1、91 FTM 9,A_VRerating Damaged Naval Ship PropulsionGearsby: R. Coblenz and C. Reeves, NAVSEASYSCOMAmerican Gear Manufacturers AssociationTECHNICAL PAPERRerating Damaged Naval Ship Propulsion Gears AR. Coblenz and C. Reeves, NAVSEASYSCOMTheStatements andopinionscontainedhereinare thoseof the authorand
2、should notbe conslruedas an officialactionoropinion of the American Gear Manufacturers Association.ABSTRACT:When naval shippropulsiongearteeth break, the affectedteeth havebeen “field dressed“, so thatthe units canbe putback into service. Then the gears are rerated to a lowerrating and used reliably
3、, at least until replacement elementsbecome available. Using sketchesor computers, an estimate is made of the instantaneoustotal length of contactas thedamaged sections go through mesh and the results used as a basis of rerating the unit. Consideration is given to endreliefs, the location of the dam
4、ageon the helix and the nature of the field dressing,and actual material properties. Theresults of this analysis, togetherwith analysis of the causes of the original failure, provide limits within which the unitcan be operated with the same reliability as new.Copyright 1991American Gear Manufacturer
5、s Association1500 King Street, Suite 201Alexandria, Virginia, 22314October, 1991ISBN: 1-55589-606-5Rerating Damaged Naval Ship Propulsion GearsRobert Coblenz, Mechanical EngineerCharles Reeves, Director, Line Shaft Equipment DivisionNaval Sea Systems Command, Washington, D.C.SCOPE an example of dama
6、ged teeth, after cleanup,It happens on occassion, that gear with this characteristic. This example wasteeth become damaged. Accidents can damage taken from a unit which was operating slowlya gear not in service. In service, damage at the time, so only one group of teeth washappens for one of two bro
7、ad reasons: faulty damaged before the problem was discoveredmanufacture, or a mismatch between the and the unit stopped. If the unit isservice conditions and the design. The operating at high speed when the material“mismatch“ can be foreign material passing passes through, it is likely that the dama
8、gethrough mesh, a blunder in the design, or an to the pinion will cause damage to eachunforseen load, alignment or environmental tooth of the gear meshing with the d_magedcondition. This paper will deal with pinion tooth, and vice versa. This willrerating a gear unit after tooth breakage quickly lea
9、d to uniform damage to all teethdue to overload, in a hunting tooth design.TYPICAL NAVAL SHIP PROPULSION GEARS IMMEDIATE ACTIONSNaval propulsion gears cover a wide After the damage is discovered, initialrange of sizes and configurations. The activity usually focuses on determining thetechniques disc
10、ussed in this paper have been causes of failure, and salvaging the damagedapplied to wide face gears, with pinions one gear. Particularly in the case of foreignto two feet in diameter, double helical, and material causes, there may be significanta length to diameter ratio up to 2. capacity left in t
11、he gear. In fact, someMaterials include steel from 200 BHN to minor damage is possible without affectingcarburized surfaces, capacity. It is necessary therefor toassess the stresses in the damaged gear andDESCRIPTION OF DAMAGE possibly rerate the unit. Then theOverload failures fall into two broad e
12、quipment owner can decide whether to usegroups: failure of all or nearly all teeth the gears in the damaged condition orin essentially the same places, or failure replace them with undamaged gears, and whatof just a few teeth in significantly use can be made of the damaged unit untildifferent places
13、. Errors in the original replacement occurs.rating result in all teeth failings_m_larly. Many systematic errors inmanufacture, such as accuracy errors or When bending failures occur, US Navymaterial properties, also result in uniform experience is that a piece of tooth isfailure around the element.
