UNCLASSIFIED

AD 438713

DEFENSE DOCUMENTATION CENTERFOH

SCIENTIFIC AND TECHNICAL INFORMATION

CAMERON SIATION. ALEXANDRIA. VIRGINIA

UNCLASSIFITED

IL

BestAvai~lable

Copy

NOTICIE: When goverzent or other drawLnp, speci-fications or other data are used for any purposeother than in connection vith a d,f.into1y reatedgoverment procurmnt operation, the U. S.Governsent thereby incurs no respsibllity, nor anyobligtion 'batsoever; and the fact that the GOvern-

sent U.7 have foualated, furnished, or in any vaysupplied the said drawings, specifications, or otherdata is not to be regrded by implication or other-wise as in any mnner licensing the holder or amyother person or corporation, or conveying any rigtsor pereisaion to manufacture, use or sell anypatented invention that xV in any vay be relatedtherer.o.

Rockets, Motor CasesTitanium FabricationTitanium Alloys

Final Report On

Research and Development of TitaniumRocket Motor Case

Volume U - Ring-Rolling and Flow-TurningCylindrical Fections

Written by: Apprfveb

October 31, 1963z

0

z

Contract No. DA-19-020-ORD-5230OCO R&D Branch Project No. TB4-004

Department of the Army Project No. 5B93-32-004

TECHNICAL REPORT NO. WAL 766. 2/1-14

Qualified requeetors may obtain copies of this report from theArmed Services Technical Information Agency, Arlington HallStation, Arlington 12, Virginia

Pratt & Whitney Aircraft o uv N AL

EAST H A R T F 0 D CON N ECTICUT

COPY NOILI_

• m • •4

IPRATT a wwTmY AIRcAaT PWA- 2267

FOREWORD

This final technical report was prepared by the Pratt &Whitney Aircraft Division of United Aircraft Corpora-tion in compliance with Army Contract No. DA-19-020-ORD-5230. It covers technical accomplishments on aprogram of research and development of titaniumrocket motor cases sponsored by the Army MaterialsResearch Agency, Watertown Arsenal, Watertown,

Massachusetts, under the technical supervision ofMr. S. V. Arnold.

IPAQ( N

PSTr a VWwMff MMC"T PWA-2Z67

ABSTRACT

Prior to this program, Pratt & Whitney Aircraft fabricated 40-inchdiameter rocket motor cases from B-120 VCA titanium alloy and formedthe center sections of these cases byflow-turning. Although these cyl-inders had adequate strength properties and satisfactory hydrotest per-formance, dimensional control during flow-turning was inadequate tomeet the required tolerances. A large amount of diametral growth andlocal bulging occurred during flow turning. The subject program wasundertaken to develop a flow-turning practice to produce cylinders whichwould meet the dimensional as well as improved strength requirements.

The, ring rolling technique and heat treatment used before flow-turningwere investigated to improve ,he quality of the flow turning blanks. Asolution treatment was developed which considerably improved the duc-tility and structure of the starting blanks.

A major improvement in flow-turning technique was developed withcubscale parts. The effects of roller geometry, mandrel speed, wallthickness reduction per pas3, and roller-feed rate on radial growth andlocal bulging were determined. It was found that roller geometry hadthe most predominant effect on dimensional control and a new roller de-sign with parabjolic contours was developed which gave positive controlof diametral growth. A correlation of flow-turning parameters wasdeveloped which permitted techniques developed with subscale cylindersto be directly applied to full scale parts. The resulting full scale flow-turned cylinde-,ts had very little diametral growth and no local bulging.

In addition to the improvement in dimensional control, processing ofthe flow-turned cylinders was investigated to improve material prop-erties. A stress relief treatment was developed which gave 50 to 60per cent reduction in residual stress. Sizing of the cylinders was ac-compii.shed after stress rolief and was follow~d by aging to the desiredstrength level. Cylinders fabricattd by the developed techniques werecapable of being aged to a minimum circumferential yield strength ofZOO, 000 psi with 4 ler cent minimum elongation and adequate toughness.

Burst testing has demonstrated that flow-turned cylinders processedby the techniques developed (minimum circumferential yield strengthof 200,000 psi) exhibit excellent performance under hydrostatic loading.Results are completely described in Volume IV, Fabrication and Test-ing of Full Scale Components.

PAOC NO. iii

1

j tkSAI a W.4r"Y *Icks*P PWA- 2267

TABLE OF CONTENTS

f PageVolume I - Forging of Front and Rear Closures

Volume Ii - Ring-Rolling and Flow-TurningCylirxical Sections

1 Foreword iiAbstract iiiList of Figures viList of Tables viii

I. Introduction 1A. Purpose and Scope of Project IB. Background 1

II. Evaluation of Material Flow-TurnedUnder Contract RM-962A. Preferred Orientation of Flov -Turned Material 3B. Fracture Toughness Methods .4

III. Ring Rolling 7A. Subscale 14-Inch Diameter Rings 7B. Full Scale 40-Inch Diameter Rings 12

IV. Initial Flow-Turning Development 15A. Flow-Turning Parameters 15B. Flow-Turning of Subscale and Fcl

Scale Cylinders 16C. Problems Requiring Further Development 18

I V. Subscale Flow-Turning Development 20A. Background 20

I B. Evaluation of a Revised Roller "onfiguration 20"- C. Development with 14-Inch Dialtteter Roll-Forged

Rings 24D. Evaluation of Subscale Cylinders 32

VI. Conclusions 49

I PAE NO. iv1

PRATT WK!'E fy AIJRCOAT PWA-2267

TABLE OF CONTENTS (Cont'd)

VI. Effect of Hydrogen Content on Mechanical Propertiesof Cold Worked B-120 VCA Titanium Alloy 39A. Inoculation Technique 39B. Test Results From Cold-Rolled Sheet and

Flow- Turned Stock 39C. Recommended Acceptance Specification for

Hydrogen Content in B- 120 VCA Titanium Alloy 42

VII. Full Scale Flow-Turning Development 43

A. Flow-Turning Practice 43

B. Aging Response of Burst-Test Components 46

C. Sizing 47

D. Tensile Property Uniformity of Cylinders F-2,F-4, and F-7 after Hydrostatic Testing 47

Appendix A - Tables 51

Appendix B - Figures 112

Volume III - Development of Welding Practice

Volume IV - Fabrication and Testing of Full ScaleComponents

Volume V - Evaluation of Candidate Titanium Alloy

PA09 NO. V

-ma w " Mac"" ai m i w vP W- Z 7'

LIST OF -FIGURESi

Polo. Cuolood Irom Twi-Io 4 1qorl P.00.0 I). "til.j SoePPI. After Vorloo,. Ho.t T"Otroot .. d Al..

*000400pkicfPojittitOA Aflt S&qo.46iasl. .1 90SF

110 1 Pol. Fig.. of 11.230 V CA TilooooAlJ-Y O-1. as IFrlorO ... of Notched.1K, * 0) ipoo. fronm

'iNbl A-.4.l A-4.l 40.10.1. thso.vIl.FosdIo ldIIW. DotolioOto .1.1. Effeof Hear.0 Tr.loo-Jo ~ ~ ~ ~ ~tij rolo Wl.o 10 VCA T41A.Ion Alloy. Cold. 1:~lP:pei.6,0 o~o.T.odC

1.14(9 3401.. d Ap.4o At 00 1 o, 30 W.. 1 g1&C"I 01.00o0 itw l- TvrlC1

4- 1ui11P.&. Sr-of 010 VCA Tilol.cii Alloy. Flo-. I.M.. it..o1o~~ $50F (or 30 3410o10 .. dAg.d

T-4.. ($40 111.4.cil..) &.4 Aged A% 050 ;tI.. 30 kdb.. At Nor7Wo Odwolo To.. Tfon. ~ 0.1.1d. Po-11o- Of -he C3 1.

-~k IS rlooe.Thoo llqofeo for F*1rI.t.i;R Cyli.Ait0.l C.o.Ior

11 1 ool Pol. Fit. a1 8.120 VCA Titst.vvt' Alloy, Flow.T..,oo (SOS lodooo 1od AprioA iso r3 Fo so0 3041.- 3.. £01010. -Fo.ol Poll.. CooI~oo00looo (Or

r-ot. M0.1o T.k1 Vfoo a.0 I.114. Po-11ow 010. CyI. Is-0 IVoCA TlonAlloy

I.A W000114q. Loft.1 of.11 Sool. 40.TklolXAO I.11

Mode lloolg 17. Igeal Ot~,lO 0.04 VCA 'Tilaoiv-% Cyliad., WIol Failed Dir.&0 to. olok).

I, Cold.1.doo.r 01.VCA 1.oo loy1.6 0 oolao.

7 AST~g Sta.Alo04 3,0.01 * 7 O .flo-io Used for T..g 3 10.14. N$.c. Of F01 3'.lo 4O.oIh lIX-0ilo 1112o

Flo-T'.oo.. 1.l0 VCA TO-o.on AlloY Cyllod.fis VrA .0.m Cylilol WAIT%. V411-4 Mol 1 t hTW

6. Toollo re p ol.... 11 Co k~00 o0 .oyed I. LooloolloOf FlowTo,.4 N.120 VCA Tlumlsm Cyll.4.00- 3 Fl...T.r If... heoli Wih Corirol Roller Co....

o,.ti. T."hee.000(00).4 vlie o..,o 3.110 30 - p.Ortlo go, lnoo.Tui.l. i8.10 VC.o T ko1. 4.

VCA P.rhll.g Fo.o.4Cylto.0lookev lWltMool Cyliedirs bld. ivni. m011.4 .. d W.W44

As1 ASrOO 3.0.04t Z x..1 R Ih N sperii.. U..4 (or I. 1 wT c16 -16W l14li ulo o~~Tog moe 7010 of Iow.Terod b4.10.1.1 .1 slow. .. o ool ,0Poool ol.Cooo

it A 1I 903, 14lock C (16 Spl... U.11 Nf. F~l~tt41 it Typloo k11#4 .0.a . Cyliod.,v Aller Fiver Flo.

Tough-.l.l . 1 T14Mra Toolo.. eo.;1'4 Co-dllom of l..ido SAojlooo vid

Yougho* * o Flo.-TwoA Moloril 4, SbclollahI.mo -10VATul.Fo

it 300300.4 CIN00p, Impo Evelsy Abslpllon - Tool lOk0 Wirer0 p..

fo PM. 1-Tr. ClISr1 S.he"%o. 14.101 Dis-tiof B.120 V.CA lti0.loo Flo-.1: Te0~o0lly110, ~o.0.~ Fow.TOO.4T,.od4Cylio.0rshOi'O T... "oWide Sool.. Allor14 34.0 CM0.py I..l~oo Iorogy AI..oopli.. Yield ii.. Fimo P..

IS l.T0lfDhoe of 14.lool Diore.loo Rolled -ad- 5.16.d 131.0

IS Uo~f~ CIA, .10)y 10011 Ofy Ah-lollo. -04 Y1114 54"sooeoo.o1. 0oon-t" 1.1,10 VCA T;Aelon Roll.

