Monday, June 21, 2021

Cooling transformers

My head is in an odd place. I have been thinking about transformers.

Not Transformers the movie, silly. Transformers in the electrical sense, devices that change current from one Amperage-and-Voltage to another Amperage-and-Voltage. Key point: Transformers cannot create energy (A in)*(V in)>(A out)*(V out). In a good transformer the two values will be close to the same but there will always be loses in the transformer and those loses become heat.

One major difference between "industrial" equipment and what is available to the the general public involves duty-cycle. A welder in industry is likely to be water-cooled while the electrical gadgetry in a Harbor Freight welder might have a fan. An industrial piece of equipment might have a duty-cycle of 60%-to-80% power-on and the Harbor Freight equipment a duty-cycle as low as 5%.


One factor that makes transformers difficult to cool is that they are massive blocks of metal and much of the heat comes from eddy currents internal to the steel core. (Fun fact: Transformer cores are laminated because the interfaces between the sheets create resistance that reduces eddy currents.) Because the heat is generated internally it must travel a distance before air or other fluid can carry off the heat and it builds up.

It is not possible to open up the transformer to allow air to flow between the coils and the core because the conductors must be close to that steel core or the volumetric efficiency falls off a cliff. One significant failure mode of a transformer is when the conductors are separated by insulators and the heat accelerates the degradation of the insulating material.

Another factor that makes cooling transformers difficult is that pumps add cost. Furthermore, transformers in applications like the ones on top of power poles are expected to last for fifty or more years without maintenance. That is why those transformers are cooled by sitting in an oil bath. Passive cooling has no moving parts that can break. Can you imagine the expense of installing a hundred-million pumps with a service life of 50 years and very, very high reliability?

Proposal


Cross-drill holes into the core so holes intersect where they bottom out.

It will probably be necessary to impregnate the holes to seal any leak paths.

Choose any of several different methods to insert-and-seal plumbing into the drilled holes. 

This concept depends on intermittent steam cells developing much like a pot at a rolling boil. That pulsation is what drives the fluid through the Tesla valves.

If the designer choses to use a sealed system, the cooling system volume is to be 65% filled with coolant (distilled water or ethanol alcohol are proposed) and then the remaining 35% volume (air) is to be evacuated.

Heat in the drilled holes will cause localized boiling. The "bubbling" will challenge the Tesla valves which will only allow one-direction flow. As depicted in the image, hot coolant will travel up the red leg and gravity will carry it down, through the heat exchanger where the Tesla valve will allow it to enter the transformer core.

The heat exchanger here is shown as a finned, aluminum extrusion but it could be as simple as a coil of steel tubing. 

17 comments:

  1. You may run into the same problems of heat transfer that occurs in nuclear reactors ... once you you get close to boiling, a vapor film forms on the heat transfer surfaces and dramatically reduces the heat transfer coefficient.

    Geometry becomes a factor, among other things.

    https://www.physlink.com/Education/Askexperts/ae112.cfm

    https://asmedigitalcollection.asme.org/heattransfer/article-abstract/83/3/351/430783/Film-Boiling-Heat-Transfer-From-a-Horizontal?redirectedFrom=fulltext

    https://www.osti.gov/pages/biblio/1797320

    https://ocw.mit.edu/courses/nuclear-engineering/22-06-engineering-of-nuclear-systems-fall-2010/lectures-and-readings/MIT22_06F10_lec13.pdf

    It's one reason why forced convection systems tend to be favored.

    ReplyDelete
  2. Plus the lifting takes a lot out of the flow rate, raising the temps as the coolant can't move as easily or quickly. Tde Delta is a bitch to get right.

    A larger failure mode for transformers is the "Concertina effect"..... this does more damage to the conductors than heat.

    ReplyDelete
  3. Better to drill top to bottom and connect the thru-passages with manifolds. Connect them to a horizontal radiator (slightly tilted) above the transformer, the height actually helps with starting the natural circulation, like the draft of a chimney. You can mount the radiator where it will have good airflow, and the slight tilt will take the cooler fluid down to the bottom of the transformer forming a natural circulation loop without needing valves.

