screens are assumed to be perfectly cross-bonded,
Fig. 3: Algorithm structure for integration of temperature calculations
which, together with negligible eddy currents, for simplicity
in load flow software. An in-depth description of the different parts is
given in the text as indicated by the roman numbers. It is furthermore means that screen losses can be neglected [12].
indicated which program handles the different parts of the algorithm.
 6

TABLE I implementing varying ambient temperature. The following

GEOMETRIC PARAMETERS OF TRANSMISSION CABLE clauses will determine the consequences of running the system
‘TYPE (A)2XS(FL)2Y 1X1600 RMS/50’ USED IN THIS STUDY [16] under the normal load cycle and during a fall out of TL2.
Conductor dc ti Screen ts* tj De
cross cross A. Normal Load Cycle
section section It is assumed that the normal load cycle in Fig. 5 is repeated
[mm2] [mm] [mm] [mm2] [mm] [mm] [mm]
1600 48.5 15/20 110 0.4 5.9 101
infinitely, thus the figure is only a snapshot of one day. In the
** *** **** simulation process, this has been modelled by applying half
*
ts being the thickness of the screen the maximum allowed steady state current for an infinite
**
Segmental aluminium conductor amount of time, followed by applying the loads of Fig. 5 for
***
The cable is equipped with semiconductive layers, which have been twenty-nine days before the results are obtained. This
assumed to be 2.5mm thick. In the thermal studies these layers are
assumed to have the same thermal properties as the XLPE insulation procedure ensures the thermal system to be in dynamic
and the insulation thus has an effective thickness of 20mm. equilibrium state before the results are produced.
****
The thickness of the screen is calculated on the basis of the screen The time step of the simulation is chosen in such a way that
area and the effective outer diameter of the insulation. In the electrical
the entire loop of dynamic calculations is repeated every ten
calculations the screen is a copper wire screen, however in the thermal
calculations it is assumed that the screen is a solid material as minutes. This means that the conductor losses of all cables are
illustrated in Fig. 2. evaluated based on electrical resistances which are up to ten
minutes old. However, as the temperatures are not expected to
The load L2 is varying daily according to Fig. 5. The
change significantly within this time the approximation is
generator G2 is mainly supposed to supply the load with
considered valid.
active power during peak hours as shown in Fig. 5.
During normal cyclic operation, the thermal limit of
transmission cables is a maximum jacket temperature of 50 oC
in order to avoid moisture migration [12]. The maximum
steady state current (Imax) is calculated based on this
temperature, here 839 A. The current in the cables of TL1 and
the jacket temperature for the centre phase for the thirtieth day
in the simulation, are shown in Fig. 6.

Fig. 5: Load (L2) and generation (G2) cycles of the components seen in Fig.
4. It is assumed that the same cycles are repeated indefinitely.

Generator G1 is forced to be slack and the reactive powers

of G1 and G2 are controlled such as to keep the voltage at the
nodes within the accepted limits.
It is assumed that the two cable systems (TL1 and TL2) are
located far enough from each other, such that they can be Fig. 6: Current and jacket temperature of cable TL1. An initial load of half the
treated as thermally independent. loadability has been applied for an infinite amount of time prior to simulation,
The thermal parameters of the study are listed in Table II. resulting in steady state, thereafter applying the loading profile of Fig. 5 for
TABLE II twenty-nine days before the shown day. Based on the conventional steady
THERMAL PARAMETERS OF TRANSMISSION CABLE state calculations the maximum allowable current would be 839A and the
maximum temperature on the jacket 50oC.
Material ρtherm ctherm
Thermal Specific As expected, it is seen from the figure that the jacket
Resistivity Heat temperature follows the current with a small time lag.
[m∙K/W] [J/(m3∙K)] In order to illustrate the benefits of using ETC as the
Aluminium (Conductor) 0 2.5∙106 operating methodology in the transmission system, instead of
XLPE (Dielectric) 3.5 2.4∙106 current based operation, the results of the simulations have
Copper (Screen) 0 3.45∙106 been normalised according to the maximum steady state
values, Fig. 7.
PE (Jacket) 3.5 2.4∙106
It is seen in Fig. 7 that the temperature lies almost steadily
Soil (Surroundings) 1 1.9∙106 around 0.57 pu, while the current fluctuates highly between
0.35-0.85 pu. This illustrates very clearly the operational
It is assumed that the ambient temperature at cable depth is freedom gained by operating transmission systems based on
steady state 15oC, the methodology though allows for temperature (ETC) instead of the current. In the specific case,
 7

the operator could be allowed to increase the current above

Imax for several hours on a daily basis without jeopardising the
reliability of the system.

