This article presents the methods and results of an empirical study that was conducted to aid process development of a carburized aerospace gear. The objective of the study was to determine the contribution of pre-machining material processing on dimensional distortion during carburizing. Five possible raw material starting conditions were evaluated. The five pre-rough machining conditions studied were: (i) normalized AMS6265 bar stock, (ii) hardened and tempered (core-treated) AMS6265 bar stock at 1,725°F, (iii) hardened and tempered (core-treated) AMS6265 bar stock at 1,550°F, (iv) normalized AMS6260 forging, and (v) hardened and tempered (core-treated) AMS6260 forging at 1,725°F.

Cost, time, and lurking variables were minimized by use of a standard distortion coupon in place of actual aerospace gears. The coupon design is shown in Figure 1. H. French used this type of coupon to study dimensional distortion during repeated quenching (Ref. 1). French’s coupon was scaled as necessary for use in this study. The diametrical changes of the coupon indicate the volume changes during hardening. The width of the slot reflects the magnitude of internal stresses set up by the volumetric changes (Ref. 1). French showed that dimensional distortion increased as the number of quench cycles increased. The distortion coupon gap width increased with each quench cycle, thus indicating that residual stresses were increasing with each thermal cycle.

Figure 1--Distortion coupon, GR-0010.

The hypothesis is that dimensional distortion increases along with the thermal and mechanical processing of the raw material prior to machining. The implication is that dimensional distortion can be influenced before the raw material enters the machining process.

Precarburizing process variables and their influence on dimensional distortion were studied previously. The Instrumented Factory for Gears (INFAC) studied the effects of processing variables prior to carburizing. The INFAC study evaluated residual stresses induced by turning and hobbing and their contribution to dimensional distortion. Mechanical and thermal processing of the raw material, however, was not included in the study.

Background and Literature Review
A dedicated manufacturing cell to produce small aerospace gears was designed and implemented. The design of the manufacturing cell and process was to minimize lead time and cost. The shaping process was used to generate the spline and gear teeth. The resultant gear and spline surface integrity produced by the newly designed process was deemed unacceptable due to machining tears that would not clean up during gear grinding. An example of the post-shaped tooth surface is shown in Figure 2.

Figure 2--Gear tooth surface, post-shaping (Ref. 3).

Surface integrity is the description and control of the many possible alterations produced in a surface layer during manufacturing. Surface integrity can be evaluated based on a minimum data set. The data set is composed of surface texture, macrostructure, microstructure, and microhardness alterations (Ref. 2). The data set for macrostructure will include surface imperfections such as pits, tears, and/or laps.

The raw material selected for use in the new manufacturing cell was normalized bar stock, which was within the engineering requirements of the finished gear. The soft normalized bar stock was viewed as a good choice for machinability. Many literature sources supported this conclusion. Mott defined machinability as being related to the ease with which a material can be machined with reasonable tool life (Ref. 5). Verzahntechnik Lorenz (Ref. 6) and Cluff (Ref. 7) indirectly used a similar definition stating that machinability (reasonable tool life) decreases as material hardness increases. The two key terms are “ease of material removal” and “reasonable tool life.” An indication of expected surface integrity is not present using these definitions of machinability.

Material hardness can be used as a machinability indicator due to the close relationship between hardness and microstructure (Ref. 8). However, hardness is an accurate representation of machinability only for similar microstructures. Mullins states that a tempered martensite matrix will exhibit superior machinability to a pearlite matrix of similar hardness (Ref. 8). Woldman studied microstructure and machinability and noted that a microstructure selected for long tool life would not necessarily produce good surface integrity (Ref. 9).

Based on literature and experience, a tempered martensitic microstructure was desired to produce the required surface integrity. The addition of the hardening and tempering operation was viewed as a risk to changing the dimensional distortion during carburizing and hardening.

A great amount of manufacturing development had been done implementing the new cell. The dimensional distortion during carburizing and hardening had been established and had been determined acceptable and manageable. The addition of a hardening and tempering operation prior to rough machining was viewed as an addition to cost, lead time, and risk of increased dimensional distortion during carburizing. Increased dimensional distortion would then require more process development time and cost.

Problem Statement
Common ground: Aerospace power transmission components must be manufactured to the highest quality standard while minimizing cost of nonquality.

Destabilizing condition: Gear tooth surfaces inconsistently have poor surface integrity (“tears”) present after finish flank grinding. The surface defects are produced during the semi-finishing, pre-hardening operation and result in deviated, reworked, and/or scrapped parts.

