transformer loss tests 1

[gwosun]

we have a 2400VAC delta distribution system with 10 single and 12 three phase distribution transformers. the total kva is about 1000, but we only average 57 kw usage. the xfmrs range in age from 30 years to 100 years old. we are billed at 2400VAC with no penalty for power factor. i would like to measure the transformer losses and distribution line losses:

cable resistance losses
cable insulation losses
xfmr primary copper losses
xfmr iron losses
xfmr seconday copper losses

we may conclude that we will save on our electric bill (currently 75,000 annually) by replacing some or all of the xfmrs.

i can estimate the cable resistance losses from the wire gauges.

We do not have very sophisticated test equipment, only an amprobe DM-II for secondary data logging and 4 1200:100 hv current transformers.

since our smallest xfmrs are 5 kva, i expect to need to measure very small currents. i think that 100 windings of hv test wire through a hv ct may work for iron losses with the secondary open.

i think that using the dm II with its cts on loops from the hv cts on the primary side and its voltage probes on the secondary side will give me power factor.

our low ohms tester only puts out 100 ma, so i think we cannot get good test results on four wire tests on the xfmr windings.

any help you can give us on xfmr and cable testing on a low budget will be greatly appreciated.

-thanks

-Craig

 

[electricpete]

I don't have much experience in this type test but it sounds like you cannot get a reliable measurement with what you have.

Off the top of my head I can think of only two approaches for reliably estimating I^2*R losses:

#1 - Precise off-line measurement of the dc resistance (you indicate you were unsuccessful).

#2 - Precise On-line ac measurement which is capable of measuring the angle between voltage and current (and therefore of computing the power). Losses would be power in minus power out. But since losses are much smaller than the power you are measuring, a small percent-error in power measurement may lead to large error in loss calc. Along with difficulties in obtaining high-voltage measurements and additional errors from instrument transformers... #1 is probably a better bet.

Is it possible that there is some paperwork filed somewhere from the time of transformer purchase which documents factory testing at that time?

Is it possible to contact the manufacturer to see if he might have that data?

 

[electricpete]

I should mention there are other categories of losses beyond I^2*R.

This would include stray losses and core losses. Obviously you would not capture these with method #1. You would capture them with method #2 if you can do it accurately. Also the factory data would have this info.

 

[gwosun]

The facility we have was the Mendocino State Hospital until 1972, when it was closed. I have done an extensive search of all maintenance records, there is nothing on the transformers. Quite a few records are gone. I suspect that they were packed up and stored in Sacramento somewhere. I will try the manufacturers. I hope that they do a better job of preserving old records that the state of California!

What were typical iron losses for old (30 to 100 year old) oil filled transformers? can I estimate iron losses by the weight of the core?

 

[busbar]

Someday, somewhere, someone will cook up an effortless, instantaneous, unattended, free magic index tool for warehouses of old paper records.


Craig— You have presented an interesting but somewhat agonizing scenario. In an older A.S. Gill book on electrical testing, he points out {likely from a CYA standpoint} that one need always have spare parts available should the stresses of field testing* initiate unforeseen and undesirable circumstances, chiefly with implied and significant contributions of age and duty in the equipment and cables under test. In many cases this advice is probably ignored, but is a valid perspective none the less. Loss testing implies a degree of insulation stress, as do transformer short-circuit tests such as described in ANSI C57** standards. I would be gravely concerned of failure in one component, much less potentially multiple failures that could occur under field conditions and close scrutiny. You could probably use some less-invasive rating or prioritization methods like oil-DGA sampling if the transformer sizes warrant the expense, and accept published calculation/modeling of cable characteristics, based on published consensus literature and plausibly tempered for their age.


Another approach may be to get an initial reference point by readings at the site’s main service entrance, with instruments connected in the secondaries of existing current and potential transformers. Evidence of this capability may be switchboard volt/ammeters and phase-selecting tap switches; or possibly protective relays at the same location.

It is crucial that the user be appropriately cognizant of the overvoltage hazards connection errors or splitting existed wiring may induce. There are portable meters intended for 3ø ‘snapshots’ that have fairly decent very-low-range amplitude and phase-angle/power-factor accuracy/resolution measurements. Also, note that portable split/probe-type current transformers often have hideous ratio- and phase-angle errors at lower currents.

Two-element 3-wire tests would be acceptable if that’s what the utility uses. Zero-sequence currents should be of no significant concern, and would require three wye-connected PTs and three CTs.

