Sizing power transformers in power systems planning

using thermal rating

Olivier Arguence, Florent Cadoux

To cite this version:

Olivier Arguence, Florent Cadoux. Sizing power transformers in power systems planning using ther-
mal rating. International Journal of Electrical Power & Energy Systems, 2020, 118, pp.105781.
�10.1016/[Link].2019.105781�. �hal-03260251�

HAL Id: hal-03260251

[Link]
Submitted on 21 Jul 2022

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.

Distributed under a Creative Commons Attribution - NonCommercial 4.0 International License

Version of Record: [Link]
Manuscript_c4654ebc12ff69473fbfcac6fcc46fa2

Sizing Power Transformers in Power Systems Planning Using Thermal

Rating

Olivier Arguencea*, Florent Cadouxa

a
Univ. Grenoble Alpes, CNRS, Grenoble INP, G2Elab, 38000 Grenoble, France

ARTICLE INFO ABSTRACT

Article history: It has already been shown in the literature that power transformers may be more accurately sized by their
- thermal limits than by their rated power limit. In practice however, thermal limits are usually considered
only in operations; but not at planning stage, where the more usual notion of rated power is used. This
Keywords: paper proposes a novel method to take into account (and benefit from) thermal limits directly at planning
Power systems planning stage. This is made possible by quantifying separately the impact of each generator and load on the
Power transformer temperature of the distribution transformer. Decoupling the effect of individual generators and loads is
Smart grid achieved by linearizing and rewriting the analytical expression of the hot-spot transformer temperature.
Dynamic thermal rating The practical value of the method is assessed using a real-world dataset, by estimating the increase in the
hosting capacity of the considered transformer for additional generators and loads. Significant gains are
obtained when the transformer is sized by generation, in particular when photovoltaic (PV) generators are
involved.

1. Introduction reinforcing or adding a transformer, DSOs should find a compromise

between the high capital cost of oversizing it, and the risks (in terms of
Nomenclature
endangering the reliability of electrical energy supply) of undersizing it.
Variables The typical steps of a planning study aiming at sizing a transformer are

Smoothed load factor, = ⁄ 1 +

Load factor (load current/rated current) pu the following. Some assumptions must first be made regarding future
loading conditions, in order to generate a set of “extreme” scenarios for
Thermal load factor, = ⁄ 1+ + ⁄ 1+
pu
load and generation that will be used as stress-tests to size the transformer.
pu
For transformers sized by generation (not consumption), a typical example
Ambient temperature °C
of such a scenario, currently used some DSOs, is to consider that all
Hot-spot temperature °C
generators will output their rated power and that the load will reach its
Hot-spot temperature at rated current °C
lowest possible value [1], while taking into account forecasts of the future

∆
Top-oil temperature °C
evolution of load and generation.

∆
Hot-spot-to-top-oil gradient °C
After loading scenarios have been defined, one of the following two
Top-oil temperature rise °C
categories of methods may be used [2]:

Parameters • The first and simpler one is to simply define physical limits, typically
Thermal model constant on instantaneous active power but possibly on other criteria, and to size
Thermal model constant the equipment so that these limits are not violated in any of the stress-
Thermal model constant tests.
• And the second, more elaborate one, is to perform a so-called “lower
x
Ratio of load losses at rated current to no-load losses
cost optimization” that aims at finding an economical trade-off between
y
Oil exponent
a smaller transformer that will be heavily loaded and undergo
Winding exponent

α
accelerated aging and increased losses, and a larger one that will cost
Transformer constant stating the relative influence of
more but not suffer from increased aging and losses [3-4], [15].

