1
Appropriate Test Conditions Proposed for
Industry Standards of Measuring Transformer Noise
Ramsis S. Girgis, Fellow, IEEE
ABB Power Transformers, St. Louis, MO, USA
Abstract: The subject of deciding on the most appropriate
method and conditions for accurately measuring the noise level
of a transformer has been discussed through the transformer
industry for decades. It is the objective of this Paper to
contribute to this effort by identifying more appropriate
conditions for both the Sound Pressure Method and the Sound
Intensity Method. The Paper presents results of comprehensive
indoor measurements made of the frequency spectrum and
total noise of a large number of power transformers using the
two measuring methods. First, the paper presents data used to
develop appropriate conditions for more accurate measurement
of transformer noise using the Sound Intensity Method. The
paper then presents data used to develop a more accurate and
reliable Sound Wall – Reflection Correction to be used with
the Sound Pressure Method. Lastly, the Paper presents
appropriate method of measuring Load noise of power
transformers. Sound measuring conditions developed and
presented in this paper are being proposed to replace existing
conditions in the IEC and IEEE Industry Standards of
transformer noise testing.
Keywords: Power Transformers, Transformer Noise, Sound Power
Measurement, Transformers Industry Standards
I. INTRODUCTION
Indoor measurement of the true noise level of a transformer
is critical, especially when requirements for low and ultra –
low noise levels are to be met. It is even more critical when the
levels of individual frequency components of transformer
noise are to be guaranteed.
Presently, the IEEE Standards [1] specify using only the
Sound Pressure Method and allowing only a correction to be
made for the ambient noise under specific ambient noise
conditions. The IEEE Standards do not allow using the Sound
Intensity Method and also do not allow a correction for the
Sound Wall – Reflections to be used with the Sound Pressure
Method. In addition, the ambient noise conditions, as specified
by both the IEC and IEEE Standards, are often not feasible to
meet when measuring Low & Ultra – Low noise power and
distribution transformers.
References [2] & [3] have demonstrated that the Sound
Intensity Method is the more accurate Method of measuring
transformer noise since it measures the true noise radiated by
the transformer, avoiding errors caused by ambient noise, near
– field effects, and Sound Wall – Reflections. This is provided
that appropriate conditions are met. This is also the conclusion
__________________________________________________
Dr. Ramsis Girgis is presently the R&D manager at the ABB
Power Transformer plant, Saint Louis, MO.
Mr. Mats Bernesjö is a Development Engineer at the ABB
Power Transformer plant, Saint Louis, MO.
Mats Bernesjö, Member, IEEE
of Reference [4]. As a result, the Audible Sound & Vibrations
(ASV) Subcommittee (SC) of the IEEE Transformers
Standards Committee recently approved implementing the
Sound Intensity Method in C57.12.90 [1], again, provided that
appropriate conditions are specified.
Section II of this paper presents results of a comprehensive
study performed by the authors of this paper to determine the
most appropriate conditions for more accurate measurement of
transformer noise using the Sound Intensity Method. These
conditions are proposed to be used by both IEEE and IEC
Standards.
Also, the IEC Standards, Ref. [5] and [6] allow using a
Sound Wall – Reflections correction to be used along with the
Sound Pressure method. Recently the ASV SC of the IEEE
Transformers Standards Committee approved correcting for
Sound Wall – Reflections in C57.12.90 [1].
Section III of this paper presents a proposal for a correction
to be used by IEC and IEEE Standards to use along with the
Sound Pressure Method to properly account for Sound wall
reflections when performing indoor sound measurements of
transformer noise.
Finally, the IEC standards allow measuring Load noise
under some conditions. The Standards also allow measuring
Load noise at less than full – load and correcting up for full –
load. In Reference [8] of this paper, the authors demonstrated
the importance of measuring Load Noise in addition to No –
Load noise, as the total noise of the transformer, including
Load noise, can be several dB > No – Load noise, especially in
low no – load noise transformers. Again, the ASV SC of the
IEEE Transformers Standards Committee recently approved
implementing Load noise measurements in C57.12.90 [1].
Section IV of this paper discusses the IEC conditions and
corrections of Load noise measurements. It offers a proposal
for improved accuracy to be incorporated in both the IEC and
IEEE Standards.
II. DEVELOPMENT OF CONDITIONS FOR APPROPRIATE NOISE
MEASUREMENT USING THE SOUND INTENSITY METHOD
IEC allows using the Sound Intensity method, provided the
difference between the measured values of the total dB (A)
using the Sound Pressure and the Sound Intensity Methods is <
8 dB. The selection of this condition was based on limited data
collected and presented in a CIGRE paper, Ref. [7]. However,
over the years, it was realized that this condition is not
accurate enough. As a result, less and less IEC customers now
approve using the Sound Intensity Method. In Reference [3],
the authors of this paper presented data that demonstrated the
inaccuracy of this 8 dB IEC condition. The data showed that
this limit should be in the 4 – 5 dB range, not 8 dB.
 2

