Gear life depends on many factors, and steel cleanliness is a critical one. This is because hard oxide inclusions act as stress concentrations for the possible initiation of gear fatigue failure. By using modern scanning electron microscopy (SEM), we can characterize steel inclusions to perform design-relevant predictions of the potential for a gear fatigue failure. We can also certify premium clean steels with these techniques. Improvement using this approach might include selecting a leaner alloy, mitigating warranty costs, and lightweighting or achieving increased fatigue performance.
Highly loaded gears are typically produced from steels that meet stringent metallurgical specifications for nonmetallic inclusion content. These limits allow gear designers and producers to specify steel that will meet the minimum expectations for fatigue performance. Unfortunately, these specified inclusion limits do not provide the data needed by designers to increase performance to meet ever-increasing demands in high power-density gearing applications.
Modern air-melted and vacuum-refined steelmaking processes, when optimized, can produce steel cleanness that is on par with vacuum arc remelted (VAR) steels at a three- to five-fold reduction in cost. Similarly, modern SEM can be employed to characterize the nonmetallic inclusion population. This can be done in a way that allows gear designers to consider the inclusions when assessing the risk of failure.
This project was initiated when a customer asked TimkenSteel to work with them and their forging supplier to find ways to improve the fatigue performance of their mechanical power transmission system. Consumers were demanding increased power capabilities up to at least 400 horsepower, which is equivalent to the horsepower of a Porsche 911, Mustang or Camaro. Performance constraints required that the gear system remained small enough to fit in a space the size of a fist.
TimkenSteel is one of the few approved suppliers of forging bar for this application. We send our SBQ steel to the customers’ forging supplier, which in turn produces gear blanks for the customer. We sat down with the teams from both companies in order to target improved gear performance.
Gears can fail for a variety of reasons, and any number of processing steps during their manufacture could contribute to the failures. We started by assessing the current manufacturing processes – from bar making, forging, cutting gear teeth, heat treatment and finish-machining and grinding. Engineering optimization efforts had been completed in the forging, machining, carburizing heat treatment and finishing of the gears, resulting in significant improvement in gear performance. Problems, however, still existed.
Anatomy of a Gear Fatigue Failure
The TimkenSteel team decided to use its premium air-melted and vacuum-refined steels to mitigate fatigue risks.
Figure 1 illustrates the progression of a gear-root bending-fatigue failure. The photo on the left shows a gear tooth that has fractured off the gear. The series of pictures to the right shows a progression of higher-magnification images of the fatigue region and initiation site at an oxide inclusion.
Linear elastic fracture mechanics (LEFM) is a science developed by the mechanics and material-science communities to understand fatigue and fracture behavior, and it can be used to understand critically sized inclusions. Through LEFM, the stress intensity at an inclusion can be calculated as a function of the principal stress (σ) and the inclusion’s projected square root area (√a) as measured perpendicular to the maximum principal stress as follows:
This equation shows that as the stress is increased, the stress intensity (KI) at the crack tip is also increased. Similarly, as the inclusion size is increased, the stress intensity increases.
Figure 2 illustrates the rate of crack growth (da/dn) as a function of crack-tip stress intensity applied in the laboratory. The inset images to the right illustrate the testing techniques used to measure crack growth rate. At low stress, the threshold stress intensity (ΔKth) is not exceeded, and no crack growth is observed. At higher stresses, cracks grow in a stable log-log linear manner until the critical stress intensity (KIc) is reached. At this point the specimen exhibits rapid crack growth. Inset on the graph are images illustrating the fatigue progression for the gear in Figure 1.
The initial stress conditions and inclusion resulted in a stress intensity that exceeded the threshold, allowing a crack to initiate and grow into the stable crack growth region. The crack propagated by fatigue until it was large enough to exceed the critical stress intensity, causing the crack to propagate rapidly and break off the tooth.
If the threshold stress intensity (ΔKth) is known, then our first equation can be rearranged to determine how large of an inclusion can be tolerated as a function of applied stresses.
Figure 3 shows the fatigue strength as a function of inclusion size using a threshold stress intensity of 4 MPa m1/2. For these heavily loaded gears, flank and root bending stresses of 1,000-1,200 MPa may be expected, yielding a critical inclusion size range of 10-20 µm. To put that in perspective, inclusions one-fourth the diameter of a human hair could be large enough to cause an inclusion-initiated fatigue in this critical application. This inclusion size range or larger commonly initiates failures of heavily loaded components. This supports the idea of using steels with a minimum number of inclusions in the critical flaw size range or larger. TimkenSteel’s Virtual Component Fatigue Model (VCFM) was also used to illustrate and confirm the impact of steel cleanness on the performance of the gears using a range of steel cleanness levels.
After reviewing this analysis, all parties agreed that optimizing steel cleanness was probably the best way to improve gear performance. Steel samples from various suppliers (including TimkenSteel) were procured for SEM steel cleanness evaluation. Each sample was produced by domestic SBQ steel producers to meet stringent bearing-industry inclusion-content specifications.
Figure 4 compares the cleanness results of TimkenSteel and other SBQ domestic competitors. The first graph shows the number of inclusions 10 µm and larger per 100 cubic centimeters, and the second shows the number of inclusions 20 µm and larger per 100 cubic centimeters.
Armed with this information, the evaluation team decided to forge some gears of our certified premium air-melted steel (CPA steel), which is produced to exacting cleanness requirements and evaluated and certified using automated SEM image analysis. Gears from CPA steel were made for testing on the customer’s test system, which simulates the harshest of service conditions. The customer tested 10 gear sets to 100 hours under accelerated test conditions, with intermittent disassembly and inspection. Based on historical test results, the customer anticipated multiple failures under these harsh conditions. As it turned out, all 10 gear sets passed with no failures or signs of fatigue.
Figure 5 illustrates one case of a loaded spur gear where TimkenSteel CPA product is compared with oxide cleanness data obtained on two other SBQ producers. The analysis results indicate that substituting the clean CPA steel into this gear application could provide for a potential downsizing opportunity of between 12-30% (by weight) over the other two steels evaluated. An alternate VCFM analysis shows that substituting CPA steel can provide an increase in torque capacity of 10-35% if the gear size is not changed.
For many customers, component (gear/pinion/bearing) downsizing or power densification are continuing goals. The impact of improved steel cleanness on the ability to successfully reduce component size, while maintaining equivalent fatigue performance, has been evaluated and verified using TimkenSteel’s VCFM program.
Lead Author Buddy Damm is an advanced steel solutions scientist at TimkenSteel. He is responsible for developing new products for customers and improved processes for the company’s manufacturing operations. He can be reached at 330-471-2703 or firstname.lastname@example.org