γ
and on
The second method involves estimating the loss of life of the
Benefit from using the thermal criterion %
transformer depending on its loading conditions — see for example [5]
Average oil time constant min
where the impact of high PV penetration on transformer lifetime is

∆
Winding time constant min
studied, and [6-7] where the impact of Electric Vehicles (EVs) on

∆
Hot-spot-to-top-oil gradient at rated load °C
transformer lifetime is investigated. However, predicting with great
Top-oil temperature rise at rated load °C
accuracy the useful lifetime of a transformer based on its loading
condition is currently considered difficult if not impossible [3], and as a
Distribution transformers and power (HV/MV) transformers are
consequence this paper will focus on the first method, namely, sizing
predominant assets in a distribution grid, and sizing them well is a crucial
transformers using physical limits.
task for distribution system operators (DSOs). When replacing,
___________
* Corresponding author at: Univ. Grenoble Alpes, CNRS, Grenoble INP, G2Elab, 38000 Grenoble, France
E-mail address: [Link]@[Link]
© 2019 published by Elsevier. This manuscript is made available under the CC BY NC user license
[Link]
 are a current limit and a hot-spot temperature limit (2).
≤ 1.3
The simplest criterion is to never exceed the rated current (or power). It

(
is the method most commonly used by DSOs, and it is used in several
ℎ ≤ 120°$
(2)
recent research papers on the topic of distribution network planning [8-9].

The limit ≤ 1.3

This method may however be too restrictive since even a current peak of

≤ 1.5
used here for power transformers becomes
short duration above the threshold would be forbidden – although such a
for distribution transformers and medium power transformers.
peak would actually have little impact on transformer temperature, and
thus on aging and deterioration. An improvement over this criterion is 2.2. Presentation of the Thermal Model
thus to use a criterion with thermal limits [3-4], [10] (most importantly on
In a planning study, in order to compute the hot-spot temperature ,
the hot-spot temperature), which makes it possible to benefit from the
one may use the model from the standard IEC 60076-7, presented on
thermal inertia of the transformer, in particularly the thermal inertia of its
Fig. 1. In this paper, we take this well-known model for granted; that is to
oil. This criterion is the core of dynamic thermal rating (DTR) and real-
say, it is assumed that the model is accurate and it is used as a black-box,
time thermal rating (RTTR), two technologies that are studied in [10-13],
without discussing the physical meaning of the equations and parameters.
[16-17] and that allow to operate a transformer closer to its thermal limit
The important question of assessing the value of the various parameters
and to extend its useful life. These two approaches have much in common
contained in this model is discussed in [7].
with the method proposed in this paper, with the crucial difference that
Three temperatures are involved in the model: the ambient temperature
they do not address planning stage but only operations. The already

& is defined as the top oil temperature rise ( & = − ) and

, the top-oil temperature and the hot-spot temperature . Parameter
existing approaches help to monitor the thermal state of the transformer

parameter & is the hot-spot-to-top-oil gradient (& = − ). Then

while operating above the nominal power, but they cannot be used in
planning stage to size a new transformer with its thermal limits – instead
can be computed using (3).
= +& +&
of nominal power. This paper aims at bridging this gap.
This document is organized as follow. A quick study of possible ℎ ℎ (3)
benefits obtained by using thermal limits is carried out in Section 2. The unknowns & and & each depend on through an ordinary
Section 3 introduces a way to formalize the commitments that the various differential equation given here by its Laplace transform (denoted by ℒ ):

1+ - 1
/
distributed generation (DG) owners should take with respect to the DSO
& =& ℒ (, . 0
1+ 1+
in order to enforce the thermal limits. It will be seen that the thermal (4)

−1
criterion needs to be linear; so, the model from IEC 60067-7 will be
& =& ℒ1 - 23 4 − 6
1+ 1+ /
linearized in Section 4.
ℎ ℎ (5)

2. Benefit of a Thermal Criterion in Planning

2.1. Thermal Limit and Sizing Criterion

Currently, transformers are often sized by setting a limit to the maximal

RMS current they can tolerate. The corresponding criterion is that the
instantaneous current , here expressed in the “per unit” system, should
not exceed the rated current:
≤1 (1)
Besides, it is possible to set less constraining limits. According to
IEC 60076-7 [3], it is possible to load the transformer beyond nameplate
rating while keeping a normal ageing rate of the transformer. To achieve
this, the transformer needs to have a “normal cyclic loading”, which is a
loading complying with the three limits from Table 1. Fig. 1. Block diagram representation of the thermal model.

Table 1 2.3. Case Study: Load, PV, Wind

Current and temperature limits applicable to loading beyond nameplate Three examples of currents and temperatures of a distribution
rating for a normal cyclic loading (extract from [3]). transformer are plotted over one day on Fig. 2. In the first case, the
Distribution Medium power Large power
Type of loading transformer is assumed to feed a typical residential load; in the second, it
transformer transformer transformer
is used to connect a wind farm; and in the last one, to connect a PV farm
Current (pu) 1.5 1.5 1.3
on a sunny summer day.

