Explain heat rise difference 60hz vs. DC 2

[ramjet]

I'm a ME designing electrical connectors and would like help understanding what I've observed in testing. The connectors under test are used to pass 3 phase current through pressure zones in oil wells and consist of large thick steel shells with 3 insulated conductors within. The surface temperature of the steel shell ends up being about 10 to 30 percent higher when using 140Amps 60hz vs. 140Amps DC. Is the increase due to skin effect? Eddy Currents? Hysteresis? If the shell was non-magnetic or non-ferrous would that make a difference? Remember, I'm an ME, not an EE, so be gentle with me

 

[BrianR]

This shouldn't happen if you really are driving 140A 60Hz sine wave on all 3 pins, compared to 140A DC on all 3 pins. At 60Hz you can forget skin effect and dielectric losses over that length and the magnetic effects should be small.

If however you are running a Variable Frequency Drive motor using 140A/phase then it's a different story. You are not really driving 60Hz but more likely up to 16KHz pulses simulating a 60Hz waveform.

My suspicion is that you are not actually comparing 3 x 140A AC with 3 x 140A DC.

 

[davidbeach]

What is between the conductors? If the conductors go through individual holes in a ferromagnetic surface you will get inductive heating.

 

[Compositepro]

I have to chastise your physics but not the electrical part. What is a 10 to 30 percent increase in temperature? What percent change is it from zero degrees centigrade to 10 degrees. Yours is not an uncommon error. As to your question the explaination may very well be eddy current heating. This would be magnified if you have any loops in your wire for stress relief or to allow for assembly.

 

[itsmoked]

Ramjet; please look at this very recent thread and see if it answers your question...

thread238-128438

 

[ramjet]

There is only insulating material between the conductors and they remain straight and parallel throught the steel shell. The 10 to 30 percent increase I refer to is regarding heat rise above ambient. For example, the surface of the shell may measure 200F at 70F ambient with DC, and 230F at 70F ambient with 60hz AC. The DC is generated by Sorensen DCR 7-300B power supply and the meter is verified using a calibrated shunt. Our AC does not come out so clean; we use the output from a California Instruments variable frequency power supply (model 4500L-1P) and put it through a transformer to get the high current we need. The current is measured with a calibrated clamp on meter. The waveform is predominantly 60hz, but a bit dirty. However, we see the same increase with AC when the testing is performed at NRTL type facilities such as FMRC and CSA. Thank you everyone for your responses so far.

 

[weh3]

The 140A value under discussion is probably the RMS (root-mean-square) value, that is, the effective DC value. From a resistance perspective, there is no difference in heating of the wires, but there would be heating attributable to inductive effects that would not be present in the DC case.

There was a recent thread about the heating of steel conduit elbows in a parallel installation, each conduit containing a single phase. The heating effect was considerable.

William

 

[Compositepro]

The data you give indicates that there is significant resistive heating. The difference you see between AC and DC is probably due to the fact that The AC current is pulsing and has higher peak value than DC current. Resistive heating is proportional to the square of the current. Perhaps using RMS values should account for this, I'm not sure.

 

[ramjet]

Typically the steel shell is 2 to 3 inches in diameter and 2 to 3 feet long. Would grounding one end of the shell make a difference? Usually the shell is ungrounded, just laying on the cement floor.

 

[BrianR]

I thought the magnetic effects would be insignificant, but the link that Itsmoked gave would suggest that this is not so.

If it is the issue and the connector body must be made of ferrous metal then you are stuck with the heating. Grounding will not help. The RMS value of the current gives the same resistive heating effect as DC, that's why they use RMS. The magnetic effects are a function of frequency.

I immagine that it's no trivial exercise to try a brass connector body, but that is worth a go if possibel, as it's non-magnetic. Also using a larger diameter shell with the cable spaced to the centre will reduce the magnetic coupling.

 

[CarlPugh]

In a transformer the bushing exits the tank case (or top). This is similar to the situatin that you have.

In oil filled transformers the following is usually done:
For low currents, do nothing.
For medium currents, cut the tank between the bushing and welded with non magnetic stainless steel.
For high currents, mounted the bushings on stainless steel.

Try doing your test with a cold rolled steel plate and then a non magnetic stainless steel plate. Just drill or punch a hole through the steel and put an insulated conductor through. A plate 1/2' thick should give good test results.
The stainless steel should have low temperature rise and the cold rolled steel should have a high temperature rise.

Use a magnet to be sure that the stainless steel is non magnetic.

Good Luck
Carl Pugh

 

[ramjet]

Thanks very much for your responses. I will plan some more experiments with a monel shell when i return to the office next week. Does anyone else have any experience with actual measured heat rise values for DC vs 60hz? I seems like if it we're really this big of a difference that this problem would come up more frequently.