    Steam in the transformer assumes it is close to 200F internally which to me is in the Not Good range.

    ReplyDelete
    Replies
    1. The old Johnny-Pop tractor, convective cooling system.

      Convective cooling depends on vertical real estate which might not be available and because of the low head requires generous piping diameters.

      Regarding temperature of boiling. One reason to seal the system and pull a soft vacuum is to lower the boiling point. Water at 13psi vacuum boils at about 125F.

      Another approach is to use a coolant with a lower boiling point like ethanol. Ethanol boils at 172 at 14.7psi. Ethanol boils at about 95F at 13psi vacuum.

      Delete
    2. Another approach is to use a coolant with a lower boiling point like ethanol. Ethanol boils at 172 at 14.7psi. Ethanol boils at about 95F at 13psi vacuum.

      Delete
  4. That's an interesting solution to a problem. The plus side would hopefully be a longer lasting transformer and slightly lower losses while operating because a cooler transformer has less resistance. Would drilling holes through the laminations allow more eddy currents when the bit goes through the insulation between laminations? Would removing iron from the core reduce the inductance and thus lower the amount of power the core could store magnetically? Could aluminum sheets or even steel sheets be incorporated in the laminations to extend past the edge of the transformer to form fins and increase its surface area? Another idea is to encapsulate the transformer in a finned heatsink and surround it with Helium. Helium is a much better conductor of heat than air, even at normal pressures. Hydrogen would work even better.

    ReplyDelete
    Replies
    1. High power generators and transformers are hydrogen cooled now. He isn't much used due to cost. As long as you are running 100% H2 there isn't an explosion problem.

      Delete
  5. What you are looking for is a heat pipe. Just search "heat pipe"

    ReplyDelete
    Replies
    1. I am not disputing what you say but the advantage of the twin holes and the Tesla valves is analogous to swirling liquid in a bottle before inverting. It organizes the flow so the vapor burping out does not impair the cooler fluid trying to flow in. https://physics.stackexchange.com/questions/150503/emptying-a-bottle-faster-by-swirl#150508

      Delete
  6. Why not use non-conductive oil and place the radiators vertically on the outside of the enclosure? Running at low or no pressure makes the hardware simple. A low pressure system (to reduce the BP) means you have to pull the vacuum in the first place then maintain it. Sealing a laminated core will be a challenge. Any metallic tubing in use will suffer induction heating and a plastic tube won't survive for long at temp and will be an insulator.

    ReplyDelete
  7. What about adapting computer CPU/GPU coolers to do the job? I have a big liquid cooler in my latest build and it runs cool and quiet. You can get pre-built units or by the system in parts. You might have to custom build the water block, but the rest is off the shelf.

    ReplyDelete
    Replies
    1. Joe is looking for a solution that doesn't need pumps and fans since the require power and break down over time. The heat pipe solutions would put the metallic pipes close to the high magnetic fields of the core making eddy current heating an issue too.

      Delete
  8. https://en.wikipedia.org/wiki/Thermosiphon

    ReplyDelete
  9. What is wrong with the standard solution of sealing the transformer in an oil bath and putting cooling fins on the outside? Alternatively you can put the transformer on a concrete slab and extend the cooling fins into the concrete.

    ReplyDelete
    Replies
    1. That is the best solution if you have the envelop to pull it off.

      Some gizmos are buried in "standard" components and the Tesla valve, intermittent pump solution is packaging friendly.

      The other thing is that I have been angling toward a post about auxiliary transformers so-as to save as much hardware in the distribution grid as possible. Those auxiliary transformers might have cooling issues.

      Dropping a few bread crumbs before my magnum opus on up-grading the domestic distribution grid seemed like a decent way to chum the water.

      I may be humble but I ain't skairt of having reach.

      Delete
  10. Absorption machine cooling....Kinda like the old Arkla technology, not the industrial absorbers like Carrier, York or Tranes which can also use waste heat sources, however they utilize pumps. Disadvantage they all require a deep vacuum and savvy mechanics.

    ReplyDelete

Readers who are willing to comment make this a better blog. Civil dialog is a valuable thing.