Nova Scotia B3K 5X5, Canada

Corrosionture, and high-cycle fatigue, were observed for the different materials and parts involved. It

9) ged byr toama

9 engines ecylinder h

shown in Fig. 1.Since the DD149 is a two-stroke engine, all four of the valves that pass through each cylinder head are exhaust

These valves are actuated in pairs by the camshaft and the exhaust from pairs of valves passes through cavities in tinder head before exiting through separate exhaust ports. In other words, the exhaust streams from the two pairs of valveson each cylinder do not mix until they exit the cylinder head.

1350-6307/$ - see front matter Crown Copyright 2013 Published by Elsevier Ltd. All rights reserved.

Tel.: +1 9024272601.E-mail address: Cameron.Munro@drdc-rddc.gc.ca

Engineering Failure Analysis 35 (2013) 499507

Contents lists available at SciVerse ScienceDirect

Engineering Failure Analysishttp://dx.doi.org/10.1016/j.engfailanal.2013.05.009engines can have up to 20 cylinders, and are so named because of their 149 in.3 of cylinder volume. DD14the pot head design, in which the cylinder heads t inside the cylinders. A schematic of one of the DD149. Thesemployeads is

valves.he cyl-In this paper, the observations and conclusions of the failure analysis are provided as a case study, and recommendations areoffered to help avoid similar failures in the future.

2. Background

The DD149s are two-stroke diesel engines commonly used in mining, marine, construction, and other industries1. Introduction

A Detroit Diesel series 149 (DD14ment Canada Atlantic was requestevey of the operational logs in ordenoteworthy due to the extent of the dwas determined that the damage sustained by the engine could be explained as the resultof severe and undetected erosioncorrosion (guttering) of one of the No. 1 cylinderexhaust valves, which caused the valve head to fracture and enter the combustion cham-ber. Possible causes of this valve guttering are discussed, and recommendations are offeredto help avoid similar catastrophic failures in the future.

Crown Copyright 2013 Published by Elsevier Ltd. All rights reserved.

nerator failed after only nine months in service. Defence Research and Develop-the equipment owner to conduct an analysis of the engine components and a sur-determine the root cause of the failure. The ensuing failure investigation wasge to the engine components, and the variety of fracture mechanisms encountered.a r t i c l e i n f o

Article history:Available online 7 June 2013

Keywords:Engine failuresValve failuresErosion

a b s t r a c t

Several components of a diesel generator failed dramatically after only nine months in ser-vice. Operators noticed the generator, a two-stroke Detroit Diesel model 16V149TIB, pro-ducing abnormal noises and smoke. Upon inspection the piston crown, cylinder head,fuel injector, and exhaust valves from the No. 1 cylinder were found to be extremely dam-aged, as was the exhaust turbine from the turbocharger assembly. Examinations of therecovered parts were conducted through visual and chemical analysis, fractography, andmetallography. Various fracture mechanisms, such as thermal cracking, intergranular frac-Analysis of a failed Detroit Diesel series 149 generator

C.D. Munro Defence Research and Development Canada Atlantic, Dockyard Laboratory (Atlantic), CFB Halifax, Bldg. D-20, PO Box 99000 Stn Forces, Halifax,

journal homepage: www.elsevier .com/locate /engfai lanal

500 C.D. Munro / Engineering Failure Analysis 35 (2013) 499507The particular generator model that failed after nine months in service consisted of 16 cylinders in a V conguration andwas equipped with four turbochargers. These turbochargers were separated, so that each was powered by the exhaust gasesfrom four cylinders. This generator was also tted with 16 pyrometers in order to monitor the temperatures of exhaust gasesemitted from each cylinder.

From installation the failed DD149 generator saw steady, though not continuous, use. Initially, operation of the generatorwas uneventful, although there were intermittent problems with the pyrometers used to measure exhaust gas temperatures.The connections of these pyrometers were prone to vibrating loose so that at times they did not give any readings at all.Although attempts were made to address the problem, repairs were not entirely successful and some loose connectionspersisted.

Six months into use of the generator operators began recording abnormal exhaust gas temperatures coming from cylinderNo. 1. The exhaust at this location was showing as 100 F hotter than usual. However, since the cylinder No. 1 pyrometerwas one that was prone to failure, operators did not trust these abnormal readings. So, although the temperature measure-ments continued to be recorded by the operators, they were not considered to be accurate and were disregarded.

