Assessment of dynamic transformer rating, considering current and

temperature limitations
Ildar Daminova,b*, Anton Prokhorovb, Raphael Cairea, Marie-Cécile Alvarez-Héraulta
a
Univ. Grenoble Alpes, CNRS, Grenoble INP, G2Elab, 21 Avenue des Martyrs, 38000, Grenoble, France
b
Power Engineering School, Tomsk Polytechnic University, 7, Usov street, 634034, Tomsk, Russia

Abstract
Today, smart grids allow system operators to manage transformers close to their thermal limits.
The state of the art in transformer thermal limits are Dynamic Thermal Rating (DTR). DTR can be
based on two temperature limits: a design temperature or maximal-permissible temperature.
Although DTR with maximal-permissible temperature provides a larger loadability, many papers
estimate DTR with design temperature only. Therefore, the true DTR is still underestimated since
temperature limit is ignored as well as current limits in some papers. Moreover, many papers rely
on the conservative assumption of typical historical load profile or net load profile (considering the
distributed generation, storage, electric vehicles). However, DSO in modern realities can control
the shape of (net) load profile. This can make a DTR estimation, assuming some typical (net) load
profile, valuable only for this particular shape of (net) load profile and not for its modifications.
This paper suggests a DTR feasible region which is constructed from current and temperature
limitations and does not rely on typical load profiles. As a case study, we investigate DTR in cold and
warm climates: one case in Russia with continental climate and another in France with temperate
climate. In these climates, we assess DTR for the most common combinations of current and
temperature limitations, used in standards and literature. As a result, DTR can ensure an additional
capacity up to 45% enabling e.g. DSO to defer primary substation upgrades which can cost from 500
k€ to 1.5 M€. DTR, based on temperature limits, are 100 % of time of higher than nominal rating.
Moreover, the quantification of limiting factors revealed that optimization formulation should include all
current and temperature limitations and not only some of them as it is used in many papers.
Keywords: Dynamic thermal rating, transformer, limiting factor, current limit, temperature limit,
capacity

1. Introduction
Today, system operators manage a power For many years, thermal ratings have been
system with a large share of distributed energy defined as Static Thermal Ratings (STR) [6]. STR
resources (DER) [1]. To avoid congestions caused is a constant limit, expressed in current or power
by new DER, system operators have to know the units, and usually calculated for daily-mean
true thermal ratings of existing network. This is ambient temperature (Tamb). For instance, Table 1
especially relevant since the lead time of DER is shows STR of power transformers used for power
much less than that of network reinforcement [2]. system scheduling in Russia [7].Similar STRs are
Besides, thermal ratings reduce a feasible region of used in other countries.
many optimization problems related to power Table 1: STR of power transformer as a function of
system scheduling [3] and a hosting capacity [4] daily mean Tamb
Mean Tamb,
among others [5]. Therefore, the more accurately -20 -10 0 +10 +20 +30 +40
℃
the thermal rating is determined, the more efficient
STR, pu 1.20 1.20 1.15 1.08 1.0 0.91 0.82
solution of optimization problem can be found.
*Corresponding author, email: [email protected] 1
 Despite the long history of using STR, it has paper life is then equal to this service life. However,
been repeatedly proven that STR is a rough if transformers operate at HST limit all the time then
approximation of true thermal ratings [8–11]. This is the insulation life will be reduced at least 12 times
explained by the fact that STR does not consider faster for non-thermally upgraded paper and 2.7
the shapes of Tamb and load profile. Consequently, times faster for thermally upgraded paper
many scientists investigate Dynamic Thermal correspondingly [24], [25]. Consequently,
Rating (DTR) [8,12–16]. DTR represents a daily transformers should operate only short time at HST
profile of admissible loadings, expressed either by limit. Thus, we can define two types of DTR : DTR,
current or power limits. DTR is usually higher than based on the design HST [28] and DTR, based on
STR but for extreme ambient conditions DTR can the HST limit [29]. The latter has a larger loading
be set lower than STR to avoid overheating the capability due to higher temperature limit. At the
network equipment and/or mechanical damage. same time, DTR based on design HST allows to
DTR is investigated for almost all network avoid any increased HST, but this can lead to less
equipment [17]: power and distribution transformers efficient optimization solution.
[16], overhead lines [18–20] and cables [21]. The existence of two DTR poses a problem:
This paper focuses on DTR of oil-immersed both DTR should be estimated in the geographical
transformers. Unlike other network elements, oil- area. At the same time, obtained DTR should be
immersed transformers have a significant loading considered as representative if DTR estimations
(overloading) capability above their nominal rating are based on long-term data (not few days or
[22]. This overloading capability allows operating a weeks). Only few studies estimate transformer
transformer above nominal ratings without limits in different areas over a long-term period. For
sacrificing its design life [23]. To ensure safe instance, authors of [30] estimate a loadability of
operation of a transformer, especially above ONAF power transformer for a typical load profile
nominal rating, one should meet current and and 30 year monthly-mean Tamb in Turkey. In [31],
temperature limitations [24]. Normally, these authors estimate DTR for ONAN power transformer
limitations are set for the current, hot spot based on the annual Tamb in United Kingdom. The
temperature (HST) of a winding and top-oil one-year DTR benefits are studied in [32] for
temperature (TOT) in the tank. For instance, Table ONAN and OFAF power transformers in Milton
2 shows current and temperature limitations given Keynes, United Kingdom. The paper [33] assesses
in international (IEC, IEEE) and Russian standards peak-load transformer capability based on three
for normal cyclic loading (i.e. without accelerated months hour Tamb and load data in Manitoba,
ageing). Canada. The data-driven approach is suggested in
Table 2: Current and temperature limitations given in [34] to estimate residential transformer overloading
standards based on two-year Tamb in West Canada. The
Limiting International Russian paper [35] investigates the loadability of ONAN
parameters IEC IEEE STO GOST GOST power transformer in Egypt based on the mean-
[24] [25] [7] [26] [27] daily Tamb at each month.
Current, pu 1.5 2 - 1.5 1.5 Although two types of DTR exist, none of the
HST, ℃ 120 120 138 140 140 papers above estimates them simultaneously.
TOT, ℃ 105 105 105 95 105 Mostly all papers [30–32,35] consider DTR based
on design HST i.e. DTR based on HST limit is
omitted. Therefore, a real DTR capability is not fully
Despite these standard limitations, many
revealed. Meanwhile, the paper [34] estimating
scientists use a design HST as an alternative
DTR based on temperature limits does not consider
temperature limit. This design HST is equal to 98 ℃
limitations of current. The lack of current limit
for non-thermally upgraded paper and 110 ℃ for
makes the DTR estimation incomplete. Thus, no
thermally-upgraded paper whereas the HST limit
study mentioned above investigates DTR in
for both papers is set from 120 ℃ (see Table 2).
accordance with temperature and current
Transformers can continuously withstand a design
limitations simultaneously. Meanwhile, the limiting
HST during the full-service life since the insulation
factor of DTR can shift from current to temperature