Contributing factor: Aerospace gears are expensive and have long lead times. A study of many variables is not always practical using actual gears.

Problem: The shaping machine used in the manufacturing cell has limited cutting parameters. Literature suggests that hardening and tempering the material prior to any machining will improve the surface integrity during shaping. The material structure is then martensitic, and hardness ranges from Rc 25 to Rc 32. However, literature also suggests that this fix could negatively influence dimensional distortion during hardening.

Solution: Material samples made of different microstructure and hardness will be fabricated and tested. Paired data studies statistically analyzing dimensional distortion will be performed on coupons of similar size and process.

Assumptions
The material samples are assumed to fully represent their population. For example, a group of normalized material samples is assumed to represent all normalized material.

The change in coupon gap width is assumed to represent the relative dimensional distortion of an actual gear.

Methods and Procedures
This section contains the details of coupon manufacturing and processing. Standard distortion coupons were manufactured for each population as shown in Figure 1. The dimensions of the coupon were proportional to the gear being developed and are shown in Figure 3.

Figure 3--Detailed specimen drawing (units = inches).

Coupons from each population were machined, stress relieved, carburized, and hardened together. The coupons were randomly located in the carburization furnace and quench basket. A summary of the populations is shown in Table 1.

A sample size of 10 was used with one additional sample used for metallurgical evaluation. The samples’ letter designations will be used from this point on to identify the populations. The raw material requiring hardening and tempering was heat treated in-house as listed in Table 2 and Table 3 prior to any machining.

The normalizing process was done per the AMS specification prior to receiving the material. The manufacturing and inspection stages of the coupons are listed in Table 4.

The coupons were carburized and hardened using a cycle common to the actual gear (see Table 5).

Dimensional measurements. A Zeiss Prismo scanning coordinate measuring machine (brass tag #253685) was used to perform all measurements. The outside diameter, inside diameter, and the gap width of every coupon was measured before carburizing, after carburizing, and after hardening. Each time, the outside diameter and etched face were scanned and set as reference. The gap width was measured at a constant radius of 0.7000 inches from the reference center. All measurements were taken in a plane 0.1500 inches (half overall length) from the reference face using a 0.054" diameter probe. The coupons were soaked in mineral spirits, wiped dry and rinsed with alcohol before each measurement. The cleaned coupons were placed in the CMM room 24 hours before measurement to thermally soak and stabilize. The CMM room temperature is held at 69°F +/– 2°F. The actual measurements are contained in Appendices A–D. A sample inspection report is in Appendix D.

Findings
This section contains all of the data and findings collected during the study. Data collected includes characterization of the pre-carburization microstructure and dimensional measurements.

Pre-carburization microstructure. To document the pre-carburized material, an extra coupon was manufactured from each population for metallurgical evaluation. The evaluation was performed after finish machining and before carburizing. The chemistry of Sample E is not reported in Table 7. The sample was lost during the metallurgical evaluation process. The hardness (see Table 6), chemistry (see Table 7), and microstructure (see Figures 4–12) were evaluated on the etched face side of the coupon in two radial locations, center and near the outer diameter.

Figure 4 (left)--Sample A center microstructure 100X, 5% nital etch (Ref. 10).
Figure 5 (center)--Sample A near OD microstructure 100X, 5% nital etch (Ref. 10).
Figure 6 (right)--Sample B center microstructure 100X, 5% nital etch (Ref. 10).

Figure 7 (left)--Sample B near OD microstructure 100X, 5% nital etch (Ref. 10).
Figure 8 (center)--Sample C center microstructure 100X, 5% nital etch (Ref. 10).
Figure 9 (right)--Sample C near OD microstructure 100X, 5% nital etch (Ref. 10).

Figure 10 (left)--Sample D center microstructure 100X, 5% nital etch (Ref. 10).
Figure 11 (center)--Sample D OD microstructure 100X, 5% nital etch (Ref. 10).
Figure 12 right)--Sample E center microstructure 100X, 5% nital etch (Ref. 11).

Dimensional measurements­ and descriptive statistics. Measurements were recorded before carburization, after carburization, and after hardening. Details of the measurement method are in the Methods and Procedures section. Serial numbers B5 and B6 were lost during carburization and serial number C10 was scrapped during manufacturing (see Appendix B). Descriptive statistics of the pre-carburization, post-carburization, and post-hardening measurements are listed in Tables 8–10.

The actual measurements are shown in Figure 13. Descriptive statistics of the paired difference between pre-carburization and post-hardening are listed in Table 11. The mean paired difference values are shown in Figure 13.