Seo.~.ol1 'pr l1-owowod Cyll~dOOO T.#1.4 .1 lOY -4 Fo-,go4 91-k N-tooS

Slook 1310.1.0p. 120 VCA lolsolom Alloy14S Wcooetoo.0. of Sob.ool. 14.1.,) Vu-tnolw lld VotivedC1i.6-v No.1r07 Alt', Fno.1 Fl-Uw.T0 P..l

Riof No..10 $ 30..0l.Tro. 0141.37 (or IS W-t0.0.Ii Doolof r..log0-A 1.lr .. io 30 Froeoen..looKh EX-boloo 8110 VCA'Tllo.Aoo Alloyit Aalo. cw;n0 (0 .l044l4.1.neo 14

Feojod Cylio4.0 N4..revor Afler 30(0014 It*o.To00 Net.

N-.oo 1.A. A., I olldy 1.t.1 OO. 000e. 11 F1Oew.ooolgof 14.boo1. Dirnew L.1to V"7 'OU-Iur4.401lcr. A -. 4 Aol* 14Sbr .. A Alod 0030 Alk-13 Cyllofoo 0).1.4 3,.4ftu4 u-ic1000o34004000 1.11.,

10 ~~ ~ ~ 4 Prolplole Slod A.A3.1,01. 3.40-110 1ef11 p#,p.to 1"14 040b X-v lol 0o 4 1.4.4..h l)..nster 0,140 VCA ioc.On AllT7 rotfto*

4lC1 ool~to lloowma .. l 01.10 mit e14l ThoogO1. p,04ipe1010"y -~d R.130VCA Pl*SC% t AdtikA014. htl .1 CIf utr.rw*.14.$c I4lktooloo

Alloy ololo.dCoo.

ArtOt Iko~l~g A0000 foor o lcw., $1o~.ly ColIoog. 41 Aqiel qop. .1 Al 4.7o1h .- .enl 0310 VCA

"043000,01000 00 (or floo 10r. F.41."4d by 4 W..,, ..r,mur Alloy 030.-I104 .00, WSM&ttood 30 1,1it w.Teoo0,

PWA-ii67

LIST OF FIGURES (Continued)

NM _ Ill* N.1.o ml.

4) Agin.g looP-.. . It .4.3.03 Ditotor ne.Tt,,,..jC 1 . S9 Aging lt.spo.. of 14.loch Pingol. h N-,..h, 1011-4. P"o1io.ly S1---...llov~at .3300, 9109kC, 0l.Tr .Si...H-1.4 (6SOF 1112) ACI .0d0S 3.01A Aged .1 1000

44 A&Co R..p-..0 .1 9.4.lath D...,o3.o 31.303 VCA 00 Tactor. T-g.h.i.. 3,) ta Yild Strength to, no,..Tita.i... Alloy Fla.-.Tred. Cylindr. St.... Atli ... 4 rood ISO% A.4.ootlo, Each P .. ) .0d Aged 5.120 VCAa33M0I. 910' (j,0on. It..o 'T il..loAlloy 0.07S lech TI-bk

4$ 0, llof 11.$4.lnch -. 4 40.,% EX..,r M,. 4I Ro..lomahip of Ao-oo . Ilydivil.. lol.docd toTamed4 Cylinde.. FA .,cliom of Float T-noot Pmt:*..iig Tin. (A, Cathodit Ilydr,.lo .Io Cold.

Rolled4 Shoot Stoch

IS-120 VCA Tit-ii, Alloy Cylinder A .-.-led t 14009 3.0 Samooth .. 4 Nolohodl (K,031 T-111.o Pr'operties. '...I*1-1S M10~ o Aa4 Av,4 lO~oo_4 900' T..k.r.Or o Cl.o1o 0%Rool, .4AoShet.. Stock With IHydroi., Coo.o,. or 70 .o4 000 pp.

4? Aging R..pom.. p4 .w 9.4loc .....io dlwToo3.300 VCA Tooal-, Alloy Cvrtsd., A..11-4 .4 34000' 61 Sooooh 0.d Notched0 (l,-$) In.di. And S~~4Notch..)

P,~ 1641-0o.. .- 4 Aged .t 601.4 NA 3 (30j.0l* Tensle. Pro.rtiva of Cold.3,lloc4 And Aged ShootStock, With 70 .. 4 Z00 ipoot .f Hydro..

40 Aging ;feeo-.. .f1.4.1a0ck .m~~ b,.Tr.30.300 VCA Tit.liert Alloy Cyli~d.r Ar,-..4 At 31009 4,4 Notched Tomtail. And Sustaineod Notchod Tensile Styngth.(Ar S -dtot.4 Aged at $00 -04 9009 of Ccl4.,Xol3.4 and Aged Sheet3 Stoch W,th T0.04 000 pp.,

41 Alt.& lioopo. of 14.1,3,Dt.,h B-110 Vt,% lit-oi.,. 3y.sfoi Voolo.. Str... Cotocooroti. Factors IK,)Alloy Flo,..T,.o4 Cylinder N~noi~. I 54o... Iioil 61 A.1.1 Tensile Pr.operties . Aging Tirm. (9003') tor Firstat390SF for 0.,. R..', sbh-ca. 14.3.oh ihta or FI.0.Tor.. Cylioder Vac...

-10 Axial .:--P.. .3 14.3..h M~arriott, 5-120 VCA TitesooA .. le 114O ooot 1-frslg

A13.j Flo.To.od Cylinder N-b.3., 0 Sir... Jlli,.4 44 Axial Utlo Properties so Agi.3 Ti,,.. for the 0Se.ondAt "OF0 Io~ ( .,,e Suksca3. 34.3.03, 1)3..,., Cylinder Vacu,n,,Amata~l.

At oo 1409d. Me.OobliaoTroat4 at 1m00r Pro o,$I ACM4 Rasp~. .1 14.3,ch Di-.0,~ B.10 VCA 111-1o-,, n-..dj

Alloy n-IoT.,. Cylinder N.H-br 1 Site.. lt.ood..atloor to, On. Mt. hi Snoolh And Notokod IK,.33 Te..il. Properties.v .t

Toiporalacs. . 0ow.Twine. 310% Rodaclm to .0.070.$0 Aging 3..po.. .1 34.h0 Dao.w. 31.30 VCA Tilao.t Inch Thid....). an4 Aged 14.-,0th Sh~omoior C0 ,a.orAl'o.y l..Ter.4 Cylinders. N-oh.r. 1, : .4 1 S,.*. With Htydroen Cm..t .170 a04000 pp..,

mell.va .or f0 or 0.. floorn-FloTxg.1.g %4tal.1th 3'.raholi C-i,.r

$3 AjS," .-pone, .1 14.13 Dia.300 3'bo.,.4Cylinder, Nurabor 4 Sir~. Xoli.,,.4 At $1OF for 30 419 llciy-l-ch M-t-wr~ I3130 VCA Tita~iot CylinderM..3.. Allt P#.Tn~~ '.pir.y 30 Tr~ming

$4 Aging Rty... .1 14 3.nch D-.1.1~ 3-10 VCA Tll.',ion 10 Tot... s roperties .1 l Io. 40.1.

LIST OF TABLES

______ 733.I'll, N.,.bo, 7333tP4.

'3 7T.1* 01330104 . B-IZOYCA I3I... 3 zi P.;.o,..;. 4. 0.3oAlloy 03.33 -4 rl-wT...d14Cyl..., U 0 oSoOl.l 9.4.3.31. 14.34..3, Ae.Ful

Sc.1. 40-1"h3 13....3. CYl1.M; 10

-d1 FI-oo....o Cy711... 453.1201VCA 00 Flo.-U..31g f.,.n03.,..447T..itol Alloy7 .4th R..P.c3 3o Ifirecit. of Ubscal.3 9.43.343, D4Name., Cylindr& 7)-d. Thiol.... s%

3 Fr&;.o r,;.. (ol3 T.om Rete~ll. 03 0S... 14.1h0 Di-ool~r CYII.4;. 14B110 VCAT7I-k- Alloy P.;.bi-& Flo..7,..4 Cy.d.4 N-to;. I .440 Z04 3321 .3 I; .;33oo F.11305-3. 40.l1"h

txo~rnow R.11.4 0433 U.34 It3.lo ..

11.120 VCA Tit.l..o Afiot Fno.To...3 0... ... OoO4.4 isNrshiM14 Cy31.do, N-tbor 0 03.o.1g Effect

.f3 )poolmo St.. st 00 73.. propoll.. i3 4.4-.3C4 Dx-..o3.C;i1.d.;o A..33ovw ToM-4 -4 After 0,1.c3

)A4..31. Ch.;py lopoom .. R...1. 0. 40. Aging at 400 to 95F 14tech4 3)340.3. YFo,.Tu-d Cyl1o4rsN-t- I -A .447..1.4, At 41 to4003F 04 4 U 9..3 ~pr3 . 4-4.3c D,-.oo.r

nl..,. Cyld.r. Afte, Sr,...bw0.,43.30304.11 0 o .. W.l. 3.... 40. 0.lr~ G$O 0).30903' -4 Agl..g -3

Oloolb'. 1-40 Z 1z? S-th33 -44 N.VC1.4 T-011' PNo3..,31. -C

1 lto...)5.~y.4 for Rolling So3..l. 03333*34. 34-..3 Kh..mr. 11o30 VCA14.1.04 h.,..3.t 03033 so 0'Ioo.r... Cyliodor N..ob., 4 After

V.10. -A..0.3oog. Att1 11-.t

Ok0... No.1 . Zl. -4 1 P.11.4 At3.Adi34 aad 141... led3. (14S011311300) S9 to 03... al. .. for 1.43.101 11... 1

I iosu. P'rope.rties of 14.313h Dtametr lw.lo, yl~r.7Iti3.',~40.g. N-..., I. 3 1 .. 3 1 0)I tar..o3. -..0.4..1 0....

Afte, A.1...loi-. a403 I ,; 30r ED M .. 6.0 Flo.73.od Cyi.4.r. 40

10 T-..3 PaoM~p.o .1 b.t.3. 14*34.4 30 Yeasilo 3'sopoollo at 1.4.3..4 D-1o313...3. .3.A. mi.&. No.00.. 4. 5 lloo.T.,..d Cy3lk..,. A ... 1.4 41 1400 to

-d 4, After M-All-I 1.14003' (or IS Ui..31 . 43 3003' -. 4 Aged At00300 03 to"3

at Cuo..sI.;.il. 7-03. 'or3. 03 3t A... 7.l. Poop.;:t;. of StS A.. 34.