= 42°$ , which is
Winding hot-spot The hot-spot temperature is computed using the IEC model. The
120 120 120
temperature (°C)
would reach 120°$ when = 1 for a long time.
ambient temperature is assumed to be constant,

Note that 40°$ is a standard maximal ambient temperature [4], [14]. One
Top-oil chosen such that
105 105 105
temperature (°C)
may notice that the temperature is “smoothed” with respect to current, as
Three limits are thus defined; one of them is however always less current peaks are damped by the thermal inertia of the transformer.

between the winding hot-spot and top-oil is 120 − 105 = 15°$. In

stringent than the other two. Indeed, the difference of temperature limits For each example, and by construction, the current reaches its

addition, if current exceeds its nominal value, then & ≥ &

maximal value according to criterion (1): according to this criterion, the

& ≥ 15°$ for every type of transformers (see Appendix, Table 2). The
and transformer is fully loaded in all three test-cases, and the remaining
hosting capacity is zero. However, for the load and the PV curves, and
winding hot-spot temperature limit is thus always reached before the top-
oil temperature limit, and there is no need to study the latter. contrary to the wind curve, some hosting capacity margin does remain
So the limits for a large power transformer under normal cyclic loading according to the thermal criterion (2): the red temperature curve does not
 3

reach the 120°$ limit. We conclude that for a transformer sized by a generators whose power output exhibits a clear plateau during sunny days
typical load or by a PV, criterion (1) may lead to slightly oversizing the (the power output seems to saturate, possibly because the power converter
transformer. However, for a wind farm (and by extension all plants except was slightly undersized with respect to the peak power of photovoltaic
PV) there is nothing to gain as the power may remain close to its modules). Also, Fig. 3 shows that the typical value of the hosting capacity
maximum for a long period. increase, for a pure mix of “normal” PV generators that exhibit no power
output saturation, would be in the range of 10 to 40 % for distribution
transformers, and 2 to 20 % for ONAN power transformers.
In a real distribution system however, PV generators would probably
be mixed with other types of generators and loads, solar irradiance would
be different from one place to another, etc. These impediments probably
explain why the hosting capacity gain associated with thermal limits
remains currently untapped: calculating the exact factor by which a given
transformer may be overloaded in order to account for its thermal inertia
would depend on many local details and prove difficult for planning
engineers. The sequel of the paper is thus motivated by the will to find a
practical way to include the thermal inertia of transformers in planning
studies.

3. Thermal Criterion: Limits in Practice

To take into account thermal limits in planning study, the hot-spot

temperature must be calculated. Two cases are considered as this
computation will not be done the same way if the transformer just feeds a
local PV plant, or feeds a distribution grid. In the first case, the
transformer feeds a pure mix of PV generators, and the transformer owner
Fig. 2. Evolution over one day of the current of a transformer and its computed knows every technical detail of the PV plant, which makes planning
temperature with (a.) a load, (b.) a wind farm, (c.) a PV farm. easier. In the second case, loads and generators are mixed, and the utility
2.4. Benefit of Thermal Criterion for a PV Farm company only has access to partial information about the current and
future mix of generators and loads that the transformer will serve, which
In this section, the benefit of using a thermal limit criterion (2) over a makes planning more complex.
limit on the current criterion (1) is computed. This study is carried out
3.1. Transformer Feeding a single PV Plant
with a dataset of 518 generation curves obtained from real-world
measurements over one year for existing photovoltaic generators and For a local PV plant, it should be possible to forecast the power
using the parameters of two different transformers: an ONAN distribution produced by the plant under optimistic solar irradiation conditions. Then,
transformer (MV/LV) and an ONAN power transformer (HV/MV). using measured parameters of the transformer or standard parameters, the
The indicator chosen to measure the benefit is the ratio γ between the IEC model [3] may be used directly to compute the future hot-spot

current under criterion (1), namely 89: = 1 . For example, if the

maximal allowable current using criterion (2) and the maximal allowable temperature and check whether the transformer reaches its maximum

maximal current using (2) is = 1.1 then the ratio ; is 1.1, and using
allowed value. The histogram from Fig. 3 gives an idea of the possible
benefit for a transformer loaded by solar panels. Provided that the
the thermal criterion yields a hosting capacity increase of 10 %. technical details of the transformer and PV plant are accurately known,
the only caveat of this approach is that the ambient temperature might be
difficult to forecast. Techniques are presented in [4] to tackle this
problem.
Frequency of occurence

