《An Assessment of Ford 3.4 Liter DOHC V8 SHO Engine Camshaft Failures.doc》由会员分享，可在线阅读，更多相关《An Assessment of Ford 3.4 Liter DOHC V8 SHO Engine Camshaft Failures.doc（47页珍藏版）》请在三一办公上搜索。
1、An Assessment ofFord 3.4 Liter DOHC V8 SHOEngine Camshaft FailuresbyBruce C. Lamartine, Ph.D.4/27/2004 Copyright 2004by Bruce C. LamartineABOUT THE AUTHORBruce Lamartine is a Technical Staff Member and Intellectual Property Coordinator at Los Alamos National Laboratory. Prior to that, he was a Techn
2、ical Staff Member with both the U.S. Air Force Materials Laboratory and Motorola Semiconductor Products Sector. He holds a Ph.D. in Physical Chemistry from Case Western Reserve University. Dr. Lamartine has led successful failure analysis teams for the U.S. Air Force in the fields of pressed metal p
3、owders and electronic emissive materials. He has 75 scientific publications, 7 issued US and foreign patents, 1 pending US patent, 3 pending foreign patents, 5 biographical citations, and 12 career awards including the Air Force Commendation Medal. His research has centered on surface chemistry, fai
4、lure analysis and production plant troubleshooting, and nanoengineered surfaces.ABSTRACTCamshaft failure in an interference engine such as that used in the 3.4 liter DOHC V8 engine of the 1996-1999 Ford SHO (also known as the Ford Gen III SHO) could be a potentially lethal event. Sufficient literatu
5、re indicating premature failure of the drive sprockets existed years before production commenced on the Ford Gen III SHO. From the literature publicly available prior to production, we were able to use both a gear tooth stress model and a valve train resonance model to predict a band of failures rem
6、arkably consistent with the camshaft failures actually occurring after production. Available Ford Gen III SHO camshaft failure data were analyzed by nonlinear least squares regression techniques. The most probable camshaft failure mileage was found to occur at approximately 70,000 miles, beyond Ford
7、 warranty, but still early in engine life. Consistent with the physics of random impacts and the assumption of the use of St 52 DOM steel tubing in camshaft assembly, the data were best fitted by a single Birnbaum-Saunders distribution function to better than the 95% confidence level usually require
8、d by courts of law. It is very unlikely that performance driving alone is responsible for all camshaft sprocket failures; rather, extreme performance driving may account only for the low mileage branch of the observed camshaft failure distribution and then only when very poor force locking and/or hi
9、gh lobe torques also occur. Insufficient spline contact area appears to be the most likely source of sprocket failure, and, given the literature of the time and reasonable engineering expertise, this defect should have been recognized and corrected before production. Lifetime predictions of two fail
10、ure remedies are also reported.TABLE OF CONTENTSINTRODUCTION 6MODELS AND CALCULATIONS 7A Physical Model of Failure 8Miners Rule .14The Birnbaum-Saunders Distribution . 14The Power Birnbaum-Saunders Approximation . 16A Priori Estimates of Failure. 18Method 1: Gear tooth stress limit model . 18Limits
11、of tubing expansion and resultant contact area. 19 Stresses applied to tubing teeth . 20 A model of tool wear . 20Method 2: Lobe torque model. 25An estimate of applied torque. 26 The relation of specific energy to applied torque. 29Consideration of valve spring resonance forces. 31Analysis of the Fa
12、ilure Data. 35 RESULTS AND DISCUSSION. 37Comparison with Failure Models. 37A Vulnerability Estimate. 39Methods to Extend Drive Sprocket Life. 39 CONCLUSIONS AND RECOMMENDATIONS. 42ACKNOWLEDGMENT. 43REFERENCES. 44 INTRODUCTIONDue to the interference design of the 3.4-liter DOHC V8 SHO engine, a camsh
13、aft sprocket failure usually leads to additional damage to valves, pistons, and other major engine parts. Such a failure often leads to loss of vehicle control since power steering and power brakes are also lost when the engine fails. At this writing, we are thankfully unaware of any associated fata
14、lities. A typical repair bill is around $6000, and a new engine can cost as much as $15,000. A complaint has been filed against Ford Motor Company 1 alleging that Ford had knowledge of the camshaft sprocket product defect early in the product cycle. It is the intent of this paper to show that engine
15、ering literature available to Ford and the public at large well before the time of manufacture strongly suggested that the sprocket of an assembled camshaft was a likely point of failure; and furthermore, that the failures were to be expected at the mileages later publicly reported. This paper will