2
within one day and especially during a year. shapes of (net) load profile, which is especially
Consequently, DTR estimation, in terms of power important in the context of smart grids. (2) An
limits, is scientifically challengeable and not effective Tamb range is determined for each limiting
evident. It is less evident if one can choose factor separately. (3) DTR are estimated for the
different temperature and current limitations (see most common combinations of current and
Table 2). Thus, DTR estimation with various current temperature limitations based on long-term Tamb.
and temperature limitations is a first problem, which The paper is organized in the following way: in
we address in this paper. section 2 we show how feasible region can be built
The second problem that we address is the and used to estimate DTR for a majority of load
issue of a typical load profile in estimation of DTR. profiles. In the same section, we present distinct
Many papers [30,33,34,36–42] assume a typical feasible regions with different limiting factors. In
shape of load profile and they upscale this shape section 3, authors estimate DTR for the most
until temperature limits are met. The upscaled common combinations of current and temperature
shape of load profile is then considered as DTR. limitations in Tomsk and in Grenoble.
However, we believe that the assumption of typical 2. The feasible region
load profile becomes outdated in the modern era of The subsection 2.1 introduces a feasible region
smart grids. In the past, typical load profiles were of loadings with two goals: (1) to consider load
indeed relevant since DSO could not actively profiles in accordance with current and temperature
control the shape of load profile and DER share limitations; (2) to distinguish DTR based on design
could not affect this load shape. However, fast HST and HST limit. In subsection 2.2, we present
integration of DER into distribution network examples of a feasible region limited by each of the
changes the shape of a typical load profile. Thus, a factors described.
load profile of consumers is not anymore a
reference for transformers but a net load profile, 2.1. The feasible region of transformer
whose shape is determined by DER and load. loadings
Whatever a (net) load profile is given, modern The term “feasible region” is introduced to
DSO can actively change a shape of (net) load consider all admissible loadings in accordance with
profile, using controllable DER: distributed current and temperature limitations. It is taken from
generation, storage and demand response [43–48]. the mathematical optimization area [51]. Generally,
Aggregators, the new market players, already the feasible region represents a set of all possible
provide such services to DSO [49],[50]. This poses solutions of an optimization problem satisfying all
another problem for estimation of DTR: DTR should given constraints. In case of a daily transformer
not target single typical (net) load profile but loading, these constraints are current and
multiple possible shapes of (net) load profiles. temperature limitations (Table 2). However, one
This paper suggests to use the DTR feasible cannot draw a DTR feasible region in one x-y axis
region depending only on Tamb to overcome two until their limitations are given in different physical
above mentioned problems. Thanks to feasible units: pu (or A) for current and ℃ for temperature.
region, we can estimate thermal ratings of a power We remind that DTR, as any thermal rating, is
transformer without using a typical (net) load traditionally measured in the units of current/power
profile. The paper estimates DTR in cold and warm or their per units. Thus, temperature limitations
climates: one in Russia (Tomsk city in Siberia) with should be also presented in the units of
continental climate and another in France current/power or their pu as a function of Tamb.
(Grenoble city in Alpes) with temperate climate. Such expression is more convenient for energy
The paper focuses on the most common specialists working with network operation and
combinations of current and temperature limitations transformer operation in particular.
given in Table 2. To harmonize the units, one should know
Contributions of this paper are: (1) we suggest thermal characteristics of a transformer. In this
to use a DTR feasible region considering both paper, we use thermal characteristics of an ONAF
current and temperature limitations. Thus, the power transformer given in the IEC standard (Table
feasible region encompasses many possible 3) [24],[52].