For the sake of this study, extensive indoor noise level 4

measurements were made on 36 power transformers of 12 – 3
2
336 MVA power ratings, using both the Sound Pressure and 1
Sound Intensity Measuring Methods. 0
-1
Now, in order to arrive at the most appropriate conditions for -2
-3
accurate noise level measurements using the Sound Intensity -4
Method, differences between measured Sound Intensity levels -5

(Im - Ic)
-6
[Im] and the corresponding theoretical estimates of the -7
accurate Sound Intensity levels [Ic] have been plotted versus -8
-9
the corresponding differences between measured Sound -10
Pressure [Pm] and Sound Intensity levels [Im]. The quantity -11
[Im - Ic] represents the measuring error in the Sound Intensity -12
-13
Measurement. The estimated Sound Intensity levels [Ic] have -14
been calculated using the measured sound pressure levels [Pm] -15
-16
corrected for the actual ambient noise level at the 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
corresponding frequency component, the IEC Sound Wall – (Pm - Im)
Reflection (Environmental) factor K, Ref. [5], and the Near –
Field effect. A value of 0.2 was used for the average acoustic (b) Total dB (A)
absorption coefficient “ ” in the calculation of the
Environmental factor K in the factory test environment. No Figure 1 – Error in Intensity Measurement of Load noise
sound wall reflection correction was made for measurements vs. [Pm - Im], measurements made at factory environment
in sound rooms. Finally, an average value of 1.5 dB was used 4
for correcting for Near – Field effects as previously concluded 3
2
by the data presented in References [2] & [3]. 1
0
Figure 1 below presents such plots for Load Noise measured in -1
-2
a typical factory test area. Figure 2 presents the same but -3
measured in a sound – room with very low ambient noise level -4
(Im - Ic)

-5
as well as walls covered with highly sound absorbing material. -6
The figures show a clear correlation between the quantity [Pm -7
-8
- Im] and the quantity [Im - Ic]. These results show an error in -9
the Intensity measurement of < 1 dB for [Pm - Im] < 4 dB. The -10
-11
error is 1 – 2 dB for [Pm - Im] between 4 & 5 dB, 2 – 3 dB for -12
[Pm - Im] between 5 & 6 dB, etc. -13
-14
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
4
(Pm - Im)
3
2
1
Figure 2 – Error in Intensity Measurement of Load noise
0 vs. [Pm - Im], measurements made at a large sound–room
-1
-2 The plots also demonstrate that a [Pm - Im] value of 8 dB, as
-3 allowed in the IEC Standards, represents an error of about 4 –
-4
-5 5 dB; the noise level measured using the Sound Intensity
(Im - Ic )

-6 Method being underestimated by this amount.