Figure 13--Gap width measurements after each processing step of carburizing and hardening.

Paired difference. The pre-carburization and post-hardening gap measurements were paired to enable a relative comparison. The gap width difference is reported as initial minus final. Thus, a change resulting in an increased gap width is reported as a negative value.

Graphical checks of the gap width change data were performed and shown in Figures 14 and 15. The checks include an individual data plot and a normal probability plot. Notice that the paired difference measurements were sorted based on each population’s mean value and plotted.

Figure 14--Data plot of gap width paired distance.

Figure 15--Probability of paired gap width distance.

An analysis of variance (ANOVA) was performed to compare the five population means (see Table 12).

Formally, the data analysis is stated as:

H0: µA = µB = µC = µD = µE
H1: µA, µB, µC, µD and µE are not all equal
a = 0.05

The F statistic is greater than the F critical value. Therefore, the null hypothesis is rejected, and the alternate accepted. The ANOVA identifies if difference is present between any of the mean values. A multiple comparison procedure is required to determine in what way they are not equal. A Fisher’s Least Square Difference (LSD) test was performed at an individual a = 0.05 to determine which of the population means were significantly different from each other. The results are listed in Table 13.

An upper value equal to or greater than zero indicates that the population is significantly less (greater distortion) than the population subtracted from. A summary of the LSD results is listed in Table 14.

Discussion
The gap width change was used to indicate differences in dimensional distortion during carburizing and hardening between five different raw material mechanical and thermal processes. Statistical analysis of the gap width change provides the following:

1.) Coupons manufactured from hardened and tempered barstock and forgings at 1,725°F had the greatest gap width change.

2.) Coupons manufactured from normalized barstock had the smallest gap width change. Coupons manufactured from normalized forgings had significantly more gap width change than those made from normalized barstock.

3.) Coupons manufactured from barstock hardened and tempered at 1,725°F had significantly more gap width change than those made from barstock hardened and tempered at 1,525°F.

4.) Coupons manufactured from barstock hardened and tempered at 1,550°F had no significant difference in gap width change than those made from normalized barstock.

The barstock used to manufacture coupons was from a common heat lot. Also, the forgings used were from a common heat lot. Additional heat lots could change the mean and/or scatter gap width change. It is recommended that future studies include multiple heat lots of materials. It is further recommended that future studies include residual stress measurements prior to carburization and hardening.

Based on the results of this study, hardened and tempered barstock at 1,550°F was selected. The surface integrity of the shaped gear and spline teeth improved greatly. Pre- and post-heat treat data collected from actual gears showed no change to the heat treat distortion.

References

1. French, H. The Quenching of Steels. American Society for Steel Treating, Cleveland, OH, 1930.

2. INFAC. “Distortion of aerospace precision gears Part I: The effects of processing variables prior to carburizing,” Report number TR-94-F-004, IIT Research Institute, Sept. 1994.

3. Evard, M. E-mail containing photos from Evard to Blake, 4/14/2005.

4. Metcut Research Associates Inc. Machining Data Handbook, 3rd Edition, Vol. 2, Zimmerman & Son, Newport, KY, 1980.

5. Mott, R. Machine Elements in Mechanical Design, Bell & Howell Company, Columbus, OH, 1985.

6. Verzahntechnik Lorenz. Gear Cutting Tools, G. Braurn, Ettlingen, Germany, 1980.

7. Cluff, B. Gear Process Dynamics, American Pfauter. Malloy Lithographing Inc., Ann Arbor, MI, 1993.

8. Mullins, J.D. Ductile iron data for design engineers, Rio Tinto & Titanium Inc., Montreal, Canada, 1990.

9. Woldman, N. “Microstructure versus Machinability of Alloy Gear Steels,” American Gear Manufacturers Association, Presented at the AGMA September 20, 1937, Semi-Annual Meeting at Lake Wawasee, Indiana.

10. Weinrich, K. “Short Form Metallurgical Routine Test,” Rolls Royce lab report number 03MTS-0712, Feb. 24, 2003.

11. Ballard, C. “Short Form Metallurgical Routine Test,” Rolls Royce lab report number 03MTS-00407, Feb. 25, 2003.

See the attached PDF for the Appendices.

This article is printed with permission of the copyright holder, The American Gear Manufacturers Association. Statements presented in this paper are those of the author and may not represent the position or opinion of The American Gear Manufacturers Association.

Gregory Blake is a senior specialist, mechanical engineer at Rolls Royce. He is responsible for technical support of gear system design and manufacturing, especially in the process development of gears.