3
Table 3: Thermal characteristics of an ONAF power To draw such dependencies (Fig.1), one can
transformer use a basic algorithm suggested below.
Parameter, units Value
Oil exponent, no unit x 0.8 Algorithm: Dependencies between loading and Tamb:
Winding exponent, no unit y 1.3
Loss ratio, no unit R 8 For each Tamb from array [-50 ℃ . . . +50 ℃]
Oil time constant, min τo 150 Set the constant Tamb & STR= 0.01 pu & Temperature limit
Winding time constant, min τw 7
Design hot spot temperature, ℃ θh 98 While Temperature is not equal to Temperature limit
Design ambient temperature, ℃ θa 20 STR = STR + ΔSTR,
Hot-spot to top-oil gradient at rated where ΔSTR can take any small value
current, K Δθ hr 35
Calculate Temperature by IEC thermal model of
Top-oil temperature rise, K Δθor 45 given ONAF transformer
Thermal constant, no unit k11 0.5
Thermal constant, no unit k21 2 end
Thermal constant, no unit k22 2 end
Fig. 2 shows a limiting factor for different
To convert temperature limitations into
combinations of current and temperature
equivalent loading limits, one should build well-
limitations. We take the most common
known dependence between steady-state loading
combinations given in standards and literature
and Tamb (Fig. 1).
(shown at left side of Fig.2).

Figure 1: Transformer loadings equal to HST and TOT limits as a function of Tamb

Figure 2: Limiting factors in the range of Tamb

4
 As we see in Fig. 1, the lines, corresponding to based on design HST). To show it, we add a line
different limiting factors, are crossing each other at (loadings), corresponding to the design HST = 98
some critical Tamb (black dots in Fig. 1). The black ℃ (the calculation is similar to HST limit). The area
dots divide the crossing lines into few Tamb below this new line is shown by green color, the
sections. The limiting factor at each section other part, above this new line, remains yellow (Fig.
represents the lowest line among lines 4).
corresponding to current and temperature
limitations. For instance, the yellow line,
corresponding to HST = 120 ℃, crosses the current
limit at Tamb -17 ℃ and crosses TOT = 95 ℃ at Tamb
+45 ℃. This means that:
 for Tamb -17 ℃, the HST = 120 ℃ is the
limiting factor
 for -17 ℃ Tamb +45 ℃, the current limit =1.5
pu is the limiting factor
 for Tamb +45 ℃, the TOT = 95 ℃ is the
Figure 4: Same feasible region, but showing the
limiting factor.
loadings with accelerated ageing (yellow area) and
We point out that these results are obtained for normal ageing (green area)
studied ONAF power transformer (Table 3) and not
for all ONAF transformers. Latter ones can have The green area represents DTR based on a
distinct critical Tamb (black dots) due to the design HST whereas the yellow area represents
transformer design variations. DTR based on HST limit. The difference between
To build a feasible region for given ONAF two DTRs is always around 0.2 pu i.e. 20 % of the
transformer, we suggest to plot power limits for nominal rating. In other words, if DTR is based on
current and temperature limitations independently design HST only then the 20 % of transformer
from each other. This means that for each given capacity is actually neglected.
value of Tamb profile we find a loading in Fig. 1 Although this feasible region conforms with
corresponding to e.g. current =1.5 pu, HST limit = temperature and current limitations, it does not
120 ℃ and TOT limit = 105 ℃. The lowest line necessary mean that all load profiles inside of it
among three lines gives us a top line of the feasible cause a normal ageing. Some load profiles, being
region. Fig. 3 shows the daily feasible region in feasible region, may exceed green area (the
(yellow area) calculated for given Tamb profile. design HST) for substantial time and therefore
cause an accelerated ageing. However, IEC
standard clarifies the accelerated ageing: “this is
not serious if there are otherwise long periods of
time (usually the case) at relatively low hot-spot
temperatures”. This citation refers to example,
given in IEC standard, where cumulative ageing
exceeds 74 times a normal ageing. Indeed, such a
high ageing rate in one day can be compensated
by low ageing during other days, week, month. IEC
suggested this assumption at the time when DSO
Figure 3: Feasible region of transformer loadings cannot actively control the shape of load profiles.
(yellow area)
Nowadays, however, DSO can control the shape of
a loading profile. Therefore, this IEC assumption
In this example, HST line is lower than the
becomes even more realistic. Thus, suggested
current or TOT line i.e. HST limit is reached earlier
feasible region is primarily determined for current
than current and TOT limit. However, it would be
and temperature limitations and implicitly for
useful to define a part of this feasible region, which
ageing.
causes the normal insulation ageing (i.e. DTR