34.3.013 D,...., Itoll.3'og0 Meg. Noo.3.3.4D..;. Flo..7;... 4 Cyliod..I. . .4 -%Aft., A ... Ilkeg It334303 .-A lNabA6 3, Z. .. 401 S,.....3 ... 4 a.3A3 . 3 .3 4003' 40 3000 3, 0.. foo.; Am.4 0t 37003 003 "Or

30 . I0331. Sj..o.L. S..I.. 14.h1 30 T ... It PPnp..3 of S.41..; 134.tot

I. £.o43 33;..0;3 .1 900 (a.is L.41.k 7o. . Vp.3. of 5.3.1.1. 34. Coe I304. A^4 Aird .3)03 00

3.010.3.; 3003131.4 O,.g .s 1 .ti..1

IN-t" 11; -T 3o....;A Cyli.d.,. Ner..o1.34 't ...3i; P~opn.. of r.31 5,I11 .. 44.4Al oo3. ,04l30003t

M).333 .g. S..opl. After V.31.. 3 33..4 1400r4.- Ax~r .1 "OF03 64

)4 1

o.].- 3';o3..463. If k.0.3 4.3aIS 7...OP".3m.. of Toll $,,I. 40.3..13 30413; o..3,.,-4 C1 3.4. N-..oI.I)o R0.r smt5.Pl.. Alt.; V.1%... 33.. 4.1.r So to...Ro.o.g . I003 100WE'

Tr. Ao3 .3.f..qt All.# 434003' 44 -. 4 Axle,% a31000 1003 "OFAl

34 R.11tal SN0.40. lot E3011 F3.11 Sc:!. 40. 30 7-0o33 3';43*;31. of S"K4..3 34.3.0

3.31. :3.. 9,3.4. 133-e I3o.. 33...trodCy3..4.; No.,.4.4 Mier, .453.5.t 40cr 401

th3.0;4 M1.A. RoIlld by L.4s.3.. 5.3.31... 3 S-11,oI- X.4 Nc.4 7;..133 3iol..7;omo.4 As 14501' for 0 U33.. .3. Aa.3.1t

C33.4.bo4 of Sub-c.. It 3.' D,41w.3 h14.0 VCATa-i-.. Allo 03o.lT,A Cyli.,

30 dw L.17.3. 3',opol.0 of e.o31 s0.3. 40. X-o 4 A

3.4 03.. o.1 3004 mA 3Iua. T4.3.3I1I.. .m0.0. 3-.1 40U3300.3043 611 D-to0 Cy33.o3g N...1..I 0,3411..

19 AMAI. 141611- .Fmo~r1. .of 13043. 9.4' 343 1....r.o0Oo.3 0... ro Cytt.&. 11 -174ft.41..

I 3'41.0 (SO%. 004.431. 3'. PA..) 70 00 7..33. I'fop.;33o 0334.4.133)34,003.Cytmt4, H-bo4. 0 n- .... 4 by .400 1-1.3 I';pto1k. of 40.3.013 3)3.oor KIMl- Pt;.... 1oo.-4- To3.. 14e(0% .0443..

I'a.34 Kili, lt.5.3,334 I ... A3 MOEO Peor Net).. 1.... .3 .i. Vo3' for Oo.w 1 41 3'1;o14"410 4%-to, Cuocrod it fllf Ica, . 4 A.4 040AM0A,

PAO& N.Viii

PWA- 2267

LIST OF TABLES (Continued)

rw N-b,

loCh .umo.Ir Cyli,.d., N-.., 9 ria.. o..o.. 9-.Y 5luo M .. d.Oi'd .. No0ckodT.-4 by 16. l',...o T-1-iii. Toircio Ir,, Te.,i. Ipcoo ito. 11-... n4

-1 OT I., 0-..ll I.., -A4 AIa -t SOOF 90 ~yid ~10140 T .. it. 11,oprliiii ot 14.1cb D-1.1 R-. of SW o.4d 1... Coods.c.4

Cylivid., N~ookr 10 lloo.Tr4 by th. aA Notclod T ... ycir. o 1'7- Ilt.-P. .. . . .lo SMite" SOOT4 -o4 AS.. 100111o2lAr Cy4.th

Hill H..? ..04 AS.4 .1 SOOF .01

at lVU.e*.Tk.. of Q,O70.boo Thick at INS.otio S,4.r 1"k D- ir DonI..anl..TYo ...4 ..04 Agri! 11-110 VCA Toit1f-l Ss.4.~k os.. plrdr 0Alloy 14.1.1, 15..;.., Cyli.4.,. 00- Ss oc M1. . A&Zici oo -I SO T.it. 1'o,i..10.049 at 0 sIpcl..o .0. Foi Sc.sI. 40.0,c

40 Tc~tl. ',op.li~o f 11.10 ~C ~il~ioc, .1l0tits F-4 0. T-io 1.4 -IQ T01 Poplt .0 012 YI H-OO60F. ll.T0oy Ikth c.o Two.

Alloy M.000.1h4 Thick Sh.irl Stock Cao Nor. Titcbiiioo, xil So-ii.R.li1-d ItR.11.4 SO%. -4 Agod 90 OO I., 0-.t Ito., 104

40 1:-Il. I'ro.W.. .oICol4.ltol.4 (S0% 04 1-11o. lI-I.iri.. of 40.1.0k D..1~~R 4"0110.1 -4 A.i (8001(l)AQC( - R.I£ogrdoal .VI I.I.stock woih. 70 2400 pymt Iydiojoo9 Rsol targ io;.re Siaoefoo t.1 C-.1

44 ?och.4 Il(* ..8lScl-d l0.*4 UM0 Ss*I.9 S-..oy of lo.pcllo R...11. of 40.1.0kyo Coi.Rollod(50% R.4.11-)o "it A;.4 Doh.I* Cylii,.r. All., 11-cw.okaSo 104(05F~jl)AC, £1.01 stock With T0 pp- *0tOO yjo- of l1v41ogeo 1s

40 It-h.4 S6o) S..l~i,4.of T..l A.W1.~ T.-I. l'.op.11ii. of

mo lcoli*Rol.4 (501 R.dcooiosl -id Aj.4 T"ll*-0.0* 404 bk Iliao.I.. Coliwo-(.ot;,IJAC I3) S. okk it 10 P--4 r. l.o.4k l.P0.1To

I0 p f1y0tO9 P.S. 1ockiuic (90% 0..01."~100 p. 1 Idroi.P .. I S3ro..Rolio-.9 ci OF (or M.

- C,14-.1l.1~i4 ISO0% Riidicli. -04 Ag.d(GSOIlI)ACI Ski.l 24 ock il1k M -04tOO0 pp.

01 lydrog.. S0 ROOF Agiog lo.pos.. of 40.Wl.k M4.o.0h.T..eCylio.. r.O 104

41 A. Toc... ifo.... of Sooc-l. It-lock i.,h o Cylil.. V".-sAoo.ldof10o _OO M.c4 ll .... k b. 15,... 08 TO.... lro.c.. o Spo.o*.o looTo.P .o. irchsioqc ISO% lI.4.o Pot i.. Fl. .4 CyIo, 'I ... I.T..I

P.. 0Copooco F.4 Whit ) i.ii.4 at S40pits 0il P ... 109

40 Ao.' T17... d. P"ro 1-rl.1 of Ooko0 14.look .c-c.I.C yloo.., V-.-ooA ... i*4 09 T-oki. Pt 4.a..liclof 11-wToroo.I bwor w.04 R.ooi io-mo4. -IOO wor Du4-Tool Coc.pso. r.? %hi;k 0.11*4(I0 s0,o-".o Prior I.. 11s-T. is A9 UV11 Prlg f1.." oil1 ... 4s,. Ito

40 5.0.01 .4 Nc.Itha ..14-. Na,ootl. .1("..oo~ (90.R4.t").. -- 4 AC-d to .. U. Pm.p-tkiis oo Ylo.Too4

j6900(IIZIACI 14-t-4k O...m.1 CvhO..Oo, Cylisdr.lcst Soobm I. 'ao, V.1 scsI.With0 IlydIo.. Coo0.1. of TJ A 1,00 p" IN0 Motor C.. Aft., II40llcDm01toIs It.

PA09 NO. iX

P*Arr & W..N.C AINCRAP PWA-2267

I. INTRODUCTION[A. Purpose and Scope of Project

As a part of the early development of the Pershing missile, a sup-porting program was conducted to determine the suitability of betatitanium alloy Ti- 13%r V- 11% Cr- 3% Al (B-120 VCA) for rocketmotor case applications. The work was performed under U.S. ArmyContract, No. DA-0-009-O...D-634, subcontract AM-96Z, and in-cluded the fabrication and testing of several subecale and full scale

cases. The oxcellent results obtained indicated that additionalresearch and development with this alloy would be justified.

In 1960, the United States Army Materials Research Agency, whichwas assigned responsibility by the Department of Defense for con-ducting the Solid Propellant Rocket Motor Materials Program,awarded the present contract to Pratt & Whitney Aircraft vo developproduction processes for general rocket motor case applications ofthe beta alloy. Although cases were made to the design establishedfor the second stage of the Pershing missile, this was a matter ofeconomy dictated by the availability of tooling and emphasis wasplaced on improving the state of the art toward possible future De-partment of Defense requirements rather than on specific motorcase applications.

B. Background

Forty-inch diameter rocket motor case center sections were

flow-turned under the RM-962 contract.

Although these cylinders had adequate strength properties and

performed satisfactorily hoth during hydrotesting and during thestatic firing tests of full scale rocket motor casec, dimensionalcontrol during flow-turning was inadequate to meet the required

tolerances. During the first of two flow-turning passes, a

diametral growth of as much as 0. 300 inch occurred. Before

being given the second pass the cylinders had to be sized. Sizingwas performed in a hydraulically operated machine which employed

pA0g NO I

I

VM*1 • jmsmCw PWA-2267

a tapered ram to force a segmented cluster assembly against theoutside surface of the cylinder and thereby reduce the diameter.The second pass produced diametral growth and also local bulging.Although a limited amount of diametral growth could be correctedby the sizing operation, bulging could not be easily corrected.Further, the flow-turning procedure used produced a non-uniformlyworked structure in the cross-section of the cylinder wall. Thecurrent program directed attention toward developing a flow-turningprocedure to produce material which would meet the dimensionalas well as the improved strength requirements.

One of the outstanding features of a B- 120 VCA titanium alloyas a rocket motor case material is its corrosion resistance.Pressure vessels of this material need no protective coating toprevent pitting or general oxidation. This was demonstrated inthe RM-962 program by storing a presiure vessel containing aflow-turned cylinder aged to the 180,000 psi yield strength levelwith a single girth weld (not stress-relieved) in a salt-sprayhumidity cabinet at 125F for 134 days. Radiographic, fluores-cent penetrant, "Dye-Chek", and visual inspection revealed nocracking, pitting or other evidence of corrosion. Because ofthis test, a program to investigate the corrosion resistance afflow-turned material was not undertaken. No corrosion orstress-corrosion probiems were encowitered diuing the subjectcontract.

Stress corrosion susceptibility of forged material and sheet weld-ments were also investigated under the subject contrat. Speci-mens from press-forged pancakes with oxygen contents rangingfrom 0. 10 to 0. 20 per cent were stressed in tension to 50 percent of the yield strength or 90, 000 psi and subsequently exposedto a salt spray without detrimental effects. Similarly, weldmentswere prestrained by bending and exposed to the salt spray, alsowithout detrimental effects. These tests are discussed inVolumes I and III of this report.

Stress-corroson cracking of materiaA in the as-flow.-turnedcondition has been encountered previously, but not in the subjectprogram. Such cracking results from the combined effects ofcontaminants, such as halides, and the high residual stresses(up to :.00, 000 psi) present in as-flow-turned material. Care-ful handling and prompt stress relieving at 850F has preventedcylinder cracking in the subject program.

