The T.M.Haja Sahib Transformer efficiency thread

[jim dungar]

But the 35% isn't a single point. It's average loading over a 24-hour period.
You'd need to have a 24-hour load profile to work out what the optimum design was for least energy losses over a typical 24-hour period.

I posted the original TP-1 reference, I am well aware that it is an average over a period.

No-load losses are primarily Fe losses. I don't quite see how reducing Cu I²R losses would or should increase Fe losses but I'm not a transformer designer.
Maybe you could explain?

Mike_Kilroy already explained there is a tradeoff between conductor and core sizing. There are physically more materials required to build a 500kVA unit then are required to build a 150kVA unit. Like most items there is no single 'always' correct answer for transformer sizing, but once you start to limit the number of variables, it becomes easier to find a correct answer. For a specific loading profile of <50% (ave over 24hr) an oversized transformer is typically a waste energy. Likewise for a loading profile of >95% fed by a transformer with little to no oversizing.

Most general purpose transformers have been wound with aluminum conductors for the past 50 years or so.
It has been decades since I was able to cost justify copper over aluminum windings based strictly on I?r losses and the resulting 'energy bill' savings.

 

[iwire]

T.M. does have a point about the transformers not being the most efficient at full load.
However, I do not agree about having a bunch of excess capacity built in as SOP.

I am mobile and have limited resources but here is one link with a graph:
http://www.copper.org/applications/electrical/energy/dist_trans.html

Just a guess here but it seems you did not bother to read thread before responding.

 

[Besoeker]

Mike_Kilroy already explained there is a tradeoff between conductor and core sizing.

Stated, not explained.
No disrespect to Mike.

There are physically more materials required to build a 500kVA unit then are required to build a 150kVA unit.

I don't think anyone would dispute that.

Like most items there is no single 'always' correct answer for transformer sizing,

Agreed.

Most general purpose transformers have been wound with aluminum conductors for the past 50 years or so

That may be regional.
I've come across just a few with Al conductors in several decades an none was general purpose.

 

[mivey]

Just a guess here but it seems you did not bother to read thread before responding.

Guilty. I thought I could get by with the most recent. It is no fun reading on a 3" screen. I thought T.M. was supporting having 65% excess capacity (ridiculous in my opinion). I also thought he was saying the max efficiency is below the 100% load point (which I believe is true). I'll have to wait till I get to the hotel to read the whole thread.

 

[mike_kilroy]

But what is the mechanism? How does reducing I²R losses result in increased Fe losses?

This is a question for an actual transformer designer for pure details. I don't do the actual design. I can repeat some comments about trading off they have said over the years. More steel in the core reduces core losses. Larger core means more wire required to go around it; hence R in I^2*R goes up, so here is one reason reducing core losses results in more I2r loss.

So say a transformer has a 10" dia coil around the core. If the core is enlarged, say it reduces the core losses by 30% and makes the coils now 15" dia, or 1.5x higher; this would change that ratio of cu to core losses by about 2:1

 

[jim dungar]

That may be regional.
I've come across just a few with Al conductors in several decades an none was general purpose.

Yep: regional as in North America.

 

[Besoeker]

This is a question for an actual transformer designer for pure details. I don't do the actual design. I can repeat some comments about trading off they have said over the years. More steel in the core reduces core losses. Larger core means more wire required to go around it; hence R in I^2*R goes up, so here is one reason reducing core losses results in more I2r loss.

So say a transformer has a 10" dia coil around the core. If the core is enlarged, say it reduces the core losses by 30% and makes the coils now 15" dia, or 1.5x higher; this would change that ratio of cu to core losses by about 2:1

Thinner lams or better steel would also reduce core loss without necessarily increasing core dimensions.

 

[mike_kilroy]

I agree there are lots of things that effect losses in a transformer. If one already uses thinnest lams made of best grain oriented steel stacked 1x1 rather than faster and lossier 2x2 or higher and any other higher efficiency designs to reduce core losses as far as possible, then i2r losses begin to get traded with core losses to change ratio between them if that is important for a given design.

 

[mivey]

T.M. does have a point about the transformers not being the most efficient at full load.
However, I do not agree about having a bunch of excess capacity built in as SOP.

I am mobile and have limited resources but here is one link with a graph:
http://www.copper.org/applications/electrical/energy/dist_trans.html

Just a guess here but it seems you did not bother to read thread before responding.

Guilty. I thought I could get by with the most recent. It is no fun reading on a 3" screen. I thought T.M. was supporting having 65% excess capacity (ridiculous in my opinion). I also thought he was saying the max efficiency is below the 100% load point (which I believe is true). I'll have to wait till I get to the hotel to read the whole thread.

OK Iwire,

I read the thread. T.M. stated that we should allow for extra capacity (actually said we should install at 35% loading), which I have disagreed with as a SOP. He also said that the maximum efficiency does not occur at 100% loading, which I agree with. So what did I miss?