5
 Figure 5: Interrelations between transformer loading and temperature

It is worth explaining why steady-state loading Therefore, the suggested feasible region is still
represents a top line of feasible region although valid for current and temperature limitations.
transformer can be subject to transient Another important observation should be
temperatures. To do that, one should address the discussed. There is a specific load profile(s) (brown
Fig. 5, showing the interrelations (circles 1-4) line in Fig. 5) with the loading higher than steady-
between load profiles (left side) and their state DTR at some time moment but their transient
temperatures (right side). Temperatures are temperatures still remain below steady-state
calculated by IEC thermal model of the studied temperature (see circle 4). Therefore, one can state
ONAF transformer. One can notice that after blue that the feasible region, obtained earlier in Fig. 4,
load steps up, the transient blue temperatures can be actually even higher in terms of loadings.
(circle 1) reach a steady-state value (purple lines, Although we agree with this statement, we point out
circle 3) without exceeding it. This interrelation that such a specific load profile can be obtained
allows us to formulate an important conclusion: if only under DSO control (reducing a transformer
loadings are always below the steady-state loading loading just after its HST reaches its steady state
then transient temperatures are also below the limit). Therefore, such load profiles have the
steady-state temperature. Therefore, load profiles advantage for short-term planning only [53],[54].
do not exceed the design HST if their loadings are Thus, we neglect them in the suggested feasible
located in the green area. Similarly, a load profile region, which we use for long-terms estimation of
does not violate the HST limit, if its loadings are DTR. Moreover, this neglection actually reduces a
located in the yellow area. feasible region, allowing us to estimate DTR with
Important remark should be discussed. The margin.
green area in Fig. 4, should be considered as area Summarizing above-mentioned results: all load
without the accelerated ageing, only for the load profiles located in green area only are always
profiles, fully located within the green area. If even feasible for both normal ageing and
a small part of a load profile is located in yellow current/temperature aspect. In contrast, all load
area, the green area should not be referred to the profiles located in yellow area only are always not
area without accelerated ageing. For instance, the feasible for normal ageing but feasible for
circle 2 in Fig. 5 shows that even if the load is current/temperature limitations. Load profiles,
instantaneously reduced, the temperatures takes located in yellow and green areas, could be either
time to reduce to new steady-state value. This feasible or infeasible from ageing point of view but
explains why the transient temperature could be always feasible for current/temperature limitations.
still in yellow areas while a loading returned back to Thus, we find the feasible region that always
the green area. Despite the fact that this complies with temperature and current limitations.
interrelation affects the ageing, it does not affect This compliance with current/temperature
the feasible region of current or HST limit. limitations allows us to state that the suggested
feasible region encompasses multiple load profiles.

6
This allows us to avoid using a typical load profile TOT limit only: Fig. 7 shows the feasible region
for DTR determination. built for limitations HST 140 ℃ and TOT 95 ℃ and
Main results of this subsection: (1) temperature Tamb profile happened on July 07, 2018 in Grenoble
and current limitations are expressed in the same France.
units – pu. This allows to (2) plot the feasible region
of admissible loadings considering majority of load
profiles of given ONAF transformer. (3) the areas of
normal and accelerated ageing are determined and
their limitations are explained.

2.2. Examples of feasible regions with

different limiting factors
In this subsection we describe feasible regions
with different limiting factors. We identify 6 possible
combinations of limiting factors: 1. Current only; 2.
HST only; 3.TOT only; 4. Current + HST; 5. HST + Figure 7: Feasible region limited by the TOT only
TOT; 6. Current + TOT. The case 2 (HST only) was
already presented in Fig. 4 when we introduced a Similarly, we use Fig. 1 to explain why TOT is
feasible region. Thus, in this subsection we present limiting factor. For given current and temperature
main examples of feasible regions (case 1, 3 and limitations, TOT limit (95 ℃) is a limiting factor for
case 4). Tamb > +2 ℃. At the same time, Tamb profile varies in
Current limit only: Fig. 6 shows a feasible the range between +10 ℃ and +18 ℃, which is
region built for limitations HST 120 ℃ and TOT 105 higher than this critical Tamb: This explains why TOT
℃ and Tamb profile happened on January 11, 2019 is a limiting factor all day long.
in Tomsk, Russia. Current+HST limit: Fig. 8 shows a feasible
region built for limitations HST limit =120 ℃ and
TOT limit =105 ℃ and Tamb profile as of January 15,
2019 in Tomsk, Russia.

Figure 6: Feasible region limited by the current only

From Fig. 6 we see that the top line of the

feasible region corresponds to the current limit Figure 8: Feasible region limited by both current and
(dash line). To explain why the current is a limiting HST
factor, one should compare the range of given Tamb
From Fig. 8 we see that once Tamb crosses the
profile with Tamb range (in Fig. 1) corresponding to
critical Tamb = -17 ℃ (Fig.1), the limiting factor shifts
current and HST limit =120 ℃. There we see that
from current = 1.5 pu to HST limit = 120 ℃.
the current limit remains a limiting factor for Tamb
The main result of this subsection: examples of
below – 17 ℃. At the same time, given Tamb profile
Tamb profiles are shown for different limiting factors.
(varying between -19 ℃ and -33 ℃) remains always
As we can see, the top line of feasible region
lower than this critical Tamb. Thus, current limit
depends on Tamb only. This allows us to avoid using
remains a limiting factor all day long.
a typical shape of load profile.

7
3. Assessment of dynamic transformer
ratings
This section presents the results of DTR
assessment in Tomsk and Grenoble. To obtain
representative results, DTR estimation are based
on long-term data of Tamb in each geographical
area. For instance, the climate science usually
recommends to consider at least 30-year-long
interval for representative estimations [55]. Thus,
we use hour Tamb [56] for the period from January
01, 1985 to March 29, 2019 (time of data
Figure 9: Hourly Tamb from 1985 to 2019 in Tomsk and
downloading) in Tomsk, Russia and Grenoble, Grenoble
France (Fig. 9). From Fig.10 we see that the black line shape
From Tamb data, shown in Fig. 9, we have 12 (limiting factor) of these feasible regions is variable.
506 daily Tamb profiles in each city. For these Tamb To quantify this DTR variability, mean DTR and its
profiles, we define 12 506 daily feasible regions maximum and minimum deviations are estimated
corresponding to different combinations of for the most common formulations of current and
temperature and current limitations (Fig. 10). temperature limitations. (Fig.11)