PAGE NO. 2

PWA-2267PRAl? a WHMMT JSMCFrWT

III EVALUATION OF MATERIAL1- FLOW-TUPNED UNDER CONTRACT RM 962

A. Preferred Orientation of Flow-Turned Material

X-ray diffraction studies of flow-turned material were con-ducted to determine the directional properties and the natureof the defrration in flow-turned material. Two 0. 35-inchlaminated cubes were prepared from a 40-inch diametercylinder which had been flow-turned by a two-pass technique(50 per rent reduction each pass) and aged at 850F for one-halfho r. One cube contained laminations taken from the outsidesurface and the other contained laminations from the insidesurface. Additional 0. 350-inci laminated cubes were preparedfrom annealed and from cold rolled (50 per cent reduction) andaged (850F(1/2)AC) 0. 062-inch thick sheet stock utilizing theentire sheet thickness. The cubes were sectioned to produce adiagonal face making equal angles with the three orthogonalfaces. The surface was then polished metallographically andlightly etched. The specimen was installed in a Norelco motor-driven gonionicter with conventional Schultz geometry so as totrace a spiral through the stereo-graphic projection, using thediagonal face of the specimen as the reflecting surface. Thereflected X-ray beam intensities were sensedbyaGeiger counterand continuously recorded on a strip chart. At discrete timeinterval, base-line reflected intensities were obtained from asample with a random crystal orientation placed -t the sameposition as the specimen. The ratio of the intensities of re-flected •adiatirn from the two samples were plotted on 'hequadrant of the pole figure and buundaries were drawn aroundareas with similar intensity ratios. Because of the specimengeometry used, only a single run was required to determinethe complete quadrant of the pole figure. The frequency ofdata points and the method of presentation are shown in Figure1. The reference direction is parallel to the axial direction ofthc flow-turned cylinder or parallel to the sheet rolling direction.

The intensity distribution of the reflections from 11 101 planeswas obtained using filtered copper radiation. The data fromthe four types of specimens (annealed and cold rolled sheet, andouter and inner surfaces of flow-turned material) are presentedin the form of 11101 pole figures in Figtres 2, 3, 4 and 5 re-spectively. With the exception of material in the mill-annealedcondition, all of the samples exhibited distinct tex.tures. The

PAGK NC

PWA- 2267PQ&V IN EWUS M*C*ST

data from mill-annealed material (Figure 2) shjws a slight in-dication of a (100) [01l] texture, but is otherwise close to ran-dom. For the deformed material, the preferred orientationswhich are nearest to the ideal, that is, the cystallographicplanes and directions which lie parallel to the deformation di-rectior,, are indicated on each of the pole figures (Figures 3through 5). The textures are summarized in Table I and illus-trated schematically in Figure 6.

As indicated in Table I, X-ray diffraction analysis has deter-mined that flow- turned B-120 VCA titanium alloy exhibits twodistinct texture components, namely, (100) [01 1 and (I 11lj[1121 . Only the (111) [112] is present at the outside surface,but both compon,-nts are present at the ia:side surface. Sincethe ( 11) LI 12] component conforms upon rotation through 90degrees to the (I 11) L01 component found in the cold-rolledsamples, it appears that the principal working direction is cir-cumferential even though the flow-turning operation results inelongation of the material without any change in the inside di-ameter. That the circumferential direction is the principalworking direction is further substantiated by tensile testingwhich indicated higher strengths and faster aging response forcircumferentially oriented specimens than for axially orientedspecimens (Table II). The mixture of textures at the inner sur-face, however, indicates that metal in contact with the rollerflows spirally to a depth of at least 0. 015 inch while the metaladjacent to the mandrel flows in an axial direction. The extentof these directional deformities depends on the flow-turningparameters.

B. Fracture Toughness Test Methods

It was necessary to determine the fracture toughness of B- 120VCA titanium alloy as a function of yield strength in order tomake the most efficient use oi the material. It was also de-sired to determine if more economical tests such as the modi-fied Charpy impact and instrumented-bend tests could be corre-lated to fracture toughnoss tests using the standard ASTM 3 byIZ-inch Gc specimens. The effect of Gc specimen size onfracture toughness values was also investigated.

Initially, material was used from 40-inch diameter cylindersnumbers I and 2 from the previous program (con'.-act RM 962).

PAGE NO 4

P WF PWA-2267PWA"¢r 6 Wt~inlmt JURCNA~w

These cylinders had been flow-turned with a reduction of about50 per cent on the last pass to a wall thickness of 0. 080 inch.They were subsequently stress relieved at 850F for one-halfhour and sectioned into axial and circumferential specimenblanks. These were aged at 800F for various times and rma-f chined into standa.-d ASTM 3 by 12-inch internally notchedfracture toughness specimens (see Figure 7) and into smoothtensile specimens with a one-inch gage length (Figure 8).Circumferential specimens were flattened by clamping duringthe 800F aging treatment. Slow crack growth was measured bythe ink stain technique.

The results from testing these specimens appear in Table IIIand indicate that for circumferential specimens the toughnessis lower and the yield strength higher than for axial specimens.The results also showed that the two cylinders had equivalenttoughness levels. As shown in Figure 9, a close correlationwas found between fracture toughness and yield strength. Thetoughness (Ge) at the ZO, 000 psi yield strength level is approxi-mately 225 in-lbs/in . Although this value is low, it is believedto be an adequate combination of yield strength and toughnesssince only very slight defects have been encountered in flow-turned cylinders. Cylinders flow-turned and aged to the200, 000 psi yield strength level and pressure tested have burstat the estimated biaxial ultimate tensile strength of the mater-ial. Further, the fracture surfaces exhibited 100 per centshear-type failures which verifies that the toughness levelobtained is adequate for flow-turned material without detectabledefects (see Volume IV of this report for a complete discussionof flow-turned cylinder burst testing and analysis).

Additional testing was conducted to determine the effect ofspecimen size on fracture toughness. The specimens em-ployed were an internally notched ASTM 3 by 12-inch standardspecimen, a 2 by a-inch externally notched specimen (Figure10), and a I by 4-inch externally notched specimen (Figure 11).All specimens were 0. 080 inch thick. The results of testingthese specimens appear in Table IV and indicate that toughnessvalues decrease with decreasing specimen size.

Modified Charpy impact specimens were machined in the axialand circumferential directions from flow-turned cylindersnumbers I and 2 and aged to various yield strength levels. Thespecimen configuration used is shown in Figure 12. Thesespecimens were tested at -35, 70, Zi, and 400F and resultsappear in Table V. As shown in Figure 13, at a given yield

I PAGE P40I

PRATT V WI IKY AIRCRAFT PWA- 2267

strength level and direction, the energy absorptions decrease withdecreasing test temperature. The transition temperature for thismaterial tested by this technique was apparently near or above400F. At a given test temperature, the energy absorption de-creased with increasing yield strength regardless of direction (seeFigures 14 and 15). This trend is readily apparent at 400F, lessevident at 215F, and barely discernible at 70 and -35F. The cor-relation of energy absorption with yield strength at 400F is simil-ar to that observed with G test results (Figures 9 and 14). Al-though the use of modified'harpy impact specimens resulted invalues which were comparable to those determined using standardspecimens, the scatter in the results, resulting either frommaterial variations or from the difficulties of accurate energymeasurement, was considered excessive for reliable correlationand investigation of this testing method was discontinued.

Instrumented-bend specimens ( 1. 50 x 0. 875 inch) were machinedfrom cylinders numbers 1 and 2 and tested at various yieldstrength levels. The results are shown in Table VI. Some cor-relation with yield strength was observed, but the correlation is notas distinct as that obtained with Gc testing. Similar to the resultsobtained using the modified Charpy impact test, the scatter in thedata obtained from instrumented-bend testing was too great forreliable data analysis and investigation of this testing method wasalso discontinued. All subsequent fracture toughness tests em-ployed standard 3-inch wide internally notched specimens.

PAGE NO 6

j PWA-2267PRATT 6 WH"6i AIRCRAFT

I

III RING ROLLING

Experience gained during the program conducted under contractRM-962 on flow-turn blanks which had been roll forged at 1850Finfrom 3 to 12 operations and sized at 1400F indicated that frequent-ly the resulting flow-turned cylinders exhibited intergranularseparations on the inside surfaces. In one instance this conditioncaused catastrophic failure during flow-turning. These difficul-ties were believed to result from a grain boundary precipitatewhich formed during ring rolling. A program was therefore con-ducted to eliminate the difficulty and to optimize the ring rollingpractice to produce high quality flow-turn blanks of B- 120 VCAtitanium alloy.

Initial investigation was conducted on 14-inch diameter rings andthe most promising technique developed was subsequently appliedto 40-inch diameter rings.

A. Subscale 14-Inch Diameter Rings

Forging Techniques and Mechanical Property Evaluation -Material for the 14-inch diameter subscale ring rolling programwas obtained from the bottom half of an ingot having the followingcomposition (in per cent):

V Cr Al Fe C H 0 Ti

13.69 11.74 3.41 0.28 0.0183 0.003 0.11 balance

Six blanks for ring rolling were extruded in single operations. Thefirst three of these were subsequently rolled in single onerationsat 2000, 1900, and 1800F respectively. The remaining three wererolled in multiple operations at temperatures from 1800 to 2000F(see Table VII). Test pieces were trimmed from the first threerings and annealed at 1450F for 30 minutes followed by a waterquench. (The selection of this annealing treatment was based onresults from previous investigations conductea over the tempera-ture range of 1450 to 1750F. No recrystalli4.,,on or grain growthwas note,' after any of the treatments and therefore for conven-ience the lowest temperature investigated was selected). Thesespecimens were tested with the results (Table VIII) indicatingsatisfactory ductility and notcher' toughness for flow-turningAll six rings were therefore annealed and sized as requiredat 1450F and water quenched. After annealing, all of the rings

PACE No 7

I

PRAT & W4MINY AIRCR*fr PWA- 2267

exhibited coarse. equiaxed microstructures such as shown inFigure 16.

Test hoops were trimmed from one end of rings numbers 4, 5,and 6. Additional material for test specimens was trepanned fromthe inside surface of each ring at the end opposite that from whichtest hoops had been removed. Smooth and notched (K t = 8)tensile testing was conducted and the results (Tables IX and X)further demonstrate the adequate ductility and notched (K t = 8)toughness of the rings and also show that all six rings were uniformand had equivalent properties.

Smooth tensile specimens were machined from rings numbers1, 2, and 3 and aged at 900F for various times. The results oftesting these specimens (Table XI and Figure 17) show that theserings had equivalent aging responses and generally low elonga-

tions. These results demonstrate that rings forged at differingtemperatures have equivalent properties after being annealed at

1450F.

The results of the subscale ring rolling program have indicatedthat the properties of annealed (1450F (1/2) WQ) flow-turn blanks-tre independent of forging prrctice within the range of techniquesinvestigated. On the basis of this study, a temperature of 1900F(the temperature of the soaking furnace) was chosen for forgingfuture roll-forged rings. Ten additional 14-inch diameter ringswere roll-forged for the subscale flow-turning development(see Table XII). The first seven of these rings were forged in asingle operation at 1900F, but tooling difficulties (resulting infinning during the forging of the last three rings) required asecond forging operation. The tensile properties of the tenadditional rings appear in Table XIII and indicate lower ductilitiesthan observed previously (Table IX and X).

