You are on page 1of 5

IEEE Transactions on Electrical Insulation

Vol. 25 No. 4 , August 1880

667

Environment a1 Aging of Insulating


Materials
Robert R. Dixon
Westinghouse Science & Technology Center,
Pittsburgh, PA

ABSTRACT
Predicting useful life of organic materials has been made possible by the use of Arrhenius plots. Recently, data have become
available to show the combined effects of humidity as well as
temperature in the aging performance of moisture sensitive materials. A method is developed here to use such information to
calculate the predicted life of a material operating in enclosed
(warm) electrical equipment in a .room of known temperature
and humidity environment.

INTRODUCTlON
REDICTIONSof long-term aging of organic materials

have been made for many years with the use of Arrhenius plots [l,21. These plots are constructed by determining the time to yield equal degrees of material degradation after aging a t a number of elevated temperatures.
Degradation information has been reported in the form
of Arrhenius plots and activation energies (slopes of the
plotted curves) for a large number of materials (e.g. [3]).
One prime advantage to the method is that long-term
predictions can be made by extrapolation of results from
short-time high temperature exposures. On the other
hand, high temperature accelerated aging eliminates the
presence of moisture because the ambient vapor pressures
in the ovens are so low. Thus, it is possible that performance of moisture-sensitive materials (such as polyesters,
nylons, polycarbonates, etc.) operating in room environments may not be predicted accurately by dry accelerated aging.
It was suggested that moisture sensitive materials used
in operating (warm) electrical equipment in a room of
known environment are actually exposed to temperatures
above room temperature and reduced relative humidities
(but not zero).

The purpose of this study was to find a method of determining the anticipated life of moisture sensitive materials
operating in such an environment, with a view toward estimating a 40-year maximum operating temperature.

INPUT INFORMATION
N order to approach this problem, it was necessary to

know the conditions of the room in which the equipment was to operate. Consequently, information was provided for room ambient conditions, and consisted of numerous temperature/humidity conditions measured in a
specific location over the period of a year. Two such
rooms were reported for the analysis discussed in this report, and the d a t a are listed in Table 1. The total number
of hours was 8752, slightly less than one year.

Results of aging exposures under controlled relative humidity are available in the recent literature (e.g. [4]), and
these results have demonstrated the extent to which some
materials are sensitive to the presence of moisture. An
Arrhenius plot of the aging curves (at constant humidities) of a flame-retardant poly( 1,4 butylene terephthalate) polyester, based on 50% retention of tensile strength,
is shown in Figure 1.

0018-9367/90/0800-667$1.00

@ 1990 IEEE

Dixon: Environmental Aging of Insulating Materials

668

----- --

/ . -I

II

E-I.1UEV
Eel.17EV
E-l.13EV
E-1.06EV

*-

0- -0
X--.3;
Y----Y

6D

1 OD

TEMPERATURE.

ii
20

OEG.

1
i

i C 2 L .A
;C C

.I- -

.
j

2D

6G
TEMPEHPTU3E.

Figure 1.

DEL.

Figure 3.

Arrhenius plot for unreinforced polyester (1,4butylene terephthalate) a t various relative humidities. Based on 50% retention of tensile strength.

Arrhenius curve for aging condition measured for


Room #l. Curve is transposed t o 50% RH curve
a t equivalent temperature of 29.8-C for unreinforced thermoplastic polyester.

Table 1.
Room Conditions a t T w o Separate Locations

h
8
78
241
434
620
866
855
759
657
625
602
624
683
693
446
561
8752

E-U.36EV
I

2 / _ L - L - I
lOGS

2D

6O

TEMPERATURE.

DEG.

Figure 2.
Arrhenius curve for equal aging calculated for
Room #1 at 60'C and 26.7%, showing intersection at 8752 h.