Figure 10: Estimation of feasible regions during 34 years

Figure 11: Mean DTR with maximum and minimum deviations during 34 years

8
 The bars in Fig. 11 estimate two types of DTR: dark green bars with other bars then DTR can
DTR, based on design HST (dark green bar) and ensure up to 45 % of additional capacity in Tomsk
DTR based on HST limit (other bars). At the same and 41% in Grenoble. This is 30-36 % more power
time, the majority of papers estimates DTR using in comparison to the DTR, based on design HST
the design HST only (dark green bars). This allows (dark green bars). It is noteworthy that this
scientists to avoid problems with accelerated additional power leads to an increased HST.
ageing which is an advantage. As a drawback, they However, DSO can control this load amplitude and
ignore a substantial part of DTR, confined by duration using the flexibility from DER.
current and temperature limitations. For instance, DTR used together with DER management
in [31,32] authors estimated that DTR provides provides an additional degree of freedom for
10% additional capacity in United Kingdom. This system operators in power systems scheduling. At
correlates with our dark green bars where mean the same time, this degree of freedom changes
DTR provides 15 % of additional capacity over during a year following DTR seasonal variations.
nominal rating in Tomsk Russia and 5% in Therefore, DTR should be estimated for all current
Grenoble France. Nevertheless, if one compares and temperature limitations per month (Fig. 12)

Figure 12: Mean DTR with maximum and minimum deviations in each month

From Fig. 12 we see that dark green bars (DTR since Tamb of studied climates is always below than
based on the design HST) exceed nominal rating of critical Tamb +45 ℃ (see Fig. 1). Therefore, HST
the transformer during almost all months. However, remains the unique limiting factor for these two
in summer months, such DTR should be set lower current and temperature limitations. Moreover, one
than nominal rating to avoid the violation of design can notice that dark red and orange bars in Tomsk
HST. As we said earlier, the dark green bars are a do not have any deviations in winter months. This
classical example of DTR, studied in many papers means that DTR of such temperature limitations is
[31,32]. However, these papers ignore other bars much higher than the current limit, which does not
shown in Fig.12 This leads to very conservative let a transformer to reach these temperature limits.
estimation of DTR. In contrast, this paper allows to In addition to DTR amplitude (bars in Fig.12), we
determine the part of DTR, which was omitted estimate DTR duration. The typical DTR duration
before. curves in Tomsk and Grenoble are calculated and
We would like to explain some particular bars in presented in Fig. 13.
Fig. 12. For instance, yellow and light green bars
are the same in both cities. This means that HST
limit = 120 ℃ is always reached before TOT limit
(95 ℃ or 105 ℃) in both climates. This takes place

9
 Figure 13: DTR duration curves

To find such a duration curve, one can sort all

values of the DTR array (taken from Fig. 10) in a
descending order. This gives us the y-data. The
DTR duration (x-axis) is obtained as following:

(1)

Where N – a numerical order of y-data (DTR

sorted in a descending order).
DTR duration in x-axis shows the amount of
time (in %) when DTR exceeds the value selected
on the duration curve. For instance, classical DTR
(dark green curve) exceeds a nominal rating of
transformer for 88,5 % of time in Tomsk and 79 %
of time in Grenoble. This also means that the
classical DTR is below nominal rating for 11.5%
and 21 % of time in these cities correspondingly.
This result correlates with conclusions of many
authors, stating that DTR can be below the
nominal rating for a short period of time [31,32].
Meanwhile, we see that duration curves of other
DTRs remain higher than the nominal rating for
100 % of time in both cities. These results quantify
that part of DTR, ignored in similar studies.
Finally, we identify the main limiting factor of
DTR for different formulations. The easiest way to
do that is to take a Tamb history in each city and see
limiting factor for each Tamb range in Fig.1. For
instance, we assume that Tamb is -10 ℃ and DTR
formulation is current 1.5 pu, HST 120 ℃ and TOT
105 ℃. For this Tamb, the limiting factor is the HST
of 120 ℃ (in Fig. 1). Fig. 14 shows limiting factors
and their occurrence, expressed in percent of Figure 14: Share of limiting factors: based on 34 years
studied period (34 years) for each DTR formulation. analysis (percents are rounded)

10
 It is interesting to recall that some papers [57– the cost of HV/MV substation can vary from 500 k€
59] assume HST as a main limiting factor. to 1.5 M€. Therefore, DSO can defer the large
Therefore, they use only HST limits as a investments into transformers by taking advantages
transformer limits in the formulation of optimization of DTR and DER.
problem. Although such an assumption can be true Another important result is that we avoid using
for the current and temperature limitations chosen a typical (net) load profile for DTR determination.
in a particular paper, other scientists can make an Instead, we built a feasible region of load profiles,
error by using such logic for other DTR which is based on Tamb only. Thus, we consider
formulations or for other Tamb. For instance, Fig. 14 multiple shapes of load profiles and not only typical
shows that DTR based on design HST has HST as ones. Moreover, DTR assessment showed that the
a limiting factor 99,9 % of time in Tomsk and 100% main limiting factor is very sensitive to current and
of time in Grenoble. Indeed, current or TOT limits temperature limitations. For instance, for majority of
do not affect the limiting factor of a studied ONAF formulations, HST partially or fully remains a
transformer whatever the climate is chosen. limiting factor but for another case (HST 140 ℃ and
However, if one uses a higher HST limit and TOT TOT 95 ℃) HST does not affect DTR at all. At the
limit, we see that current becomes a main limiting same time, many papers considered HST as the
factor from 9% to 51 % of time in Tomsk. Notably, main limiting factor, which is not necessarily true.
the HST is no longer a limiting factor for The results of this paper could be valuable for
formulation: current 1.5 pu, HST 140 ℃ and TOT the industry and academia experts dealing with
95 ℃. Therefore, the assumption that HST is network constraints and transformer thermal ratings
always a limiting factor can be fully and partially in particular. The following companies are firstly
true but also totally wrong for particular current and targeted: TSO, DSO, transformer manufacturers,
temperature limitations. network planners and research facilities.

