1、2000FTMl Minimization of In-Process Corrosion of Aerospace Gears by: A. Manish including the investigation of an adapted on-line Digi-Galv probe as a predictive tool. Identify and implement preventive practices to reduce in-process corrosion. O o Approach, Scope, and Methodoloqv: This study was cond
2、ucted in two phases: Phase I was to identify root causes of in-process corrosion by conducting a comprehensive study, using manufacturing process bench marking, at two gear production facilities. The Phase I study was augmented by controlled laboratory experiments. The desired end product of Phase I
3、 was to understand the sources of corrosion on gears during the manufacturing process and to identify solutions to minimize the corrosion problem. Phase II was to implement the identified solutions to a restricted area of the manufacturing floor to test them in a production environment. Phase I invo
4、lved defining the boundaries of the gear making process, developing detailed process maps that describe all of the steps required to produce the gears, veriiing the actual inputs at each process step, and perfoming the Failure Mode and Effects Analysis (FMEA) to assess priority of the process parame
5、ters. After the process conditions that could potentially be contributing to corrosion initiation were identified and prioritized, the design of experiments (DOE) were planned and performed. 1 For the process mapping, the boundary of the study was established from raw material handling through final
6、 inspection of the part. Typical manufacturing process steps for gears included turning, cutting, hobbing, deburring, grinding, heat treating (Cu plating, carburizing, striping), lapping, shot peening and inspection. Three gear parts (spur, helical, and planetary pinion) were selected from two separ
7、ate gear manufacturers as a result of the process mapping described above. These parts provided the basis for detailed evaluations of the significant manufacturing steps that were likely to initiate in- process corrosion. In addition, parts selected were made out of common base materials that are wi
8、dely used in the aerospace industry to manufacture gears. The materials selected were 9310 (AMS6260, AMS6265) and Pyrowear X-53 (AMS6308) steels. A 1 B I C I D I E 1 F I G IE*G Hrs to Initiation 1848 1848 1847 1847 1847 1847 1843 1843 1843 1843 1842 1842 -1 -1 -1 1801 1801 1395 1-1 11 1394 11 1392 1
9、392 1392 1392 1392 -1 -1 1392 -1 -1 1392 7 338 1319 1319 1220 1204 When the part selection and verification of the manufacturing steps were completed, the FMEA was used to evaluate the process inputs. The FMEAs allowed the team to prioritize process inputs that could contribute to corrosion initiati
10、on. Process control charts and the FMEA were used to identify and rank the suspect inputs to be tested. After a comprehensive investigation, the coolant was identified as one of the key process inputs that may be contributing to corrosion initiation. Multiple potential failure modes were identified
11、for the coolant along with multiple possible interactions. Key factors identified after control chart and FMEA analyses were: coolant concentration, coolant temperature, exposure time, coolant type, coolant contamination and material heat treat condition. , 1224Treatment Run Example Once the team ha
12、d narrowed the list of potential corrosion initiation factors, the first DOE design strategy began. The strategies for the design and factor level settings are detailed in Figure 1. Factors selected for testing are: (A) coolant condition (virgin and reclaimed), (B) specimen alloy (9310 and Pyrowear
13、X-53), (C) specimen heat treat condition (carburized and uncarburized), (D) exposure time to coolant (30 and 90 minutes), (E) coolant concentration (1 and 8 percent), (F) coolant temperature (709 and 150QF), and (G) coolant type (soluble oil and synthetic). Refractometer readings were used as a meas
14、ure of coolant concentration. Factors LEVEL= + LEVEL= “-“ A Condition Virgin Reclaimed B Alloy Pyrowear 53 9310 C HT Cond Carburized Core Material D Exposure Time 90 minutes 30 minutes E Coolant Concentration 8% 1 %o F Coolant Temperature 150 F 70 F G Coolant Tvpe Svnthetic Soluble Oil Phase II comm
15、enced in parallel with the DOEs as information became available from Phase I to speed up the verification of benefits resulting from implementation of the identified solutions. Examples of the success of this approach were the implementation of a new coolant in one gear facility and the resolution o
16、f the chemical attack problem noted on the gear line. Figure 1. Partial Experimental Matrix with Screening Analysis. Discussion of Results: The DOEs were conducted at Honeywell Engines & Systems facility. The steel specimens were subjected to prescribed conditions identified in Figure 1. An example
17、of a treatment run in the matrix was to soak a carburized Pyrowear X-53 specimen in an 8 percent concentration virgin soluble oil coolant at 709F for 30 minutes. The specimens were then loaded into a humidity chamber that was set at 8O9F and a relative humidity level of 70 percent. These samples wer