Roughly three months later (or after nine months in total of generator operation) the temperature reading from cylinderNo. 1 changed again, this time abruptly dropping by almost 200 F so that it was running signicantly colder than normal.Again, these readings were not trusted and so were ignored. Just a few days later operators noticed a grinding noise andexcessive white smoke and oil leakage coming from the turbocharger serving cylinder No. 1. The generator was then shutdown, and extensive damage to the turbocharger and cylinder No. 1 components was discovered.

3. Methods and results

3.1. Visual analysis

Fig. 1. Schematic of a cylinder head from a DD149 engine [1].When the turbocharger serving cylinder No. 1 was disassembled, extensive damage to its exhaust turbine was found. Theturbine blades, which were fabricated from an aluminum alloy, were worn away due to what appeared to be impact. Severallarge pieces of metallic debris were also recovered from the exhaust side of the turbocharger housing.

Tracing the exhaust pathway upstream from the failed turbocharger, technicians began nding exhaust valve heads. Therst of which, Head A shown in Fig. 2, was found near the turbocharger housing, where the exhaust enters the turbocharger.Further up the exhaust stream another valve head, Head B of Fig. 2, was found in the exhaust manifold. By comparing these

10 mm

Head A

Head BNew Head

Fig. 2. Valve heads recovered from the exhaust pathway of DD149 generator. A new valve head is shown for comparison.

valve heads to a new, unused component (Fig. 2) it was clear that the recovered valve heads had each experienced at leasttwo chordal fractures and signicant impact damage. Additionally, on Head B a distinct region of material loss was visible.This region is shown circled in Fig. 3. It was clear that this material loss was not due to fracture, based on the areas curvedsurfaces and well-dened edges. Furthermore, this region had a cracked, black surface, suggesting signicant oxidation oc-curred at this location.

Since the failed turbocharger served cylinders Nos. 14, all of these cylinders were disassembled and closely inspected fordamage. Cylinders Nos. 24 exhibited no noticeable damage. However, several components from cylinder No. 1 were se-verely damaged. The underside of the No. 1 cylinder head, pictured in Fig. 4, showed serious impact damage. Likewise therewas signicant impact damage to the top of the piston crown, to the point that there were three holes clearly punchedthrough it. The fuel injector (not shown in Fig. 4) was also destroyed, apparently by impact damage.

In Fig. 4, the No. 1 cylinder exhaust valves are shown still in their guides. In this image one can see that three of the ex-haust valve heads (numbered 13 in the gure) were fractured from their stems. The remaining valve stem (valve 4) exhib-

Fig. 3. Recovered valve Head B detail. A region of material loss is visible in the circled area.

1

4

C.D. Munro / Engineering Failure Analysis 35 (2013) 499507 5012

3

25 mm

Fig. 4. Underside of No. 1 cylinder head. Fractured valve stems are numbered, still in their guides. The circled area highlights a region of distinct materialloss on the No. 2 valve seat insert.Valves 1 and 2 Valves 3 and 4

Fig. 5. Side view of No. 1 cylinder head showing the state of its two exhaust pathways.

ited a chordal fracture and was driven up into its seat. There was no damage to any of the valve springs that would otherwisehave affected proper valve closure.

Although three exhaust valves from the No. 1 cylinder were fractured, only two valve heads were found in the engine.There were, though, a number of large pieces of debris recovered from the exhaust side of the failed turbocharger as wellas from the crankcase. Some of these large pieces were chemically identied as being from the exhaust valves. However,these pieces were all heavily abraded and misshapen due to impact, such that their fracture surfaces retained no useablefeatures.

The four valve stems were removed from their guides and cleaned in a hot detergent solution. After cleaning, large regionsof corrosion damage were revealed on valve stems 1 and 2, near their respective fracture surfaces. By comparison, valvestems 3 and 4 exhibited no such corrosion. Perhaps related to this observation, the exhaust port serving valves 1 and 2was found to be much cleaner than that serving valves 3 and 4 (Fig. 5). As shown in the gure, the latter port was coatedwith a thick layer of incomplete combustion product.

Like the underside of the cylinder head, the exhaust valve seat inserts from the No. 1 cylinder also exhibited considerableimpact damage. However, there was a region of distinct material loss on one of the valve seat inserts. This region is showncircled in Fig. 4 on the valve seat corresponding to valve stem 2. The surface of this region was darker in colour than the restof the valve seat, and was covered in many ne cracks.

