Why the column totals lie
On a single-phase panel the two legs are 180 degrees apart and the arithmetic is nearly ordinary. On a three-phase panel it is not. A 208 volt load connected between A and B draws its full current on A and its full current on B — not half on each. A second 208 volt load between B and C also draws its full current on B. And those two currents on B arrive 60 degrees apart, so they add to less than their arithmetic sum.
At the defaults the panel comes out at 57.0 A, 67.3 A and 53.8 A on A, B and C, with 6.6 A on the neutral and 13.4 percent imbalance. Add the same circuits up by column in watts and you get a different picture, because the welder and the booth heater both land on B and the arithmetic that combines them is not addition.
What the neutral does and does not carry
Only line-to-neutral circuits put current on the neutral. A line-to-line circuit sends its current out on one line and back on the other, so it contributes nothing. A balanced three-phase load contributes nothing either, because its three line currents sum to zero exactly.
Even the line-to-neutral circuits mostly cancel. Three equal 120 volt loads on A, B and C give exactly zero neutral current, which is the whole reason a four-wire wye system exists. Two equal ones on A and B give a neutral current equal to either of them, not twice it — that is the 120 degree geometry, and it surprises people every time.
One large caveat, and the page repeats it. That neutral figure is the fundamental only. Third harmonic current from switch-mode supplies and LED drivers does not cancel in the neutral — it adds arithmetically across all three phases — and on a panel of modern electronics the neutral can carry more current than any line while this page reads close to zero.
Moving the biggest load is not a rule
The page tries the largest line-to-neutral circuit in each of the three positions and prints all three results, because moving the heaviest load to the lightest line is not reliably the best move. The line-to-line circuits already on the panel pull the totals in directions that a column of watts does not show, and the phase that looks lightest by plain addition is frequently not the one that ends up best.
The defaults show this happening. The 1,440 W east lighting circuit is the largest line-to-neutral load and it sits on A, which is the middle line at 57.0 A. Line C is the lightest at 53.8 A, so the rule says move it there — and doing that takes the imbalance from 13.4 percent to 23.9 percent and the neutral from 6.6 A to 19.3 A. Moving it to B is worse again at 33.4 percent. Leaving it where it is wins, because the welder on A-B is already pulling A up in a direction the watt column never showed. That is why the block computes all three rather than naming a rule.
Power factor is in here and it matters
Each circuit can carry its own power factor, and the two motor circuits at the defaults are set to 0.85. That does two things: it raises the current for the same watts, and it rotates the current phasor behind the voltage, which changes how it combines with everything else on the same line. A panel of purely resistive loads and a panel of motors with the same watt totals do not produce the same line currents, and the difference is not a scale factor.
Where this stops
At the currents. It does not tell you whether a line, a neutral, a busbar or a breaker can carry them, which is an ampacity question with a code table and an inspector attached to it. It assumes the circuits you list are all drawing at once, which no panel ever does — real diversity is what makes an installed panel run cooler than its schedule says. And moving circuits between phases inside a live panel is precisely the work that produces arc flash injuries, which is why no procedure for it appears anywhere on this page.
Questions people ask
How do I calculate amps per phase on a three-phase panel?
Not by adding watts down a column. Line-to-neutral circuits go on their own line at the line-to-neutral voltage; line-to-line circuits put their full current on both of their lines; balanced three-phase loads put equal current on all three. Then the currents on any shared line have to be combined as phasors, because they arrive at different angles rather than in step.
Does a 208 volt load put half its current on each phase?
No. It puts its full current on both of the lines it is connected to, because the same current leaves on one and returns on the other. That is the single most common mistake in hand-balanced panel schedules, and it makes a panel with several line-to-line loads look far better balanced on paper than a clamp meter finds it.
Why is my neutral current lower than the sum of the phase currents?
Because the three line-to-neutral currents are 120 degrees apart and largely cancel. Three equal ones cancel exactly. Two equal ones give a neutral current equal to either of them rather than twice it. Line-to-line and balanced three-phase loads contribute nothing to the neutral at all.
Should I move the biggest circuit to the lightest phase?
Sometimes, and the page tries all three positions rather than assuming it. Line-to-line circuits already on the panel push the line totals in directions plain addition does not show, so the phase that looks lightest in the watt column is often not the one that produces the lowest imbalance once the move is made.
Is a low neutral current on this page safe to rely on?
Only if the loads are linear. This page computes the fundamental, and third harmonic current from switch-mode power supplies, LED drivers and electronic ballasts does not cancel in the neutral — it adds arithmetically across the three phases. On a panel of modern electronics the neutral can carry more than any of the lines while this figure reads near zero.