21, rue d’Artois, F-75008 PARIS A2 – 117 CIGRE 2018

http : //www.cigre.org

Thermal comparison between mineral oil, natural and synthetic esters at

largest single-phase 420 kV green transformer

M. Cuesto1, C. González-García1, M. Vaquero1, D. Vukovic2

1
ABB s.a, Spain
2
ABB AG, Germany

SUMMARY
Besides commonly used mineral oil, natural esters and synthetic esters are being more
frequently used in liquid-filled power transformers. These fluids differ in aspect of dielectric,
thermal, chemical properties and consequently in handling routines and material compatibility
[1], [2]. Thermal properties differences refer to: specific heat, thermal conductivity, thermal
expansion coefficient and kinematic viscosity; being probably this last property the one that
provides more significant differences on the thermal performance; mainly oil and winding
temperature rises due to different local and global fluid velocities.

A R&D project was agreed with one of the main Spanish utilities consisting on the feasibility
evaluation of using biodegradable fluids (natural and synthetic esters) in their existing 420 kV
mineral oil filled fleet of 200 MVA standardized single-phase transformers. The result of this
evaluation has been positive, meaning that these units can operate also with both natural and
synthetic ester fluids. The main purpose of this paper is to present the transformer thermal
performance differences when operating with the three mentioned fluids.

The feasibility study focused on thermal aspects, including state of the art computational fluid
dynamic simulations with the three fluids, allowing to predict the expected differences
regarding thermal behaviour is presented. Regarding cooling stages, it is important to remark
that the cooling equipment of the transformer object of this project was kept as the original one.

Finally, the paper is completed with the comparison of other transformer parameters that have
been measured with mineral oil, natural ester and synthetic ester along the project. This is the
largest 420 kV single-phase set of transformers ever tested with three different fluids: mineral
oil, natural ester and synthetic ester.

KEYWORDS
Power transformer 420 kV, mineral oil, natural ester, synthetic ester, mineral-natural ester-
synthetic ester thermal comparison, largest single-phase.

miguel.cuesto@es.abb.com
1. INTRODUCTION

Along 2016 and 2017, two 420 kV units have been successfully tested respectively with natural
and synthetic ester. As already presented in [2], this has been a R&D project which main driver
was a feasibility evaluation of using natural and synthetic ester fluids in existing 420 kV units
that were initially designed only for mineral oil, and with the additional advantage of the future
application of the lessons learned in new units.

The main characteristics of these transformers are described below:

- Single-phase autotransformer.
- Outdoor type for continuous service operation.
- Rated voltages:
- HV winding: 400kV. BIL 1425 kV.
- LV winding: 230 kV ±10x1.5% OLTC. BIL 950 kV and 170 kV neutral.
- Tertiary winding: 33-26.4-24 kV. BIL 170 kV.
- Rated power and cooling: 200 MVA at ODAF, 160 MVA at ODAN/ONAF and 120 MVA
at ONAN.
- Frequency: 50Hz.
- Vector group: YNa0d11 (for a complete three phase bank).
- Temperature rises (as requested for mineral oil):
- Top Oil: 55ºC.
- Average Winding: 55ºC.
- Hot Spot Winding: 65ºC.

In this paper, the focus is the thermal comparison between three different fluids in three
transformers of the same design. In that sense it is important to remark that some adaptions
were needed to prepare the units to be tested with both esters: high voltage bushing and exit
insulation, low voltage bushing and on load tap changer, but none of those adaptions are related
to the cooling equipment that was kept exactly the same, meaning same number of pumps, fans
and radiators.

The fact of having performed temperature rise tests at different cooling stages allows also to
evaluate which of those components of the cooling equipment can be more influenced by the
different type of fluid and therefore which cooling stage is more or less sensitive to the type of
fluid, specially due to different kinematic viscosities as anticipated in the beginning of this
document.

One more advantage of the units selected for this exercise was the fact that these units have
fiber-optic probes originally installed inside the winding for direct hot-spot measurement and
therefore the amount of measured data available for the comparison is even more valuable.