{Typically} utility-owned 2-element metering installed for three-wire delta service, especially if older. Utility metering may indeed only be fed from two phase-to-phase voltage connections and two current transformers. Id est, if you’re not getting billed for it, the measurement becomes academic.

It’s not clear how the 57kW average was determined, but is the peak value known or anticipated to be significantly different? Some units, eg, “75,000 annually” could use clarification. No doubt you are aware of fundamental differences of AC versus DC losses and their corresponding measurements. The term ‘ohms tester’ is not apparent if of AC or DC excitation. Low system loading does not guarantee no overloading on a particular component, but you are likely aware that logger data may be of lesser value if loading is consistently as you have described.

The 57 versus sum-totaled 1000 comparison suggests fairly light use, although insulation deterioration can result from mere energization. Generally diverse loading does not completely dismiss possible small “bottlenecks” or “hot spots.” Has any of the equipment or cables, switchgear or transformers had heavier legacy loading periods? Your close observations, and those of others having regular contact with the system are most critcal for making others understand the relative significance that you and your immediates place on the gear, or merely the “service” it provides. Realize that each part has its own mode of aging. Particularly in your situation, it would be effort well spent soliciting RFQs from as many consulting firms as time permits, weighed against the opinion of the system’s value to upper tiers.


If you decide to tackle this, it can be a big job trying to prioritize a long list of observations and comments, but do not forget to include input by people who have and may continue to have daily contact with the system components. Their understanding of present and future system iterations have a big effect on service continuity and basic operational safety. Because cutover from old to new will probably not occur at an immediate point in time, scheduling will be increasingly important, as will your sense of local recipients’ perception of electrical service.

There will be some obvious resistance to unbolt so much as a cabinet cover, so a mandatory, documented walkaround of potential vendors to demonstrate as fully and honestly as possible what they are up against [and listening for feedback on several levels.] It may be blatantly obvious, but extensive communication between you, your crew, managers and vendors will be necessary for this to work. This may be invaluable to take to your procurement and maintenance managers, to best appeal to their concepts of common-sense with deference to the immediate and, if applicable, historic worth of the system [and their budget projections.]

If the inspection tour or preparations made by your crew exposes any otherwise enclosed live surfaces, include PPE as carefully suitable for the application. Absolutely acute and unquestioned knowledge of broad and specific operational hazards by your personnel encroaching or nearing legally-defined minimum approach distances becomes literally a life-or-death matter. [Be cautions of remarks stemming from the sometimes deep-seated ‘done-it-that-way-for-forty-years’ rationalization.]

Written descriptions and photographs made available to potential participants in the project are usually of great value. If you have not already discovered slate, cotton or maybe some asbestos, it will probably turn up on closer inspection.

For some served facilities, short-tem procurement, nursing and rotation of portable, isolated gensets may be desirable during modifications where disruption could be costly or compromising.

Be well aware that in the US, regardless or your affiliation, OSHA practices are unwise to skirt. Critically, practices for >600V systems—known to be energized or deenergized—are extremely different and most unforgiving if the associated personnel are not thoroughly and respectfully aware of and rehearsed in accepted, established procedures. Ungrounded-delta distribution adds still further need for solid understanding of how theory is manifested in operational procedures. Use of modified or experimentally conceived electrical assemblies should be avoided if at all possible. Load-monitoring/analyzing equipment is often rentable/loanable, eg, contact Arbiter Systems or Reliable Power Meter in NorCal or Sensorlink in Acme, Washington.

You are translating your understanding of the importance of electric service to your facilities, and the significance of doing that with the present gear, or that of radically overhauled version.

Consider in the rework approach that a new, higher operating voltage may have economies in the form of standardized equipment applicability and availability. As ad rem, a grounding zig-zag or grounded-wye/uncommitted-delta transformer bank usually limits overvoltage extremes and permits more effective use of surge protection. This typically provides more-sensitive, faster and associated increasingly effective ground-fault protection. It is remotely possible that a 4160Y/2400 high side could ease systematic-swapover burden, if parts of existing are not found in immediate peril.

If no one is able to present typical transformer numbers here, you may want to call 541.826.2113 in Medford and explain your situation—alternately Tom Steeber 541.826.8840 Alstomtransformers.com formerly Balteau-Standard {Casually hint that you may need quotes on assorted replacement distribution transformers soon.}

Note that there are immediate and long-term consequences of electrical-distribution failure, and how that fits into your expectations. Also, if removal, repair or modification of any existing components is needed, as you see fit take time to become oriented—however unavoidably—to environmental, not-always-rational restrictions that may surface with operation or modification of the old system. Expensive accommodations that may be needed can become a budgetary landmine.

