RK 618

Technical Report =:

LIGHT HOUSINGS FOR DEEP-SUBMERGENCE

APPLICATIONS—PART III.

GLASS PIPES WITH
CONICAL FLANGED ENDS

March 1969

Sponsored by

NAVAL FACILITIES ENGINEERING COMMAND
NAVAL CIVIL ENGINEERING LABORATORY
Port Hueneme, California

This document has been approved for public
release and sale; its distribution is unlimited.

LIGHT HOUSINGS FOR DEEP-SUBMERGENCE APPLICATIONS—
PART Ill. GLASS PIPES WITH CONICAL FLANGED ENDS

Technical Report R-618
Y-FO15-01-07-001
by

K. O. Gray and J. D. Stachiw

ABSTRACT

The objective of this study was to evaluate commercially available
glass pipes with conical flanged ends for application as transparent housings
for underwater lights and instruments. Flanged-end glass pipes in diameters
ranging from 1 inch to 6 inches and in lengths ranging from 6 inches to
36 inches were imploded under short-term and cyclic external pressure
loading. Collapse pressure and recommended operational depth data are
presented for one-way trip, round-trip, and cyclical applications. Recom-
mendations for end-closure systems are also presented. One prototype
light design and one prototype instrument housing design are described.

distribution is unlimited.

fic & Technical
dringfield, Va. 22151

CONTENTS

page

RROBBEN a STPATHEMIEN Ts me cnor tars cise l i iech ye) oe Me a ie 1
STU DVGOBYECTME nie Gee Saar 2° NN, Men ae A Saas sel keer 1
RW RiR@S Ea @ eS N i mey camera, ct ee arni Se Meo aeretys pace aR Cr 2
EX RE RIMENTAES PRO CEDURES S: fai alse ee tee) bale 3
Phase |: Short-Term Critical Pressure of Glassware. . . . . 3
fainaselllienGaskemiSStenmblkestouivecehe me siiteseer i wet niente ates ateen atin 9511(0
phaser PrOtotnyOcus OUSIMGMRCSTS <i) eee eee ern et oO
NIGDESPO are AIRE ete eat Mi tec se Ree Cale ae WO
PRED CONOR GRiIlmlCALSP RIESSUIRIES ieee sites nub soi ei aeSG
SETIVINA VAN IN Gate ae Olea a eyed Weer ee Ca NR PU ME I tt At gta
CON CW SHOIN eaten cee tected eee se AUN Meme. oobagr a Mieun G71)
RIECOMINIEIN ED ANd @ INS? Satara yeast eo einen ky elo Ae a wea Ane tA
Fries faut trait NCA Sri lm aM tue ara alli ece, scm ai ede MeN te eae GBS aN Maclay
SUPE WG NEE Sie UNeieraeinies du tect tame: lc nia gee amseatt tn. AA

mM (WN MN

0301 OO4O4O?
iil

PROBLEM STATEMENT

The study, exploration, and utilization of hydrospace require the
use of instruments and devices which must be ‘‘packaged”’ in sealed housings
to protect them from the high-pressure seawater environment. While a few
such housings are available as off-the-shelf commercial items, the variety is
small, the cost high, and delivery time is frequently long.

Lights for hydrospace illumination are a fundamental requirement
for either direct or indirect visual observation of any type of undersea
activity. Only in the shallowest portions of the sea, and then only during
sunny days, is the natural light adequate for most observational purposes.
The majority of lights and lighting systems require a transparent, pressure-
resistant housing for the proper functioning of the light-producing element.
Certain other devices such as light sensing or measuring instruments also
require pressure-resistant transparent housings. These requirements immedi-
ately point to the possible application of glass as a housing material since it
is transparent, has high compressive strength and in certain formulations is
resistant to thermal shock. Glass has an additional attraction in that it is a
nonmagnetic material and is transparent to many types of electromagnetic
radiation that are attenuated by metallic housing materials. The more
obvious advantages of these properties are the applicability of magnetic
and photo cell techniques for operating instruments in such packages and
for the transmission of data through the instrument housing walls without
the requirement for physical penetrations of the housing or end closures.

STUDY OBJECTIVE

This study is a continuation and extension of an investigation into
the applicability of commercially available off-the-shelf, dome-shaped glass
containers to the design of underwater lights.''2. Additional potential
applications for pressure-resistant glass housings caused the investigation to
be broadened to include larger commercial tubular glass shapes for both
lights and instrument housings. After evaluation of selected glass pipes of
various diameters and lengths, prototype light and instrument housings were
to be designed and fabricated.

