Thermal fatigue of electrical components. 1

[mwmonty]

I am currently working on an RCFA of an electrical cabinet that is having premature component failure due to thermal fatigue. I have heard in several training courses that an increase of 10 degrees C will result in a decrease in electrial component life by 50%. However, I am having problems locating a reference that I can use to support my investigation. If anyone knows of a online reference I can use, it would be of great help.

MWMonty

 

[IRstuff]

The prediction is a generalization of the Arrhenius rate equation, which essentially Life(t)=rate*exp((-Ea/k)(1/t-1/t0)). Ea is the activation energy and the rule of thumb only applies for a particular activation energy and a particular t1.

Most reliability books and references will discuss this in more detail

A search for arrhenius and reliability pulled up:

http://nppp.jpl.nasa.gov/asic/Appendix.7.html

which is discussing integrated circuit reliability, but the concepts are the same for anything with this type of rate model.

TTFN

 

[electricpete]

The 10C rule applies generally to thermal aging of insulation (usually organic insulation in cables, transformers, motors, perhaps also found in smaller components). I will find you a reference if you can confirm this is the relevant area of your study. But....

I don't think it will have anything to do with thermal fatigue. If I'm not mistaken you're probably talking about fatigue of either a fuse element or perhaps a bolted or soldered connection. Either way it's metal, not insulation. The 10C Arhenius rule has nothing to do with it.

Can you tell us more specifically what you are looking at?

 

[mwmonty]

IRstuff,
Thank You for the link. It is just what I need to support my conclusion.

 

[mwmonty]

Electricpete,

The system I am working on is an electrical control cabinet for a hot oil system. I am seeing an MTBF of 3 months on components (wire, contactors, and fuses). Most failures have been multi-component and result in extended downtime. We have similiar systems on several other lines, but these lines are not subject to the same temperatures. The cabinet we are having problems with is subject to temperatures in excess of 60 degrees C for about 4 months a year. During the cooler months it drops to 40 degrees.

 

[electricpete]

I'm sorry I can't find a reference at the moment but I'll keep looking (perhaps you can search on Arhenius electrical aging).

To repeat there are at least 2 different temperature-related aging mechanisms
#1- thermal aging of insulation (unrelated to fatigue) which doulbes by the 10C rule and depends on time at temperature (but not cycles)
#2 - Fatigue type failures of metal components exposed to temperature cycles which impose changing mechanical loading.

 

[IRstuff]

Is the 60C the ambient or the actual equipment temperature?

The actual temperature of the electrical components could be substantially higher than the ambient, which would certainly degrade their reliability.

TTFN

 

[mwmonty]

IRstuff,
60C is the ambient temperature. Using thermography, I have found components with a Delta T of 55C. My recommendations to prevent reoccurance is to cool the cabinets using chilled air from our comfort and cooling (this system does not have a heating system.) I am convienced of the failure mechinism. Management is just wanting solid science (i.e. someone from outside the plant) to justify my findings.

 

[electricpete]

I have left a link to this thread at the power engineer's forum. You should see a lot of responses show up here shortly.

 

[redtrumpet]

Standbook Handbook for Electrical Engineers, Fink and Beaty, 11th Edition, McGraw Hill, p.10-42 Art. 143

"The aging of insulation is a chemical process which occurs more rapidly at higher temperatures according to the Arrhenius reaction-rate theory, as expressed in Eq. (10-63),
h = e^[K1 + K2/(C+273)] where C = degrees C temperature of insulation, K1, K2 = constants determined by test, and h = hours of life."

Hope this helps.

 

[electricpete]

thanks redtrumpet.

I didn't notice the first time that it appears mwmonty is happy with the link from irstuff.

I'll see if I can remove my post from the power forum.