18、e monitored twice daily for corrosion initiation. When corrosion initiation (response) occurred, the time (hours) to initiation was recorded and used for the experimental analysis. 2 Upon establishing the hours to corrosion initiation on the specimens, analyses of the experiment results were perform
19、ed. A screening analysis was performed by sorting the hours to initiation in ascending order and evaluating the experiment matrix for any patterns. The visual pattern shown in figure 1 that appeared to be significant is: coolant type, coolant condition, and an interaction between coolant concentrati
20、on and coolant type. A statistical analysis was performed next. The normal probability plot and the pareto of effects shown in figure 2 and figure 3 respectively, identified the same factors as being significant. Figure 4, the interaction plot, showed the coolant type by coolant concentration as the
21、 most significant interaction in the model. The main effects plot, figure 5, indicated that coolant concentration and coolant type had the greatest effect. An analysis of variance (ANOVA) confirmed the results repotted above as statistically significant. Based on the data, low concentration and virg
22、in coolants are predicted to initiate corrosion more quickly than reclaimed and higher concentration coolants. Heat treat condition also affected the initiation of corrosion in that high carbon (carburized) surfaces were more resistant to initiation as compared to core (uncarburized) surfaces. It wa
23、s also synthesized from the data set that usedreclaimed coolants offered better corrosion resistance than the virgin mix in both synthetic and soluble oil coolants. The data also supported using the soluble oil coolant as the preferred cutting fluid. As a result of the experiment, the soluble oil, w
24、ater based coolant was implemented in the gear production area as part of Phase II work. Based on the results of the first DOE, the strategy for the second experiment was developed. The factors selected for this experiment were: coolant concentration (3 and 6 percent), base material magnetism (O and
25、 10 gauss), specimen surface finish (6 Ra and 36 Ra), iron particles in coolant (clean coolant and particles added), specimen raw material heat lot (heat lots 1 and 2), specimen heat treat condition (carburized and uncarburized), specimen alloy (9310 and Pyrowear X-53), degreasing solvent condition
26、(virgin and used), and preservative oil application (4-minute soak application and spray application). At the suggestion of manufacturing personnel, the additional factor that was added to the DOE was the method of preservative oil application. Previous process mapping analysis revealed that a typic
27、al lot of gears could go through the preservative soak cycle 24 times. If the soak cycle could be replaced by a spray/quick immersion with the same result, sicinificant savings in cycle time and cost could With the factors and level settings established, the DOE was executed as described previously.
28、 However, there was no corrosion initiation observed on the samples after 6 months. The experiment indicated that the factors tested would not initiate corrosion provided a coating of preservative oil was applied, regardless of application method. Based on this result, the manufacturing specificatio
29、n was modified to only require a quick immersion in lieu of a 4-minute soak. The practice was quickly implemented in Phase II. Nonnal Probability Plot 01 the Standardized EHeCIs (Responos %Time io. Alpha = 0.10) .E I .C A A Condition B B Alloy C C HT Condition D DYinuie E E Refacometer F F Temperatu
30、re G G Coolant Type I I I -10 O 10 G8- standardized Ettecl Figure 2. Normal Probability Plot. Normal Probability Plot of the Standardized Effects (Response isTime to, Alpha = 0.10, Only 30 Largest Effects Shown) E EG G A C ADE AG DG F AE DE AEG ABF AF EF BG A: A Condition BC FG C: C HT Condition AD
31、AB D: D Minute D CD E: E Refracometer F: F Temperature BD BF G: G Coolant Type BE AC CG CE DF B 6: B Alloy O 2 4 6 8 10 12 GgnaJ1 be realized. figure 3. Pareto of Effects. 3 A Condaim .I .-I mm32 , loa, 500 1500 loa, 500 1500 loa, 500 1500 loa, 500 I500 loa, 500 1500 loa, 500 I I I I G Coolml -1 Fig
32、ure 4. Interaction Plot. . / F T.i.Run Figure 5. Main Effects Plot. Chemical Attack: During the study, a large quantity of gears o cook-l were exhibiting a pitted condition at the machining operations in various areas. After metallurgical examination, it was determined that the pits were a result of
33、 chemical attackkorrosion. Since the machine cutting fluids, cleaning and preservation factors had been previously tested in Phase I, they were quickly dismissed as the root cause. This allowed the team to focus the investigation in the plating facility. The key factors identified through a thought