3.2. Chemical analysis

The debris recovered from the turbocharger housing and the exhaust valve heads, Head A and Head B, were cleaned andanalyzed for chemical composition using energy-dispersive X-ray spectroscopy (EDS). The heads and debris proved to besimilar in composition, corresponding to a nickel-based precipitation hardened superalloy known as Pyromet 31. Thisaustenitic alloy is often used for diesel exhaust valves and is well suited to this application [2].

The regions of material loss on Head B and the No. 2 valve seat insert were also analyzed via scanning electron microscopy

3.3. Fractography and metallography

502 C.D. Munro / Engineering Failure Analysis 35 (2013) 4995072 mm

Smearing

Shear Lip

Fig. 6. Valve stem 1 fracture surface macrograph.The fracture surfaces of the three valve stems shown in Fig. 4 were analyzed visually and via SEM. The fracture surface ofvalve stem 1, Fig. 6, was largely undamaged after fracture. On the macroscale it appeared to be a predominantly brittle frac-ture, although there were small shear lips present. Under the SEM (Fig. 7), the fracture of stem 1 proved to be generally trans-

Table 1Average concentration (wt.%) of impurity elements on guttered region of Head B, as detected by EDS.

Mg P S Ca Zn

0.8 0.8 0.5 6.9 1.7(SEM) and EDS. Both surfaces were found to be covered with a cracked oxide layer and exhibited similar surface chemistry.Present on these oxide surfaces were elements commonly found in fuels and oil additives: magnesium, phosphorous, sul-phur, calcium, and zinc. The concentrations of these elements present in the oxidized surface are given in Table 1. As shown,calcium was present in the highest concentration of the impurity elements.

Fig. 7. Electron micrograph of valve stem 1 fracture surface (450 original magnication).

C.D. Munro / Engineering Failure Analysis 35 (2013) 499507 503granular, but some intergranular fracture was apparent. There was also some secondary cracking perpendicular to the frac-ture surface, and this cracking was mostly intergranular.

The fracture surface of valve stem 2, which shared an exhaust cavity with valve stem 1, was also largely undamaged afterfracture. It also showed a very brittle fracture on the macroscale with little in the way of plastic deformation, and thereforeresembled the surface shown in Fig. 6. Unlike stem 1, however, shear lips were absent from the fracture surface. A consid-erable amount of secondary cracking was visible even under relatively low magnication (Fig. 8). Under the SEM (Fig. 9), thefracture had a strong intergranular character, and the secondary cracking appeared to be intergranular as well. Ductile tear-ing was also visible on this fracture surface, but to a lesser degree than the intergranular character. A cross-section of thefracture surface (Fig. 10) provided corroborating evidence of a predominantly intergranular fracture. It also showed thatthere was notable surface corrosion on the valve stem, and that there were many cracks beginning at areas of corrosion.These cracks, however, did not always propagate in an intergranular manner and were relatively short (

504 C.D. Munro / Engineering Failure Analysis 35 (2013) 499507apart at the centre of the valve stem. Based on this striation spacing the stemmust have been under high cyclic stress, as it isestimated that fracture would have occurred in less than ten thousand cycles after initiation. At an engine speed of 1800 rpmfracture would have therefore occurred in a matter of minutes. Unfortunately the fracture surface was otherwise too dam-aged to determine where and how the fatigue crack started, but the fractographic evidence shows that stem 3 fractured un-der cyclic loading at high stresses.

Fig. 9. Electron micrograph of valve stem 2 fracture surface (300 original magnication).

Fracture Surface Corrosion

Cracking

Fig. 10. Cross-section of valve stem 2 fracture surface.

Cracking

Corrosion

Fig. 11. Cross-section of valve stem 2, showing cracking beginning in region of surface corrosion.

C.D. Munro / Engineering Failure Analysis 35 (2013) 499507 505Fig. 12. Electron micrograph of valve stem 3 fracture surface (700 original magnication).As visible in Fig. 2, the headstem fracture surface on Head A had suffered severe impact after fracture. There were noregions of the original fracture surface that remained to be examined even by SEM. The headstem fracture surface of HeadB, as seen in Fig. 3, also experienced extensive damage, making it difcult to see any fractographic features on the macro-scale. However, there were small regions of this fracture surface that were undamaged and could be inspected via SEM. Thisrevealed that the fracture had a strong intergranular character (Fig. 13), and was therefore similar to the fracture surface ofvalve stem 2. Unfortunately, an exhaustive attempt to match fractographic features between Head B and stem 2 could not beconducted, due to time constraints.

