Transformer reliability data 2

[PowerfulStuff]

Hi,
Has anyone got a good source of transformer reliability data? I look after a 22kV distribution system in a tropical area and am looking for data to benchmark against.

Thanks in Advance!

 

[electricpete]

Take a look at the IEEE Gold Book

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[cuky2000]

It is difficult to give you specific data do to the variety of conditions in tropical areas such as lightning activities, maintenance practice, and diversity of conformance standards.

Transformer failure rate are negatively impacted by long season of hot weather, scarce maintenance resources, minimum reliability demand and high level of electrical disturbances often found in many tropical areas.

As general guidelines, the cause of transformer failures rate is expected to be around the following figures:
• Electrical Disturbances (30%)
• Lightning (20%)
• Deterioration of Insulation (17%)
• Inadequate Maintenance (15%)
• Loose Connections (8 %)
• Moisture (7%)
• Overloading (2%)
• Others: _{(Misapplication, poor workmanship-Manufacturer, sabotage)} 1%

For additional source on generic information on transformer failure rate may found in the enclose site

http://www.hsb.com/thelocomotive/Story/FullStory/chart4.jpg

Beware that many of these failure effects, however, will increase in probability due to ageing.

See the enclose site for supplementary references could help you

http://www.hvlab.ee.upatras.gr/iclp2000/proceedings/599_604.pdf

 

[xabproject]

Dear PowerfulStuff,
Are you looking for data in the form of
<<Unit Size-No of Units-Forced Outage Rate-MTTF-MTTR-Scheduled Maintenance wks/year-Loading pattern(daily/weekly/yearly)>>?
MTTF: Mean Time To Failure
MTTR: Mean Time To Repair
FOR: (MTTR)/(MTTR+MTTF)
Not that I have the figures readily with me, but I can look around to find them if that is what you are looking for! I feel that it would be best to look into the records available with you to create a benchmark for yourselves.
best regards,

 

[xabproject]

Dear PowerfulStuff,
Following extract from an article makes a point -


Determination of the probable condition of today’s operating, and somewhat aged, power transformers is a complex and arduous exercise. This requires a rigorous methodology in order to benchmark and rank the units on any given system.
For example, due to the insulation aging factor, a 25 year old transformer today, is not the same as a 25 year old transformer was fifteen or more years ago!
Determining the probable condition of power transformers which have reached the majority of their service life is a complex matter involving many variables:
•??All transformers are NOT created equal. Historically there has been little standardization, even within any given manufacturer, over the past 50 years.
•??Most units are custom-designed to meet individual utility specifications involving significant difference in design methodology, features, safety factors and use of materials. Economic and environmental requirements, such as no- load and load loss evaluation factors and noise levels, can have a significant impact on the design of any two units with “identical” nameplate ratings.
•??Transformer insulation systems, particularly for EHV units, are complex structures which require thorough analysis to determine electrical, dielectric and thermal stress levels. Without the detailed design skills and tools, this can result in a significant difference in design integrity for units in the same voltage class.
•??Utility purchasing and local P.U.C. practices have over the years resulted in major differences in loading practice with some being ultra conservative and others very aggressive.
•??Transformers are consumable assets and can be loaded in a variety of ways. Due to deterioration of the insulation system resulting from temperature, moisture level and the possibility of oxygen ingress in the oil, two units of the same design and chronological age can have a totally different “service age” or residual life expectancy.
•??No two operating environments are the same. The position of the unit on the system (and protection system), the service load and power factor, physical location to sunshine and airflow, system impedance and probability of over- voltage from switching and lightning strikes, corrosive elements. The leading cause of failure for power transformers is listed as “external”. The frequency and magnitude of short circuit faults can shorten the life or catastrophically fail even the “best” transformers on a T & D system.
•??Maintenance is arbitrary. Historically, maintenance practices and frequency were dictated by individual transformer manufacturers. As previously discussed, maintenance expenditure is low compared to the asset cost. This is, in part due to the fact that power transformers are (have been) very reliable with the typical system failure rate being 0.5 to 0.8 percent.
We are now entering a maintenance regime where “less is more”. Condition based or predictive maintenance relies heavily on testing, diagnostics, monitoring and the management of data resulting in action plans. A major part of this maintenance approach is to understand the demographics and probable condition of units on the system as the key to managing risk and reliability.
There is no single scientific method available to determine the condition or end-of- life of an operating power transformer. Experienced engineers, chemists and technicians are required to conduct analysis, tests, inspections and review historical data to help form the decision.

 

[PowerfulStuff]

Thanks for all the information so far. I should add some more detail. I am very fortunate to work on an industrial site built around 30 years ago. This has meant that 75% of the 100 22kV:433V transformers that I look after were all made at the same time by the same manufacturer and there are only a few different sizes! Most have circuit breakers rather than fuses and cable feeds. With good maintenance over the years everything is very reliable however I'm looking for an 'average' figure to compare out performance with. I would expect that the tropical location, installation on an industrial site (fallout and no overhead lines) and any MTTF's relevant to this particular situation or with a list of adjusting factors would be ideal.

 

[wrong100]

PowerfulStuff,

Go back to original requirements for min requirments.
There should be min specs, while I have seen these off by
as much as 1000 to 1 in reality. There should also be some
test reports.

Build to products should give insulation etc.
Performance specs should give mtbf mumbers direct.
Mech one condersation (vibe etc), electric is another,
but they both must be considered.
As far as starting from scratch "good hunting"

The thermo design, the base wire, insulation,
the mag structer "magnetic condiserations",
How much was the
wire stressed, when it (the trans) was built,
And on and on.
In specific to much detail to specify,
and even if you could,
This is only a number which states, if you build
1 billion units than 50% will fail in this time period.

