Failure analysis of HNBR oil seals.

At the heart of what appears to be a purely metallurgical application in the aluminum industry lies a crucial elastomeric component, an oil seal. Physically, the seals have a beveled geometry with outside diameter dimensions of 29 and 22 inches at the top and bottom, respectively. The total height of the seal is approximately 4.5 inches, with a fabric reinforcement on the upper edge and a continuous wound metallic spring within a groove around the bottom. The seal is situated within a gear coupling, which is used to drive a standard aluminum rolling mill. During operation, a motor causes the seal to flex at a nominal speed of 60 rpm. A cross-sectional diagram depicting the seal in its extended and compressed state is shown in figure 1. The upper, fiber reinforced edge of the seal is largely constrained during operation, as indicated in the diagram. During service, one surface of the seal remains at ambient conditions, while the opposite surface is exposed to 3-5 psi of pressure and is coated with a grease-type lubricant. The seals were manufactured in a two piece compression mold from a peroxide cured, carbon black filled HNBR rubber. Premature failures occurred during normal service operation for seals installed on the drive end of the gear coupling, as described in table 1, while no corresponding failures occurred on the output end. The non-failed seal was identified as Sample A, while the failed seals were designated B, C and D. This article will document the failure investigation performed on the oil seals.

[FIGURE 1 OMITTED]

Test procedures and experimental results

Visual examination

The seals were examined visually and with the aid of an optical stereomicroscope in order to document the overall condition of the parts and the nature of the failures. Sample B exhibited a three inch tear on the oil side at a transition region between the fiber reinforced flange and the beveled face. This tear is depicted in figure 2 from the oil side, after removal of the reinforced flange. The teat--was completed in the laboratory and examination of the mating surfaces indicated a potential tear origin on the oil side surface, with features characteristic of rubber tearing extending toward the air side. Inspection of the completed tear also indicated that the tear occurred within a relatively thin wall section of the part.

[FIGURE 2 OMITTED]

Sample C exhibited a six inch tear which extended somewhat diagonally across the beveled face on the air side surface. A corresponding tear was detected on the oil side surface, and numerous linear features were observed in the vicinity of the oil side tear, oriented circumfrentially on the seal surface, as illustrated in figure 3. These features were suggestive of permanent deformation likely due to repeated mechanical stress. Completion of the tear in the laboratory revealed evidence of an apparent tear origin oil the air side. Sample D, which had been previously sectioned through the failed region, tidied in a similar fashion as Sample C, although the extent of tearing was not as extensive. Linear features were detected on the oil side surface, and the tearing appeared to originate on the air side. Both Samples C and D exhibited mechanical damage along the beveled face on the air side surface, as represented in figure 4 for Sample D. The location of the damage was clearly associated with the tear origins. The relative location of the tear origins for the three failed seals is illustrated in cross-section in figure 5. By comparison, the Sample A seal exhibited no obvious signs of mating or mechanical abrasion and the smooth surface condition was characteristic of a molded rubber product.

[FIGURES 3-5 OMITTED]

Scanning electron microscopy (SEM)

Portions of the Sample B and Sample C tear surfaces were chosen for further analysis using scanning electron microscopy (SEM). Prior to the examination, the tear surfaces were ultrasonically cleaned in mineral spirits to remove any residual grease present. The cleaned surfaces were then sputter coated with gold in order to facilitate the examination. The Sample B tear surface exhibited a clear origin on the oil surface side. An electron image near the origin, illustrated in figure 6, indicated a relatively smooth morphology without evidence of chemical attack or thermal degradation. Adjacent to the origin, arrest marks were observed, suggestive of cyclic stress application. Remote from the origin, the features were consistent with standard robber tearing in tension.

[FIGURE 6 OMITTED]

The gold coated Sample D tear, shown in figure 7, exhibited an origin on the air side, as indicated. The majority of the surface morphology was characteristic of rubber tearing in tension. However, mechanical damage was clearly indicated at the tear origin, as shown in figure 8. Abrasion of the surface resulted in multiple tear initiation sites which, through the cyclic application of stress during operation, resulted in extension of the tear into the wall of the seal.

[FIGURES 7-8 OMITTED]

Fourier transform infrared spectroscopy (FTIR)

Material representing each of the four seals was analyzed using Fourier transform infrared spectroscopy (FTIR) in order to compare the nature of the elastomeric formulations. Spectra were obtained on material from corresponding locations in the four seals using micro-FTIR in the attenuated total reflectance mode. The results obtained were very consistent for the four samples, with no evidence of contamination in the spectra representing the failed seals. Interpretation of the results indicated that the spectral absorbances were characteristic of a hydrogenated nitrile rubber. Identification of the seal material is illustrated in the spectral comparison shown in figure 9. The nature of the spectra also indicated that the elastomers contained a relatively high loading of carbon black. Overall, the FTIR analysis indicated that the seals had been manufactured from similar elastomeric compounds.

