# Why don't we use the third harmonic in power transmission?

For some high power high voltage lines DC is used with the advantage that sinus wave AC obviously doesn't use the line to it's maximum capacity at all times. To correct this, we could use square wave instead of sinus wave, but the multitude of high frequency harmonics contributes little to the overall power transmitted and brings a bunch of problems. However we could compromise on not using all harmonics, but only the third i.e. instead of V1(t) = Vmax * sin(2*PI*f*t) we'll be using V3(t) = Vmax * ( sin(2*PI*f*t) + 1/3 * sin(6*PI*f*t) ). This theoretically should provide a gain of a little more than 10%.

On the power station side you can use a generator built for producing V3. For solar power generators it is a trivial task to reprogram them for V3. Now I'm aware that the infrastructure is optimized for f not 3*f, e.g. transformers might block the third harmonic but this might actually be an advantage, i.e. if you want to increase the capacity of the wide area synchronous grid, you only need to install transformers for 3*f in parallel on the receiving side so you can access the second harmonic and the rest i.e. the middle and low voltage grid runs on f as usual. The transformers double as a frequency separating filter. You can use your 10% gain coming out of the 3*f optimized transformer for internal power needs, sell it to industrial customers who can handle it or convert it to f and reinsert it into the middle and low voltage grid.

So ... why not?

• One "minor" drawback is that it would require an overhaul of the entire electrical grid, as well as many industrial machines as well as many consumer products. Commented Aug 5, 2023 at 17:27
• Why does "AC not use the line to its capacity"? The main issues I think are capacitive leakage and skin effect, both of which get worse with higher frequency, so transferring part of the total power at a higher frequency will make the efficiency worse, no? Commented Aug 5, 2023 at 18:25
• If your end result (of using the 3rd harmonic in addition to the fundamental) is a10% power transmission gain, then surely you could achieve pretty much the same effect by just raising the voltage by 10% (assuming the insulators can withstand the extra)? Your plan would depend on the conductors being rated to carry a 10% higher RMS current, which is likely to eat into the safety factor they were spec'ed at when installed. Commented Aug 5, 2023 at 19:04
• @Math Keeps Me Busy If the transformers themselves don't do a sufficient job of separation, a frequency filter at the substation should suffice. You can convert line by line, no need to do everything at once. Also 180Hz is still rather low, no spikes or surges.
– Čäřý
Commented Aug 5, 2023 at 20:36
• @tobalt If I do understand the article about electric power transmission on Wikipedia correctly the main loss is joulean, i.e. unwanted heating. Also the proposal is transporting additional power. Even if this additional power has a higher loss due to the capacity of the lines, the energy transported by the main frequency should be unaffected.
– Čäřý
Commented Aug 5, 2023 at 20:42

I recommend you to take a look at IEEE Std 738, which provides guidance in the calculation of bare overhead conductor temperature. The basic idea for calculating the steady-state temperature is to solve a heat balance equation such that the heat lost by convection and radiation is equal to the heat gained due to the sun and I^2*R losses for a given current and set of weather conditions and conductor parameters. The AC current frequency is related only insofar as it affects the effective resistance of the conductor. The time frame over which this steady-state heat balance is applicable is much greater than one power system cycle, so the waveform of the current is irrelevant. AC fundamental, AC with harmonics, square wave, DC... doesn't matter, all you need is the RMS current. (Except that frequency affects resistance, as already mentioned.)

Many different factors can limit the power capacity rating of transmission lines, and although in most cases the end result is a fixed rating in Amps, the assumptions used to arrive at this rating are generally worst-case such that they are not actually true at any given time. However, without the operational infrastructure to estimate a more accurate line rating applicable to some interval of time, this is how it has to be done to ensure safe operation of the transmission network.

Dynamic line ratings are a way to address this shortcoming. For one reference, see this report by the United States Department of Energy to Congress: https://www.energy.gov/sites/prod/files/2019/08/f66/Congressional_DLR_Report_June2019_final_508_0.pdf

I would also suggest that you read up on the reasons that High Voltage DC (HVDC) transmission lines are used. There are a few reasons, but utilization of the maximum conductor capacity is not one of them. As a starting point, you could read the relevant Wikipedia page: https://en.wikipedia.org/wiki/High-voltage_direct_current

• If conductor temperature is the limit, then of course no fiddling with the wave form can help this. I fully concur in this. However it sound strange to say that wave form is irrelevant, only RMS counts, because different wave forms need not but can have different RMS and especially my proposal has a higher one. It is like saying the size of the car, weight, medium speed you drive etc. are irrelevant for mileage, only gasoline intake is relevant.
– Čäřý
Commented Aug 6, 2023 at 21:12

A trivial reason why not, is because we use a three-phase AC system.

If each conductor carries a 3rd harmonic, and each current is 120° phase shifted from its companion, then regardless of the relative phase of fundamental and harmonic, the phase of the 3rd harmonic is (120°) × 3 = 360° = 0°. That is, if we insist that each wire has the same waveform (a symmetry condition), then all harmonic currents will necessarily be in phase!