Metallographic Investigatio: of Effect of Heat Treatments onMicrostructure - G rain boundary constituents were occasionallyobserved in the microstructures of the 14-inch diameter flow-turnblanks. Similar grain boundary conditions had previously beenobserved in roll-forged rings produced under contract RM 962.These observations led to an investigation of the precipitationcharacteristics of the subject alloy since it was believed that,

PAGE NO 8

PIA,? , ITNGE ,ICRAT PWA-2267

Ialthough the 14-inch diameter flow-turn blanks had considerable

Tductility, the elimination of the constituent would increase theductility and permit a more severe flow-turning technique. Theuse of increased reduction, feed, and roller radius would improvefthe inside surface condition of the flow-turned cylinders.Samples of a 40-inch diameter roll-forged ring, 0. 250-inchthick place-stock, and a particularly poor 5-inch diameter forgedblank were used for these studies. The 40-inch diameter ringwas one of two transferred from the previous program and hadbeen successfully flow-turned for use in studying the effects ofhydrogen. The 5-inch diameter blank received mu tiple forgingupsets at approximately 200OF and subsequently ruptured duringflow-turning. The plate-stock was used for comparison purposesand also because it exhibited a grain boundary constituent in themill-annealed condition.

Samples of the above materials were first heated for two hoursin an argon atmosphere at tenperatures in the 900 to 2100Frang- and water-quenched. These treatments were intended todeter'mine the temperature range or ranges in which grainboundary and matrix precipitation occurred and to determine inwhat ranges these constituents could be placed in solution.Micro-exam!nation of the specimens after treatment showed th.following:

1. In the 900 to 1200F temperature range, matrix ,%ndgrain boundary precipitation occurred in all sampleswith the constituont particle size and spacing increasingwith increasing temperature. The particle shape wasacicular when formed between 900 and 1200F, whichis characteristic of alpha phase. Samples of the40-inch diameter ring heated at 12OOF showcd astructure very similar to the puor 5-inch diameterblank in the as-forged condition (see Figures 18 and 19).

2. In the 1300 to 1400F tempera.ure range, ;. .- all amountof both grain boundary and matrix precipitate wasobserved in the relatively clean 40-inch diameter ringand plate stock (see Figures 19 and 20). The 5-inchdiameter blank showed a smaller amount of constituent

PAGE NO 9

1

PRATT & WHITNY AIRtCRAFT PWA- 2267

than in the as-forged condition indicating that somere-solution had occurred (Figure 18). The nature ofthis precipitate 's not known since it persisted when thespecimens were heat treated above the beta trans-formction temperature (appro>imately 1325F) andwater quenched.

3. In the 1600 to 2000F temperature range, almost completere-solution of all of the constituent occurred in the 40-inch and 5-inch diameter blanks (Figures 18 and 19).After the 20O0F treatment, both pieces were essentiallyfree of precipitate. In cortrast, the plate stock sampleshad relatively clean structures with some grainboundary precipitate after treatments at 1600F and1800F, but had matrix and sub-grain boundaryprecipitation at 2000F (Figure 20).

4. At 2100F, the 40-inch and 5-inch diameter blanks hada coarse grain boundary precipitate (Figurs 18 and19). The plate stock sample has an increased amountof the matrix and subgrain boundary constituentobserved after the 20OF treatment.

These results indicated that the precipitate occurring in roll-forgings apparently formed during cooling through the 900 to140OF temperature range. Grain boundary precipitation at2100F was discounted because no forging operations wereconducted above 2000F. To determine the effect of coolingrate on precipitation, similar samples were heated attemperatures in the range of 1400 to 2000F for two hours andcooled in the center of a 3 l/2-inch diameter, 6-inch longsection of titanium bar stock. It waa felt that this woulusimulate the cooling of a forging during fabrication. Theresultant microstructures showed that grain boundary andmatrix precipitation occurred in all instances where there wasslow cooling from the 1400 to 2000F range (Figures 18 and 20).When cooled slowly from 800F or 20OOF, the resultant grainboundary precipitate was coarse with a more spheroidalmatrix precipitate. To affirm the previous indication tatre-solution could be accomplished by ,vater-quenching Lom1800F, a sample of the 40-inch diameter ring was heated at

PAOI NO 10

L PAr 41 WHIfNR" AIclFrr PWA- 2267

a temperature of 2000F for two hours, slow-cooled to producea heavy matrix and grain buundary precipitation, reheati.d :at1800F for two hours and water-quenched. The resultantstructure. shown in Figure 21, was essentially clean.

Samples from the flange sections of the two 40-inch diarneterrings were then heat treated and machined into smooth andnotched (Kt = 8) tensile specimens to determine the relativeductility of the structures discussed above. The heat treat-ment used was as follows:

1. As-received (roll-forged and solution-treated at 1400Ffor 30 minutes),

2. 1200F, 140OF and 2000F for two hours and water-quenched, and

3. 1400F, 180OF and 2000F for two hours and furnace cooled.

In addition, except for -he treatment at 1200F, smooth tensilespecimens were similarly heat treated a:,d aged for 64, 72and 96 hours at 900F. The results of these tests are shown inTables XIV and XV, and in Figure 22. This data shows thatreheating to 180OF followed by water-quenching produces thebest ductility both bef3re and after aging at 900F. In comparison,reheating at 1400F or 2000F followed by water quenchingproduces lower ductility before and after aging, and reheatingat 1200F followed by water quenching produces extremely lowductility. Furnace cooling in all instances produces lowerductility and more sluggish aging response. An illustrationof the relative ductilities as indicated by nothed (Kt = 8)specimen fracture surfaces is shown in Figure 23.

The study indicated a method for improving or restoring theflow- turnability of B- 120 VCA titanium roll-forgings containingan unacceptable structure that would otherwise produce surfacetears or fracture during the flow-turning operation. Further-more, the study showed that slow cooling during processing ofring forgings should be avoided.

PA0OI N

PRATT , WHITrY AIRCRAFT PWA-2267

B. Full Scale 40-Inch Diameter Rings

Seven 40-inch diameter rings were forged at 19001(furnace soakingtemperature)uaing the technique developed with 14-inch diameterrings. Although the results from forging 14-inch diameter ringsindicated that two operations would be sufficient, tooling difficul-ties encountered during the preliminary operations necessitateda total of four operations to produce the required diameter. Theforging sequence used for these pieces is outlined in Table XVI.After rolling, the rings were sized at 1450F and water quenched.

Test pieces were cut from one end of each ring. Typicalcompositions (in per cent) are:

RingNumber V Cr Al Fe 0 H N C Ti

2 12.90 10.03 2.89

PMATr WHIfh AIRCRAF PWA- 2267

ITo determine the effect of the 180OF solution "reatment, two9.4-inch diameter cylinders, one annealed at 1450F and theI other solution treated at 1800F, were flow-turned in two passeswith 50 per cent reduction per pass. The cylinders were thenevaluated for aging response after stress-relieving for one-half hour at 850F. The results (see Table XIX and Figure 24)showed a rapid and equivalent aging response for both cylindersbut erritic and low tensile elongation for the cylinder solutiontreated at 1800F prior to flaw-turning. These results wereconsidered anomolous and therefore a 14-inch diameter ring(number 6) was solution treated at 180OF and flow-turned intwo passes. Unfortunately this ring was not properly annealedbeLween flow-turn passes and the test results, which indicatedpoor tensile properties and unusual aging response, were notconaidered representative of a cylinder solutiun treated at1800F before flow-turning. As a result, four additional 14-inchdiameter rings were solution treated at 1800F for 15 minutes,water quenched, and flow-turned by the two-pass technique.The results from testing these cylinders are reported in detailin section V-D, page 37 of this volume and indicate that the1800F solution treatment does not deleteriously affect the agingresponse or tensile ductility of the subsequently flow-turnedcy.n ders.

Forty-inch diameter rings numbers 4 and 5 were then solutiontreatet at 180OF for 15 minutes and water quenched. Ringnumbei 4 was flow-turned in two passes and stress relievedat 850F !or one-half hour. A test piece was cut from the ringand ealuted for aging response and tensile ductility. Theres..,',t (diicussed in section VII of this volume) wereconsi 4a rd o. tisfactory and the remaining rings were there-fore rol,-tion xeated at 180OF for fifteen minutes and wa:erquencl % . T, st specimens solution treated with the rings weretested (s, v Tz.blc XX) and again demonstrated that the 180OFsolution trt:-iment -nproved ductility (compare Table XX withTable XVWL,. The ,lIks were subsequently flow-turned.Each ring displayed cxcellent flow-turnability and there were nomaterial falures or intee'-ranuar separations on the insidesurfaces of the reOting cylinders.

PAO. N01 3

P"ATT WHITNEV AIfCRAT PWA- 2267

The solution treatment at 1800F for fifteen minutes followedby a water quench has been demonstrated to be an effectivemethod for improved flow-turnability correlating with a dramaticincrease in tensile reduction in area of the blank material. Theimprovement is believed to result from the solution of an undesirablegrain boundary constituent formed by slow cooling from tempera-tures above 1400F.

PACC NO 14

L POAIT WNrIthV AICA rT PWA- 2267I.

IV. INITIAL FLOW-TURNING DEVELOPMENT

A. Flow-Turning Parameters

Initially, the flow-turning development program provided typical• flow-turned cylindo 9 for the material and heat treatment inves-

tigations previously described, and for determining effects ofring rolling and heat treatment factors on flow-turning behavior.In order to evaluate these factors with a minimum of variables,it was decided to use the same flow-turning technique for eachblank of the first series. The flow-turning practice that showedthe best results during the preceding program (Contract RM-962)

was selected for use. This practice utilized the forward flow-turning technique which requires a flanged blank and producesdeformation in the direction of roller feed. The blanks wereflow-turned in two passes, using approximately 50 per centreduction in wall section thickness (and simultaneously increasingthe blank length by a factor of approximately two) during eachpass. The flow-turning was done on a two roller (opposed)4 by 50 inch Cincinnati Hydrospin machine. The machine para-meters were set for 950 or 630 inches per minute mandrel sur-face epeed and a roller feed of 0. 015 inch per mandrel revolutionper roller. These parameters were established so that a constantworking rate could be maintained for each blank regardless ofdiameter.

The flow-turning starting blank was machined from a roll-forgedring to dimensions commensurate with (I) the required wallthickness, (2) the length of the finished cylinder, (3) the numberof passes, and (4) ihe magnitude of reduction during each pass.The cylinder design for this series required a wall thickness of0. 075 inch. Since two passes were to be used with approximately50 per cent reduction during each pass, the thickness of themachined blank was established at 0. 300 inch. The blank lengthwas not of prime importance during the initial and developmentalphases of the program since it was not intended to use the result-ing cylinders for rocket motor case parts. All blanks, subscaleas well as full scale, were machined to the 0. 300-inch wallthickness eo that the amount of deformation would be the samefor blanks of the various diameters. The machined blankincorporated a chamfer, machined at an angle approximately

PAO9 NO 15

I

PRATT W"MW..Nr AINCRAT PWA- 2267

equal to the lead angle of the working surface of the roller to thefull depth of the redu Aion to facilitate maximum depth ofplastic deformation as soon as plastic deformation commenced.A sketch of a flow-turning starting blank is shown in Figure 25.