3.2. Transformer Feeding a Distribution Grid

A transformer in the public distribution network may aggregate a lot

obviously impossible to forecast the current - that will flow through

of different producers and consumers, who change over time, so it is

the transformer over its entire lifetime, with sufficient accuracy to directly
compute the temperature using IEC model [3]. As observed above, this is
probably the reason why the hosting capacity gains that may be unlocked
by using thermal limits at planning stage are currently untapped: DSOs
generally do not use the thermal criterion (2) today, but use a limit on
rated current or power.
We argue that the current impossibility to use a thermal criterion at
Fig. 3. Histrogram of the benefit of thermal criterion γ for 518 PV.
planning stage fundamentally results from the following fact: the
The histogram of γ spreads over a wide range of values. Looking into expression of hot-spot temperature from the IEC model [3], described by
the details of the simulations that yielded extreme values of γ, we Fig. 1, is not proportional to the current. Indeed, with a linear expression,
observed that the lowest values of γ were obtained for photovoltaic it would be possible to study separately the impact of each generator and
4

load, and then sum up everything to check if the constraint is satisfied. In

particular, individual “thermal constraints” could be set to individual
customers in addition to their maximum power or current limit, and
implementing these individual limits would guarantee that transformer
temperature remains acceptable.
In Section 4.3 below, we investigate how a DSO could set a new
limit within the grid connection contract of each individual customer. This
new limit would be less constraining than the original limit on RMS
current or active power. In order to fulfil the new constraint, customers
could decide to implement load or generation curtailment, although this
would by no means be mandatory.
To get a linear criterion that is equivalent to a limit on the
temperature, a 1st order Taylor expansion will be used. The mathematical
proof of this criterion and its analysis is the subject of Section 4.
Fig. 4. Variation of the load factor , the thermal load factor -> and the hot-spot
4. Linear Thermal Criterion temperature during a PV daily load cycle for a distribution transformer of type
“ONAN”.
4.1. Calculations to get a Simple Linear Criterion
In the end, the sizing criterion (2) for power transformers becomes (10)
To get a simpler and linear criterion, calculations are performed and
which is called “linear thermal criterion”.
≤ 1.3
can be separated in three steps (each one is detailed in appendices):
(
-ℎ ≤ 1
The expression of is simplified and is neglect, see Appendix 6.1. (10)
1. is linearized around its nominal working point, see Appendix 6.2.
2. The thermal limit studied is changed to incorporate the ambient 4.2. Interpretation of the Thermal Model
temperature , see Appendix 6.3.
Three parameters determine the value of using a thermal criterion
In the end, a new indicator defined by (7) is introduced. This new
instead of a limit on instantaneous current:
≤ 120°$:
indicator is called the “thermal load factor”. The new thermal criterion is
• Firstly, the shape of loading curve. The longer the current curve stays
- ≤ 1
(6) which is a replacement of the limit in temperature
ℎ (6) close to its maximum value, the higher , which is detrimental, the
In the definition (7) of , the new variable is a “smoothed load worst-case scenario being that the current remains high for so long that
would reach its steady-state value. All benefits of using a thermal
factor” (subscript “o” stands for oil) defined by (8) and is a constant
criterion would then be lost. This is what happened for a pure mix of

1
specific to the considered transformer, defined by (9).

ℎ: = +
wind generators on Fig. 2.

1+ 1+
(7) • Secondly, the oil-time constant, or more exactly the product : a

:=
larger time constant makes smoother and is thus beneficial.