14、Some released from the bulk. We are aware thatmanufacturing errors result in many adjacent an entire tooth can come off in one piece,teeth failing in a similar manner; for but we have not experienced this. Perhapsexample, when service loads cause an element this is due to the large helix angles,to d
15、istort non-uniformly, such as typically 30 to 45 degrees, and large lengthprogressive failure of a fabrication weld. to diameter ratios, often 1.5 to 2.0, usedin Navy gears which result in a relativelyHowever, when foreign material passes short instantaneous contact lines, asthrough mesh, adjacent t
16、eeth can be damamged compared to the face width.in somewhat different places. Figure 1 isAconfined to the axial extent of the damag_Field dressing damaged gear teeth on the most damaged tooth. However, fo|usually involves grinding or stoning off any helical gears with perhaps five or more“proud“ met
17、al and removing any crack teeth in contact, this approach will beindications, if any, so as not to leave any overly conservative if only one tooth isstress risers or local high spots which will damaged or not conservative enough if damagebe subject to increased contact load when to several teeth occ
18、urs in one axial strip,put back into service. Figure 1 shows the with all damaged sections passing throughsmoothed surfaces left after such a cleanup, mesh at the same time. Also, in some cases,In extreme cases it may be necessary to the axial location of the damage band may becompletely machine off
19、 a section of tooth significant to the value of Lmin, theflank from every tooth around the gear, minimum instantaneous total length ofespecially if rotor balance is critical. In contact, after damage. As discussed later,addition to smoothing the fracture surface Lmin forms the basis for rerating the
20、 gearand associated upset metal, an area of tooth set.flank adjacent to the damaged edge will berelieved slightly to restore the end relief In order to estimate the length ofnecessary in large face width gears to avoid contact with damaged elements, one can, byend loading, hand, seek the position of
21、 mesh for whichcontact length is minimum using a sketch ofAfter cleaning up the damaged teeth, the damaged teeth and an overlay of the zonethe results are measured in place and of action. With a little practice, thisdocumented for analysis. Due to the need to method works well. However, it is laborg
22、et the equipment back into servicequickly, intensive, subject to error, and very timeand the high cost of removing the gears from consuming, especially if one wishes to do athe ship for more accurate results, these sensitivity analysis of the damage estimate.measurements are typically done with the
23、aid Moreover, if both meshing elements areof hand held scales and reported with an damaged and of hunting tooth design, thenaccuracy of perhaps plus or minus 1/16 inch. several cases are necessary to study allAdditional information is gathered from the possible combinations of mesh conditions.operat
24、ors and operating logs in order tounderstand the conditions surrounding the Since the damage often happens to afailure, and samples of pieces broken off ship in service, or very soon to be i_are used for analysis of the cause of service, assessing the stresses in th_failure. Casts of the damaged are
25、a are damaged gear is a matter of great urgency_often used to record the damage or final It therefore has become natural to considerconfiguration, or both. using a computer to expedite the analysis.If foreign matter can be identified, DESCRIPTION OF THE COMPUTER PROGRAMthe unit will likely be put ba
26、ck into A computer program was written in GWservice based on the assumption that this is BASIC to study the effects of damage on thethe only cause, and a follow up total length of contact. The programinvestigation will be persued as calculates the instantaneous total length ofconfirmation only. Othe
27、rwise, a complete contact, L*, that is, the sum of the contactfailure investigation will be done before lines on all teeth, or parts of teeth, inattempting to rerate to the full remaining the zone of action at the instant beingcapability of the unit. analyzed. Referring to Figure 3, showing adamaged
28、 tooth, lines AC and DF are the linesTOTAL LENGTH OF CONTACT of contact on that tooth at some instant.Rerating the unit after damage is based Without damage the contact line would extendon an examination of the total length of continuously from point “A“ to point “F“.contact, L. The quickest method
29、would be to Starting at the position of mesh just beforecalculate the stress as if all the teeth are the damaged area enters the zone of action,damaged exactly like the worst tooth, both and stepping through equal fractions of basein length of tooth lost and axial location pitch, the program calcula
30、tes L* for eachof the damage. If one further simplifies step change of mesh position, correspondingthe problem to ignore the effects of the essentially to changes in time. A graphicalactual axial location of damage, a display of the contact lines at each step isrecalculation with a simple reduction
31、of shown, with damaged segments not in contactface width yields the result. This is shown in contrasting color. An output dataadequate for wide face gears and when the file is made of the positions of mesh, anddamage is in fact in a circumferential band the corresponding L*, for later study.pattern,
32、 or if the damage to all teeth isASAMPLE RESULTS predictions to those calculated using theFigure 2 shows the results of equations of Niemann and Winter given inpredictions of L* for three conditions of a reference I. Figure 4 also shows the sameparticular gear mesh. Table 1 gives the mesh with one s
33、ection of one tooth gone.pertinent geometry data for this example. Here, the same formulae were used, and theThe first condition is without damage. The results reduced by the length of contact onother two conditions are both with a the missing section.circumferential band of teeth 0.4 inches Figure
34、5 shows the effects of damage aswide damaged. In one case the damaged band shown on figure i. Table 2 gives thewas modelled to start 0.32 inches from the geometry data for this example.end of the helix, and in the other case thedamaged band was modelled to start 1.26inches from the end of the helix.