[FIGURE 9 OMITTED]

Thermogravimetric analysis

In order to further characterize the nature of the seal formulation, material representing each part was analyzed using thermogravimetric analysis (TGA). A thermal program was devised which would allow for separation of the formulation ingredients. This is accomplished through a combination of heating rate and controlled atmosphere. The weight loss profile obtained on Sample A is shown ill figure 10. The initial weight loss event is associated with the evolution of lower molecular weight species, including plasticizer. Between approximately 300 and 550 [degrees] C, degradation of the polymer will occur in nitrogen. Conversion to an air atmosphere at that temperature results in rapid combustion of a carbon-based char formed during breakdown of the polymer. Further heating in air results ill combustion of the carbon black filler and a final residue of inorganic, noncombustible material remains at the completion of the analysis. Subsequent interpretation of the TGA thermograms yielded quantitative information about the elastomeric formulation of the four seals, which is summarized in table 2. Overall, the TGA data, coupled with the FTIR results, indicate that the four seals had been manufactured from carbon black-filled HNBR compounds having very similar formulations.

[FIGURE 10 OMITTED]

Physical properties

Sections were removed from corresponding locations on the four seals in order to obtain durometer hardness measurements. The hardness values were obtained in accordance with ASTM D 2240 on a durometer A scale at ambient laboratory conditions. Five individual readings were taken from each sample and an average value of 74 A was obtained in each case.

The mechanical properties of the seal materials were determined in tension in accordance with ASTM D 412. Sample preparation consisted of removing large sections at corresponding locations and machining standard specimen shapes in which the long dimension aligned with the circumference of the seal. Five specimens were tested from each seal using a crosshead speed of 20 in./min, at ambient laboratory conditions. The obtained average data are summarized in table 3. A review of the data indicated that the properties fell into two categories associated with either Samples A and D or Samples B and C. On average, the tensile strength and ultimate elongation of Samples A and D were approximately 24% higher than for Samples B and C. Also, the stress at 100% elongation was, on average, 17% higher for Samples B and C, as compared to A and D. The results indicate that the materials representing Samples B and C were mechanically weaker and exhibited higher stiffness. The trends observed in the mechanical data correlated well with the observed time-to-failure of the seals, in that the seals which exhibited the lowest mechanical strength failed in the shortest amount of time.

Crosslink density

With any application involving a thermoset rubber, the state of cure is a significant factor in the performance of the product. The rubber's cure state will have a profound effect on the part's mechanical properties, including strength and fatigue life. During compound development, maximum tensile strength is often used as a measure of the optimum cure state.

However, it is important to remember that different physical properties will achieve optimum values at different cure states. In the seal application, flex fatigue and tear strength are more significant than tensile strength. One means of measuring the cure state of the molded part is by determining the crosslinking density. This was accomplished using a swelling method in which the sample material was treated with methylene chloride for 96 hours at room temperature. The density values were derived using a Flory-Rehner equation, assuming a 0.39 polymer/solvent interaction parameter. The average values obtained are summarized in table 4. A review of the data indicated good correlation between the crosslinking density and the observed trends in the mechanical data. The lower crosslinking densities exhibited by the Sample A and B materials corresponded to superior mechanical strength properties and therefore a longer service life.

Conclusion/discussion

It was the conclusion of the failure analysis that the oil seals failed via two distinct mechanisms. Seal B failed as a result of the exertion of cyclic bending stresses beyond the strength of the material. Tearing initiated within a constrained sharp corner region and propagated through the seal, with the relatively thin wall section providing a significant contributing factor. The relatively poor mechanical strength of the Seal B material, which correlated to a higher crosslinking density, was also believed to be a significant material factor in the failure. Seals C and D failed in a similar mode as a result of the application of cyclic stresses within regions which had experienced mechanical abrasion during service. The locations of mechanical damage served as initiation sites, with tear extension resulting from the application of cyclic stresses beyond the strength of the material. The failure of Seal C prior to D correlated to the inferior mechanical properties of the Sample C material. Seal A, which had not failed, had been in service on the output end of the system for approximately one year. The successful operation of Seal A was attributed to several factors, including the mechanical strength of the material and the service conditions experienced, which did not result in either mechanical damage to the surface or the application of bending-type stresses. The analysis further indicated that the formulation of the seals was consistent, and that neither material contamination nor degradation had played a role in the observed failures. As a result of the investigation, additional monitoring was implemented regarding the cure conditions utilized to manufacture the seals.

Table 1 - sample description

Sample   Service history         Location

  A      12 months, no failure   Output end
  B      1 week, failed          Drive end
  C      1 week, failed          Drive end
  D      12 weeks, failed        Drive end

Table 2 - TGA data summary

Constituent                   Sample A   Sample B   Sample C   Sample D

Volatiles content, %             3.9        3.3        3.5        3.5
Polymer content, %              69.4       69.8       68.9       69.6
Carbon black content, %         19.4       19.3       20.0       19.3
Inorganic filler content, %      7.6        7.6        7.6        7.6

Table 3 - mechanical test summary

Property                     Sample A   Sample B   Sample C   Sample D

Peak load, lbs.                 175        124        130        172
Tensile strength, psi         3,806      2,896      2,910      3,792
Ultimate elongation, %          323        245        253        338
Stress at 20% elong., psi       188        206        203        188
Stress at 100% elong., psi      608        713        681        548
Stress at 300% elong., psi    3,467        --         --       3,234

Table 4 - crosslinking
density (moles per [cm.sup.3])

Sample      Average value

  A      1.59 x [10.sup.-4]
  B      1.70 x [10.sup.-4]
  C      1.69 x [10.sup.-4]
  D      1.42 x [10.sup.-4]