Heat Treating: Every Part in the Load Sees a Different Cycle
The recipe says the load goes to temperature, soaks, and quenches. The parts don’t all agree. In a dense load, the parts on the outside of the rack heat first and cool first. The parts buried in the middle of a basket, or on the bottom tier, lag behind on the way up and hang on to their heat on the way down. Every part in the load runs its own version of the cycle, and the hardness, case depth and distortion you measure afterward are the record of which version it got.
Experience covers most of that. Good operators know which positions run cold, which parts need extra soak, and which fixtures cause trouble, and that knowledge is hard won. What experience can’t easily do is show you the inside of a load you have never run before. That is why a new part can take a few tries to clear first article, and why soak times tend to carry a little extra margin.
What the thermocouples don’t see
Temperature gets measured in a handful of places. Load thermocouples are attached to a few parts, or test pieces, in a production load. A temperature uniformity survey (TUS) under AMS2750 pyrometry places thermocouples at fixed points across the work zone, often with the furnace empty, and checks that the zone holds tolerance at temperature. Both are essential, and neither tells you how every part in a dense, real load heats, soaks and quenches. That is where the variation lives.
A thermal and flow simulation of the furnace fills in the rest. It models the chamber, the heating elements, the quench gas, the racks, boats and fixtures, and the parts together, and follows the whole cycle over time.
Surface temperature across a four-shelf gear load through a gas quench cycle. The lower tiers hold heat after the upper ones have let go.
In this gear load, the top shelves cool quickly while the bottom tiers stay hot well into the quench. On paper, every gear went through the same cycle. In the furnace, they went through very different ones, and that is the kind of difference that shows up later as scatter in hardness and distortion. The flow field shows why.
Quench gas velocity on a cut plane through the loaded chamber. Red is gas moving at 20 m/s or more; dark blue is nearly still.
Quench gas comes in fast through the nozzles, but vertical rack posts block some of the jets, and parts of the load sit in nearly still gas. Those dead zones are where the slow-cooling parts are.
What a simple fix is worth
Seeing the problem points to the fix. In this case it was a modest change to the rack: the same load, the same temperatures and the same quench flows, with the obstructions moved out of the way of the gas.
Temperature spread across the load through the quench, original rack against the modified rack.
The peak temperature spread across the load fell from about 378 °F to about 310 °F (210 °C to 172 °C), and the load settled out sooner: more consistent parts from the same furnace, with no new equipment. The furnace and load here are representative, because client studies stay confidential, but the pattern is one I see often: the biggest gains come from the fixturing and load arrangement, not from the furnace itself.
Why most shops don’t simulate
If this is so useful, why isn’t every heat treater doing it? Because the economics don’t work for a job shop. Doing it in-house means a commercial simulation license, which typically runs tens of thousands of dollars a year, plus a simulation engineer who knows how to drive it and keep the models honest. That is a full-time salary aimed at questions that come up a few times a year. Against job-shop margins, and lots that change every week, it doesn’t pay back. The tools were built for equipment makers and large manufacturers, and they are priced that way.
The industry reflects that. I keep a running survey of heat treaters across the US and abroad, based on what each shop shows publicly. Of 217 reviewed, 154 (71%) show no sign of in-house simulation, and only 13 clearly have it.
Most are small: of the companies with a known headcount, 106 of 137 have 50 people or fewer. The median founding year is 1976. That says nothing about skill. It is a matter of what makes sense to own. A question that comes up a few times a year is better answered by buying a study when you need one than by staffing for it.
What’s coming from aerospace and defense
The question may start coming from customers, too. In 2025 the Navy put out a small business research topic, N251-D02, asking for modeling and simulation tools that predict the microstructure and properties of aerospace gear and bearing steels from their melt and heat treatment processing. The topic is candid about where things stand: modeling tools exist, but “there has not been a dedicated effort to apply these to the subject steel alloys,” and qualifying by empirical testing “consumes large amounts of resources and time.”
Metallurgical models like these are complex, and every one of them starts from the same input: the time-temperature history of each part. A microstructure prediction built on the furnace setpoint, instead of on what each part actually saw, inherits every cold corner and slow quench in the load as error. Getting that history right is the first step toward controlling variation in microstructure, and for most shops it is the biggest return: much of the scatter in hardness and case depth traces back to it before any metallurgy is modeled. If prediction like this moves into how critical parts get qualified, the shops that already know what their loads actually see will be well placed to support it.
See inside your furnace, ship more consistent parts
Your customers judge you on consistency: every part in every load hitting the same hardness, case depth and dimensions, lot after lot. That is what a study helps you deliver, by showing which parts see a different cycle and what change to the load, rack or fixturing brings them back in line.
You don’t need CAD drawings or a simulation department to get there. Send us photos of the load and fixturing, a few key measurements, the make and model of the furnace, your cycle data and the problem in your own words. We build the CAD and the simulation from that, drawing on decades of designing, building and running high-temperature furnaces and the materials that go inside them, so the model behaves like your equipment and comes together quickly. You get a fixed scope, a fixed price, and a written report in about two weeks. You know your furnace; a study just lets you see inside it. Here is how one works.
Mathew J. Leis, P.E., founded Reactech Engineering Group. He has spent fifteen years designing, building and operating furnaces above 2,000 °C, and developing the CFD and FEA models that predict how they behave. He works alongside a principal engineer with over thirty years in high-temperature equipment: furnace design at an OEM, prototype furnace development, and hot-zone materials engineering for vacuum heat treating.