4. Conclusion Acknowledgments
In summary, the paper assesses DTR for We deeply appreciate Conference des Grandes
various current and temperature limitations. In Ecoles, who first supported our research and laid
contrast to similar studies, the paper takes into the ground for its further development. The
account all limiting factors (current, HST and TOT) research is also funded from Tomsk Polytechnic
and their combinations. This allowed us to assess University Competitiveness Enhancement Program
those parts of DTR, which were usually omitted in grant. Project Number 97/2017. The authors thank
similar studies. At the same time our results the IDEX for funding Ildar DAMINOV travel grant.
showed that this omitted DTR represents a large We gratefully appreciate the funding received from
transformer capacity in the range from 25% to 45 % Grenoble INP in the framework of ATER position of
nominal rating. Moreover, DTR duration curves Ildar DAMINOV. We thank Egor GLADKIKH for his
prove that this additional DTR capacity is higher help with financing Ildar’s Ph.D thesis. Authors
than nominal rating during 100 % of time in contrast thank MeteoBlue company for provided data on
to classical DTR which is 88,5% in Tomsk and 79 ambient temperature in Grenoble, France.
% in Grenoble correspondingly. However, we pay
attention to the fact that this additional capacity is References
operated at increased HST. Nevertheless, modern
[1] Nawaz A, Wang H, Wu Q, Kumar Ochani M.
DSO can control the shape of transformer loadings TSO and DSO with large-scale distributed energy
by using DER. This allows to control the amplitude resources: A security constrained unit
and durations of transformer loadings and therefore commitment coordinated solution. Int Trans
make them feasible from both sides: current Electr Energy Syst 2020. doi:10.1002/2050-
7038.12233.
/temperature limitations and ageing. This provides
[2] Hoff TE, Herig C, Awerbuch S, Preston A.
additional degree of freedom for system operators Managing Risk Using Renewable Energy
to manage the power systems. This additional Technologies. n.d.
capacity can be especially relevant if we recall that [3] Miura M, Satoh T, Iwamoto S, Kurihara I.