Room #1
Room #2
T'C RH% T * C RH%
36.0
34.8
34.5
33.8
33.5
33.1
31.9
34.3
31.5
28.7
25.9
23.7
23.7
23.7
23.7
23.7
total

36.0
35.5
36.0
35.0
37.0
37.0
37.0
28.0
29.0
29.0
33.0
23.0
20.0
14.0
12.0
< 10
hours

the form
From these d a t a and from additional data, constants
were calculated for a multiple regression equation [5],
which relates temperature and moisture effects. It takes

logt=A+-

46.3
45.2
44.8
44.2
43.8
43.4
42.2
44.6
41.8
39.1
36.3
34.1
34.1
34.1
34.1
34.1

20
20
20
20
21
21
20
16
16
17
18
13
11
< 10
< 10
< 10

-ClogP

where t is time (h), T is temperature (K), and P is the

IEEE Transactions on Electrical Insulation

140/110t E-C. 32EV


140/80 I 4BSEV
.
1 IO/BO I E 4 . 3 9 E V

Vol. 25 No. 4, August 1000

660

temperature is assumed (e.g. 60'C), holding the absolute


humidity constant allows the vapor pressure to be calculated [6] at the elevated temperature. Thus Tz and P2
are known for the assumed case, and t z can be calculated.
The resulting value of t z is the time required at the higher
temperature and lower relative humidity to produce the
same amount of aging that would have occurred a t the
room ambient in 8 h.

i
4
j

YR

__---

D3;----.6D

- -20

TEMPERATURE.

DEG.

Figure 4.
Arrhenius curves for aging condition measured in
Room #1 and calculated at three sets of temperatures to show effect of temperature selections.

partial pressure of water vapor in Pa, and A, B , and C


are constants developed for each material. For the material mentioned previously, 1,4 butylene terephthalate, the
constants were calculated as A = 4.411, B = 2189, and
C = 1.608.

1 GO

60

TEMPERATURE.

20

DEG.

Figure 5.
Arrhenius curves for aging condition measured in
Room #l. Three temperature sets, when transposed to 50% RH curve at their equivalent temperatures.

THE ISO-AGING CURVE


multiple regression equation represents the locus
of points of equal aging (in this case 50% retention
of tensile strength) over a range of temperature/humidity
conditions. Thus,

HE

logtl = A +

B
- ClogP1

(2)

Tl

Repeating this procedure for each of the conditions of


Room 1 over the period of 1 yr gives the total number of
hours at 60'C required to produce an extent of aging equal
to that which would result a t the actual room conditions
during one year.

represents the same amount of aging as

B
logti = A + - -ClogPZ
TZ

(3)

Subtracting these Equations yields

1% t z =

Lz XI
---

- c(lOg pz - 1%

+ log t l

(4)

From the data for Room 1 in Table 1, tl = 8 h, TI =


36% and RH = 36%; converting TI to K and RH to PI
in Pa produces values for t l , TI and PI. If an elevated

This entire calculation can then be repeated a t some


other assumed temperature (e.g. 26.7-C) to generate another number of hours required to produce the identical
amount of aging. Plotting the hours a t 60'C and the hours
at 26.7% on an Arrhenius plot allows a line to be drawn
through the time ( t l ) at room conditions, and the effective temperature/humidity point lies a t this intersection.
The line of equal aging and its intersection with 8752 h
is shown in Figure 2, with the equivalent temperature
calculated as 29.8.C.

Dixon: Environmental Aging of Insulating Materials

6 70

TRANSLATION OF THE
CURVE
LTHOUGH the curve calculated for Figure 2 represents
points of equal aging, the amount of aging of the material during a 1 yr exposure would be very small. In order
to achieve a significant amount of deterioration, the aging
curve must be moved upward to represent > 1 yr of aging. Of necessity, for this translation it must be assumed
that successive years of aging will occur under conditions
represented by the data in Table 1. In other words, it
must be assumed that the temperature/humidity data
were collected during a typical year.

$- -I 00

__

--

--

--I-

20

60

TEMPERATURE.

DEG.

Figure 7.
Arrhenius curves for aging condition measured in
Room #2. Curve is transposed to 25% RH curve
at equivalent temperature of 41.1-C for unreinforced thermoplastic polyester.

TEMPERATURE
ASSUM PTlONS
1
1 c3

J
20

60
TEMPERATURE.

DEG.