2
2. THERMAL PERFORMANCE COMPARISON

As already explained, the original cooling equipment was not changed in the transformers,
remaining the same components as for mineral oil. The main criteria at the moment of agreeing
the power rating of each temperature rise test at different cooling stages, was to obtain
temperature values with esters as similar as possible to the values measured with mineral oil,
so that the comparison is more fair. In that way, heat run test counting with ONAN cooling
needed from a test power adaptation.

2.1 ODAF

According to calculations and manufacturer know-how, it was expected to have similar

temperatures between the three different fluids at ODAF cooling: 200 MVA, full power of the
transformer. For that reason the test was run at 200 MVA for the three fluids, providing the
comparison presented in Table I below:

Table I. Measured comparison at ODAF 200 MVA with the three fluids (*)
Parameter Mineral oil Natural ester Synthetic ester
Top oil rise (p.u.) 1 0.88 0.98
Average winding gradient HV (p.u.) 1 1.66 1.36
Average winding gradient LV (p.u.) 1 1.44 1.26
Average winding rise HV (p.u.) 1 1.038 1.039
Average winding rise LV (p.u.) 1 1.075 1.064
Hot spot rise HV (p.u.) 1 1.012 1.058
Hot spot rise LV (p.u.) 1 1.097 1.15
(*) In all the parameters the values measured with mineral oil are considered the reference:
1.0, and the measured values with natural and synthetic ester in per unit.

From the above table some comments can be made:

- Top oil temperature is lower for esters than for mineral oil, especially in the case of
natural ester.
- On the other hand winding copper gradients are higher for esters.
- Both previous effects somehow get balanced on the windings rises, presenting similar
values for the three fluids.
2.2 ONAN

On the other hand, again from preliminary calculations and experience it was expected a higher
impact of esters higher viscosity when the pumps were not running, so the tests were performed
each one at the power that would provide similar temperature values:

Table II. Measured comparison at ONAN with power ratings to achieve similar temperatures
Mineral oil @ Natural ester Synthetic ester
Parameter 120 MVA @ 77 MVA @ 86MVA
Top oil rise (p.u.) 1 0.91 0.82
Average winding gradient HV (p.u.) 1 3.26 3.13
Average winding gradient LV (p.u.) 1 1.79 1.57
Average winding rise HV (p.u.) 1 1.017 0.92
Average winding rise LV (p.u.) 1 1.041 0.92
Hot spot rise HV (p.u.) 1 1.002 0.86
Hot spot rise LV (p.u.) 1 1.007 0.86

3
Previous table II shows also interesting conclusions:
- Despite reducing the power rating in esters, winding average gradients are higher than
for mineral oil, being similar the differences in both natural and synthetic esters.
- High voltage average winding gradient seems to be more sensitive to the absence of
pumps than low voltage winding average gradient.
- The lower top oil rises again get balanced with higher winding gradients so that the
differences on the winding and hot spot rises are much smaller.
- In the case of synthetic ester there is some potential to increase the power rating
keeping similar temperatures.

In summary in both cooling modes the results confirm the expectations: in OD cooling the
temperatures are similar for the same power rating while in ON there is the need of going to
lower power ratings with esters to keep the same similar temperatures to mineral oil. The values
are also in line with [3].

4
3. COMPARISON BETWEEN SIMULATIONS

Finally as part of the comparison experience presented on this paper, it has been considered
interesting to include an exercise of the application of state of the art simulation tools, in this
case computational fluid dynamics (CFD), in order to compare results based on the different
characteristics of the three different fluids.