Fig. 8: Conductor temperature and current of cable TL1 when a failure occurs
in cable TL2 after 17:00h. Even though the current exceeds the limit by 50%
for several hours each day, the temperature is not critical, and the line can thus
be kept operational.
Fig. 7: Normalised current and jacket temperature of the data shown in Fig. 6.
It is seen that even though the current exceeds its limit by
The current has been normalised based on the current that, during steady state
conditions, would give a jacket temperature of 50oC. The jacket temperature 50%, the system is in a safe state because the conductor
has been normalised based on the maximum allowed temperature of 50oC. It is temperature at most reaches approximately 60 oC within the
clear that the temperature is much below the maximal acceptable level for all first hours and 75oC within the next three days. An extension
times, whereas the current varies a lot and almost reaches the steady state
operating conditions of 839 A.
of the time scale in Fig. 8 shows that the temperature will not
reach 90oC within the first twenty days after the failure. This
B. Failure of Transmission Line time frame can in many cases be sufficient to arrange the
repair of the faulted line, or at least find other measures to
This clause will discuss the benefits of operating reduce the extra loading of TL1.
transmission systems based on ETC instead of currents when
experiencing a failure in the grid. It is from this example recognised that the transmission
Again it is assumed that the load and generation patterns of system becomes more flexible when operated on the basis of
Fig. 5 have been applied for infinitely long time (twenty-nine temperature than when operated on the basis of current. In
days) prior to the day of the fault. At 17:00:00 a fault occurs in current based operation, the operator is forced to disconnect
TL2 which forces the line to be disconnected where after the TL1 at 18:00h, since the current still exceeds 1 pu after one
load is immediately taken over by TL1. hour, while the temperature based operation does not require
The Danish transmission system is operated based on disconnection of TL1.
current schemes and current levels up to 2 pu is normally
allowed for maximally 1 hour during such emergencies, where In market based systems, which is how transmission
after the line must be disconnected. systems in EU are operated today, the system would be in an
For transmission systems operated based on ETC, it is the economically optimal state prior to the failure of TL2.
conductor temperature which is the limiting factor during Disconnection of TL2 will not change this state considerably;
emergency operation. The transmission system is allowed to however disconnection of both TL1 and TL2 will change the
run uninterrupted until the critical conductor temperature is system to an economically non-optimal state. The benefits of
reached and the transmission line must be disconnected. implementing ETC in the system are thus obvious, as it
During emergency operation, the limiting conductor enables flexible operation and optimisation of the costs.
temperature has here been set to 90oC, and it is assumed that
this conductor temperature is allowed for as long as the V. CONCLUSION
emergency operation is necessary. This limitation may result
in jacket temperatures higher than 50oC, however as It is shown in this study how dynamic calculations of the
emergency operation is expected to be short and because thermal conditions in transmission cables can be combined
moisture migration of the surroundings is a process requiring with load flow simulations of a power grid, enabling
some time to evolve, higher jacket temperatures may be electrothermal coordination (ETC). This paper shows how
allowed during emergency operation. dynamic temperature calculations can be implemented into
The ETC tool is seen to be flexible as any temperature limit commercially available power systems software, obtaining a
can be used. Thus if the TSO e.g. only allows a jacket strong tool for ETC based system planning and operation,
temperature of 50oC under emergency operation, this allowing for a significantly increased loadability of the
limitation can easily be implemented. transmission lines. This implementation strategy was chosen
Fig. 8 shows how the temperature and current of TL1 in order to allow for easy integration of ETC into software
evolves before and after the failure, when the system is which is already widely used by distribution and transmission
operated based on ETC. companies. The software development has been closely
 8