So you're just shifting the harmonics to the common mode (between all lines collectively, and neutral or earth). Which, is a way to transmit power -- but it's basically a single-phase system again, and, it introduces complications when we actually want to wire this to a transformer, and add a ground point for safety and lightning protection: now we have either CM voltage on the neutral, or CM current through it.

If we propose a single-phase harmonic system instead, we still have the problem that fundamental and harmonic(s) must be treated separately, because they will undergo different phase shifts and impedances along the line, necessitating many more line reactors to maintain waveform (and now we need to maintain not just amplitudes, but the phase relation between fundamental and harmonic(s), to keep the peak voltage consistent as well). And we lose the handy feature that 3-phase power is always available (it never drops to zero, power is available continuously), and has a rotating direction (making motors much easier to use, and more efficient).

And the whole thing is even further complicated by loads that either abuse the harmonic content (nonlinear loads like rectifiers), or literally don't know what to do with it (the harmonic energy just causes useless heating in motors, and maybe torque pulsation as well).

Transformers also dissipate more power at higher frequencies; there is some pressure already to reduce harmonic content for this reason. Mind, it's a small effect -- fractional percent efficiency -- but we're talking massive grid-scale effect here, and that means saving billions of dollars over the installed lifetime of the grid and components (a transformer might last half a century or more in continuous operation!).

We could also propose DC for distribution (not just HVDC links), but besides being incompatible with literally everything it touches (how do you make such a change without starting over completely from scratch?), this brings further complications of safety, from the static fields that birds landing on wires would experience, to the incredible challenge of breaking high voltage DC arcs (for purposes of basic line switching, on up to safety in case of line breakdown). Not to mention annoyances like electrolytic corrosion (when a conductive path through water or salt film is possible). AC has the advantage that, arc power dips to zero twice per cycle, making it much more likely to self-extinguish, and much easier for a fuse or switch to clear the arc. It's also electrolytically balanced: apparently underground substations (vaults) can become flooded as a normal operating experience, and continue to operate, just with a lot more power dissipation due to the water only moderately shorting things out.

Still, with all these issues, we do have the technology to adopt a DC distribution system; it's just that it's immensely more expensive to do so. Switching will vary between specialized mechanical types, and semiconductors (high voltage SiC MOSFETs and IGBTs are available); arc fault detectors could be deployed; transformation can be done with semiconductors and power converters; fault currents and surge voltages can be handled by just using a hell of a lot of semiconductors (cascode stacks in parallel); and better insulation can be employed on wires and connectors, and sacrificial or noble electrodes can be placed near connections where corrosion may be an issue (like the ringed insulators supporting wires on poles). These are all known, understood and available materials and techniques -- it'll just vastly increase the cost of the system, for nearly no improvement in efficiency, and a decrease in usability.

In summary: whether we go up or down in frequency, or use a mixture thereof, there are many more issues that pop up, ranging from theoretical necessity to economic convenience. The 50/60Hz range seems to be good enough; there would be merit in considering a relatively small change, perhaps 40Hz to reduce transformer losses, or 80 or 100Hz to reduce transformer size (give or take the loss ratios of newer materials available for transformer construction), or perhaps certain special ratios in that range that would be mechanically convenient (maybe 83.3.. Hz instead of 80).

In any case, making any kind of sweeping change at all, requires updating the whole system, so these are largely historical-hypothetical questions: what if the system had evolved towards a different base frequency, back when change was still feasible?

• A good place to look for answers is Japan. They stick with 50/60 Hz in half the country each with lackluster means of distributing power between both zones. Because even the change from 50 to 60 Hz would be harder. Commented Aug 6, 2023 at 16:08
• My understand of "no current through neural" was always that this is just what we hope for i.e. we connect this sockets to this phase and that to that and hope, that it mostly evens out. Anyway there are still three phases so there still is the possibility of keeping the neutral free of current.
– Čäřý
Commented Aug 8, 2023 at 19:57
• Rectifiers of the old diode-capacitor kind should even benefit from a waveform closer to square, but neither am I in favor of using them nor do I intend to necessarily distribute to the last socket. Concerning your note about transformers, are we talking about higher dissipation of transformers built for f at the 3rd harmonic or about transformers built for the 3rd harmonic? Shouldn't a transformer for higher frequency be cheaper to build?
– Čäřý
Commented Aug 8, 2023 at 20:14
• @Čäřý At a consumer, neutral current might not cancel out, but after the first distribution transformer, it's isolated, ready for a new ground reference; there can be line imbalance, but this doesn't necessarily imply neutral currents, IIRC. Transformer core loss can often be expressed as $p_c = c_0 f^\alpha B^\beta$, with α near 1 and β near 2. When equal to 1 and 2, we have plain old eddy current losses, and losses decrease with frequency (for constant V). But real materials often have these parameters a little bit larger, and it's still added loss regardless. Commented Aug 9, 2023 at 6:21