 

[mivey]

The simple point I was making is that the notion of over rating a transformer to obtain better efficiency, as you have suggested simply, doesn't hold water.

That I totally agree with.

What it shows is that the efficiency doesn't vary very much with load over the likely operating range and is best at rated load thus negating the argument that operating below full load rating delivers better operating efficiency.

But it varies a little more with TP-1 transformers. Between 35% and 100%, the efficiency drops almost 2%.

Still does not justify sticking in a transformer that is oversized to have 65% extra capacity as T.M. suggested.

How about you provide a reference to support that?
...
Given that J&P, a comprehensive work, doesn't give ratios greater than 4:1 I'd be inclined to the view that anything beyond that wouldn't be very common and not really classed as general purpose.

I have J&P as well and it was certainly comprehensive at the time it was written but things do change.

Not only are we having to comply with TP-1 but dry-types are close to, if not already, going to have to comply with the NEMA premium efficiency rules.

Here is another link with a graph showing the TP-1 efficiency peaking at 35% load:

http://ecmweb.com/news/electric_select_right_energyefficient/

 

[mivey]

That could be full load for 35% of the time and zero load for 65% of the time or an unvarying 35% load all the time.

I'm pretty sure they specified a minimum efficiency at 35% loading not 100% for 35% of the time.

Is there a point I am missing? Sorry if I am missing it so help me out here.

I understand that T.M. is wrong to put in a transformer with 65% extra capacity as SOP, but how is T.M. wrong in stating that the maximum efficiency point is at 35% (at least by TP-1 rules)?

 

[mivey]

But the 35% isn't a single point. It's average loading over a 24-hour period.
You'd need to have a 24-hour load profile to work out what the optimum design was for least energy losses over a typical 24-hour period.

I don't think that is how it is spec'd.

I posted the original TP-1 reference, I am well aware that it is an average over a period.

Are you sure? I thought it was a minimum efficiency at a set point (the 35% and 50% loading points based on the transformer).

 

[mivey]

Thinner lams or better steel would also reduce core loss without necessarily increasing core dimensions.

I agree there are lots of things that effect losses in a transformer. If one already uses thinnest lams made of best grain oriented steel stacked 1x1 rather than faster and lossier 2x2 or higher and any other higher efficiency designs to reduce core losses as far as possible, then i2r losses begin to get traded with core losses to change ratio between them if that is important for a given design.

I don't have any of my references with me but what Mike is saying sure sounds familiar. There are trade-offs that I remember reading about but I do not recall the specifics. I won't be back at my office until the end of the week but maybe someone can find additional details before then.

 

[iceworm]

I was surprised that the best efficiency was not as 100% load. To my simple thinking, the core losses are fixed - independent of load, and the coil losses are porportional to load. This math model puts the best efficiency at max allowed loading.

But others are telling me best efficiency occurs at lower loading - and per TM, much lower loading.

So I came up with a new math model (new to me - I didn't do any research) that lincorporates the following:
1. Heat generated in the transformer (Ploss) is porportional to (Loading X coil resistance) + fixed core loss

2. Coil temperature is porportional to Ploss x (Tcoil - Tambient) + Tambient

3. Coil resistance goes up with temperature 0.00323 ohms/degC

4. Neglected voltage drop at output terminals

I used the data from the mike_kilroy post

... One of our last designs:

Our 400 kva transformer watts loss:
core loss: 3960.7
coil loss 4636.9
total loss: 8597.6

I set the ambient at 30C and picked full load coil temp of 150C

The results were fairly interesting (to a technogeek or enginerd - to quote my daughter)

From 50% load the efficiency is 97.7%. The efficiency peaks about 83% load - to 98%. The efficiency then drops off to 97.7% at 100% load. This model appears consistent with test results listed in this thread (Bes I think). The attached graph shows the results. This model appears to be a good second-order approximation.

So, what is the point? I feel like I understand the physics a bit better. And the exercise has reinforced my thinking there is no money in underloading to gain efficiency.

ice

 

[topgone]

I was surprised that the best efficiency was not as 100% load. To my simple thinking, the core losses are fixed - independent of load, and the coil losses are porportional to load. This math model puts the best efficiency at max allowed loading.

But others are telling me best efficiency occurs at lower loading - and per TM, much lower loading.

So I came up with a new math model (new to me - I didn't do any research) that lincorporates the following:
1. Heat generated in the transformer (Ploss) is porportional to (Loading X coil resistance) + fixed core loss

2. Coil temperature is porportional to Ploss x (Tcoil - Tambient) + Tambient

3. Coil resistance goes up with temperature 0.00323 ohms/degC

4. Neglected voltage drop at output terminals

I used the data from the mike_kilroy post



I set the ambient at 30C and picked full load coil temp of 150C

The results were fairly interesting (to a technogeek or enginerd - to quote my daughter)

From 50% load the efficiency is 97.7%. The efficiency peaks about 83% load - to 98%. The efficiency then drops off to 97.7% at 100% load. This model appears consistent with test results listed in this thread (Bes I think). The attached graph shows the results. This model appears to be a good second-order approximation.