followed by the Danish TSO, Energinet.dk, and ETC is from Soil Specific Heat, Thermal Resistivity and Precipitation,” IEEE Trans.
Power Delivery, vol. 28., no. 3, pp. 1909-1917, July, 2013.
Energinet.dk’s point of view seen as ready for implementation
[7] Faruk Aras, Cüneyt Oysu, and Güneş Yilmaz. "An assessment of the
into the Danish transmission system. methods for calculating ampacity of underground power cables,"
Electric Power Components and Systems, vol. 33, No. 12, pp. 1385-
ETC based operation is in this paper studied and compared 1402, October 2005.
[8] Ajit Hiranandani. "Calculation of conductor temperatures and ampacities
to the presently used current based operation of transmission
of cable systems using a generalized finite difference model," IEEE
grids. A two line transmission system is used as a case study Power Engineering Review, Vol. 11, No. 1, pp. 15-24, January 1991
to demonstrate how the temperature in all cables can be [9] G. Anders, "Rating of Electric Power Cables - Ampacity Computations
monitored during normal cyclic operation by utilising dynamic for Transmission, Distribution and Industrial Applications", IEEE Press,
1997.
thermal calculations, thus giving a real time picture of the
[10] IEC Calculation of the cyclic and emergency current rating of cables.
thermal state of the transmission system. By determining the International Electrotechnical Commission Standard Series 60853, July,
real time thermal state of the transmission grid, ETC can 1989
predict the future thermal evolution of the system, and thus the [11] R.S. Olsen, J. Holboell and U.S. Gudmundsdóttir, "Dynamic
Temperature Estimation and Real Time Emergency Rating of
reliability of the transmission system, based on predicted
Transmission Cables", IEEE General Meeting 2012, July, 2012.
loads, can be foreseen. This is especially of great importance [12] IEC Electric Cables - Calculation of the Current Rating. International
during faults in the system, where the loadability of all Electrotechnical Commission Standard Series 60287, December, 2006.
transmission lines can quickly be estimated via ETC, given the [13] Z.S. Tseng, ”Systems of First Order Differential Equations,”
Pennsylvania State University – Department of Mathematics, 2012,
actual thermal conditions, and ETC can show the operator
[Online], Available:
which actions should be taken. http://www.math.psu.edu/tseng/class/Math251/Notes-LinearSystems.pdf
[14] T. Kusada and P.R. Achenbach, “Earth Temperature and Thermal
It is emphasised that the ETC tool has been applied to larger Diffusivity at selected stations in the united states,” U.S. Department of
Commerce/National Bureau of Standards, Tech. Rep. NBS Report 8972,
systems with equal success, however in order to maintain a
June, 1965.
clear and comprehensive overview, the simple two line system [15] H. Brakelmann and G. Anders, “Ampacity Reduction Factors for Cables
was used in this paper for illustrating the effectiveness of Crossing Thermally Unfavorable Regions,” IEEE Trans. Power
ETC. Delivery, vol. 16, No. 4, pp. 444-448, October, 2001.
[16] NKT cables, “High Voltage Cable Systems - Cables and Accessories up
to 550 kV,” [Online], Available:
In summary, the transmission system operator can benefit http://www.nktcables.com/support/download/catalogues-and-
from ETC in cable based transmission grids due to access to: brochures/high-voltage-and-
1. The thermal history of all cables offshore/~/media/Files/NktCables/download%20files/com/HighVolt_e_
200309.ashx, Apr., 2012
2. The present state of the system, including the
consequences of possible control actions
3. The future evolution of the transmission system state, VI. BIOGRAPHIES
based on predicted load patterns Rasmus Olsen received his master of science in
electrical engineering in 2010 from the Technical
The predictions mentioned in 2 and 3 are only possible via
University of Denmark (DTU). His PhD project,
modelling. Monitoring by measurements is thus not sufficient which he began shortly thereafter, is concerned with
for ETC applications. dynamic load optimisation of cable based
transmission grids and it is a collaboration between
Energinet.dk and DTU. His main interests are within
In addition to the above three elements, ETC might also thermal performance of transmission cables and grids.
benefit the planner of transmission grids, since especially R. Olsen is a student member of IEEE.
cable based systems will experience a significantly increased
loadability, both during the normal daily load cycle and during
Joachim Holboell is associated professor and deputy
emergency operation.
head of center at DTU, Department of Electrical
Engineering, Center for Electric Power and Energy.
REFERENCES His main field of research is high voltage
components, their properties, condition and broad
[1] H. Banakar, N. Alguacil and F. D. Galiana, "Electrothermal
band performance, including insulation systems
Coordination Part I: Theory and Implementation Schemes," IEEE Trans.
performance under AC, DC and transients. Focus is
Power Systems, vol. 20, No. 2, pp. 798-805, May. 2005.
also on wind turbine technology and future power
[2] Energinet.dk, “Cable action plan,” Tonne Kjærsvej 65, 7000 Fredericia,
grid applications.
Denmark, Tech. rep., March 2009.
J. Holboell is Senior Member of IEEE.
[3] Århus, Odense og Roskilde Miljøcentre, By- og Landskabsstyrelsen og
Energinet. Dk, “Improvement of the visual impact of the 400 kV grid,”
Unnur Stella Guðmundsdottir received her M.Sc.
Tech. rep., Tonne Kjærsvej 65, 7000 Fredericia, Denmark, ISBN: 978-
and PhD degree in 2007 and 2010 respectively, from
87-90707-72-9, April 2010.
Aalborg University in Denmark. Her PhD studies
[4] J. Ausen, B.F. Fitzgerald, E.A. Gust, D.C. Lawry, J.P. Lazar and R.L.
focused on modelling of underground cable system at
Oye, "Dynamic Thermal Rating System Relieves Transmission
the transmission level. She was a guest researcher at
Constraints", IEEE 11th International Conference on Transmission &
SINTEF in Norway in November 2008 and at
Distribution Construction, Operation and Live-Line Maintenance, Paper
Manitoba HVDC Research Centre in Canada during
25, 2006.
June-October 2009. Currently she holds a position as
[5] M.J.C. Berende, J.G. Slootweg, and G.J.M.B. Clemens. "Incorporating
a senior cable specialist and group leader at
weather statistics in determining overhead line ampacity," 2005
Energinet.dk where she is a deputy manager for
International Conference on Future Power Systems, 2005.
DANPAC, a research project focusing on
[6] R. Olsen, G.J. Anders, J. Holboell and U.S. Gudmundsdóttir,
undergrounding of almost the entire Danish transmission system. U. S.
“Modelling of Dynamic Transmission Cable Temperature Considering
Gudmundsdottir is a Member of IEEE.