# Parallel busbars: how to orient?

As part of my research, I'm doing calculations on a hypothetical high-current (4000 A) medium-voltage (5000 V) DC power transmission system using two parallel busbars. However, I need to decide how the busbars are oriented relative to each other, and I can't find information on what best practices are for this. Are there best practices, or does it really depend on the situation?

The main question I have is this: Should the busbars have their long or short sides facing each other; that is, should they be arranged with cross-section like this?

or like this?

(assume the ground plane is to the right, if that makes a difference.)

Edit: Come to think of it, should the ground plane be to the right? I assumed that would be best because we want symmetric impedance to ground on both the positive and negative busbars, but if there's a compelling reason not to, do let me know.

• One of those ascii arts doesn't look particularly parallel to me Commented Jul 11, 2018 at 15:26
• @PlasmaHH What do you mean? These are cross-sections of the busbars, if that's unclear. Commented Jul 11, 2018 at 15:30
• it indeed is, maybe up from ascii art to mspaint? Commented Jul 11, 2018 at 15:32
• at 45 degrees to the phase angle/ :) Commented Jul 11, 2018 at 15:33
• That was unclear. Add it into your question maybe with "wide edge horizontal" and "wide edge vertical". Commented Jul 11, 2018 at 15:33

Four things to consider

1. Stray inductance.
If you are after minimising stray inductance than option #1 is prefered

2. heat dissipation.
If you are are dissipating as much heat as possible, the larger the exposed surface area the better. thus #2 is prefered

3. proximity effect.
If you are after maximising the amount of copper you are after, you do not want the effective conductive area to be reduced. Akin to skin effect with AC but the general cause: proximity effect due to the magnetic field will reduce the usable area. Thus #2 is prefered

4. Attractive force. If you are wanting to minimise the attractive force (due to normal conduction or during a fault) and cannot mitigate it via restraining bolts then #2 is best

To the extent that the system is really steady state DC, it probably does not really matter apart from attempting to limit the forces on the mountings produced by the fields due to the Lorentz force (Remember to ensure that the things survives a short circuit, which can briefly produce huge fields, it is likely mainly the magnetic component that will matter).

Inductance is clearly not important in the steady state DC case, but if running DC with large current variations it can become a critical issue, in which case minimising the loop area will help with the inductance of the system. Remember your load impedance is ~1 ohm, so in the presence of high frequency AC current components (which does NOT mean AC current, necessarily!) it will not take much inductance to cause issues.

Parallel plates are probably better from a inductance and mechanical perspective, but may be worse from a cooling perspective, especially is bolted either side of a dielectric to minimise line impedance at high frequency.

• Inductance may actually be important, as the goal of this research is to use the normally-present ripple in order to find where a ground fault has occurred (either positive or negative getting shorted to ground; neither should normally be grounded)--perhaps I should have mentioned that in the question. So the AC component on the line is important, as it's the transmission line characteristics where that will show up. Commented Jul 11, 2018 at 15:48
• With only a single connection to ground, there will be little fault current flowing given the source is ground free, so you would probably not see much in the way of transmission line effects. I would be measuring the current flowing in a pair of known impedance's to ground, if they unbalance then something is pulling one leg towards ground and you can trivially alarm on that. This is old school technology in battery powered submarines where cable leakage was often a problem. Commented Jul 11, 2018 at 16:28
• There is always the capacitive connection to ground, though, which is what we'll be looking at. Our goal is to be able to use this with zero modification to the system, as well as to be able to locate the fault (though there will probably be wide tolerances on that location). I suppose the parasitic capacitance is essentially that known impedance to ground, since it can be measured fairly easily. Commented Jul 11, 2018 at 16:32