4. Discussion

4.1. Sequence of failure

The failure of the turbocharger serving cylinders Nos. 14 was the event that alerted operators of the DD149 generator tothe wider damage. However, the metal particles recovered from the failed turbocharger were of the same alloy as the ex-haust valves, and severely damaged exhaust valve heads were found upstream in the exhaust pathway leading to the tur-bocharger. These observations suggest that the failure of the turbocharger was merely a secondary consequence offracture of the three cylinder No. 1 valve stems. Therefore, in order to understand the wider failure it is necessary to examinewhat caused the fracture of the exhaust valve stems in the rst place.

Fig. 13. Electron micrograph of Head B fracture surface (500 original magnication).

4.2. Erosioncorrosion of exhaust valve

The regions of material loss on Head B and valve seat insert No. 2 likely hold the key to the fracture of the valve stems.Properly functioning exhaust valves and seats would be expected to exhibit some minor oxidization during their operatinglife, but certainly not to the extent, and with the degree of material loss, seen on the recovered parts. This type of excessivedamage does occur though, and is referred to as valve guttering. Guttering initiates when a valve fails to seat correctly, dueto, for example, valve distortion or the interference of combustion product deposits. This creates an opening through whichhot, pressurized exhaust gases can leak, leading to material loss in the valves and valve seats via an erosivecorrosive mech-anism [3]. An example of a guttered valve taken from a DD149 engine is shown in Fig. 14. It is clear that the appearance of theburnt area in this example is very similar to the circled region of Fig. 3. The guttering of Head B, however, appears to be muchmore extensive than in the example gure, as the guttered region on Head B extends almost all the way to the valve stem.

Unmitigated exhaust valve guttering, such as evidenced on Head B, would have had several effects on the operation of theDD149 generator. First, one exhaust cavity (if only one valve was guttered) inside the affected cylinder head would be per-manently opened to the cylinder, even during the combustion portion of the power cycle. This would raise both the temper-ature of exhaust gases coming from the cylinder and the temperature of the exhaust valves sharing the affected port.Furthermore, these valve stems would be bathed in a stream of corrosive exhaust gas, leading to increased valve corrosion

506 C.D. Munro / Engineering Failure Analysis 35 (2013) 499507and perhaps impairing the mechanical properties of the valves. Finally, the guttered valve head itself would have been weak-ened due to excessive, unbalanced material loss.

The observations made during this analysis support not only the presence of valve guttering at cylinder No. 1, but alsothat this guttering occurred on valve 2. First, the pyrometer measuring the exhaust gas temperature at cylinder No. 1 showedelevated temperatures in the months leading up to the failure. Second, the exhaust port serving valves 1 and 2 was scouredclean of incomplete combustion product, indicating that it was one of these valves that was open to the combustion cham-ber. Third, the stems of valves 1 and 2 exhibited notable corrosion, while those of stems 3 and 4 did not. Finally, the onlyevidence of valve guttering on any of the valve seat inserts was found on seat insert 2, and this region exhibited the samesurface chemistry as the guttered region on Head B. Therefore, although it cannot be said for certain that valve Head B camefrom valve stem 2, the observations strongly support this conclusion.

4.3. Valve guttering and fractography

If guttering of valve 2 is acknowledged, as the observations suggest, then this helps to explain the differences in fracturesurfaces seen on the valve stems. Valve stem 2 exhibited a predominantly intergranular fracture, indicating that the grainboundaries of this stem had been weakened. Studies of Pyromet 31 valves have shown that the abnormal exposure to ex-haust gases as a result of valve guttering can promote intergranular attack of this alloy [3]. This attack has been found tooccur more readily when lubricating oils containing calcium-based, rather than magnesium-based, detergent additivesare used. Relatively high concentrations of calcium were detected on the guttered surfaces of Head B and valve seat 2, show-ing that the particularly damaging calcium-based detergents were used in the failed generator. Thus guttering of valve 2would have exposed its stem to a steady stream of exhaust gases that preferentially attacked and weakened the grain bound-aries of the alloy. So, when fracture of the stem did occur, it generally proceeded in an intergranular manner.