16、also show that corrective measures already known by at least as early as 1992 could have been implemented either before or during the time of manufacture to avoid this product defect. Since early 2000 at least, an association of Gen III Ford SHO owners (V8SHO.com) has recorded the camshaft sprocket
17、failures of their members 2. As of mid-April 2004, some 385 sprocket failures (approximately 295 with mileages at failure) have been publicly documented by this organization. While internal Ford records of these failures would most likely be a larger data sample, the public data are nevertheless a s
18、ufficiently large sample to recover significant information. We have chosen to follow the analysis practices set forth by the National Institute of Standards and Technology (NIST) known as Exploratory Data Analysis (EDA). EDA is a combination of graphical and numerical techniques 3. For court compat
19、ibility where appropriate, we have also included the conventional chi-square statistical tests at the 95% confidence level 3,4. Crucial to further analysis is the assumption of the actual species of steel tubing used, and we have made speculations only. Ford, however, is known to have used an assemb
20、led camshaft on the Mondeo, a V6 European model 5. The supplier for that engine was identified as Thyssen-Krupp Presta. No particular grade of steel tubing was revealed in that reference; but we do know from public import relief request documents 6 that only 26Mn5 and St52 (both DIN 2393 grade) were
21、 specified as the highest European standard for drawn over mandrel (DOM) tubing intended for use in assembled camshaft manufacture. St52 is a low carbon steel of high yield strength and well-known strain-life properties 7. Regardless of the actual assembly process used, the steel tubing is deformed
22、to form features which engage the camshaft elements such as drive sprockets, bearing sleeves, and cam lobes. The patent literature prior to 1995 points out a number of limitations of this kind of manufacturing, particularly the potential weakness of drive sprockets, although no quantitative physical
23、 arguments appeared. An assembled camshaft patent 11 issued in 1992 and assigned to the company Emitec in Germany specifies St35 (an older, weaker steel composition) or St52 as the tubing used in the preferred embodiments. A second patent (issued in 1991) also assigned to Emitec 12 specifically poin
24、ts out the potential for failure at the drive sprocket. There are similarities between the drawings in the Emitec patents and the pictures shown on the member website 2. Essential differences, however, are the internal spline design for the drive sprocket and the internal deformation pattern. It is
25、here that the contact area is obtained upon tubing expansion (see below). An assembled camshaft and its method of manufacture wherein a star-shaped mandrel is used for internal plastic expansion was patented by Bendoraitas and Clark 13,14 in 1989. The patent was assigned to the Torrington Company. T
26、he internal expansion of the Gen III Ford SHO camshaft shows evidence of the use of a star mandrel or similar device. Consequently, it appears likely to this author that Thyssen-Krupp Presta was the supplier of camshafts for 3.4-liter DOHC V8 SHO engines and that some variation of the Emitec or Torr
27、ington assembly process was used. The methods of assembling camshafts were also well reported in the patent literature of the late 1980s and early 1990s. All assemblies of the time appear to have been made in an elaborate jig where all lobes and sprockets are aligned before force locking. Force lock
28、ing is achieved by either hydraulic expansion of selected tubing regions 11, ballizing; i.e., forcing an extremely hard oversize metal ball (usually tungsten carbide) down the inside of the camshaft tubing 15,16, or splined mandrel expansion where a forming device is passed through the inside of the
29、 tubing 13,14. As a result of any of these methods of manufacture, the tubing sees internal pressures in the range of 2000-3500 bar (200-350 MPa) 11 and deforms plastically into the groove region between the element (sprocket, bearing sleeve, or lobe) splines. Expansion of the outer diameter of the
30、tubing in this region is typically no more than 10-15% of the tubing thickness 11. This means that the depth of the element must be designed appropriately in order to produce a contact area sufficiently large so that the stress applied on the tubing is low enough that the strain produced insures suf
31、ficiently long service life. It should be noted that a service life on the order of 100 million-10 billion reversals is sought and that early fatigue designs for steels assumed infinite life beyond approximately 1 million reversals 17. Even more recent strain-life models can overestimate the lifetim
