Usually, at least in traditional things like a triac firing inverter/silicon controlled rectifier,  the higher order terms are progressively weaker, and beyond some point, the modest error from neglecting them is acceptable (and avoids trying to sum all frequencies from DC to daylight, when there really is no useful energy above a few hundred Hz present.)

Until quite recently no-one bothered that much with anything about this - H50 is 2.5KHz in the UK, and a single transient with  a risetime of that rate (looking like a half cycle of that frequency) or indeed any other fast-ish risetime profile, that only occurs  once or twice per 50Hz cycle period does not amount to  a row pf beans energy wise and is not normally a big issue,

Problems arise with modern electronic VSDs and so forth where there is non-stop full amplitude switching at some much higher frequency- perhaps creating 100s of volts peak to peak  of a near-square wave at tens KHz.

EMC standards have rather had to play catch up with this in recent years, and the filtering and testing requirements have become significantly harder.
However, in your case the equipment has filtering, and you can probably look at the figures you have already got up to H25  to discern a trend - assuming this is downward with rising frequency, then I'd be very surprised if the lines you have  got no results for were any higher than the average of the highest few harmonics you have data for, would it throw things far out of kilter if you assumed the missing data was a flat line? If this predicts a transformer similar to the one being replaced, you are on safe ground.

Note that the idea of 'K being a fixed limit is also an approximation - you could achieve the same total 'K 'in different ways, with different mixtures of harmonics, and any given actual transformer heating will not be the same, as the mix of winding vs core vs copper eddy losses changes.The standardised test waveforms are only part of the battle

There are special 'high K' tolerant transformers, that use multi-core windings and strategically placed magnetic shunts to try to reduce the losses, but it is far from an exact science.

Mike

Usually, at least in traditional things like a triac firing inverter/silicon controlled rectifier,  the higher order terms are progressively weaker, and beyond some point, the modest error from neglecting them is acceptable (and avoids trying to sum all frequencies from DC to daylight, when there really is no useful energy above a few hundred Hz present.)

Until quite recently no-one bothered that much with anything about this - H50 is 2.5KHz in the UK, and a single transient with  a risetime of that rate (looking like a half cycle of that frequency) or indeed any other fast-ish risetime profile, that only occurs  once or twice per 50Hz cycle period does not amount to  a row pf beans energy wise and is not normally a big issue,

Problems arise with modern electronic VSDs and so forth where there is non-stop full amplitude switching at some much higher frequency- perhaps creating 100s of volts peak to peak  of a near-square wave at tens KHz.

EMC standards have rather had to play catch up with this in recent years, and the filtering and testing requirements have become significantly harder.
However, in your case the equipment has filtering, and you can probably look at the figures you have already got up to H25  to discern a trend - assuming this is downward with rising frequency, then I'd be very surprised if the lines you have  got no results for were any higher than the average of the highest few harmonics you have data for, would it throw things far out of kilter if you assumed the missing data was a flat line? If this predicts a transformer similar to the one being replaced, you are on safe ground.

Note that the idea of 'K being a fixed limit is also an approximation - you could achieve the same total 'K 'in different ways, with different mixtures of harmonics, and any given actual transformer heating will not be the same, as the mix of winding vs core vs copper eddy losses changes.The standardised test waveforms are only part of the battle

There are special 'high K' tolerant transformers, that use multi-core windings and strategically placed magnetic shunts to try to reduce the losses, but it is far from an exact science.

Mike