Valve stem 1 shared an exhaust port with valve stem 2, and so it was also subjected, albeit to a lesser degree, to the cor-rosive action of the exhaust gas. For this reason its fracture surface exhibited a slight intergranular character. Meanwhile,valves 3 and 4 did not share the affected exhaust port, and so these valves were not weakened by corrosive action. Valvestem 3 therefore only fractured after being fatigued due to thousands of cycles of high stress loading, and valve stem 4did not fracture at all when its head was driven up into its seat. In this way, the mechanism of guttering at valve 2 corre-sponds well with the manner in which fracture of the exhaust valves occurred.

5 mm

Fig. 14. Example of a guttered valve taken from a DD149 engine.

Although the fractographic evidence indicates that valve 2 was subjected to intergranular attack, the question still re-mains as to what caused the valve head to nally fracture from its stem. The fracture surface was not entirely intergranularso some loading must have been present to cause fracture. The most plausible explanation is that the normal valve closingforces, which can be signicant both in magnitude and in loading rate, caused a fracture in the weakened and unbalancedHead B. It is likely, given the degree of guttering and the necessary associated material loss, that the rst fracture to occurwas a chordal fracture of this valve head. This having occurred, one or more large chunks of material would have subse-quently fallen into the combustion chamber, to be pulverized between the piston crown and cylinder head and to cause fur-ther fracture of the exhaust valves.

4.4. Causes of valve guttering

C.D. Munro / Engineering Failure Analysis 35 (2013) 499507 507Given the degree of damage to the cylinder No. 2 valves and seat inserts it was impossible to detect any evidence as towhat initiated the guttering. However, there are a few recognized causes of exhaust valve erosioncorrosion that may beapplicable in this situation.

Valve guttering is, in simplest terms, caused by the incomplete closing of a valve against its seat. This creates a radial leak-age path through which exhaust gases can begin to ow and, in some instances, erode/corrode the surrounding material. Thecontact surfaces between valves and seats are fabricated with precise tolerances, and so anything that interferes with thesemating surfaces can lead to leakage. Gross valve distortion, either permanent or thermally-induced during operation, canunderstandably disrupt valve sealing, but more commonly the initial cause of leakage is more subtle.

Studies into the characteristics of ash deposits, or scales, on the sealing faces of valves and seats have shown that theseplay an important role in the guttering process [3]. Such scales are ubiquitous, consisting of, for example, sulphates, phos-phates and oxides of inorganic fuel and oil constituents. Not only do these scales interfere with the tight geometric t ofvalves and seats, but they also restrict heat ow out of the valves. Around 7580% of the heat absorbed by exhaust valvesexits through the contact between valve and valve seat [4]. Valves can therefore be insulated by excessive scale, raising theirtemperature and making themmore susceptible to distortion or damage. Finally, uneven aking or spalling of the scale itselfunder the closing action of valves can create sufcient localized leakage paths to initiate valve guttering.

5. Conclusions and recommendations

Based on evidence found on one of the recovered valve heads and a valve seat insert from the No. 1 cylinder, extensiveguttering of a diesel exhaust valve from a failed DD149 generator had occurred. The initial cause of guttering could not bedetermined, due to the resulting damage to the valve head and seat. However, records of engine temperatures support theexistence of guttering as they showed elevated exhaust temperatures at the affected cylinder beginning roughly six monthsinto operation of the generator. After three more months of operation the guttered valve was so weakened due to the ero-sivecorrosive action of the exhaust gas that it fractured, likely under normal valve opening and closing forces. Debris fromthis fracture travelled to different areas within the engine causing signicant secondary damage.

Two recommendations can be drawn from this failure. First, pyrometer readings should not be ignored by operators andtechnicians. It should not be assumed that a pyrometer is malfunctioning when it gives an abnormal reading. Elevated ex-haust temperatures should instead be investigated as soon as possible, as they may help to avoid catastrophic failure. Sec-ond, it may be worthwhile to avoid using calcium-containing oil detergents in this type of engine, as these have shown topromote guttering of Pyromet 31 valves. Magnesium-based detergents may be a better option, as these have led to fewerinstances of guttering in controlled engine tests [3].

References

[1] Detroit Diesel series 149 service manual, Detroit Diesel Corporation, Detroit, 1993.[2] Woldman NE, Frick JP. Woldmans engineering alloys. 9th ed. Materials Park: ASM International; 2001.[3] Scott CG, Riga AT, Hong H. The erosioncorrosion of nickel-base diesel engine exhaust valves. Wear 1995;181183:48594.[4] Lewis R, Dwyer-Joyce RS. Automotive engine valve recession. London: Professional Engineering Publishing Limited; 2002.