32、e if insufficient data (or an insufficient data interval) is taken in the high fatigue cycle range.MODELS AND CALCULATIONSA table of approximately 1200 data pairs was generated for each equation in order that its graph could be presented with an equivalent resolution of 590 miles at 70,000 miles. At
33、 least five significant figures were carried through all exponential and logarithmic calculations. Common software packages available today for such work are Mathematica, Matlab, and Mathcad. Mathcad was chosen for this work because tabular data corresponding to a single graphical point could be ext
34、racted by the “trace” function. Similar operations using a lookup table could have been programmed with ease by workers in the late 1980s using the FORTRAN and BASIC computing environments. BASIC software of the time also had an internal graphics package sufficient for this work. A physical model of
35、 failure.We treat our expanded tube as if it were a gear with individual teeth that mesh with the camshaft element splines and fail in the manner of cantilevered beams. Fig. 1a shows an actual cross-section of a Gen III camshaft drive sprocket 2 and Figs. 1b-d show the conceptual detail of the compl
36、ementary tubing expansion region. Figure 1a. A section of a failed Gen III camshaft drive sprocket. Photo by Bob Gervais. Downloaded from 2 with webmaster permission.Figure 1b. Cross-sectional model of Gen III camshaft tubing after force locking by star mandrel to the drive sprocket. The outer splin
37、ed edge of this tubing meshes with the concave grooved region of the sprocket shown above in Fig. 1a. The radial expansion was exaggerated in this calculation for clarity 18. Figure 1c. Three-dimensional model of Gen III camshaft tubing after force locking by star mandrel to the drive sprocket. The
38、outer splined edge of this tubing meshes with the concave grooved region of the sprocket shown above in Fig. 1a. The radial expansion was exaggerated in this calculation for clarity 18. Figure 1d. Detail of force locking tubing expansion. (a): Before expansion. (b): After expansion. The radial displ
39、acement dr is exaggerated for clarity. Although movement of material is shown, conservation of volume is assumed (see Method 2 below). Several models of gear tooth stress are available 19-21, but the simplest of these (available in 1981 19) suffices for further analysis if the ultimate strength and
40、the elastic limit of the tubing material are known. Fortunately, the entire stress-strain curve for St52 steel was known at least as early as 1972 22, and we have digitized data from that source for further use in lifetime prediction. It should be noted that the slope of the elastic region of the st
41、ress-strain curve is almost invariant with temperature. Thus we have confidence that a low-temperature curve will suffice to describe small camshaft element stresses at all operating temperatures. Fig. 2 shows the stress-strain curve used.Figure 2. Data digitized from the stress-strain curve of St52
42、 steel at 20 degrees C 22. Note the logarithmic axes.The strain-life of metals was investigated systematically in the early 1970s 23, and four parameter strain-life coefficients of St52 steel can be derived from the data reported at least as early as 1987 7. A strain-life equation for the total stra
43、in e is , (1)where sf, b, ef, c, E, and Nf are the fatigue strength coefficient, the fatigue strength exponent, the fatigue ductility coefficient, the fatigue ductility exponent, Youngs modulus, and the number of fatigue cycles, respectively. The data presented in 7 show evidence of an endurance lim
44、it typical of triangular load waves of constant amplitude and frequency. In a camshaft application, however, random load amplitudes and frequencies are applied; and consequently, a continually decreasing strain as a function of life rather than an endurance limit is expected 8. A more appropriate re
45、presentation of high cycle fatigue in this case is obtained when the coefficients of the strain-life equation are calculated from single frequency test data representing both low and high cycle fatigue (beyond transition life) but not including noisy data typical of an endurance limit. Figure 3 show
46、s the combined data for duplicate runs of St52 steel as well as the strain-life curves calculated 9 for single frequency and random frequency loading. The random frequency case compares well to a later curve, also shown in Fig. 3, given by Ilzhofer, et al. 10 for multiple frequency impacts on a moment loaded round bar. The coefficients, , obtained from a four parameter fit of