 

[cuky2000]

The 10oC Arhenius rule to decrease electrical component life by 50 % could be difficult to explain
With the editing limitations However, I will take the risk to luck foolish with the following simplistic explanation:

a- Assuming that the component l follow the Arrhenius life expectancy:

Log[E(t)] = A + B/T_abs

T_abs = Ths+273

_{Where:
E = Life expended in during time interval t
Ths = Temp hot spot (oC) = Ambient Temp + Hottest- spot winding temp.
t = Time in hours
Tabs = Absolute temp (oK)
A, B = Constant for life expectancy curve of each insulation as follow:
Max Hottest Constant
Spot Temp A B
150 oC -8.270 5,581
185oC -7.941 5,907
220oC -10.453 7,582}

b- Using a graph (or equations) for the Percent Relative Life Expectancy (P) versus the Insulation hottest-spot temperature (Ths) the following data are determine
_{P is defined as follow
P = (E/Er).100
Where: See E defined above, Er = Life expended when operated at rated conditions.}

For Example, let’s use Fig 3 ANSI/IEEE C57.96-1989 for Dry Type transformers.

Insulation Hottest Relative Life Expectancy (Approx)
Spot Temp (oC) T@ P=100% T@ P=50% DELTA
150 oC 140 150 10oC
185oC 178 188 10oC
2200C 210 220 10oC

 

[buzzp]

Keep in mind the ambient temperature ratings of the components you are using as well as the operating load. UL508 only specifies controls up to 40C unless otherwise specified on the manufacturers data sheet. I believe this applies to contactors as well. Also, resistance changes with age and operating temperature. This also applies to fuses, etc. How are your fuses failing, open, failing to open? Are wires melting or how is the wire failing? This sounds like a more serios problem other than MTBF of components. Are the failures only in the hot months?

 

[IRstuff]

The 50% rule applies only to failures that have a certain activation energies, roughly between 0.5 eV and 0.65 eV. However, many electronic components have significantly higher activation energies, which means that they could have significantly higher acceleration rates.

for example, if the activation energy is 1.2 eV, the life reduction is more like 80% for a 10 degree change.

TTFN

 

[hotspot]

To MWMONTY,
Great question! The responses generated are a learning experience! I thought I might have a bit of usefull info for you. In 1990 at IRINFO'90, Mr. Alan C. Pierce of Factory Mutual Research Corporation presented a paper on derating electrical and electrical/mechanical components using tables, a simple equation and temp data. The report can be obtained from Infraspection Institute at

http://www.infraspection.com

, contact Jim or Chris Seffrin. The derating info applied to your equipment may assist you in convincing the powers to be of what kind of harm is occuring to their equipment. The Derating information was developed to assist in priortizing maintenace based on thermal excesses to the equipment. The data used to develop component thermal limits for this report is listed and consist of multiple existing standards.
Best of luck, and to those responding to MWMONTY's question, Thanks!

 

[mwmonty]

I wanted to thank all of you for all of the help. I have learned alot from all the the posts. I have only been in the reliability field for 2 years and this RCFA has shown me that I have so much more to learn. I have also found that this discussion board is a vitial resouce with all of it's member and thier technical expertise.

Thanks Again.

 

[jbartos]

Suggestion: Visit

http://www.silicavalley.com/archives/Reliability.pdf

http://www.spie.org/web/abstracts/2700/2780.html

http://www.cem.msu.edu/~cem852/lecture_03_25.pdf

etc. for more info

 

[Bung]

Interestingly, I first came across this rule in senior high school biology, applied to the decomposition of proteins to amino-acids under heating (frying your steak, f'rinstance - every 10deg rise in cooking temp halves the cooking time (and don't hassle me with comments about rates of heat conduction into the middle! assume ideal cooking of an ideal steak))

So maybe the biologists or bio-chemists can assist?
Bung
Life is non-linear...

 

[electricpete]

Well now I'm curious about something.

I can appreciate that many reactions rates will double for a fixed change in temperature.

But is the amount of that fixed change in temperature always 10C for any material and reaction? I kind of thought is was different for different materials/reactions,but Bung's discussion makes me wonder.

 

[cuky2000]

Electricpete,

I made a quick mathcad template applicable to 150, 185 & 220 oC Insulation for dry type transformer and the 55 & 65 oC for oil filled transformer. Although the result do not show exact loss of life of 50 % with 10 oC increment, the values are close enough for practical purposes. That me suspect that the 10 oC rule is applicable to different materials and medium.

An interesting observation of the Arrhenius equation plotted in a log-log scale shows that this relation behaves as a quasi-straight line within a practical temperature ranges.

This characteristic is perhaps the answer of the consistency of the 10 oC rule for any material and insulating media.