 

[ramjet]

I did some more carefully controlled testing using the exact same test parts and the results are startling! in some cases the heat rise with 60hz is double that of dc. Three .250 inch insulated copper rods running through a steel cylinder 1 3/8 ID with .188 inch wall thickness were tested with 154 amps. At 25C ambient temperature the 60hz test stabilized at 144C. The power was then changed over to DC and the temp then settled down to 82C. Amazing. Any similar experiences out there?

 

[MagneticFlux]

This may be a bit elementary, and please correct me if i wrong on any of this.

Here goes....

a) Whenever current flows thru a conductor, a magnetic field is produced around that conductor.

b) Magnetic Induction is the reverse of that law; whenever a conductor cuts through magnetic lines of flux, a voltage is induced into the conductor.

AC Issue:
You have a conductor (the ferromagnetic shell) which has a magnetic field passing through it (the three phase conductors) AND its alternating 60 times per second. The alternating field would induce some level of current into the "shell" causing a temp rise. Also, there is friction, current flow is electron flow and each time a electron strikes another, heat is produced and the energy of the striking electron is divided. Next, there is the magnetizing and de-magnetizing of the ferrous "shell". The magnetic domains are in a state of disarray while de-magnetized, and are spun into order (lined up) when magnetized, again, friction. And this is repeated 60 times per second.

Another thing worth noting is that there is 200 amps on each conductor which would mean there is 600 Amps of current flow thru the “shell” vs only 200 Amps for the DC wired (+) (-) or 100 wired (+) (-) (+)

DC issue:
Same as above however the current is constant, equal and opposite in polarity, conducted by 2 conductors with a 3rd conductor being used as ??? My guess is unconnected as a grounded conductor would be redundant. Copper is a good thermal conductor so its probably acting as a heat-sink thus lowering the temp by some level. The magnetizing of the "shell" takes place only once and is steady.

 

[ramjet]

Thanks MagneticFlux. Your explination make a lot of sense to this mechanical engineer.
To clarify the setup - the current is daisy-chained through the three conductors in the shell, for both AC and DC. In other words, current goes in on the left end, lead #1, while lead #2 and #3 are joined on the left end. On the right end, lead #1 and #2 are joined and the current returns to the supply on lead #3.

Since my last post, i've experimented some more and found that the AC heat rise remained the same whether the shell was grounded or not. I was hoping, and kind of expecting that grounding the shell would reduce the heating, but it did not.

I also tried holding 2 steel blocks about 4"x4"x2" each, with a 1.5" dia half cutout on each such that they can be held around the shell with only .001 clearance. The blocks pull together with a fairly strong magnetic force when the gap is closed to around .125 inch. The current level drops about 3 amps, indicating that the circuit resistance has gone up, simply by adding steel around the shell.

 

[zeitghost]

Have you experimented with a non magnetic shell?

Such as brass or aluminium?

Or stainless steel?

 

[BigWallyB]

If u suspect magnetic induction heating there is a method used in some older systems. Use a peice of copper or brass or alum. sheeting (inside) if possible where ever you can (insulated if necessary) to break that magnetism.

Life is what happens while we're making other plans.

Wally

 

[MagneticFlux]

Ramjet,

Its possible that the way you are testing the assembley is the reason for a good deal of the heating. You mentioned the cables are connected as follows:

(1)----------------(2)
|
(3)----------------(4)
|
(5)----------------(6)

Power supply at point 1 and return path on 6, is this correct?

If so, the looping of the conductors will increase the amount of flux induced in the shell. The shell would be act as a "core" like that of a transformer with AC applied.

Also, your using one-half of a single phase signal for the AC test ( A - N) ? This eliminates any field interaction that would normally take place between conductors, which has a canceling effect on the field.

I believe if you were to test the assembly connected in circuit as if it were actually installed for operation, you would see a decrease in the temp rise. My thinking is that you have a Pin and Sleeve type connector for a 3 phase, 3 wire circuit. With all 3 phase signals present, field interaction will reduce the flux available for induction, thus reducing the amount of induced current into the shell.

This is why electricians never group feeder cables, of like phases, in the same metallic raceway. Its always A-B-C-(N) or A-B-(N). There are some exceptions, but that's for another discussion.

Again, if I am in error, please correct...

 

[ramjet]

Sorry about the delay, i was on vacation last week. MagneticFlux - yes, you have illustrated the circuit arrangement correctly and we have speculated that the heat rise would be less with 3 phases cancelling each other's flux to some degree, rather than all in phase and additive. If i can figure out a way to do this test i will.
BigWally - will a thin sheet of brass or aluminum under the shell really break the magnetic effect? i can try experiment and i'll let you know.

 

[MJR2]

This is not a magnet effect exactly. It is eddy current heating of the tube. And brass or aluminum will heat faster or better/worse depending on your view point.

Magnetic fields will pass through brass or aluminum.

Eddy current heating only takes place with AC not DC fields as generated by the conductors.

Interesting idea the phase cancelling concept.