So, what is the point? I feel like I understand the physics a bit better. And the exercise has reinforced my thinking there is no money in underloading to gain efficiency.

ice

There's nothing new on these, IMO. You will only achieve maximum efficiency when the Cu losses equal the Fe losses. Per example, the total losses = 3960.7 X 2 = 7,921.4 watts; maximum efficiency = 400/(400+7.9214) = 0.98.058 or 98.06%.

The load at maximum efficiency will be equal to the sqrt(Pc/Pl) where Pc = transformer iron loss at full load = 3960.7 watts and Pl = transformer copper loss at full load = 4636.9 watts. The transformer load at maximum efficiency will be = sqrt(3960.7/4636.9) = 0.9242 or 92.42% of full load.

Any loading lower or higher than 92.42% of 400 kW (369.68kW) will be an operating point that is less efficient.

 

[iceworm]

There's nothing new on these, IMO. ...

I didn't think there was. I said it was only new to me. I was, and am still, sure it would be old hat to someone such as yourself, that fully understands the physics. The reason I brought it up was no one was discussing the math models.

.... You will only achieve maximum efficiency when the Cu losses equal the Fe losses. ...

At my level of understanding, this is not readily apparent to me by inspection. I'm sure it is a simple derivation - its nothing I would ask you to do here. I'll see if I can work this one out.

QUOTE=topgone;1387356].... You will only achieve maximum efficiency when the Cu losses equal the Fe losses. Per example, the total losses = 3960.7 X 2 = 7,921.4 watts; maximum efficiency = 400/(400+7.9214) = 0.98.058 or 98.06%. ...[/QUOTE]
I don't get this one at all. The efficiency calc you use is for full load - 400kw. But the coil losses you are using are less than that listed for full load. ???

... The load at maximum efficiency will be equal to the sqrt(Pc/Pl) where Pc = transformer iron loss at full load = 3960.7 watts and Pl = transformer copper loss at full load = 4636.9 watts. The transformer load at maximum efficiency will be = sqrt(3960.7/4636.9) = 0.9242 or 92.42% of full load. ...

Again, at my level of understanding, this is not readily apparent to me by inspection. This may be a simple, but I suspect, tedious derivation. Perhaps you could list a reference.

I'm a bit surprised the maximum efficiency is not influenced by the ambient. Is that a third order effect and not part of your model? Or not even relavent?

ice

 

[mike_kilroy]

...
But it varies a little more with TP-1 transformers. Between 35% and 100%, the efficiency drops almost 2%.

Still does not justify sticking in a transformer that is oversized to have 65% extra capacity as T.M. suggested.

...

THIS is the problem TM could not see past: it is NOT 2% difference in efficiency! It is 2% when losses are compared to the transformers full rated kva, but NOT the the actual efficiency at lower loads! At lower loads the efficicency seldom changes by even 0.2%!

See actual TP-1 transformer load efficiencies: http://www.eaton.com/Electrical/Consultants/ConsultingApplicationGuide/index.htm

Look at page 19.

What TM insists on is comparing a transformer losses to its rating, NOT to the load attached. So Yes, comparing the lower losses at 50% load to the xfmrs full load rating IS maybe 2% less, but not when you compare power out/(power in +losses). Therein was TM's shortsightedness.

 

[jim dungar]

Are you sure? I thought it was a minimum efficiency at a set point (the 35% and 50% loading points based on the transformer).

A case of remembering wrong.
The research into the original "TP-1" justification found a 24hr average loading of 35%.
The current version of TP-1, is a specific point of 35%.

 

[iwire]

OK Iwire,

I read the thread. T.M. stated that we should allow for extra capacity (actually said we should install at 35% loading), which I have disagreed with as a SOP. He also said that the maximum efficiency does not occur at 100% loading, which I agree with. So what did I miss?

My read on your post was that you had come to save the day,.... set us all straight with some new info ..... etc. and I saw nothing new in your post.

From your link it showed very little difference between 35% and 100% loading.

 

[iwire]

I understand that T.M. is wrong to put in a transformer with 65% extra capacity as SOP, but how is T.M. wrong in stating that the maximum efficiency point is at 35% (at least by TP-1 rules)?

Maybe the difference comes from knowing where TM started.

A thread was opened and someone had asked how much they could load with a transformer Charlie B had suggested 100% was OK.

TM said no, you must have spare capacity

When no one bought into that he changed to claiming there would be a savings by installing a transformer that was too large.

So if you want to say a specific transformer may be more efficient by a small margin at 35% loading I have no issue with that.

If you want to say that installing a transformer with 65% extra capacity to save energy as TH seems to be suggesting I disagree with that.