34、process mapping exercise, as possible initiators of the attack were alkaline pH, exposure time, and the ounces of copper in the alkaline strip solution. An experiment was designed to test these factors on 9310 steel specimens as shown in Figure 6. Analysis of the experiment identified pH as the fact
35、or with the highest significance. Data also indicated when the pH level was greater than 10, chemical attack did not occur. IhnOrder Cooper ozigal DH Exposure minutes Attack ,a 7 1 1 2 6 8 1 -1 -1 2 4 -1 -1 -1 2 5 1 -1 1 3 3 -1 -1 1 4 Figure 6. Alkaline Strip Experiment Matrix. For phase II implemen
36、tation, the team installed a controller on the alkaline strip tanks to maintain the tank pH levels above 10.3, by automatically adding anhydrous ammonia when the pH level dropped below 10.3. It was noted that during the anhydrous ammonia addition, an exothermic reaction occurred, heating the solutio
37、n to temperatures above 100“. The problem identified with the elevated temperature was that when the alkaline stripping solution reached temperatures above 77“, it became corrosive to carbon steels. The team installed a chilling unit in the stripping tanks to stabilize the solution temperature at 60
38、“. After installing the chiller and maintaining stable pH levels at 10.3-1 0.7, no chemical attack on an entire lot of gears has occurred due to pH imbalance as a root cause. Residual Coolant: Early in the study, it was determined by extensive laboratory testing that water-based machining coolant wa
39、s an effective corrosion inhibitor. Low carbon alloy gears coated with a film of in-process coolant did not corrode after 2 months in the high heat and humidity test cham ber. The results of this testing were: Used coolant is not a corrosion enabler. The coolant solutions ability to form a protectiv
40、e film barrier on the surface of the part determined its ability to inhibit corrosion. Whether this film came from virgin or in- process coolant was irrelevant. 4 If a film was not left on the part by the coolant after the machining operation and it was not properly treated with a preservative oil,
41、then corrosion was enabled. Such corrosion resulted in increased manufacturing costs and lead times. A corrosion-protective film could be formed after processing through either the coolant or preservative solution typically applied during the manufacturing process. Test results indicated that residu
42、al coolant would leave a short-term corrosion inhibiting film on the gears similar to the ones resulting from immersion in an oil bath. This residual film could reduce the incidence of corrosion even if the preservative oil was not applied. If the proper fluid attributes could be measured and mainta
43、ined, not only would costly corrosion initiation be greatly reduced, but the corrosion prevention steps (preservative baths) could also be decreased or eliminated. On-line PCI-Win Probe: A significant barrier to relying on residual coolant as an effective short-term corrosion inhibitor is determinin
44、g if the coolant has the proper attributes. Standard corrosion testing takes too long to give information usable in an on-going production environment. For example, this testing procedure would take approximately 48 hours to complete at added cost for coolant analysis. The teams goal was to find a t
45、ool to provide real time sump-site testing to predict the coolants corrosion inhibiting properties. After further research, the team decided on the use of an adapted Digi-Galv probing system, PCI-Win Probe, similar in operating principle to the ones used in the oil pipeline monitoring industry. The
46、probe has the potential to provide quick indications of a coolants corrosion-inhibiting capabilities. The PCI-Win Probe System shown in figure 7, consists of a mini test cell (specimen housing and electrode), a digitally controlled dual range precision instrument unit, and a standard Pentium desktop
47、 PC with the AMPLICON PC3OAT card and associated software installed. In addition, three accessories are added: a glass sleeve sample vessel, a constant temperature water bath with tubing to allow flow through the glass sleeve, and a pump to provide circulation of the sample. The accessories create a
48、 testing environment that simulates the actual conditions under which the coolants perform in the specimen while both are submerged in the coolant being analyzed. The resistance represents the film formed by the coolant on the specimen. The higher the resistance, the more complete the coverage and t
49、he thicker the film. This resistance is measured and graphed by the PCI-WIN Probe System, in the form of a “sweep” shown in figure 8. The sweep consists of two parts, cathodic and anodic. The anodic portion (above the X-axis, or positive voltage values) of the sweep indicates the rust preventative (RP) packages of the fluid. The longer the curve remains in the vertical upswing, the better RP packages in the fluid. Also, the further to the left that the inflection point occurs, the better RP in the fluid. In general, an inflection point at or to the left of 1 O indicates a fluid that