The following Figure 1 shows the thermal pattern in half of the winding (symmetrical) for the
three fluids at ODAF condition, full 200 MVA load:

Mineral oil Natural ester Synthetic ester

Figure 1. Qualitative comparison with CFD simulation between the three fluids

In terms of comparison (p.u.) for the three simulations referred to top oil, please see the table
III below:

Table III. Comparison of simulation results at ODAF 200 MVA with the three fluids
Parameter Mineral oil Natural ester Synthetic ester
Top oil rise (p.u.) 1 0.97 0.99

From the previous exercise two mains comment can be made:

- Figure 1: thermal pattern is not significantly affected by the type of fluid at ODAF.
- Table III: the simulations provide similar ratios and same order in terms of top oil than
measurements: top oil natural ester < top oil synthetic ester< top oil mineral oil.

5
4. OTHER INTERESTING COMPARISONS

Finally it has been considered interesting to share with the readers of this paper other
comparisons between measurements of the three different fluids in three units with the same
design.

Measured capacitances between windings and to ground provided different values between the
three fluids:
- The capacitances of natural ester and synthetic ester are higher than for mineral oil, in
the range of 20% higher.
- The difference on the capacitances between natural and synthetic esters is much lower
that between these fluids and mineral oil.
- In this particular case, the fluid that provides higher capacitances has been the synthetic
ester, on average 3% higher than for natural ester.

An immediate consequence of the different capacitances is the difference on FRA wave shapes
at certain frequency range. In particular the frequency responses are very similar, having same
amplitude in every peak but with a displacement after the first resonance peak. This learning is
important when considering the footprint of the transformer.

Regarding gas generation, being in all the cases low and acceptable, the pattern was not the
same for all the fluids.

Finally in terms of tan delta (power factor), values of ester fluids were higher than for mineral
oil, which was expected and already considered as acceptable by international standards.

6
5. CONCLUSIONS

The first conclusion from this project is that the feasibility evaluation of using mineral oil,
natural ester and synthetic ester in three units with the same design has been successfully
confirmed by thorough factory testing according to international standards.

At this point it is important to remark that this conclusion is related to the particular case of the
unit described in this paper. For any other case it would be recommended to repeat the exercise
based on the manufacturer know-how. Some other designs may require much more changes to
meet targeted expectations.

On top of that, the lessons learned could be applied to the request of future new units that would
be fully compatible with both type of insulating fluids (mineral and ester).

Regarding thermal performance the conclusions are summarized below:

- Without any change on the cooling equipment, ODAF power ratings can be kept the same
as for mineral oil, since measured values were similar. This would be applicable also to
ODAN as in these stages the pumps are the main driver for the thermal behavior inside the
windings.

- On the other hand ONAN power ratings are limited when keeping same temperatures as
for mineral oil. In this case, synthetic ester allows slightly higher power rating than natural
ester. This would be applicable also to ONAF as in these stages the fans only mean a
difference on the fluid temperature and not on the windings gradients.

- Results presented of computational fluid dynamic (CFD) simulations are in line with the
measured differences between fluids.

Other interesting difference presented is the different capacitances and FRA between the three
fluids, being around 20 % higher for natural and synthetic ester than for mineral oil and with
much more similar values among the two ester fluids, in the rage of 3 % (in this particular unit
higher capacitances for synthetic ester that for natural ester). This should be considered
especially for the transformer footprint recordings.

Gas generation pattern, being in all the cases low and acceptable, was not the same for all the
fluids. In terms of tan delta (power factor), values of ester fluids were higher than for mineral
oil, which was expected and already considered as acceptable by standards.

As far as the authors know this is the largest world reference of single-phase transformer ever
tested with mineral oil, natural ester and synthetic ester by now.

BIBLIOGRAPHY
[1] J. Hajek et al. “Considerations for the Design, Manufacture, and Retro – filling of
Power Transformers with High Fire Point, Biodegradable Ester Fluids”. Cigre SC A2-203 Paris
session 2012.
[2] M. Cuesto et al. “Largest single-phase green transformer with natural ester at 420 kV and
tests comparison with mineral oil”. CIGRE A2 Colloquium, Krakow, 2017
[3] R. Girgis et al. “Detailed performance of a 50 MVA transformer filled with a natural
ester fluid versus mineral oil”. Cigre SC A2-107 Paris session 2010.