1+
(8) • A third crucial parameter is α. It is also determined by the physics of the

2
& : −& = −1
considered transformer; the higher its value, the more is weighted in

:= 1+
& =
ℎ (9) , which is beneficial.
ℎ 4.3. The Linear Thermal Criterion in Practice
Parameter α expresses the relative weight of both variables and Most importantly, the new thermal criterion is a linear expression, an

with weights 1⁄ 1 +
(see typical values of in Table 2). Equation (7) expresses that is the indispensable property to decouple the study of various loads and

⁄ 1 + , and it should remain below 1

arithmetic weighted mean of and and generators connected to the transformer under consideration. Indeed, in
according to (6). the common situation where information about the total transformer
The dynamics of , and is illustrated by Fig. 4, showing the current is available as a sum of individual generation and consumption
close dynamic behaviour of and . profiles ? , the individual thermal load factor ,? may be evaluated
separately and then summed to check constraint (10). For further
simplification, (10) may be conservatively replaced by (11).

DE 89: ≤ 1.3
B ?
?

CE 89:F ℎ,? G ≤1
(11)
B
A ?
For DSOs the latter formulation (11) is even easier to use than (10):

89: ? and a maximum thermal load factor 89:F ,? G, that only have
to each generator or load, we simply associate a maximum current

to be evaluated once for each generator separately. Then criterion (11)

boils down to summing them; a very quick check that may be sufficient to
ensure the absence of a constraint. Another advantage of the decoupled
formulation (11) is that it provides a basis on which the behaviour of
individual generators could be specified in a contract with the DSO. While
 5

it is impractical to contract with all customers as a whole and require them calculated by linearizing in the vicinity of its nominal value. The
to collectively ensure that constraint (10) will be satisfied, it is much more method is dependent on the IEC and IEEE models and its accuracy. The
practical to contract with each generator individually, which is what linearity of the proposed criterion is a crucial characteristic in practice, as
formulation (11) makes possible. it allows the DSO to contract with each generator and load individually. A
case study gives an example of how to use this linear criterion in practice,
In practice, computing ,? would be straightforward and could be a
and the expected benefits. The proposed criterion can also be used for
new feature embedded inside a standard smart meter without requiring
existing transformer in order to size a new PV installation. The paper also
any thermal sensor. The meter could be set to open its internal circuit
provides several figures on the hosting capacity gains which can be
breaker whenever constraint (11) is reached, allowing for a practical
expected by sizing transformers with thermal limits. Benefits are
enforcement of the thermal limit. Observe that this scheme would allow
substantial when the transformer is sized by generation including PV,
enforcing the global constraint — not overheating the transformer — by
which will be more and more common in the future. In the end, using this
coordinating the actions of individual customers without communication.
criterion would facilitate the installation of low-carbon technologies like
This is in contrast with the technology of real-time thermal rating [12-13]
PV in locations dominated by generators, as it would reduce the cost
where thermal constraints must be managed in real-time; implementing
induced by transformers.
such a scheme requires that some “active resources”, loads and/or
generators, are available to the DSO upon request, and thus calls for a
6. Appendices
communication link with these resources.
6.1. Simplification of the thermal model
4.4. Case Study: Transformer Sized by Generation
According to the value of and from Table 2, for every type of
A HV/MV power transformer has several feeders. The first is medium and large power transformer, equation (12) holds.

⁄ =
connected to a 10 MW plant with wind generators, the second to a 10 MW
plant with PV generators, and all the other feeders are only connected to (12)
loads, with a maximal consumption of 15 MW, and an estimated minimal
For a distribution transformer, (12) no longer holds but, as is equal
consumption of 3 MW. To simplify, a constant voltage will be assumed so to 1, the fraction (in the upper-right block on Fig. 1) using N⁄ 22
K can be assimilated to a power. Then, the two sizing methods can be vanishes. It can be deduced that N⁄ 22 may be replaced by 11 N for all
compared: types of transformers.

• The current method, based on power limits. Both plants would commit Table 2 Example characteristics related to the load ability of transformers

flowing through the transformer would be ∑ = 10 + 10 − 3 =

to limit their maximal power to 10 MW. Then the maximal power (extract from [3], Annex E).

