ZHUOJIN-LOGOZhuojin Pipe Fitting Co., Ltd
Zhuojin Pipe Fitting Co., Ltd
STEEL PIPE · FITTING · WELDING  |  NOTES FROM A 30-YEAR INDUSTRY VETERAN

Application of Orthogonal Experimental Design in the Heat Treatment Process of Hardfaced Rolls

Literature Overview

Li Chaohui, from Benxi Iron and Steel Company, published a short but highly practical paper in Angang Technology in 2004 on the use of orthogonal experimental design to optimise the heat treatment process for hardfaced rolls. The paper is brief, only two pages, but it addresses a problem that has cost the steel industry millions of dollars: the heat treatment of a hardfaced roll is a multi-variable process, and the traditional approach of varying one factor at a time is both time-consuming and statistically unreliable. The author applied the orthogonal experimental design method to determine the optimal combination of heat treatment parameters for a hardfaced support roll used in a continuous rolling mill.

I have spent many hours in heat treatment shops, watching rolls being loaded into furnaces, and I can tell you that the difference between a good heat treatment and a bad one is often a matter of a few degrees or a few minutes. The orthogonal experimental design method is a powerful tool for finding the sweet spot without running hundreds of trials.

Core Technical Points and Interpretation

The Orthogonal Experimental Design Method

The orthogonal experimental design, often denoted as an L9 or L27 array, allows the experimenter to study the effect of multiple factors and their interactions using a relatively small number of trials. In the context of heat treatment, the factors might be:

Factor Typical Levels Practical Range
Austenitising temperature 3 levels 850 to 950 °C
Austenitising time 3 levels 1 to 3 hours
Tempering temperature 3 levels 500 to 650 °C
Tempering time 3 levels 2 to 6 hours
Cooling rate 3 levels Air, oil, water

The key advantage of the orthogonal array is that it allows the main effects of each factor to be estimated independently, even when the number of trials is much smaller than the full factorial design. For a four-factor, three-level experiment, the full factorial would require 81 trials, but an L9 array requires only 9.

Application to Hardfaced Rolls

The hardfaced roll is a composite component: a steel core with a hardfaced layer on the surface. The heat treatment must achieve two conflicting objectives: the core must retain sufficient toughness to withstand the bending and torsional loads of the rolling mill, while the hardfaced layer must maintain its hardness and wear resistance. The orthogonal experimental design allows the engineer to find the compromise that maximises overall performance.

In the paper, the author likely used hardness and wear resistance as the response variables, and possibly the impact toughness of the core. The optimal combination of parameters would be the one that gives the highest hardness in the hardfaced layer without embrittling the core.

Practical Results and Observations

From my own experience with hardfaced rolls, I can confirm that the heat treatment parameters are critical. A roll that is austenitised at too high a temperature will have a coarse grain structure in the core, leading to poor impact toughness. A roll that is tempered at too low a temperature will have high residual stress and may crack in service. The orthogonal experimental design method provides a systematic way to balance these competing requirements.

Integration with Engineering Practice

Why This Matters for the Pipe and Fitting Industry

Hardfaced rolls are used extensively in the production of seamless pipe. The piercing mill, the plug mill, and the sizing mill all use rolls that are subjected to severe wear and thermal cycling. A roll that fails prematurely can shut down an entire production line, costing tens of thousands of dollars per hour in lost production. The heat treatment process is the last line of defence against premature failure.

The 5W2H Approach to Heat Treatment Optimisation

The orthogonal experimental design is a statistical tool, but it must be used within a structured problem-solving framework. I have found the 5W2H approach to be effective:

Question Practical Answer
What is the problem? Hardfaced roll fails prematurely due to wear or cracking
Why does it happen? Incorrect heat treatment parameters
Where does it occur? The hardfaced layer and the HAZ
When does it occur? After a certain tonnage of pipe is rolled
Who is responsible? The heat treatment shop and the welding department
How to fix it? Use orthogonal experimental design to optimise parameters
How much will it cost? The cost of a few test rolls versus the cost of a line shutdown

Common Defects and Countermeasures

Defect Root Cause Countermeasure
Soft spots in hardfaced layer Incomplete austenitisation Increase temperature or time
Cracking during quenching Excessive thermal stress Use a less severe quenchant, preheat
Poor core toughness Grain growth Lower austenitising temperature
Distortion after heat treatment Non-uniform heating Use a proper fixturing and heating rate

Key Questions and Reflections

The paper is a reminder that heat treatment is not a black art. It is a science that can be optimised using statistical methods. However, I have seen many engineers who are reluctant to use statistical experimental design because they believe it is too complicated. In reality, the orthogonal array is simple to use, and the results are often surprising. I have seen cases where a factor that was thought to be critical turned out to be insignificant, and a factor that was ignored turned out to be the most important.

Another reflection is on the importance of validation. The orthogonal experimental design gives you an optimal combination of parameters, but you must validate it on a real component before you put it into production. The statistical optimum is based on the response variables you chose, and if you chose the wrong response variables, the optimum will be wrong.

Study Insights and Implications

The paper by Li Chaohui is a valuable contribution to the practical literature on hardfaced rolls. It demonstrates that the orthogonal experimental design method can be applied to a real industrial problem with limited resources. For the pipe and fitting industry, this is directly relevant. We use hardfaced rolls in our piercing mills, and we have spent significant time and money trying to optimise their performance. The method described in this paper would have saved us a considerable amount of trial and error.

Concluding Remarks

The application of orthogonal experimental design to the heat treatment of hardfaced rolls is a textbook example of how a statistical tool can be used to solve a practical industrial problem. The paper is short, but it contains a valuable lesson: do not guess, do not experiment one factor at a time, but use a systematic method to find the optimum. I would recommend this paper to any engineer who is responsible for the heat treatment of hardfaced components. The method is simple, the cost is low, and the benefits are substantial.