The flow-turning rollers used in the initial phases of thisprogram were the same as those previously used to flow-turnB- 120 VCA titanium cylinders. These rollers had also beensatisfactory for producing steel cylinders. The roller configura-tion contained a conical lead surface inclined 25 degrees fromthe bla.k axis and a flat surface approximately 0. 100 inchlong at the tip. (See Figure 26)

B. Flow-Turning of Subscale and Full Scale Cylinders

Two 9.4-inch and two 40-inch diameter roll-forged rings, avail-able from a previous program, were flow-turned by the initialpractice to provide cylinders for evaluation. These cylinderswere used for the material and heat treatment investigationspreviously described. In addition, the 40-inch diametercylinders were used for sizing experiments. It is interesting torote the problems which occurred during the processing ofthese cylinders since they were indicative of the areas requiringfurther development.

The two 9. 4-inch diameter roll-forged rings were flow-turnedin two passes (approximately 50 per cent reduction per pass) toa wkll thickness of about 0.075 inch. The dimensions of thecylinders and the parameters employed are tabulated inTable XXI. Between flow-turn passes the cylinders were stress-relieved at 850F for 30 minutes and a section trimmed fromeach for determination of an intermediate anneating treatment.

Annealing at 150OF for 15 minutes produced the optimumcondition with complete recrystallization and limited graingrowth through the wall thickness. The parts were given thistreatment. After annealing, the cylinders were pickled in anaqueous solution of 3.5 per cent HF (70 per cent concentrate),and 35 per cent HNO 3 to remove 0. 002 inch per surface ofcontaminated (oxygen rich) material. Because of the radialgrowth during the first pass, both cylinders were shrunk by a

PAGZ NO 16

MATT a wMy MMcawr PWA- 2267

sizing operation to mandrel size for the second pass. Bothcylinders again grew radially and also bulged locally during the

second pass. Dimensions were not taker. after this pass because

the cylinders were immediately sectioned in the as-flow-turned

condition for determination of residual stresses and the optimur.

stress-relief treatment.

The two 40-inch diameter roll-forged rings (numbers 7 and 8 inTable XXI) were given the first flow-turn pass after being annealedat 1450F for 30 minutes. The cylinders were then stress-relievedat 850F for 30 minutes and an intermediate annealing treatmentwas determined. Microexamination of annealed test samplesshowed that the optimum treatment was heating at 1550F for 15

minutes. After annealing, the cylinders were pickled in an aquaeous3.5 per cent HF (70 per cent concentrate), 35 per cent HNO 3 so-

lution to remove 0.004 inch per side of contaminated (oxygen-rich)skin. Both cylinders grew radially approxdmately 0.250 inch on thediameter during the first flow-turn pass and cylinder number 8bulged locally in addition to growing radially. Visual and fluores-cent penetrant inspection of cylinder number 8 before and afterpickling revealed several cracks on the ou.side surface in the lo-cally bulged area. These cracks were removed by blending priorto pickling but reappeared after pickling. Additional blending to adepth of 0.050 inch was required to remove the defects. The depthof these; blended regions was considered too great for subsequentflow-turning so the locally bulged area war trimmed off.

Cylinder number 7 ruptured during shrinking to mandrel size in

preparation for the second flow-turn pass (Figures 27 and 28).The cylinder had been shrunk approximately 0. 210 inch on thediameter when failure occurred. No defect was apparent toaccount for failure. Cylinder number 8 was then instrumentedwith strain gages to evaluate the stress conditions duringshrinking. During shrinking, the strain gages were continuallymonito-ed to determine if local highly stressed regions appearedaround the cylinder circumference. The gages indicated noabnormal stress distributions and gave essentially equivalentstrains around the circumference after yielding had occurred.

As an additional precaution, it was decided to stress-relievethe cylinder after shrinking approximately half the desired amounton the diameter (0. 125 inch). The cylir.de- was therefore re-annealed at 1400F for 15 minutes and shrunk the remaining0. 125 inch on the diameter to mandrel size. The cylinder was

PAGE NO 17

I

PRATT V wwMY AmCAO," PWA- 2267

then given the second flow-turning pass to a wall thickness ofabout 0.080 inch as outlined in Table XXI. Radial growth but nolocal bulging occurred during the second pass. The cylinderwas cut axially to provide material for the various investigations.The sectioning was done in the as-flow-turned condition do thatthe cylinder would open circurnferentially to provide flat materialfor tensile specimens. The cylinder was also instrumented withstrain gages to determine the residual stress level in the as-flow-turned condition.

C. Problems Requiring Further Development

The cylinders described in the previous section were flow-turnedusing the roller configuration and practice developed during aprior contract (RM-962). This flow-turning practice and rollerconfiguration produced radial growth and local bulging on the cyl-inders which was similar to that experienced during previous work.The radial growth relative to flow-turning parameters (roller feedand per cent reduction) is shown in Table XXI and Figure 29. Ra-dial growth is expressed as the diametral change per inch of di-ameter and the flow-turning feed-reduction parameter is expressedas a product of effective roller feed (feed in inches per mandrelrevolution per roller) and per cent reduction. The diametralgrowth expressed as growth per inch of cylinder diameter shouldbe applicable t all cylinders regardless of diameter. The flow-turning parameter expressed as a product of feed and reduction isused since flow-turning performance is most responsive to changesin these variables.

The results of testing the series of cylinders discussed in thissection has shown that excessive radial growth (as much as 0. 026inch/inch of diameter) and local bulging occurred as a resultof the flow-turning practice utilized. The shrinking experiencedwith two 40-inch diameter cylinders indicated that 0. 125-inchdiametral shrinking could be accomplished after the first pass pro-viding the material was properly annealed. Diametral shrinkinggreater than 0. 125 inch, although successfully achieved withoutre-annealing in the previous program, in some instances resulted inbuckling failures during the shrinking ope. -tion. This experience

PACE NO 18

I PRATTO IWIsv A.CRA PWA- 2267

indicated that the development program should be directed towardimproving the flow-turning practice whereby the diametralgrowth of full scale cylinders would be reduced to 0. 100 inch orless (0. 0025 inch per inch of diameter), and all associated localbulging eliminated. On the basis of the tensile findings, itwas decided to study the flow-turning process in order todetermine and control operational variables to attain structuralquality and close dimensional tolerance.

rAGX NO 19

1

PRATY V WHNE AIMCMA7T PWA- 2267

V. SUBSCALE FLOW-TURNING DEVELOPMENT

A. Background

Major problems encountered during previous experience in theflow-turning of B- 120 VCA cylinders were radial growth andlocal bulging which were unpredictable and of considerablemagnitude. Although the cylinders produced by early flow-turningmethods had adequate strength properties, dimensional controlwas not acceptable for production of rocket motor cases andthe amount of shrinking required to offset radial growth wasintolerable. The objective of this development program wasto define operating parameters and to develop a revised flow-turning practice which would reduce radial growth, (not toexceed 0. 0025 inch per inch of diameter), and to eliminatelocal bulging.

Since the experience with the initial flow-turning practice(desc ribed in the previous section) indicated that similar problemswere encountered with cylinders ranging in diameter from 9. 4to 40 inches, it was reasoned that a revised flow-tuxningpractice could be developed with subscale cylinders and appliedsuccessfully to full scale 40-inch diameter cylinders.

An analysis was made of all previous experience in the flow-turning of titanium at Pratt & Whitney Aircraft, and the para-meters expected to affect radial growth and bulging wereevaluated. This analysis indicated that the largest gains wouldbe obtained by changes in the roller geometry. A conical rollerconfiguration whichwas satisfactory for flow-turning steelcylinders had been used for most of the early B- 120 VCAtitanium flow-turning. No combination of available machinesettings showed promise of providing dimensional stabilitywith titanium cylinders using this roller configuration. Alimited amount of experimentation had been done with a sharpparabolic roller contour which produced less radial growth thanother roller contours on titanium cylinders.

B. Evaluation of a Revised Roller Configuration

Based on preliminary indications that roller geometry wouldhave a marked influence on flew-turning results with B- 120 VCA

PAGE NO 20

MSATT 6 wwiMs AmcRCAP PWA- 2267

titanium, a series of experiments were prepared to evaluate theoriginal parabolic roller. Both 9.4-inch diameter and 14-inchdiameter starting blanks were prepared by rolling and lor.gitudi-nally welding 0. 375-inch thick plate stock. Flanges were weldedto one end of the rolled and welded plate stock and the assemblywas machined into flow-turn blanks with approximately 0. 300-inchwall thicknesses. (It was not the intent of this experiment to de-velop rolled and welded blanks for this application. This methodof fabrication was used only for expediency to obtain basic flow-turning information prior to flow-turning development using roll-forged starting blanks.)

The initial program included evaluation of a series of eight 9.4-inchdiameter cylinders flow-turned on a single-roller 60 by 74-inch Lodge &Shipley machine. The purpose of this series was to compare two-pass and three-pass methods and to evaluate the effect of varia-tions in per cent reduction, mandrel speed, feed rate, and rollertilt angle on the radial growth and local bulging of cylinders flow-turned with a parabolic roller. The roller was 14 inches in diam-eter and had a parabolic lead contour with a tip radius of 0. 062inch (See Figure 26). Lubriplate 630AA was used as a lubricantbetween the blank and mandrel and a low viscosity sulfurized fattymineral oil was sprayed ahead of the roller as a coolant.

The flow-turning parameters and resulting cylinder dimensionsfor this series of tests are shown in Table XXII. The first twoblanks were flow-turned in the as-welded condition. The firstof these blanks, which was flow-turned at 25 per cent reductioncommensurate with a three-pass sequence, failed immediatelyafter the start of flow-turning. The failure originated in thecircumferential flange weld in an area of incomplete weld pene-tration. The second blank was flow-turned for approximately25 per cent of its length with a 38 per cent reduction when flow-turning was intentionally terminated. When flow-turning wasresumed, failure occurred at the point of termination. Thefracture propagated circumferentially along the roller path.

Blanks numbers 3 and 4 were flow-turned successfully with firstpass reductions of 44 and 42 per cent and second pass reductionsof 63 and 54 per cent, respectively. Blank number 3 wasannealed at 1825F for 30 minutes after welding and prior toflow-turning while blank number 4 was flow-turned in the as-welded condition to determine if heat treating oi the welds wasnecessary prior to flow-turning these rolled and welded blanks.

PAGE NO 21

I

PRATT A WNIrNsY AIRC.APT PWA- 2267

Both cylinders were annealed and completely -- .tystallized at1600F for 30 minutes in an argon atmosphere and vapor blast-cleaned prior to the second flow-turn pass. Cylinder number3, which had been annealed prior to flow-turning, was freeof defects on the inside surface of the axial weld, whereaRcylinder number 4 which had not been annealed after weldingcontained small surface tears in the weld. As a result of thesetests, all subsequent blanks fabricated from rolled and weldedplate stock were annealed prior to flow-turning. Figure 30shows typical blanks in the as-welded, machined, and flow-turned conditions after the first and second passes.