17 JK. So a transformer with a nominal power of at least 17 MW

Distribution
Power Transformers
Transformer
would be needed.
• The proposed method, based on thermal limits. The wind plant would ONAN ONAN ONAF OF OD

x
commit to a 10 MW capacity on both and , but the PV plant
0,8 0,8 0,8 1 1
y
would rather commit to a 10 MW capacity on and to a 9 MW capacity
Recommended Values

on . The capacity would be computed beforehand, based on the 1,6 1,3 1,3 1,3 2
time-profile of the power generated by the plant, and on the transformer 1 0,5 0,5 1 1
parameters (given by the DSO). The value of 9 MW is an example
1 2 2 1,3 1
∑ = 10 + 10 − 3 = 17 JK and ∑ = 10 + 9 − 3 = 16 JK. So,
value consistent with Fig. 3. Then the power balance gives:
2 2 2 1 1
min
a transformer with a nominal power of at least 16 MW would be enough
180 210 150 90 90
min
to satisfy criterion (11).
4 10 7 7 7

R
In the end, 1 MW are gained on the sizing of the transformer by
considering its thermal capacity instead of its power capacity. In this 5 6 6 6 6
∆ °$
example, gains are only linked to the PV generation. Further gains might
Example

23 26 26 22 29
∆ °$
be achieved, but they are expected to be either negligible, either too
complicated: 55 52 52 56 49

• The of the load might also be taken into account and would α 2 0.55 0.55 2.4 1.5
potentially be strictly lower (e.g. -3.1 MW instead of -3 MW).
• Criterion (10) would give higher gains than (11), but it would be too Two time constants still need to be managed: and . It can
complicated, if not impossible, to use in planning studies — this is be noted that for all transformer types, the winding time constant is
because as stated above, it does not decouple individual loads and much lower (a few minutes, see Table 2) than . In addition, for the
generators. purpose of sizing transformers, we consider that load variations over a
timeframe of a few minutes are too short regarding the dynamics involved
5. Conclusion (see e.g. Fig. 2); its impact would be negligible. For these reasons,

1
is neglect by approximating

≅ 1,
This paper proposes a method to size a power transformer using its

1+
thermal limit directly at planning stage. The method relies on the linear (13)
criterion (11) which is based on so-called thermal load factor . is
and the expression of (14) is obtained.
an image of the hot-spot temperature and its expression has been
6

−1
≅& ℒ1 - 23 4 − 6
1+
• to further simplify the mathematical expression of the constraint,
ℎ ℎ
• and to take the ambient temperature into account only implicitly,
1+ - 1
/
+& ℒ (, . 0
1+ 1+
through the fact that the rating of the transformer is assumed well
(14)

1 −1
chosen for the geographical area.

+ & = ⋅ ℒ1 − 13 4 − 6
1+ 1+
2 1
ℎ

+& : ℒ1 − 13 ≤0
(18)
1+ 1+
6.2. Linearization

ℒ1 − 13
ℒ1 − 13 + ≤0
As a second step, (14) is further simplified by linearizing it. Since, in

1+
the context of sizing a transformer, it is important to get a good behaviour (19)

= 1 (15), see illustration on Fig. 5. 1

for close to 1, so the 1st order Taylor expansion is used in the vicinity of

- + - ≤1
−1 1+ 1+
(20)

≅& ℒ11 + = - − 1 34 − 6
1+ - ≤1
2 1
ℎ ℎ

+& ℒ T1 + : - −1 U
ℎ (21)

1+ 1+
(15)
1
References
+
1+ [1] ‘Enedis-PRO-RES_06E’, Enedis Technical Document [Online]
[Link]
RES_06E.pdf, assessed 08 August 2018
[2] Willis HL. Power Distribution Planning Reference Book, Second
Edition. CRC Press; 2004.
[3] Power Transformers Part 7: Loading Guide for Oil-Immersed
Power Transformers, International standard IEC 60076-7, 2005.
[4] IEEE Guide for Loading Mineral-Oil-Immersed Transformers and
Step-Voltage Regulators - Redline. IEEE Std C5791-2011
(Revision of IEEE Std C5791-1995) - Redline 2012:1–172.
[5] Pezeshki H, Wolfs P. Impact of high PV penetration on
distribution transformer life time. 2013 IEEE Power and Energy
Society General Meeting (PES), 2013, p. 1–5.
doi:10.1109/PESMG.2013.6672074.
[6] Turker H, Bacha S, Chatroux D, Hably A. Low-Voltage
Transformer Loss-of-Life Assessments for a High Penetration of
Plug-In Hybrid Electric Vehicles (PHEVs). IEEE Transactions on
Power Delivery 2012;27:1323–31.
doi:10.1109/TPWRD.2012.2193423.
[7] Gao Y, Patel B, Liu Q, Wang Z, Bryson G. Methodology to assess
distribution transformer thermal capacity for uptake of low carbon