Figure 6.
Arrhenius curves for aging condition measured in
Room #2. Curve is transposed to 50% RH curve
at equivalent temperature of 41.1C for unreinforced thermoplastic polyester.
Lacking a definitive place to re-locate the curve, it was
necessary to make a conservative assumption that the
equivalent temperature had an associated relative humidity of 50%. This value was considered to be conservative
because the values listed for Room #1 in Table 1 were
< 50% at the equivalent temperature of 29.8C. When the
constant aging line was re-constructed a t 29.8C and 50%
R H , it intercepted the 40-yr line a t 73.9.C on the Arrhenius plot in Figure 3. Thus, it is estimated conservatively
that the material would reach 50% loss in tensile strength
in 40 yr if the operating temperature in the equipment
were 73.9-C.

URING the calculations to establish the iso-aging


curve, two temperature values were assumed (60 and
Also, it was assumed that the line connecting
the resultant points was a straight line. In fact, the line
is not straight, but is concave upward, as indicated in
Figure 4. The curves in this Figure were determined by
assuming three sets of temperatures (60/26.7, 60/43.3,
and 43.3126.7). It is seen that the 60/43.3 assumption
gave the flattest slope (lowest activation energy) and the
43.3126.7 selection gave the steepest slope.

D
26.7.C).

Transposing the 60143.3 curve to 50% RH (at an equivalent temperature of 26.7C) yielded an optimistic 40 yr
extrapolation of 91C (see Figure 5). The 60126.7 and
43.3126.7 pairs produced approximately equal effective
temperatures (29.8 and 29.6.C). However, the 60126.7
pair gave the more representative curve for elevated temperature operation, because the slopes decrease as the
temperature values increase. This last set of temperatures was used in the analysis described in this study,
but other selections may be more appropriate in other
applications.

IEEE Transactions on Electrical Insulation

Vol. 25 No. 4 , August 1000

HUM IDlTY ASSUMPTIONS

I#1

N a review of the humidity conditions existing in Room

of Table 1, it can be determined that the effective

relative humidity is less than 50% at the equivalent ternperature (29.8C). Therefore, the assumption of 50% RH
for the point of translation of the constant aging curve is
conservative.
When this same 50% value is used for Room #2 (Table l ) the resultant predicted life temperature a t 40 yr
is 44.4C, only slightly above the effective room temperature. This result is shown in Figure 6. However, the
relative humidities are lower in this warmer room, and a
translation to a 25% RH curve is calculated to be a more
realistic conservative value. With this conservative assumption, the 40 yr extrapolation yields 73.2C, as shown
in Figure 7.

CONCLUSIONS
LTHOUGH several assumptions must be made in this
development of predicted life, guidelines are indicated
for selecting reasonable assumptions. First, an explicit
assumption is made, that the period selected for measurement of room temperature and humidity is typical of
the room conditions over the life of the equipment under
consideration.

Secondly, the assumed temperatures for the iso-aging


curve calculations should span the room temperature and
the equipment operating temperature. Finally, the assumed relative humidity for translation of the iso-aging
curve should be somewhat above the humidity a t the calculated equivalent temperature, in order to allow a conservative prediction.

6 71

It should also be pointed out that, although thermal aging at controlled humidity has been conducted for some
of the materials known to be sensitive to the presence of
moisture, testing needs to be done on additional materiaging predictions On a wide range
as, in Order to
Of

REFERENCES
[l] T. W. Dakin, Electrical insulation deterioration treated as a chemical rate phenomenon, AIEE Transactions, Vol. 67, pp. 113-122, 1948.
[2] T . W. Dakin, Electrical insulation deterioration,
Electro-Technology, Vol. 66, pp. 123-130, 1960.
[3] R. R. Dixon, Thermal Aging Predictions from an
Arrhenius Plot with Only One Data Point, IEEE
Trans on Elec. Ins. Vol. 15, pp. 331-334, 1980.
[4] R. J . Gardner and J. R. Martin, Effect of Relative Humidity on the Mechanical Properties of Thermoplastic Polyesters, Society of Plastics Engineers
Annual Technical Conference Preprints, Vol. 25, pp.
831-834, 1979.
[5] W. F. H. Borman, The Effect of Temperature and
Humidity on the Long-Term Performance of Poly( Butylene Terephthalate) Compounds, Poly. Eng. &
Sci., Vol. 22, pp. 883-887, 1982.
[6] 0 . T . Zimmerman and I. Levine, Psychrometric Tables and Charts, 2nd edition, Industrial Research Service, 1964.
was received

On

in final form 1 May

You might also like