The good thing about the K system is the built in headroom. If you are working at a K -14 ref level then it means your rms part of the signal can actually go almost 14 db above your nominal signal level. But that also depends on the size of the transient that is attached to the front end of it.
 
But even so a shift of +4 over is not going to harm anything.  It is just your rms component jumped up 4 dB but there is still 10 dB of headroom above that.  Don't let the +3 dB max scale of the VU meter throw you. There is still 11 db of headroom sitting above that mark even. (at K -14 that is, at K -20 there is whopping 17 dB of headroom above that + 3dB mark!) Hey, get to love K -20!
 
I work with a magician here and I do all his music/sound effects. The music all sits at K -14 actually but I have a full 14 db of headroom.  I can put in an explosion that might go 12 dB over the music but it will still do so without clipping into 0 dB FS. Only for a short time but through a big PA it sounds pretty amazing.
 
But your rms levels should really never be peaking +8 or even +4 db over much at any time. With care you can control the dynamics of signals on tracks and busses so they still carry all their dynamic range and transients but mainly hit 0 dB VU and not go over by more than + 1 or 2 dB. That is achieved with careful setting of dynamics processors in these places.

I am trying to calibrate my monitoring environment with an SPL meter tonight but a little confused despite this excellent thread from a few months ago.
 
Despite reading the thread again just now, I am still being a little stupid and would appreciate some input on the following?
 
I introduce a -20dB pink noise signal into an audio track.
The pink noise source is the Sonitus Surround plug in, the help documentation stating that the test noise signal is generated at -20db
The audio track gain is set to unity, gain is 0.0, and it goes direct to the Master bus.
The Master bus is set to unity, gain 0.0
 
However, the pink noise signal on my Sonar track meter (and also my Master bus) reads 6.9, set to RMS and peak.
 
So in order to calibrate correctly, do I bring the fader on the pink noise audio track down to -20 before calibrating each speaker to 83dB, or leave as it is? 

Check out this informative article from  Sound on Sound Magazine from last May:
 
http://www.soundonsound.com/sos/may14/articles/reference-monitoring.htm

Hey, thats a great article, and my specific question was neatly answered in the following paragraph - 
 
'We now need to replay the appropriate pink noise from the DAW to just one of the speakers. However, there is a potential trap here, since most pan controls alter the level slightly at the extremes relative to the centre. This is where that ‑20dBFS sine‑wave tone signal becomes useful because it's the level at the output meter that matters. So play the tone, and offset the channel fader slightly to compensate if necessary for any panning loss or gain. You need to get the output meter for the selected channel to read precisely ‑20dBFS'
 
Thank you for that...