Just to make your day all of these probility numbers
are false. They are based on a rate of degratiation that may or may not be valid
If when building the trans the wire used was under
too much stress it "may" have developed a weak spot,
so it could die in 1 yr, or 1000 yr, compaired
to the rest of the system.
Untill you have valid failure numbers, the answer is
"it will last forever".

NEMA has some info, but their numbers are based on
slow "less performance" nunbers, rather than on faults
(shorts).
Millitary is much worse.

If I rember there is a max life for oil filled trans.,
That is why they must be changed.
It relates to a chem conversion (ion transfer) I think.

A thought,
a first sign of failure should a temp rise above amb.
If you develope a chart of power verses temp rise including wind , sun etc . the first sign of failure should be
a leakage (using more power making a abnormal temp tise).

Note do not depend on UL ,CSA, etc., after geting unwritten
approval for a product (waiting for paperwork) they found it could be used for the military, 5 design "major" changes
two years $100k to them $200k to us on a $1k item cost, for 20 units, we said go away, than we got a pass. You got it right we loss 280k, anyone find something wrong.
In real mumbers, fire mech elec gen safety improvement
was zero. They (personal openion) were scared and wanted money.

My first mil design was 18 yrs ago the mtbf was 10,000 hrs,
units were being replaced once per 6 mo with the same mtbf
specs. After 18 yrs "NONE" of my units have failed. I only had a 12,000 hr mtbf. Going to mtbf numbers, they should have been replaced 6 times, min, they are used 365 days a year 24 hrs a day. (did 6 others same result).
Moral numbers are as valid as a weather report 10million
years ago.
O just for info some years ago it was deterinined
out of "ALL" failures in the mil that only .01% , that is
100 per million, could not be determimed for a cause of failure, so they decited to lump mtbf into this cat.
To restate "NO MEANING" sorry,

100WRONG

 

[wrong100]

I FORGOT
we built 250.
250*18*365*24=39.4 million hrs.
with no failures.
Than multiply that times 6 (different units).
That means real mtbf past the life of the universe.

Consder a consumer mfg.
100 million units 10 million fail year 1.
Mtbf is not 10/100 years.
I would be supprised if any were working in 10yrs.
"This is good design now days"
Only 10 % in the first yr.
Quality is in the design, not in some 50yr old
inate mumber system.

How about if I sell you a car that has a 50%
failure in 2 yrs(50% built will crash in 2 yrs).
Now how about if I "CHANGE THE RESTRICTIONS"
on the numbers and say "WHEN YOU USE THIS" AND YOU CHECK (THIS ITEM ONCE A MONTH)" than it should not fail for
10 years (1% ? will survive).

 

[prc]

Powerfullstuff:
Your question is source for reliability data of distribution trfs .
To understand reliabilty estimation and mathematics ,please refer to IEEE C57.117 Guide for reporting failure data of power transformers -Annexure A

Failure rate of trfs is normally expressed as % of total number of failures so far divided by total trf -years in service .Mean time between failures (MTBF) is reciprocal of failure rate.Failure rate of any engineering product follows a bathtub curve ie failure rate will be high at the beginning and end of the life .

Reliability ie chance for failure, decreases with age of trfs .This is equal to e raised to minus failure rate *time in years.

To specific query,normal failure rate of distribution trfs today is:
Single phase 0.3 % ,three phase 0.6 %.
There is also another IEEE std on the subject C57.125 Guide for failure investigation ,analysis of transformers.

 

[wrong100]

PowerfulStuff
Sorry but I tried to make a point.
Only good design and building in the real world
matters, if the specs are good.
Regardless of the specs. If one
supplier choses to handle the wire as it should be,
and another did not (too much tension, turns too sharp,kinks,etc), or the insulation temperature
was ok but the mech was not very good to shock, moisture,
or abrasion, than the first suppliers product will
last a lot longer (on the avg.)
This includes all handeling (original mfg,
dist, product mfg, even transport.)

Go to "NEMA,WIRE" to start looking, this will give you the
base knowledge, do not forget the pratical side.

The units I ref are all exposed to salt water
at temps of -55c (ice that moves things, friction,
breaks contacts, pushes wire to wire ), to
temps at 85c (rusts iron in 24hrs, expands coils,
in gen odd things to equiptment).

When I say that MTBF or any other standerds numbers
do not have a real meaning, go to the standerds.
The military as well as other org have not had to
justify the failure numbers.
About 15 yrs ago the mil published a report ,
something like 90% of failures were from misuse,
9% from wrong application, testing errors,etc,

NO FAILURES COULD BE RELATED TO "MTBF" NUMBERS.

this was out of a failure report that included
tens of millions if units in the U S gov.

If you want the numbers, the advice above is good.

wrong100

 

[leoliu]

I am doing a project on transforme reliability analysis too. The project will incoporate the reliability performance of different protection relays and circuit breakers. If anybody can provide the source of the reliability data of these safeguards, it will be appreciated a lot.

 

[wrong100]

leoliu

NEMA, UL, MIL T 28, CS, all have sections addressing
transformers. These are all based on heat and insulation
classes (materials). All assume both mech and elect mfg,
will not degrade the insulation material, (all of the general design rules followed).
Therefore they address the insulating properties verses
temperature and voltage.
Hope this helps.

wrong100