11
 Application of dynamic rating to increase the [16] Bracale A, Carpinelli G, Pagano M, De Falco P.
available transfer capability. Electr Eng Japan A Probabilistic Approach for Forecasting the
2009;166:40–7. doi:10.1002/eej.20537. Allowable Current of Oil-Immersed Transformers.
[4] Papathanassiou S, Hatziargyriou N, IEEE Trans Power Deliv 2018;33:1825–34.
Anagnostopoulos P, Aleixo L, Buchholz B, doi:10.1109/TPWRD.2018.2791181.
Carter-Brown C, et al. Capacity of Distribution [17] Yang J, Bai X, Strickland D, Jenkins L, Cross
Feeders for Hosting Distributed Energy AM. Dynamic Network Rating for Low Carbon
Resources. Cigre C624 2014. Application—A U.K. Application. IEEE Trans
[5] Numan M, Feng D, Abbas F, Rahman U, Wattoo Smart Grid 2015;6:988–98.
WA. Impact assessment of a co‐optimized doi:10.1109/TSG.2015.2389711.
dynamic line rating and transmission switching [18] Kim SD, Morcos MM. An Application of Dynamic
topology on network expansion planning. Int Thermal Line Rating Control System to Up-Rate
Trans Electr Energy Syst 2020. the Ampacity of Overhead Transmission Lines.
doi:10.1002/2050-7038.12457. IEEE Trans Power Deliv 2013;28:1231–2.
[6] Degefa MZ, Humayun M, Safdarian A, Koivisto doi:10.1109/TPWRD.2012.2234940.
M, Millar RJ, Lehtonen M. Unlocking distribution [19] Karimi S, Musilek P, Knight AM. Dynamic thermal
network capacity through real-time thermal rating rating of transmission lines: A review. Renew
for high penetration of DGs. Electr Power Syst Sustain Energy Rev 2018;91:600–12.
Res 2014;117:36–46. doi:10.1016/j.rser.2018.04.001.
doi:10.1016/J.EPSR.2014.07.032. [20] Saied MM. Assessing the dynamic rating of
[7] FSK. STO 56947007-29.180.01.116-2012 The overhead transmission lines. Eur Trans Electr
guidance for transformer operation 2014:52. Power 2007;17:526–36. doi:10.1002/etep.151.
[8] Viafora N, Morozovska K, Kazmi SHH, Laneryd [21] Huang R, Pilgrim JAA, Lewin PLL, Payne D.
T, Hilber P, Holbøll J. Day-ahead dispatch Dynamic cable ratings for smarter grids. 2013 4th
optimization with dynamic thermal rating of IEEE/PES Innov Smart Grid Technol Eur ISGT
transformers and overhead lines. Electr Power Eur 2013 2013:1–5.
Syst Res 2019. doi:10.1016/j.epsr.2019.02.026. doi:10.1109/ISGTEurope.2013.6695230.
[9] Teh J, Lai C-M, Muhamad NA, Ooi CA, Cheng Y- [22] Tripathy SC, Lakervi E. Evaluation of transformer
H, Mohd Zainuri MAA, et al. Prospects of Using overloading capability. Eur Trans Electr Power
the Dynamic Thermal Rating System for Reliable 2005;15:455–64. doi:10.1002/etep.59.
Electrical Networks: A Review. IEEE Access [23] Simonson EA, Lapworth JA. Thermal capability
2018;6:26765–78. assessment for transformers. IEE Conf. Publ.,
doi:10.1109/ACCESS.2018.2824238. 1995, p. 103–8.
[10] Zarei T, Morozovska K, Laneryd T, Hilber P, [24] IEC. IEC 60076-7 Power transformers – Part 7:
Wihlén M, Hansson O. Reliability considerations Loading guide for oil-immersed power
and economic benefits of dynamic transformer transformers Transformateurs. 2018.
rating for wind energy integration. Int J Electr [25] IEEE Power & Energy Society. IEEE Guide for
Power Energy Syst 2019;106:598–606. Loading Mineral- Oil-Immersed Transformers and
doi:10.1016/j.ijepes.2018.09.038. Step-Voltage Regulators. vol. 2011. 2012.
[11] A. Douglass D, Grant I, Jardini JA, Kluge R, [26] GOST 14209 - 85. General-purpose oil-
Traynor P, Davis C, et al. A Review of Dynamic immersed power transformers. Permissible loads
Thermal Line Rating Methods With Forecasting. 2009:36.
IEEE Trans Power Deliv 2019;34:2100–9. [27] GOST-14209-97. Loading guide for oil-immersed
doi:10.1109/TPWRD.2019.2932054. power transformers 1997:82.
[12] Douglass DA, Edris A-A. Real-time monitoring [28] Norris ET. The thermal rating of transformers. J
and dynamic thermal rating of power Inst Electr Eng 1928;66. doi:10.1049/jiee-
transmission circuits. IEEE Trans Power Deliv 1.1928.0095.
1996;11:1407–18. doi:10.1109/61.517499. [29] Sealey WC, Hodtum JB. Overloading of
[13] Kim D-M, Bae I-S, Kim J-O. Determination of Transformers---Cases Not Covered by the
available transfer capability (ATC) considering General Rules. Trans Am Inst Electr Eng
real-time weather conditions. Eur Trans Electr 1944;63:149–53. doi:10.1109/T-
Power 2011;21:855–64. doi:10.1002/etep.481. AIEE.1944.5058912.
[14] Daminov I, Prokhorov A, Caire R, Alvarez- [30] Biçen Y, Aras F. Loadability of power transformer
Herault M-C. Receding horizon algorithm for under regional climate conditions: The case of
dynamic transformer rating and its application for Turkey. Electr Eng 2014;96:347–58.
real-time economic dispatch. 13th IEEE doi:10.1007/s00202-014-0301-6.
PowerTech, 2019. [31] Yang J, Chittock L, Strickland D, Harrap C.
[15] Daminov I, Sazonov A. Two-stage algorithm to Predicting practical benefits of dynamic asset
solve the economic dispatch problem with ratings of 33KV distribution transformers. IET Int
dynamic transformer ratings. Electr. power Ind. Conf Resil Transm Distrib Networks 2015
eyes youth, Irkutsk: 2019. 2015;60354:1–6. doi:10.1049/cp.2015.0898.