-7
-8 The following two examples are given to help understanding
-9
-10
the above results and Figures:
-11
-12
Example 1: Low difference between Pm and Im
-13
-14 Measured Sound Pressure level, Pm: 62.2 dB
-15 Measured Sound Intensity level, Im: 58.4 dB
-16 (Pm - Im) = 62.2 - 58.4 => 3.8 dB
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Measured ambient level: 49.5 dB
(Pm - Im)
IEC wall – Reflection correction “K”: 1.3 dB
Near – field effect: 1.5 dB
(a) 125 Hz Component
Ic = [62.2 - 49.5] - 1.5 - 1.3 => 59.2 dB
 3

[Im - Ic]= 58.4 - 59.2 => - 0.8 dB 4

3
So, in this example, the measured Sound Intensity level is only 2
1
0.8 dB lower than the theoretical value. 0
-1
Example 2: High difference between Pm and Im -2
-3
Measured Sound Pressure level, Pm: 72.1 dB -4

(Im - Ic)
-5
Measured Sound Intensity level, Im: 63.7 dB -6
[Pm - Im] = 72.1 - 63.7 => 8.4 dB -7
-8
Measured ambient level: 65.5 dB -9
IEC wall correction factor K: 1.8 dB -10
-11
Near – field effect: 1.5 dB -12
Ic = [72.1 - 65.5] - 1.5 - 1.8 => 67.7 dB -13
-14
(Im - Ic) = 63.7 - 67.7 => - 4.0 dB 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
The measured Sound Intensity level in this case is 4 dB lower (Pm - Im)
than the theoretical value. (c) – 400 Hz component
Figure 3 presents data, of frequency components and total 4
Core noise of the same 36 power transformers, presented in the 3
2
same manner as that of Figure 1 for Load Noise. Again, these 1
transformers were tested in a factory environment. 0
-1
4 -2
3 -3
-4
(Im - Ic)

2
-5
1
-6
0
-7
-1
-8
-2
-9
-3
-10
-4
(Im - Ic)

-11
-5
-12
-6
-13
-7
-14
-8
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
-9
-10 (Pm - Im)
-11
-12 (d) – Total dB (A)
-13
-14 Figure 3 – Error in Intensity Measurement of Core noise
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 vs. [Pm - Im], measurements made at factory environment
(Pm - Im)

(a) – 125 Hz component The figures show that, except for the data points in
yellow, the same correlation exists between the error
4 quantity [Im - Ic] of core noise and the quantity [Pm - Im]
3
2 as observed in Figures 1 & 2 for Load noise. Again the
1 figures show an Intensity measurement error of < +1 dB
0
-1 for [Pm - Im] < 4 dB. The error is 1 – 2 dB for [Pm - Im]
-2
-3
between 4 & 5 dB, 2 – 3 dB for [Pm - Im] between 5 & 6
-4 dB, etc.
(Im - Ic)

-5
-6 The data points in yellow in Figures 3 (b) and 3 (d)
-7
-8 correspond to Ultra – Low core noise transformers, where
-9 the 250 Hz component and the total Core noise are in the
-10
-11 range of, or below, the ambient noise. In such cases, the Ic
-12 calculation is not valid and, hence, the quantity [Im - Ic] is
-13
-14 not valid. However, in the majority of these cases, the
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 quantity [Pm - Im] is > 6 dB.
(Pm - Im)

(b) – 250 Hz Component

 4

Figure 4, below, presents core noise data measured at a large 4

semi – anechoic test facility. Again, the same relationship K = 10 ⋅ log10 1 + , A = αS v
exists between the quantity [Im - Ic] and the quantity (Pm – A/ S
Im). In such a facility, the difference between Pm & Im is Where:
always low and hence the error [Im - Ic] is very low.
4 = Average Acoustic Absorption coefficient, Table – I below
3
2
1
Sv = Total Area of the Surface of the Test Room
0
-1 S = Transformer Measurement Surface Area
-2
-3 The maximum value allowed for this Environmental factor
-4
correction is 7 dB, otherwise measurements would be
(Im - Ic)

-5
-6 considered invalid.
-7
-8 Table I – Values of Acoustic absorption coefficient per IEC
-9
-10
-11 Average
-12 acoustic
-13
-14 Description of room absorption
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 coefficient,
(Pm - Im)