It may seem inappropriate to say, and discouragement is not intended, but be very careful and best of luck in your efforts. At the upshot, it could be a very valuable electrical-operations episode. If you suspect so, I hope that if some record of the existing system would be of historical benefit to a larger community, that you will be allowed to adequately record its present and past significance for others.



*Inanimate anthropic bias/quantum theory: “observation affects reality”

**Besides insulation quality, there are basic no-load {‘core,’ iron, low-pf, generally voltage-influenced} and full-load {‘winding,’ copper, higher-pf, short-circuit, mostly current-influenced} loss tests. Make the determination that instrumentation for core-loss checks is suitable at very-low current range and low PF.

 

[jbartos]

Suggestion: Probably, you may have to check all affected equipment nameplates and try to obtain information from the equipment manufacturers. Also, software, modeling your power distribution system, would help.

 

[gwosun]

I would like to thank the particpants for their long and well thought out responses. Here is additional information on our system, as well as a few more questions.

Spare parts. We currently have three 5000 ft. spools of XLP-PVC 5 KV wire mounted on spools in a truck ready to drive up to a pull hole. We only have 6 spare 5 kva transformers and one single phase 50 kva on hand for xfmr replacement. Our supply of transformers was severely depleted 15 years ago when we paid for disposal of all our PCB xfmrs. If xfmr replacement will have a reasonable payback period, we plan to leave the old xfmrs in the vaults and manholes as backups. Since we do not want to be caught short, it does not seem like a good idea to strenuously test all the transformers, perhaps losing a few, and then order replacements for those that failed as well as those with unacceptable losses. If i could get some numbers for typical losses for transformers over 30 years old and in sizes from 5 to 150 kva, I might be able to make a replacement decision even though I do not have precise numbers for our particulat xfmrs.

Our total bill for 2001 was $74,661 for 541500 kwh (13.8 cents per kwh) If xfmr copper losses were 2% at 100% load, then maximum copper losses are $1493, minimum losses are (62 kwh average (57 was incorect) ) $6. So the mean, $95 is a very rough estimate, enough to see that copper losses are not enough to justify replacement.

However if iron losses are 1% of xfmr capacity, then annual iron losses are $12,089. This would be enough to fund transformer upgrades and downsizing.

Are there rules of thumb, or better estimates of xfmr losses that I can use?

How much is a DGA test? Can I mix xfmr oil from many xfmrs of the same vintage and usage history to get an inexpensive average for the condition of the lot? If the xfmrs are at the end of their useful lives, then replacement will be easier to justify.

On the subject of cable, where can I find "published calculation/modeling of cable characteristics"? We have been using the "replace the wire when it blows up" model so far, but this usually means pulling cable in the middle of the worst winter storms. I am interested in partial discharge monitoring to choose which wires should be replaced, but since I have not finished reading all the threads on this topic, perhaps I will do that and then see if I still have questions.

There is quite a lot more to explain, but time has run out.
Thanks again for your well though out responses. I will explain more later.

-Regards

-Craig

 

[electricuwe]

Judging from the amount of cable you are stocking as spare part I assume that your system contains not only 22 transformers but also several 100000 feet of cable. May be thats the main reason for your losses. Cable draws an significant ammount of capacitive current which leads to resistive losses in the cable als well as in the transformer windings. See thread238-21218 for a discussion on that topic.

Regarding the transfomer no-load losses I will look for some old books to give you some information on that topic.

 

[electricpete]

The DGA test requires the sample to be carefully controlled to prevent gas escape. I don't think there would be any easy way to mix samples on site. Even if you could think of a way (have the lab mix upon arrival), I just don't think that's a good idea. DGA results are tough enough to interpret as it is, without mixing several transforme samples together.

 

[electricuwe]

I just want to cite two design examples from the first half of the last century I found in some old books.

10 kVA single phase, 6000V/220V 50 Hz air cooled: 100 W no load losses, 200 W load losses
(Rziha, Seidener: "Starkstromtechnik, Taschenbuch für Elektrotechnik"; Berlin, 1921)

100 kVA three phase, 6600V/400V 50 Hz oil-cooled:
576 W no load losses, 2060 W load losses
(Kehse, W.: "Der praktische Transformatorenbau"; Stuttgart, 1934)

So the data you have choosen for your calculation seem to be quite reasonable.

 

[busbar]

I haven’t combed IEEE C57 on this, but possibly iron and copper losses could be tested on one transformer, and scaled, based on {safer} low-side excitation.

[Does] Labeled per-unit IZ correlate to copper losses only or copper+iron losses?

