Comparison of Through-Induction Hardening with Conventional Hardening

Two methods are often used to through harden and low alloy steels.  The well recognized conventional hardening method involves heating to an appropriate austenitizing temperature for a specific length of time in a furnace with or without a protective atmosphere. This is followed by quenching in oil or water, and then tempering at a specific sub-critical temperature in order to achieve desired mechanical properties and hardness.  A second method involves rapid heating to the austenitizing temperature in an induction coil for a length of time sufficient to obtain a uniform temperature followed by rapid cooling.  The mechanical properties and hardness are usually dictated by the cooling conditions following austenitizing, since sub-critical tempering is not generally used.  While comparable mechanical properties and hardness can often be achieved with both methods, microstructures can vary significantly.

Of interest is a comparison of the fatigue properties that can be obtained with either method.  The AISI Bar Steel Fatigue Database contains data for SAE 4140 low alloy steel which was through hardened using both conventional and through-induction hardening.  While the data was developed for bar stock as opposed to fabricated parts, it does provide a means of comparing the fatigue properties for the two heat treating methods.

The table below summarizes the mechanical properties and hardness values obtained for the two processes.

post 16 table

Somewhat higher values of yield strength and tensile strength were achieved with the induction hardened process.  The microstructure obtained after conventional quenching and tempering was tempered martensite, whereas the microstructure observed in the induction hardened condition contained a large amount of bainite and a lesser amount of ferrite.

Figure 1 shows the strain-life fatigue curves obtained for both heat treating conditions.  The strain-live curve for conventional quenching and tempering is given by Iteration No. 68, and the strain-life curve for induction hardening is given by Iteration No. 93.


No. 16 Figure 1
Figure 1

It can be seen that comparable fatigue properties are obtained for both hardening processes.  The fitted strain-life curves show a slight advantage for through-induction hardening at fatigue lives from 103-105 reversals and somewhat better performance for conventional hardening at lives greater than 106 reversals.

The decision as to which process to use can be based on manufacturing considerations and other property concerns, e.g. notch toughness, rather than on fatigue requirements.  It should be noted that, while parts of almost any configuration can be hardened using the conventional process, the through-induction hardening process is constrained by part geometry.  Induction coils cannot be designed for hardening many complex parts.

This entry was posted in News and tagged , , , , , , , . Bookmark the permalink.

6 Responses to Comparison of Through-Induction Hardening with Conventional Hardening

  1. Mary Starkey says:

    I am a little concerned at how the lines fit the long life points. Are you following the practice of ASTM E739 and treating stress / plastic strain as the independent variable and reversals as the dependent variable for the linear regression fits? You can get quite different results if life is the independent variable in the regression.

  2. Mary Starkey says:

    There are some very interesting results being discussed on this blog so it is very disappointing to find the the graphs are such poor quality one cannot interpret the results. It is impossible to distinguish between the two sets of data. Please can they be better resolution and bigger, showing the colour of the line in the key would also help.
    I notice that the blog is not showing the cyclic stress strain curves. This is also relevant as the same applied load might produce a higher strain in one material.

  3. Mary Starkey says:

    Yesterday I queried the curve fits on iteration 68. I found the tabulated data on your website and ran through an analysis. I can confirm
    1. Manson Coffin fit uses life as a dependent variable and is fitted to given plastic strains = loop width from checks on values. Apparent modulus changes from specimen to specimen
    2. Cyclic fits Ramsberg Osgood to total strain less elastic assuming a constant modulus of 199000MPa (NOT to loop width)
    3. I cannot reproduce Basquin line anyway I try. I note the published constants give 1e6 cycle value as 611MPa whereas the value in the tabulated material constants is 679MPa so big discrepancy.

    • Your comments are being referred to Professor Tim Topper at the University of Waterloo. He conducted the testing and is the person who can best respond.

      Professor Tim Topper’s response: Your first two observations correctly describe our data fitting procedures. The fitting of Basquin’s relationship to steel data merits some explanation. Steels tend to show a fatigue limit attributed to dislocation locking by carbon and nitrogen atoms leading to the truncation of the elastic strain life Basquin equation at the point where the strain life curve goes flat. For mild structural steels this is usually about one million cycles. For harder steels the life at which the truncation occurs is less. The strain life data for the present steel suggests a fatigue limit starting from about one hundred thousand cycles as shown on the graph. If this were the case and extension of the Basquin relationship would fall below the data beyond this fatigue life. When we look at the elastic strain life data for this material we see that because of the sudden and severe cyclic softening the stress range (and the elastic strain range) remain almost constant over a range of total strain amplitudes corresponding to fatigue lives of about five thousand to one hundred thousand cycles making any fitting of the Basquin relationship questionable. However because it is used to derive the strain life relationship used in fatigue prediction constants are included to allow the calculation to be made. As shown in the figure calculations using the equation should not be made in the life region beyond one hundred thousand cycles where the strain-life curve reaches the fatigue limit and becomes flat.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s