Figure 4 – Error in Intensity Measurement of Core noise vs. Nearly empty room with smooth hard walls
0.05
[Pm - Im], measurements made at a large Sound Room made of concrete, brick, plaster or tile.
Partly empty room, room with smooth walls. 0.10
Based on above data, the authors propose using the following
corrections when measuring transformer noise using the Sound Room with furniture, rectangular machinery
0.15
Intensity Method: room, rectangular industrial room.
No correction for (Pm - Im) 4 dB Irregularly shaped room with furniture,
0.20
Add 1 dB to measured value of Im for 4 < [Pm - Im] 5dB irregularly shaped machinery or industrial room

Add 2 dB to measured value of Im for 5 < (Pm - Im)
If (Pm - Im) > 6 dB, measurements using the Intensity Method
should not be allowed. Instead, the Sound Pressure method
would have to be used; correcting for Sound Wall –
Reflections and ambient noise. Obviously, this is provided that
conditions of ambient noise, as specified in the IEC and IEEE
Standards, are met.
III. THE SOUND WALL – REFLECTION CORRECTION
[ENVIRONMENTAL FACTOR]
Correction per present IEC Standards
When a transformer is tested indoors for noise level, sound
reflections from walls, and other large objects, adds to the
measured noise level when using the Sound Pressure Method.
Transformer noise contains discrete tones that are affected by
room acoustics, room geometry, and reflecting objects. In
order to account for the effect of these sound reflections, the
IEC Standards, Ref. [5], provides the following calculation
formula for what is termed as the “Environmental factor K”.
The calculated value is to be subtracted from the measured
value of the noise level:
6dB Room with upholstered furniture, machinery or
industrial room with a small amount of acoustic
0.25
material (for example, partially absorptive
ceiling) on ceiling or walls
Room with acoustic materials on both ceilings
0.35
and walls
Room with large amounts of acoustic materials
0.50
on ceilings and walls
Typical Values of Sound Walls – Reflection Correction
In order to have an appreciation for typical magnitudes of
this correction and the impact of the different parameters on
the magnitude of this correction, this factor was calculated for
a transformer with a measuring surface area of 100 m2. The
calculations were made for values of the Wall Sound –
Absorption Coefficient of 0.10, 0.15, 0.20, and 0.35 which
corresponds to a large variety of test room types. Also, the
calculations were made for three different test room sizes from
2000 m2 to 8000 m2. Table II below presents the calculated
values of the wall sound reflection correction using the IEC
formula.
 5

Table II – Example values of sound reflection correction measure Load noise of a transformer if the transformer is to be
loaded to any significant levels during operation.
Wall Sound – Absorption
Transformer Area Coefficient -
100

100 m2 90

0.10 0.15 0.20 0.35 80

Sound Power level, dB (A)

70
2000 4.8 3.7 3.0 2.0
60
Area of
5000 2.6 1.9 1.5 1.2
Room, m2 50

40
8000 1.8 1.2 1.0 0.6
30

Reiplinger's Equation
Proposed changes to IEC Sound – Reflection Correction 20
Measured
10
First, and for most, the table above shows that typical values of 0
this correction for a reasonable ratio of Area of Room / Area 30 60 90 120 150 180 210 240 270 300 330 360
Rated Power, MVA
of Transformer are in the 1 – 3 dB range. In fact, a wall
reflection of 3 dB means that the wall reflections are of a Figure 5 – Measured values of Load noise versus calculated
magnitude equal to the total noise of the transformer. using Reiplinger’s equation of Ref. [4]
Therefore, it is proposed that a limit of 3 dB is applied to this
correction (Environmental factor “K”). Measuring load noise at reduced load current

Secondly, Table – II shows that it would be sufficient to use
values of 0.10, 0.20, and 0.35 for the coefficient “ ” to
account for three different types of rooms. The error involved
in using these 3 values, versus the original 7 values in the IEC
Table – I, is within a fraction of a dB. It is, therefore, proposed
According to the IEC Standards, if the measurements can only
be performed at a reduced current (Current lower than the
rated full – load current of the transformer), the sound power
level at the rated current shall be calculated by the following
equation:

LwRated = LwRe duced + 40 Log [I Rated I Re duced ]

to use the IEC formula as is, but select one of the 3 values
proposed above for coefficient “ ” that represents the type of
the room used in the noise testing of the transformer. The equation is valid for a reduced current of 70 % of the
rated current.
IV. PROPER CONDITIONS FOR MEASURING LOAD NOISE As shown in Figure 6 below, the measured levels of load noise
Condition of Magnitude of Load Noise are compared to the levels calculated using the theoretical
relationship above. It is evident that this relationship is
Presently, the IEC Standards recommends to, first, perform a
accurate within 1 dB for a reasonable range of load [3].
calculation of the Load Sound Power level of the transformer
100
according to the following formula, Ref. [4]:

LW = 39 + 18 log(MVA) , dB (A)
95

90

According to IEC, “If Lw is found to be 8 dB or more below

Sound level, dB

85
the guaranteed no – load Sound Power level of the
transformer, load current sound measurements are not 80

appropriate.” 75
The above formula presents magnitudes of load noise as a
70
function of the transformer MVA rating regardless of the
Measured
design of the transformer or the physics of the electromagnetic 65
40 Log (I)
forces which produce this noise. It is, therefore, expected to
60
not provide accurate values of load noise. This is demonstrated 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3
in Figure 5 below where measured load noise levels are I, P.U.
compared with those calculated using the above equation for Figure 6 – Measured Load noise levels at different loads
power transformers of a wide range of power ratings. It is
evident that the formula mostly provides an average value of Note that measuring Load noise at lower currents can be
Load noise level but at the same time, for some designs, it can difficult as load noise levels at such currents can be in the
be as much as 5 – 10 dB in error, Ref. [8]. Hence, the above range of, or even lower than, ambient noise.
equation can not be used to judge whether Load noise is high
enough to warrant measuring it. It is, therefore, proposed to