35、 Note that TABLE 2more knees, or changes of slope, show up in Pinion OD 19.153Figure 2 than really exist. This is because Pinion PD 18.5751the positions of mesh in the analysis did Gear OD 158.973not correspond exactly to the positions mesh Gear PD 158.3949producing the real knees, where some tooth
36、Pitch helix angle 31.2674just begins or ends meshing. Thus, too Normal pressure angle 17.5large a step will cause the analysis to miss Normal diametral pitch 3.464the point of Lmin. Face width 14.25TABLE 1 Once the length of contact line is(dimensions in inches or degrees) known as a function of rot
37、ation, it can beused to recalculate the appropriate toothPinion OD 15.134 stress. This can be done in several ways.Pinion PD 14.7526 One can simply use the minimum lengthGear OD 39.771 calculated after damage the way the minimumGear PD 39.2474 length was used in the original ratingPitch helix angle
38、27.4939 calculations. This is reasonable where theNormal pressure angle 17.5 original rating was actually based on Lmin,Normal diametral pitch 7.734 as AGMA does. However, rating for US NavalFace width 4.05 ship propulsion gears is based on theaverage total length of contact, La_. Usingthe following
39、 formula, this is the timeDamage is considered to be bounded by average, or the average over one base pitch.transverse planes. This ignores thepossible contact between the assumed planes L_vg=F*Z/P_and the actual boundary of the removed orrelieved area, as shown in Figure 3. The whereresulting appro
40、ximation is conservative, F=active or effective face widthrecognizes the approximate nature of damage Z=length of line of actionmeasurements, and greatly simplifies input. P_=normal base pitchThe precision of the definition of damagewill limit the precision of the analysis. In order to be consistent
41、 with the standardHowever, we have not attempted to US Navy rating approach, using thecharacterize the fracture surface associated allowables, we could use anmathematically, since the surface is hand average length. This can be averaged overworked as mentioned above and since the one base pitch, sin
42、ce it is the basis forcomplex nature of gear tooth and bulk rotor the undamaged calculation. Another approachdeflection makes precise description of the would be to average over the actual timeend of damage illusory. Also, we have used that the profile is loaded, i.e., the lengththe idea of buttress
43、ing to take credit for of the line of action. Observing thestrengthening bytooth material remaining in effects of damage on the shape of the L*the damaged area when damage was only on the curve, as well as on the magnitude, led ustooth surface. In that case, repair didnt to consider the literature o
44、n the effects ofacually remove the entire thickness of random and complex loads. This is not totooth, and the length of damage in the model imply that the occurrence of damage suddenlywas slightly reduced, causes the relationship between load andstress in gear teeth to be less determinate,Figure 4 s
45、hows the results of using the but simply to acknowledge that in marineTable 2 gear set, comparing the computer gears, and particularly in combatant ships,d-the load profile is widely variable and Evaluation of the case history an_mostly unpredictable, even if the peak load metallurgical factors will
46、 help determine ifis reasonably well bounded. This is due to the gear failed due to a one time event orthe variable nature of Naval ship commanding a condition likely to cause a recurrence.officers, the widely different tempos of Lastly, knowledge of the long term and shortoperation that occur from
47、times of relative term plans for the ship in which the gear isquiet to relative crisis, and, sometimes, to operate will be necessary to establish ajust plain luck. new rating with the appropriate life.Tooth load distribution is critical onThere is evidence in the literature wide face marine reductio
48、n gears, and a morethat fatigue failure under random loads is accurate prediction of local stresses as theindeed dependent on both the peak and the damaged area goes through mesh shouldmean stress. Topper, et al, in reference 2 account for this. We have not calculatedshows the effect of mean stress
49、on fatigue the effects of damage on load distribution,failure, trusting relief at the ends of damage toavoid end loading.In summary, the new rating is developedbased on the ratio of a total length of EXPERIENCE TO DATEcontact as designed to a total length of The US navy has been very sucessful incontact after damage. Three ratios were establishing operating limits for damagedconsidered: propulsion gears which preserve theirreliability. Relatively few have actually- Lmin(after)/Lmin(design) operated restricted as a result of thedamage, with some continuing to operate- Lmin(after)/Lavg(de