Blank number 5 was given the first flow-turn pass with a re-duction of 25 per cent, commensurate with a three-pass sequence,to repeat evaluation of a three-pass method since the results offlow-turning blank number 1 were not considered valid. Thisblank cracked severely on the inside surface after flow-turningapproximately three inches of its length. The cracking wasindicative of insufficient plastic deformation on the inside surfacewhich was considered to be the result of the relatively lightreduction of 25 per cent during the first pass. The reductionsintended for the three-pass method were 25, 33. 3 and 50 per centrespectively. These reductions were established to provideapproximately 50 per cent reductiora during the last pass toassure maximum cold working through the section thlckness.It is very important that ample and uniform cold working through-out the section thickness be accomplished during the last passwith B-120 VCA titanium cylinders in order to achieve the de-sired aged properties. As a result, the first two passes wererequired to have relatively low reductions with the startingblank thickness chosen. Since small reductions may fail toproduce plastic deformation through the section thickness (as wasthe case with blank number 5), the three-pass method wouldapparently require a greater starting blank thickness to allowlarger reductions on the first two passes. On this basis, thetwo-pass method was favored since the same objective could beachieved with a fewer number of operations. The two-passmethod with approximately 50 per cent reduction per pass wastherefore adopted for all the subsequent flow-turning conductedfor this program.

Cylinder number 6 was flow-turned in two passes with the rollertilted in the direction of feed by the amount of helix angle gener-ated by the roller circumferential path. Strain gage analysisof this cylinder suggested a lower residual stress in the

PAGE NO 22

PArr wftr ,g AINCRAP- PWA- 2267

as-flow-turned condition but the results are considered incon-clusive because of the unknown effect on residual stress im-posed by the axial weld in the cylinder wall. No improvementin radial growth was observed to result from the roller tilt sofurther stndy of this parameter was dropped for this flow-turningprogram.

Cylinde.'s numbers 7 and 8 were flow-turned in two passes withincreased roller feed. This resulted in reduced diametralgrowth during both passes, with negative growth occurring inboth cylinders during the second pass, indicating that signifi-cant reduction in diametral growth was possible using the newparabolic roller configuration during both the first and secondpasses. Figure 31 shows effects of varying the flow-turningfeed-reduction parameter on the diametral growth of the 9.4-inchdianeter cylinders. With proper selection of this parameter itshould be possible to prevent radial growth. It can be seen thata curve drawn through an average of the data passes throughzero unit growth at a feed-reduction parameter of 0.010.Therefore at a reduction of 50 per cent, a feed of 0. 020 inch permandrel revolution per roller should be the optimum value forpreventing radial growth ari no local bulging should occur.

The cylinders in this -=:.es of tests were free from local bulgingafter the first pass and the diametral growth was less than thatexperienced during previous work. Metal tearing occurred on theinside surface of the axial weld in most blanks during the firstpass. These were generally shallow surface tears w4'hin thelength of the flow-turned portion and one or two deep tears atthe beginning and end of the flow-turned section. These tearsdid not extend into parent material and did not propagate duringsubsequent flow-turning. Because of the presence of these tears,the blanks were not sized to correct the loose fit of the blankson the mandrel prior to the second pass. The second flow-turnpass at the higher feeds id reductions produced little or noradial growth. However, because of the loose fit of the blank,local bulging occurred on some blanks. The higher feeds, how-ever, were also effective in reducing the amount of bulging.Figure 32 shows the inside surface of a typical 9. 4-inch diameterflow-turned cylinder from this series of experiments.

The next phase of the flow-turning program was t evaluate (1)the effect of scaling from 9. 4-inch to 14- inch diameter cylinders(2) the effect of changing from a single roller to a two-rollerflow-turning machine and (3) to determine whether the feed-

PA09 NO 23

I

PRATT & WHITHsV AINCRAFT PWA- 2267

reduction parameter optimized in the previous series of testswould hold constant with the same roller configuration,

The change from a single-roller to a two-roller machine was a nec-essary step toward flow-turning full scale parts. The roll forcesnecessary to flow-turn B- 120 VCA titanium alloy produce a.cveremandrel deflection on the single roller machine to the extentthat a negative roller-mandrel gap setting is necessary to pro-duce a 50 per cent reduction during the second pass. Thisobjectionable machine deflection can be reduced by using ahydrospin machine with two opposing rollers. Reducedmandrel deflection allows bett, r control of reductior and wallthickness. In transferring the flow-turning from the one-rollerto the two-roller machine, the optimized roller feed of 0.020-inch per mandrel revolution per roller was applied to both rollers,thereby doubling the carriage feed in inches per minute. Theremaining 9.4- and 14-inch diameter blanks fabricated fromrolled and welded plate stock failed in the circumferential flangeweld when flow-turning was attempted on the two-roller machine.Failure was attributed to the increased axial flow-turn load im-posed by the two-roller action. Since the use of rolled andwelded blanks was only for expediency, there was no furtherattempt to develop this type of blank.

C. Development with 14-inch Diameter Roll-Forged Rings

The next series of experiments were conducted on ten 14-inchdiameter roll forged rings. The intent of these experimentswas to evaluate the effect of scaling-up in diameter, to evaluatethe performance of the parabolic contoured roller with a Lwo-roller machine, and to optimize flow-turning parameters inpreparation for fabricating 40-inch diameter full scale cylinders.

The ten 14-inch diameter roll-forged rings were solution treatedat 1450F for 30 minutes, water quenched, and machined intoflow-turn blanks. Six of these rings were flow-turned utilizingthe optimum technique as determined by previous experiments,including the parabolic roller configuration. All of the flow-turning was accomplished on a two-roller 42 by 50-inchCincinnati Hydrospin machine. The flow-turning parametersand cylinder dimensions are listed in Table XXIII.

The first blank was flow-turned close to t:e previousl deter-mined optimum feed-reduction parameter of 0.010, using afeed of 0. 020 inch per mandrel revolution per roller and

PAG9 NO 24

P*AITY WINTN9V AINCRAWT PWA- 2267

as-flow-turned condition but the results are considered incon-clusive because of the unknown effect on residual stress im-Iposed by the axial weld in the cylinder wall. No improvementin radial growth was observed to result from the roller tilt sofurther study of this parameter was dropped for this flow-turningprogram.

Cylinders numbers 7 and 8 were flow-turned in two passes withincreased roller feed. This resulted in reduced diametralgrowth during both passies, with negative growth occurring inboth cylinders during the second pass, indicating that signifi-cant reduction in diametral growth was possible using the newparabolic roller configuration during both the first and secondpasses. Figure 31 shows effects of varying the flow-turningfeed- reduction parameter on the diametral growth of the 9. 4-inchdiameter cylinders. With proper selection of this parameter itshould be possible to prevent radial growth. It can be seen thata curve drawn through an average of the data passes throughzero unit growth at a feed-reduction parameter of 0. 010.Therefore at a reduction of 50 per cent, a feed of 0. 020 inch permandrel revolution per roller should be the optimum value forpreventing radial growth and no local bulging should occur.

The cylinders in this series of tests were free from local bulgingafter the first pass and the diametral growth was less than thatexperienced during previous work. Metal tearing occurred on theinside surface of the axial veld in most blanks during the firstpass. These were generally shallow surfaco tears within the

length of the flow-turned portion and one or two deep tears atthe beginning and end of the flow-turned section. These tearsdid not extend into parent material and did not propagate duringsubsequent flow-turning. Because of the presence of these tears,the blanks were not sized to correct the loose fit of the blankson the mandrel prior to the second pass. The second flow-turnpass at the higher feeds and reductions produced little or noradial growth. However, because of the loose fit of the blank,local bulging occurred on some blanks. The higher feeds, how-ever, were also effective in reducing the amount of bulging.Figure 32 shows the inside surface of a typical 9. 4-hich diameterflow-turned cylinder from this series of experiments.

The next phase of the flow-turning program was to evaluate (1)the effect of scaling from 9. 4-inch to 14-inch diameter cylinders,(2) the effect of changing from a single roller to a two-rollerflow-turning machine and (3) to deter ne whether the feed-

PAGE NO 23

I

PRATr WHITNY AIRCRMAT PWA- 2267

reduction parameter optimized in the previous series o testswould hold constant with the same roller configuration.

The change frcm a si-g1e-roller to a two-roller machine was a nec-

essary step toward flow-turning full scale parts. The roll forces

necessary to flow-turn B-120 VCA titanium alloy produce severemandrel deflection on the single roller machine to the extentthat a negative roller-mandrel gap setting is necessary to pro-duce a 50 per cent reduction during the second pass. Thisobjectionable machine deflection can be reduced by using ahydrospin machine with two opposing rollers. Reducedmandrel deflection allows better control of reduction and wallthickness. In transferring the flow-turning from the one-rollerto the two- roller machine, the optimized roller feed of 0.020-

inch per mandrel revolution per roller was applied to both rollers,thereby doubling the carriage feed in inches per minute. Theremaining 9.4- and 14-inch diameter blanks fabricated fromrolled and welded plate stock failed in the circumferential flangeweld when flow-turnine ",aF ,attempted on the two-roller machine.

Failure was attributed to the increased axial flow-turn load im-posed by the two-roller action. Since the use of rolled andwelded bianks was only for expediency, there was no furtherattempt to develop this type of blank.

C. Development with 14-inch Diameter Roll-Forged Rings

The next series of experiments were conducted on ten 14-inchdiameter roll forged rings. The intent of these experimentswas to evaluate the effect of scaling-up in diameter, to evaluatethe performance of the parabolic contoured roller with a two-roller machine, and to optimize flow-turning parameters inpreparation for fabricating 40-inch diameter full scale cylinders.

The ten 14-inch diameter roll-forged rings were solution treatedat 1450F for 30 minutes, water quenched, and machined intoflow-turn blanks. Six of these rings were flow-turned utilizingthe optimum technique as determined by previous experiments,including the parabolic roller configuration. All of the flow-turning was accomplished on a two-roller 42 by 50-inchCincinnati Hydrospin machine. The flow-turning parametersand cylinder dimensions are listed in Table XXIII.

The first blank was flow-turned close to the previously deter-mined optimum feed-reduction parameter of 0.010, using afeed of 0. 020 inch per mandrel revolution per roller and

PAGE NO 24

PRATT 0 wwffNgY AggcmRr PWA-2267

reductions of 51.8 per cent and 49.5 per cent on the first and secondpasses. This cylinder exhibited good dimensional stability on bothpasses which showed good correlation with previous results andindicated that the feed-reduction parameter was valid with increasingdiameter and a change frorm one to two rollers. However, metaltearing occurred on the inside surface during the first pass, mostlywithin one inch of each end of the flow-turned section, as shown inFigures 33 and 34.