V. CONCLUSIONS
The extensive data and discussions presented in this paper
suggest that the IEC and IEEE Transformers Standards
incorporate the following changes / additions:
A. Appropriate conditions for measuring Transformer noise
using the Sound Intensity Method
No correction for (Pm - Im) 4 dB
Add 1 dB to measured value of Im for 4 < [Pm - Im] 5dB
Add 2 dB to measured value of Im for 5 < (Pm - Im) 6dB
If (Pm - Im) > 6 dB, measurements using the Intensity Method
should not be allowed. Instead, the Sound Pressure method
would have to be used; correcting for Sound Wall –
Reflections and ambient noise. Obviously, this is provided that
conditions of ambient noise, as specified in the IEC and IEEE
Standards, are met. The presently used limit for the quantity
(Pm - Im) in the IEC Standard is 8 dB. The data presented in
this paper has demonstrated that such a value represents a
negative error of 5 dB in the measured noise level.
B. Wall Sound – Reflection Correction
The present IEC Formula for this correction should be used.
However, a limit of 3 dB is to be allowed for this correction
instead of the presently used limit of 7 dB.
Also, the value for the reflection coefficient “ ” should be
selected from one of the following values: 0.10 / 0.20 / 0.35,
such that the value selected should represent, as close as
possible, the type of the room used in the noise testing of the
transformer. These 3 values replace the original 7 values
presently used in the IEC Standards.
C. Measuring Load Noise
Presently, the IEC Standards recommends to, first, perform a
calculation of the Load Sound Power level of the transformer
“Lw” according to an old very approximate formula. Load
noise is to be measured only if Lw was found to be of a
magnitude within 8 dB of the guaranteed No – Load Sound
Power level of the transformer.
Data presented in this paper demonstrates that for some
designs, the formula can be as much as 5 – 10 dB in error.
Hence, it is proposed to measure Load noise of a transformer if
the transformer is to be loaded to any significant levels during
operation.
According to the IEC Standards, if the measurements can only
be performed at a reduced current, the sound power level at
the rated current shall be calculated using the “40 Log”
relationship between Load current and Load Sound Level.
Results presented in this paper confirmed that this relationship
is accurate within 1 dB for a reasonable range of load.
6
VI. REFERENCES
[1] IEEE Std C57.12.90-2006 “IEEE Standard Test Code for
Liquid Immersed Distribution, Power and Regulating
Transformers”.
[2] Ramsis Girgis, K. Garner, M. Bernesjö, J. Anger, and D.
Chu, “Measuring No – Load and Load Noise of Power
Transformers using the Sound Pressure and Sound
Intensity Methods”, Part – I: Outdoors Measurements,
IEEE PES Conference Proceedings, July 23, 2008, Paper
# 08GM0992, 8 pages.
[3] Ramsis Girgis, K. Garner, M. Bernesjö, J. Anger, and D.
Chu, “Measuring No – Load and Load Noise of Power
Transformers using the Sound Pressure and Sound
Intensity Methods”, Part – II: Indoor Measurements, IEEE
PES Conference Proceedings, July 23, 2008, Paper #
08GM1388, 8 pages.
[4] “Study of Noise Emitted by Power Transformers based on
Today’s Viewpoint”. E. Reiplinger, CIGRE Session 1988,
Paper # 12 -08, 8 pages.
[5] IEC 60076-10:2001, “Power transformers – Part 10:
Determination of sound levels”.
[6] IEC 60076-10-1: 2005, “Power transformers – Part 10-1:
Determination of transformer and reactor sound levels –
User guide”.
[7] “Transformer Noise: Determination of Sound Power
Level using the Sound Intensity Measurement Method”
Report by Cigre Working Group 12 of Study Committee
12. Electra No 144 October 1992.
[8] Ramsis Girgis, M. Bernesjö, J. Anger, “Comprehensive
Analysis of Load Noise of Power Transformers”, IEEE
PES Conference Proceedings, July 29, 2009, Paper #
09GM0439, 7 pages.
VII. BIOGRAPHIES
Dr. Ramsis S. Girgis (F' 93) is presently
the R&D manger at the Power
Transformer Division of ABB located in
St. Louis, Missouri. He is also the leader
of the global ABB R&D activities in the
area of “Transformer Core Performance”.
He is also the Co – Leader of the R&D
activities in the “Transformer Noise”
area. Ramsis received his Ph.D. degree
from the University of Saskatchewan,
Canada, in Electrical Engineering in 1978. Dr. Girgis has 40
years of R&D experience in the area of power, distribution,
and high frequency transformers, rotating machines, and pulse
power components. He has published and presented over 70
scientific papers in IEEE, IEE, CIGRE, and other international
journals. Until recently, he has been the chairman of the IEEE
Transformers Standards “Performance Characteristics” S.C.
Dr. Girgis is presently the Chairman of the Audible Sound &
Vibrations WG to revise IEEE Standards of Transformer noise
 7

Measurement, C57.12.90, Section 13. He co-authored chapters

in two electrical engineering handbooks on transformer design
and transformer noise. He is the past Technical Advisor
representing the US National Committee in the IEC Technical
Committee (14). In the past 5 years, Dr. Girgis has been the
leader of the team that developed the ABB Low and Ultra –
Low Power Transformer Technology.

Mr. Mats S. Bernesjö was born in

Säffle, Sweden. He received his B.Sc.
from Mid – Sweden University, Sweden
in 1998 and his M.Sc. degree from
Narvik University College, Norway, in
2000, all in electrical engineering. Mats
has worked in the power transformer
industry for almost 7 years and is
presently a development engineer at the
ABB Power Transformer plant in St
Louis, Missouri, USA. His past experience includes
positions at ABB Power Transformers division in Ludvika,
Sweden, as an electrical designer. In the past 5 years, Mats
has been one of the main contributing members of the team
that developed the ABB Low and Ultra – Low Power
Transformer Technology.