This sounds to me like an extrapolation of the common “fact” that orchestral peak levels reach 110dB. And yes they do if you stand directly in front of the brass section in a live performance. But no one listens to orchestral music like that. The front row might be 50 feet or more back from that. Volume level ten yards away isn’t the same as volume level an inch from your ear. Recordings are mixed to give a natural perspective from a theoretical best seat in the house. It isn’t mixed to sound like you’re sitting in the musician’s lap.

An orchestra is not a small point source. When you stand "directly in front of the brass section", you might be 5 feet from the closest brass player, but some other brass players are much further away from you. You can't be 5 feet away from all players of the orchestra at the same time. That's why when you integrate the acoustical power at your ears, it gets "only" to 110 dB SPL. When you walk away to the "best seat in the house", your distance to those players who where originally furthest away from you increases much less relatively (relative distance matter in distance attenuation calculations). Your distance to the closest player perhaps 10-folds ( -20 dB), but your distance to the furthers players maybe only doubles ( -6 dB) or even less. That's why the integrated acoustical power at your ears doesn't drop 20 dB. Also, the hall has reverberation. It is far from a free-field situtation. The reverberation (without early reflections and direct sound) is the same level everywhere in the hall and increases the overall sound pressure level significantly.

Reverberation radius rH is the distance where the sound energy densities of the direct sound and the reverberation are equal. It can be calculated using the formula

rH = SQRT ( Q * R / 𝜋 ) / 4,

where SQRT is simply square root, Q is directivity index of the sound source and R is room constant related to total absortion area A:

R = A / (1 - ā),

where ā is the average absorption coefficient.

Let's assume a concert hall of the size 40 m x 30 m x 15 m (133' x 100' x 50' for Americans). Desirable value for reverberation RT for an orchestral music hall of this size is about 1.8 seconds. Reverberation time can be calculated using the simple formula (there are better formulas, but this will do for this example):

RT = 0,161 * V / A,

where V is the volume of the hall. So, from this formula we get that total arborption are A = 0,161 * 40 * 30 * 15 / 1.8 = 1610 m². Now we can calculate the needed average absorption coefficient ā for the hall: ā = A / S = 1610 / (2 * 40 * 30 + 2 * 30 * 15 + 2 * 30 * 15) = 1610 / 4500 = 0.36. Now we can calculate the room constant : R = 4500 / (1 - 0.36) = 7031 and finally the reverberation radius assuming Q = 1.5 (almost omnidirectional sound source): rH = SQRT ( 1.5 * 7031 / 3.14) / 4 = 14.5 meters (almost 50 feet).

This means that 50 feet away from the orchestra we are in the border of near field and free field. I'd say the peaks for the listeners are 90 - 95 dB.

The most dynamic commercial recordings have a dynamic range of about 55dB. That reflects how it sounds in a concert hall. In fact, that is probably more dynamic than most concert halls. 110dB is past the flinch point and pushing the threshold of pain. When was the last time you attended a classical concert or listened to a symphony CD and experienced discomfort from volume spikes? That just doesn’t happen.

The peaks for very loud music are between 80 and 85 dB, not anywhere near 110.

I wouldn't call music with peaks of 80 dB "very loud". Loud maybe. Peaks of 85 dB is perhaps "very loud", but yes, 110 dB is "damage your hearing in minutes" loud.