“By controlling the important variables, dimensional changes in heat treatment can be controlled.” Patrick McKenna
Variability is the enemy in our precision machining shops, and reducing variability is a key to sustaining our businesses and improving our capabilities.
When I talk about statistical process control with someone,  I listen closely to see if they are focused on the average (where the process is performing) or the standard deviation ( how the process is performing.)
If they are fixated on the average, I know I need to look at the data myself.  On the other hand if they are talking about standard deviations, I generally take their word on the data…
In the latest  issue of Production Machining Magazine, PMPA Technical Member Patrick McKenna from Nevada Heat Treating Inc., and Daniel Herring, the Herring Group Inc.  teach a nice class on how to reduce process variation in heat treat to minimize the post heat treat variability that all of us face.

Good advice here...

This is important if we are not to waste our production time trying to remove excess material because we left too much stock  for cleanup, or worse, finding the parts have shrunk in some critical dimension, rendering all of the parts ‘scrap.’
This article lists 9 variables NOT in control of the heat treater, and 14 that are under their control (furnace temperature uniformity, load configuration) or shared by the customer ( process selected, batch size, part size).
Not every order we produce is part of a long running job where we can control every input variable, but this piece does a great job of providing sensemaking on what can be a complicated and confusing subject.
I predict that you’ll keep this article in your “great to know” file.

Hardness of Quench and Tempered alloy steels is a function of the tempering temperature. The higher the tempering temperature, the lower the hardness.
This is called an inverse relationship.
And it’s why some people call tempering “drawing.”
The temper “draws the hardness out of the steel.”

Normalized at 1600F, Quenched in oil 1550 F, Tempered 2 hours

 These curves are a rough approximation of the as tempered brinell hardness for the grades shown. For example, I have other data for 4340 that shows 440 BHN at 800F; 410 at 900F; 380 at 1000F; 340 at 1100F, and 310 at 1200F temper temperature.
Fire! Can't do a blog on heat treat without a picture of fire.

Your mileage may vary, in other words, but this graph is close enough for ‘considered judgement.’
Additional 4140 data that I have from my notes suggests 397 BHN at 800F; 367 at 900F; 335 at 1000F; 305 at 1100F and 256 at 1200F.
If you have better data from your process – USE IT.
Better yet, if you have time, send a sample to your heat treater for a pilot study.
In the absence of data from your process, the above figure and data will give you “a place to stand” in understanding what is possible when heat treating .40 carbon alloy steels- the steels most commonly encountered in our precision machining shops for Automotive, Aerospace, Agricultural and general applications.
Here is a video from PMPA member company Nevada Heat Treating to give you an inside look at what goes on at a heat treat service provider.