Fig. 5. Δθo and Δθh Taylor expansions for a distribution transformer.

technologies. IET Generation, Transmission & Distribution
2017;11:1645–51. doi:10.1049/iet-gtd.2016.0722.
[8] Rupolo D, Jr BRP, Contreras J, Mantovani JRS. Medium- and
6.3. Change of the considered thermal limit low-voltage planning of radial electric power distribution systems
The temperature that would be reached with a constant load = 1
considering reliability. IET Generation, Transmission &
Distribution 2017;11:2212–21. doi:10.1049/iet-gtd.2016.1478.

denoted by :
(and that still depends on through a 1st order differential equation) is [9] Dias FM, Canizes B, Khodr H, Cordeiro M. Distribution networks
planning using decomposition optimisation technique. IET

- := & - +& - + ℒ -1 4 6
Generation, Transmission & Distribution 2015;9:1409–20.

1+
(16) doi:10.1049/iet-gtd.2014.0860.
ℎ ℎ
[10] Lachman MF, Griffin PJ, Walter W, Wilson A. Real-time
dynamic loading and thermal diagnostic of power transformers.
Then, (17) is obtained by combining (15) and (16). IEEE Transactions on Power Delivery 2003;18:142–8.
≅ doi:10.1109/TPWRD.2002.803724.
−1
+& = ⋅ ℒ1 − 13 4 − 6
ℎ ℎ
[11] Douglass DA, Edris AA. Real-time monitoring and dynamic
1+
thermal rating of power transmission circuits. IEEE Transactions

2 1
ℎ
(17) on Power Delivery 1996;11:1407–18. doi:10.1109/61.517499.
+& : ℒ1 − 13
1+ 1+
[12] Degefa MZ, Humayun M, Safdarian A, Koivisto M, Millar RJ,
Lehtonen M. Unlocking distribution network capacity through
real-time thermal rating for high penetration of DGs. Electric

temperature is the constant value 120°$. Assuming that the rating of the
We started from constraint (2) in which the maximum hot-spot Power Systems Research 2014;117:36–46.
doi:10.1016/[Link].2014.07.032.
transformer was well chosen, this hot-spot temperature limit should not be [13] Greenwood DM, Gentle JP, Myers KS, Davison PJ, West IJ, Bush
JW, et al. A Comparison of Real-Time Thermal Rating Systems in

≤ 120°$.
exceeded for any plausible ambient temperature curve in the considered the U.S. and the U.K. IEEE Transactions on Power Delivery

≤ 120°$ can be dropped and conservatively replaced

geographical area. Said otherwise, it can be assumed that 2014;29:1849–58. doi:10.1109/TPWRD.2014.2299068.

≤
Thus, constraint [14] ‘Power and Distribution Transformers’, in Low Voltage and
by (18) or more compactly (21). This approximation allows Medium Voltage Products, Schneider Electric, Catalogue 2015.
[15] Biçen Y, Aras F, Kirkici H. Lifetime estimation and monitoring of
both: power transformer considering annual load factors. IEEE
 7

Transactions on Dielectrics and Electrical Insulation [17] Yip HT, An C, Lloyd JG, Taylor P, Michiorri A, Jupe S, et al.
2014;21:1360–7. doi:10.1109/TDEI.2014.6832284. Dynamic thermal rating and active control for improved
[16] Degefa MZ, Humayun M, Safdarian A, Koivisto M, Millar RJ, distribution network utilisation. Managing the Change, 10th IET
Lehtonen M. Unlocking distribution network capacity through International Conference on Developments in Power System
real-time thermal rating for high penetration of DGs. Electric Protection (DPSP 2010), 2010, p. 1–5. doi:10.1049/cp.2010.0213.
Power Systems Research 2014;117:36–46.
doi:10.1016/[Link].2014.07.032.