[Does] Non-zero-sequence {ø-g-ø or “series” zero-sequence} cable capacitive reactance partially offset transformer magnetizing reactance?

These two comments are open to correction or discussion. Understood these would not yield precise measurements, but possibly close enough to determine running costs ±10%.

Craig—if you need a loaner low-range instrument to conduct loss tests, could you furnish your mailto: for further offline?

 

[gwosun]

Thanks again for thorough and useful responses.

Here is some more info:

Safety: Fortunately, we have a "never work live" policy. We lockout the hv power, lockout the secondary mains ( A generator hooked to the building wiring can generate hv if you do not do this. ) test for the presence of power just to be sure, and work with gloves and insulated tools whenever possible anyway. 2400vac at 100 amps can ionize the air and sustain an arc six feet long!

Based on the feedback I got, I now plan to use Sangamo watt meters with CTs and PTs (The secondary of the transformer under test should suffice for a PT.) and a dummy load to measure iron losses. A dummy load about ten times the size of the iron losses should be enough to bring the meter up into a level where it is fairly accurate, and yet not be so large that it overwhelms the iron loss power. I will measure the dummy load with the same meter, and the difference should be iron losses. Copper losses do not seem large enough at present to warrrant measuring. EPA has a program called distribution transformer cost evaluation model (DTCEM) for making decisions about replacement:

http://yosemite.epa.gov/Estar/consumers.nsf/content/utility_distribution_transformers.htm

Here are more questions:

Can I use a capacitance meter to accurately estimate the inrush current losses? Is the deterioration of the dielectric properties of the cable over time large enough to warrant replacement of cable?

Partial discharge monitoring would seem to give a shorter warning period for higher voltages, and a longer warning period for lesser voltages. Would partial discharge monitoring on 2400 vac be effective for yearly testing to determine which cables should be replaced that year?

-Thanks

-Craig

craigcas@jps.net

 

[busbar]

Craig--one link is partialdischarge.com posted by others here. Cutler-Hammer seems to be a big proponent of PD testing and interpretation/evaluation. I cannot speak firsthand, but consensus seems to be that considerable skill is necessary to judge insulation quality (and quantity, of sorts) with this method.

 

[busbar]

Based on ANSI/IEEE C57.12.90, power/distribution transformers’ no-load/iron/excitation losses are not generally load dependent, where load/copper losses are. In either case, a dummy load should be unnecessary. No-load tests are referred to as ‘open-circuit’ tests. Load teats are also termed ‘short-circuit’ tests. Temporary isolation and energization through the transformer-under-test low side should give reasonably accurate measurements, and increase safety and convenience in testing. Transformer label information should provide some information about losses to determine preliminary instrument range.

One brief comment is

http://www.control-transformer.com/tests.htm

· NO LOAD AND EXCITATION CURRENT TEST:
No load losses (excitation losses) are the losses of a transformer that is excited at a rated voltage and frequency, but which is not supplying load. No load losses include core loss, dielectric loss and loss in the windings due to exciting current. The no-load of a transformer consists principally of the iron loss in the transformer core.
∙ IMPEDANCE VOLTAGE AND LOAD LOSS TEST:
Load losses are those losses in a transformer that are incident to the carrying of the load. Load losses include I²R loss in the windings due to load current and stray loss due to stray fluxes. The voltage required to circulate the rated current under short-circuit conditions when connected on the rated voltage tap is the impedance voltage, and the watt loss measured is the load loss.

http://standards.ieee.org/reading/ieee/std_public/description/dtransformers/C57.12.90-1993_desc.html

Refer to "§8. No-load losses and excitation current," and "§9. Load losses and impedance voltage."

Post if further information is desired.

 

[tgott]

Re: Dissolved Gas Analysis (DGA) I want to say one performed by Doble would run around $400. I don't think you would be able to mix the oil from different units, and as electricpete points out, the collection process needs to be tightly controlled. The "problem" with a DGA in your situation is that DGA's are really used to establish a trend. That is, one will generally tell you the condition of the insulation, but won't tell you the rate at which it's changing. So you might get a test result that is really bad, but it may have been really bad for a long time and is still working! You just didn't know how bad it was so you weren't worried about it. There may be other tests that would tell you more about the condition of the units than DGA. See

http://www.doble.com/services/labs.htm

Back to Doble: They have probably the largest database of transformer problems and failures of anyone in the world. I suggest you give them a call and see if they have any data on sister units to yours.

http://www.doble.com/index.html

 

[busbar]

gwosun, would you contact me offline at tmp173@@nikola.com ?