Their effect on a large complex electrical system would be negligible. If you are talking a large commercial or industrial load, lighting is typically a small percentage of the load. I have looked at hundreds of waveforms from large customers and the only harmonics producing loads that can affect their voltage waveforms are large drives, not lighting.
I used to think the same way. My background is primarily in variable speed drive systems for industrial applications. I had pretty much discounted lighting and small power as significant sources of harmonic voltage distortion.
A couple of cases a few years ago made me look at it a bit differently. The first was a pumping station. There were eight VSDs. A couple of the larger units were 12-pulse and the others were 6-pulse. The station supply transformer was rated at 11kV/400V, 1500kVA. The 11kV was the point of common coupling (PCC). Typically for such a project, we had to do before and after harmonic measurements over a seven day period to show compliance with a contractually agreed standard following installation.
In the event, harmonic voltage distortion was outside the agreed limits with no drives running and no worse with all of them running in any combination. I could make no statistical correlation whatsoever between supply voltage distortion (at the PCC) and VSD loading.
My thoughts were that it was contrary to my expectations and initially seemed counter intuitive. Eventually the brain kicked in. The harmonic pollution already on the 11kV supply swamped anything we could do to it.
The pumping station was in a mainly residential area and the combined effect of all the non-linear domestic from all the houses was far greater than our eight VSDs because of the sheer number of them.
In UK and, I suppose, other countries too, power usage is very roughly one third industrial, one third commercial, and one third residential. Commercial premises dirty up the supply too. Like residential, no single item makes a huge impact, but a huge number of single items does.
And the supply system has to accommodate that.
The second case that made me review my thoughts about lighting and small power was a problem in a large prestigious hotel in the Middle East. It was the overheating neutral thing.
Total supply capacity of the transformers supplying the building was 15 MVA.
Electronically controlled lighting was about 3MVA. Although distributed around the phases, it was all single phase and rich in third harmonic. Bad for the neutral conductor.
Sure, it is just one very exotic building. But put together a number of residences fed from the same LV supply and you have the same issues albeit on a different scale. Add a number of these connected back at the HV and you have a problem of the same magnitude.
Low power non-linear loads are a growing problem.
If electric vehicles gain significant market share it could get a whole lot worse.