12
[32] Yang J, Bai X, Strickland D, Jenkins L, Cross M, Millar RJ, Lehtonen M. Unlocking distribution
AM. Dynamic Network Rating for Low Carbon network capacity through real-time thermal rating
Distribution Network Operation—A U.K. for high penetration of DGs. Electr Power Syst
Application. IEEE Trans Smart Grid 2015;6:988– Res 2014;117:36–46.
98. doi:10.1109/TSG.2015.2389711. doi:10.1016/j.epsr.2014.07.032.
[33] Li X, Zielke G. A study on transformer loading in [45] Haque ANMM, Shafiullah DS, Nguyen PH, Bliek
Manitoba - Peak-load ambient temperature. IEEE FW. Real-time congestion management in active
Trans Power Deliv 2003;18:1249–56. distribution network based on dynamic thermal
doi:10.1109/TPWRD.2003.817734. overloading cost. 2016 Power Syst. Comput.
[34] Dong M, Li B, Nassif A. A Data-Driven Conf., IEEE; 2016, p. 1–7.
Residential Transformer Overloading Risk doi:10.1109/PSCC.2016.7540985.
Assessment Method. IEEE Trans Power Deliv [46] Swaminathan Prasanna B. Gestion
2018;PP:1. doi:10.1109/TPWRD.2018.2882215. prévisionnelle des réseaux actifs de distribution -
[35] Salama MMM, Mansour DEA, Abdelmaksoud relaxation convexe sous incertitude. 2017.
SM, Abbas AA. Impact of Long-Term Climatic [47] Yip HT, An C, Lloyd GJ, Taylor P, Michiorri A,
Conditions on the Ageing and Cost Effectiveness Jupe S, et al. Dynamic thermal rating and active
of the Oil-Filled Transformer. 2018 20th Int. control for improved distribution network
Middle East Power Syst. Conf. MEPCON 2018 - utilisation. 10th IET Int. Conf. Dev. Power Syst.
Proc., 2019. Prot. DPSP 2010, vol. 2010, AREVA TandD UK
doi:10.1109/MEPCON.2018.8635184. Ltd., Stafford, United Kingdom: 2010, p. 1–5.
[36] Nguyen TT. Constrained optimisation procedure doi:10.1049/cp.2010.0213.
for evaluating cyclic loading of power [48] Ali M, Degefa MZZ, Humayun M, Safdarian A,
transformers. IEE Proc - Gener Transm Distrib Lehtonen M. Increased utilization of wind
1995;142:240. doi:10.1049/ip-gtd:19951746. generation by coordinating the demand response
[37] Savaghebi M, Jalilian A, Gholami A. A new and real-time thermal rating. IEEE Trans Power
approach for transformer loading capability Syst 2016;31:3737–46.
assessment under non-linear load currents. IEEE doi:10.1109/TPWRS.2015.2498899.
Int. Conf. Ind. Technol., IEEE; 2008, p. 1–5. [49] IRENA. Aggregators Innovation Landscape brief.
doi:10.1109/ICIT.2008.4608354. 2019.
[38] Shahbazi B, Savaghebi M, Shariati M. A [50] Kaheh Z, Baradaran Kazemzadeh R,
Probabilistic Approach for Power Transformer Sheikh‐El‐Eslami MK. A trilevel programming
Dynamic Loading Capability Assessment. 2007. model for flexiramp and reserve procurement in
[39] Zhang M, Liu Y, Tylavsky D. Dynamic loading of high penetration of wind farms and participation
substation distribution transformers in a of a large industry and a DR aggregator. Int
production grade environment. North Am. Power Trans Electr Energy Syst 2019;29.
Symp., IEEE; 2012, p. 1–6. doi:10.1002/2050-7038.12105.
doi:10.1109/NAPS.2012.6336380. [51] Beavis B, Dobbs I. Optimisation and Stability
[40] Pasricha A, Crow ML. A method of improving Theory for Economic Analysis. Cambridge
transformer overloading beyond nameplate University Press; 1990.
rating. 2015 North Am. Power Symp., IEEE; doi:10.1017/cbo9780511559402.
2015, p. 1–6. doi:10.1109/NAPS.2015.7335148. [52] Susa D, Nordman H. IEC 60076-7 loading guide
[41] Alvarez DL, Rivera SR, Mombello EE. thermal model constants estimation. Int Trans
Transformer Thermal Capacity Estimation and Electr Energy Syst 2013;23:946–60.
Prediction Using Dynamic Rating Monitoring. doi:10.1002/etep.1631.
IEEE Trans Power Deliv 2019;34:1695–705. [53] Wang MXM-X, Han XSX-S. Study on electro-
doi:10.1109/TPWRD.2019.2918243. thermal coupling optimal power flow model and
[42] Bunn M, Das BP, Seet BC, Baguley C. Empirical its simplification. IEEE PES Gen. Meet. PES
Design Method for Distribution Transformer 2010, School of Electrical Engineering,
Utilization Optimization. IEEE Trans Power Deliv Shandong University, Jinan, Shandong Province,
2019. doi:10.1109/TPWRD.2019.2926328. China: 2010, p. 1–6.
[43] Degefa MZ, Koivisto M, Millar RJ, Lehtonen M. doi:10.1109/PES.2010.5589644.
Dynamic thermal state forecasting of distribution [54] Wang M, Han X, Zhang H, Jiang Z. Advanced
network components: For enhanced active thermal rating and its application. 1st Int. Conf.
distribution network capacity. 2014 Int. Conf. Sustain. Power Gener. Supply, SUPERGEN ’09,
Probabilistic Methods Appl. to Power Syst. School of Electrical Engineering, Shandong
PMAPS 2014, Department of Electrical University, Jinan, Shandong Province, China:
Engineering and Automation, Aalto University, 2009, p. 1–5.
Espoo, Finland: Institute of Electrical and doi:10.1109/SUPERGEN.2009.5348354.
Electronics Engineers Inc.; 2014. [55] World Meteorological Organization. Guidelines
doi:10.1109/PMAPS.2014.6960607. on the Calculation of Climate Normals. 2017.
[44] Degefa MZ, Humayun M, Safdarian A, Koivisto [56] MeteoBlue. history+ - Historical weather data

13
 since 1984 2019. [58] Kostin VN, Minakova TE, Kopteva A V. Urban
https://www.meteoblue.com/en/historyplus Substations Transformers Allowed Loading
(accessed November 5, 2019). 2018:692–5.
[57] Arguence O, Cadoux F. Sizing power [59] Savaghebi M, Gholami A, Vahedi A, Hooshyar H.
transformers in power systems planning using A fuzzy-based approach for transformer dynamic
thermal rating. Int J Electr Power Energy Syst loading capability assessment. Proc Univ Power
2020;118:105781. Eng Conf 2008:1–5.
doi:10.1016/j.ijepes.2019.105781. doi:10.1109/UPEC.2008.4651476.

14