Forged rings numbers 2 through 5 were flow-turned with changesI. in the various flow-turning parameters to determine which para-meter was most effective in reducing inside surface tearing duringthe first pass. The changes in parameters during the flrzt passwere, 1) increased roller tip radius, 2) less reduction, and 3)reduced roller-feed rate. These changes were applied singly andevaluated on the basis of incidence of inside surface tearing. Theresults of this experiment showed that reduced roller-feed ratewas most effective in reducing surface tearing although the sur-face tearing was not completely eliminated.

It appeared that the flow-turning practice was not the basic causeof the surface condition and thus attention was diverted to thestarting material. The forged rings had been solution treated at1450F. Early studies uf ring-forging annealing practices hadshown improved ductility after solution treatment at 180OF andwater quench. This latter treatment was not ,sed for the forgedrings, however, since tests conduL.ted on 9.4-inch rolled andwelded cylinders had indicated poor aged ductility after 1825F solu-tion treatment and subsequent flow-turning. As described in sec-tion I1-B, test specimens from full scale rings were solutiontreated at 1300, 1450, and 180OF for from 15 to 30 minutes andresults from testing these specimens (Table XVIII) indicated thatgreater tensile elongations and reductions in area result fromhigher solution treatment temperatures. Additional tensile pro-perties were then determined for full scale test ring materialwhich had been solution treated at 1450F and subsequently re-solu-tion trcatcd at 180OF for various times. The results (Table XXIV)again demonstrated the excellent ductility attained with an 1800F-solution treatment and indicated that maximum ductility with mini-mum grain coarsening was achieved after 15 minutes of treatment.

PAME No 25

SATY W"gTPITY AIRCRAFT PWA-2267

On the basis of the annealing treatment evaluation, 14-inch dia-meter ring number 6 was given an additional annealing treatmentat 1800F for 15 minutes followed by a water quench, and wasflow-turned during the first pass with a reduced roller-feed ratesimilar to that used with ring number 5. Visual and nondestructiveinspection showed cylinder number 6 to be free from surface tearsafter the first pass. The improved ductility of the starting blankfrom the 1800F aztnealing treatment was apparently required toeliminate surface metal tearing with the feed rate used to mini-mize diametral growth.

The effect of the changes in flow-turning parameters on diametralgrowth with cylinders I through 6 is shown in Figure 35. Thetearing condition during the first pass did not affect the flow-turningresponse to these changes, and all cylinders were given the secondflow-turning pass to evaluate changes. With a roller tip radius of0. 062-inch (the same as that used for 9.4-inch diameter cylinderexperiments), a feed-reduction parameter of about 0. 010 stillappeared to be optimum for eliminating diametral growth duringboth flow-turning passes. Rollers with larger tip radii of 0. 100and 0. 220 inch had been fabricated to evaluate methods for elimi-nating the severe thread pattern produced by the sharp 0. 062-inchtip radius, These rollers had the same parabolic lead contour(see Figure 26) blended with a larger radius to produce a blunt tip.The rollers with a tip radius of 0. 100-inch were used for the firstpass of cylinder number 2 in an effort to eliminate surface tearing.This did not correct the problem and the diametral growth increasedconsiderably. The rolle-s with larger tip radii were uld in allother cases during the second pass for the purpose of improvingthe prior surface finish. These rollers did produce an improve-ment, although they did not completely eliminate the thread pattern.The diametral growth, however, increas-d with an increase inroller tip radius, giving a unit growth above 0. 003 inch per inchof diameter at a feed-reduction parameter close to 0.010 with aroller tip radius of 0. Z20 inch. Local bulging of the cylinders wasfound to be a direct result of an increase in blank diameter frommandrel size as a result of diametral growth. In the absence ofdiametrai growth, local bulging did not occur. Based on theobjectional growth trend, further evaluation with the 0. 100 and0. 220 inch tip radii rollers was discontinued.

PAG9 NO 26

PRATT W01HaNE AIMCPAPT PWA-2267

[

Tensile tests on specimens from cylinder number 6 following thesecond pass and a stress-relieving treatment showed low ductilitydue to an inside surface condition (yielding and intergranularseparations) throughout the flow-turned area. Microexaminationof material from the flow-turned ring showed that incompleterecrystallization had existed adjacent to the inside surface priorto the second pass and grain boundary precipitate was abundantthroughout, with evidence that the precipitate existed prior tothe second pass. It was apparent that cylinder number 6 had notreceived a satisfactory annealing treatment between flow-turnpasses. An improved annealing practice was developed on sub-sequent cylinders.

Numerous pits were found on the inside surface of cylinder number6. and to various degrees on cylinders I through 5. These pitswere considered to be caused by metal being transferred from theblank to the mandrel, Figure 36. To prevent metal transfer, allsubsequent cylinders were coated with a colloidal graphite suspen-sion (Fel Pro C-2O0). Pitting was significantly reduced and inmost cases eliminated.

Ring number 7 was annealed at 180OF for 15 minutes followed bya water quench prior to flow-turning. The first flow-turn passwas performed with a reduced roller feed rate similar to thatused for cylinder number 6. The inside surface was coated witha colloidal graphite suspension to prevent the transfer of materialfrom the inside surface of the blank to the mandrel. Visual andnondestructive inspection showed this ring to be free of tears onthe inside surface after the first pass. Between the first andsecond passes the microstructure of a section of the ring wasexamined after various annealing treatments, and 1500F for 30minutes followed by air cooling was established as the optimumannealing practice prior to the second pass. After the firstflow-turn pass, an area was blended to a depth of 0. 025 inch tosimulate removal of a surface defect that might occur during thefirst pass operation. Cylinder number 7 was then given tlesecond 'flow-turning operation, (see Figures 37 and 38). Thedepression was completely filled and no cracking or unusual be-havior was noted. This cylinder was used to evaluate the effectsof the 180OF anneal on the aging response characteristics.

PAGE -40 27N* ,,o 7.

S

PMATT WHITN9 AINCRAP PWA-2267

Ring number 8 was also annealed at 1800F for 15 minutes and waterquenched. The first flow-turn pass was performed with an in-creased roller feed rate necessary to obtain the desired feed-re-ductior, parameter and to determine whether the modified annealingcycle wonld prevent surface tears at this condition. The insidesurface vas coated with colloidal graphite as before. Visual andnondestructive inspection of the inside surface of the ring after thefirst pass showed no tears. However, there were some cracks onthe outer surface of the ring at the start of the flow-turned area.These cracks were attributed to a machine malfunction at thebeginning of the flow-turning cycle and were not found to be detri-mental through the remainder of flow-turning operations on thisring. The modified intermediate annealing cycle developed forcylinder number 7 was applied to cylinder number 8. No difficultywas encountered during the second flow-turn passes.

A roller tip radius of 0.062 inch was used in flow-turning cylin-ders 7 and 8 and there was essentially no diametral growth ofeither. Each second pass included contour flow-turning to pro-vide a heavier section for the weld joints as shown in Figure 25.Cylinder nunaber 8 demonstrated that the material condition wasnow optimum and that flow-turning could be accomplished withparameters that produced very little or no radial growth and nolocal bulging.

Subsequent evaluation of the tensile proper.ies of cylinders num-bers 7 and 8 (discussed in section V-D) showed erratic propertieswhich were believed to be a result of roller misalignment duringflow-turning. It was found that the Hydrospin machine was notcapable of keeping the two opposed rollers opposite one anotherunder the loads required to flow-turn B-120 VCA titanium. Ana-lysis of the machine behavior indicated that one roller was axiallydisplaced from the other in excess of 0. 125 inch under extremeoperating conditions. Although roller misalignment does not or-dinarily have adverse effects on most steels, it is detrimental tothe aged properties of B-120 VCA titanium alloy. The optimumpractice established under this program requires an effective feedrate of 0. 020 inch per revolution per roller. An aiignment errorof 0. 010 inch or greater would cause the lagging roller to deformmaterial that had been cold-worked by the leading roller. In addi-tion, the lead roller would carry the entire working load, resultingin the flow-turning operation oeing performed at double the effec-tive feed rate. The combined effects could cause microcracks onthe outside surface as were noted in cylinders numbers 7 and 8.It is this condition that was considered to be the cause of erratictensile behavior in the axial direction with these cylinders.

PAGE NC 28

I~ f WMnrM junc*-, PWA-2267

'IA study was made to determine the revisions necessary to theHydrospin machine to provide better roller alignment and thus amore accurate control of effective roller feed. To analyze themachine behavior during the flow-turning operazion, pressuregages were installed in the hydraulic loading systems on bothroller carriages. These gages indicated that the master c-rriagewould consistently assume the major portion of the flow-turningload although the rollers were accurately aligned under no-loadconditions. A dial indicator was mounted on a support extendingbetween both roller carriages and b-idging the mandrel to mea-sure roller misalignment under load conditions (see Figure 39).The dial indicator showed that the slave roller would lag behindthe master roller under feed ioads. The amount of misalignmentwould vary with feed load with a maximum of a 0.200-inch ris-alignment recorded under maximum feed loads. Using the bridgedindicator to measure the relative position of the rollers, it waspossible for the operator to advance the slave roller into alignmentwith the master roller during the flow-turning operation throughthe use of the machine controls provided. This method was foundparticularly helpful in maintaining roller alignment during contourflow-turning of reinforced sections for weld joints when feed loadswere changing. The effectiveness of this roller aligning metl.odwas indicated by the pressure gages which showed approximatelyequal hydraulic loading of the carriages.

The last two subscale forged rings, numbers 9 and 10, were flow-turned to determine the effects of improved roller alignment anda modified roller configuration. These rings were given the initialsolution treatment and between-pass annealing treatment establishedwith the previouf. cylinders in this test series. Both cylinderswere flow-turned with a slightly high feed-reduction parameter of0.011 due to reductions of 55. 2 and 51. 5 per cent respectively.A modified roller configuration was designed to improve finalsurface finish. The pronounced feed lines produced by the pre-viously used rollers with a sharp tip radius of 0. 062 inch had notcaused any apparent reduction in axial tensile strength, butimproved surface finish was desired for full icale parts. Sincerollers with increased tip radius had caused too much diametralgrowth, new rollers were prepared consisting of the same para-bolic configuration but with a modified tip as shown in Figure 26.

PAGE -o 29

I

PeAlT a W14TNC JARcArar PWA-2267

Blank number 9 was given the first flow-turn pass with the modifiedrollers and the original sharp 0. 062-inch tip radius rollers wereused for the first pass with blank number 10. The results showedthat the modified rollers significantly improved the surface finishwithout affecting dimensional control. It was expected that themodified rollers might give some increase in diametral growth,but the data showed the opposite trend. The unexpectedly highgrowth with cylinder number 10 was attributed to roller misaligL-ment. During the first pass of cylinders 9 and 10, an attempt wasmade to solve the roller misalignment problem by reducing rollercarriage feed. This was done because it was believed that the car-riage hydraulic system flow capacity was marginal at the previouslyused 4 inches per minute per roller carriage feed. Since the mar-ginal hydraulic flow capacity was believed to contribute to rollermisalignmenta lower feed rate of 3 inche- per minute per rollerwas used during the first pass with the mandrel speed reduced to

150 to maintain the 0.020 inch per revolution per roller effectivefeed. A mandrel spe