PURPOSE OF STUDY

Commercially available off-the-shelf glass pipe used in the chemical
industry is available in a variety of lengths and diameters (Figure 1). Due to
the wide use of this material in the industry, it is mass produced-.and is much
less expensive than either custom-fabricated or limited-production housings
of similar dimensions fabricated from either metal or glass. In addition, the
widespread and continuing use of standard glass piping assures a continuing
supply of standardized sizes and formulations. Thus, if successful designs
of underwater light and instrument housings based on these types of glass-
ware can be developed, low acquisition and replacement costs can be
guaranteed for the future.

The purpose of this report is to document the experimental study
of readily available glass pipe for use in underwater housings, to present
data on its performance, and to offer a limited number of designs for
instrument and light housings utilizing glass pipe as the main component
of the housina.

Figure 1. A sampling of the variety of glass pipe diameters and lengths
tested in this program.

EXPERIMENTAL PROCEDURE

In order to satisfy the objectives of the overall study, it was necessary
to embark on several phases or substudies. These were (Phase |) determination
of the relative critical pressure of various diameters and lengths of glass pipe,
(Phase Il) determination of the best end-closure-to-glass bearing and sealing
system, and (Phase II!) the testing of prototype light and instrument housings.

Phase I: Short-Term Critical Pressure of Glassware
Relative critical pressure was determined by preparing test assemblies

of four specimens of each selected diameter and length combination. Table 1
shows the dimensions and quantity of specimens evaluated in this first phase.

Table 1. Dimensions and Number of Specimens Tested
in Phase | of Study

Inside Pipe Length* (in.)
Diameter
(in.)

* Lengths of up to 120 inches are available.

Figure 2 shows the component parts of a test assembly. Figure 3
shows one specimen of each length of 4-inch-ID pipe made up into a test
assembly and Figure 4 shows 6-inch lengths of five different diameters of
pipes made up into test assemblies. Due to the complete destruction of
the metal flange assemblies, which generally took place when the specimens
imploded, a simplified, less costly hold-down system was devised. This

system consisted of three small pieces of Plexiglas which clamped the glass
to the end plates. This simplified system (Figure 5) or a tie-rod system was
used after the first few tests.

The test assembly end plates, fabricated from 6061-16 aluminum,
were designed to be of sufficient thickness (a) to provide a rigid end closure
capable of withstanding pressures up to 20,000 psi, (b) to prevent end-plate
thinning (resulting from the metal removal in refinishing the seating surfaces
after each use) from weakening them excessively before the experiment was
completed. Remachining was necessary after each use because of the damage
done to the sealing surface when the specimens either imploded (Figure 6)
or failed by cracking without implosion.

Figure 2. Component parts for a short-term critical pressure test assembly.

The initial thickness of the aluminum end plates for each diameter
of pipe was as follows:

Inside Pipe Diameter (in.) Plate Thickness (in.)

1 0.910
1/2 1.365
z) 1.820
3 2730
4 3163
6 5.45

Figure 3. Four-inch-ID glass pipe test assemblies made up of 6-, 12-, 18-,
24-, and 36-inch lengths.

Figure 4. Six-inch-long glass pipe test assemblies made up with 1-, 2-, 3-,
4., and 6-inch-ID pipes.

The assembly process was accomplished by first applying room
temperature vulcanizing (RTV) silicone rubber paste to the glass pipe flange
(Figure 7) then inverting it and placing it on the end closure (Figure 8) using
the bolts as alignment pins to center the pipe on the end closure. Figure 9
shows the finished seal just before the bolts were torqued. The bolts were
torqued to a uniform load as shown in Figure 10. After about 24 hours,
the glass pipe was partially (85% to 90%) filled with water (to reduce the
force of the implosion) and the top end closure was assembled in a similar
manner. The assembly was then allowed to stand for a minimum of
24 hours before testing.

The standard procedure for testing specimens which would fit into
the laboratory pressure vessels was to suspend them from the pressure vessel
head (Figure 11) and then place them in the vessel. The vessel was filled with
water and then pressurized by means of air-operated piston pumps at a rate
of 100 psi per minute. Pressurization was continued at a constant rate until
the test assembly imploded, leaked, or the pressure reached 20,000 psi.
Pressure was then relieved, the specimen removed from the vessel, and a
detailed examination made of the remains.

Figure 5. Simplified hold-down system
used in majority of short-
term critical pressure tests.

Figure 12 shows a 4-inch-|D x
12-inch-long (90% water-filled) speci-
men assembly and the remains of a
similar (except air-filled) assembly after
failure by implosion. The violence of
the implosion of the air-filled specimen
is shown by the granular nature of the
remains of the glass pipe (the small
pile of white material seen between
the end closures).

The 4- and 6-inch-diameter
pipes that were too long to fit into the
NCEL pressure vessels were assembled
with tie-rods (used between the end
plates) and were left completely filled
with air (no water inside). These pipes
were tested by taking them to sea and
lowering them into the ocean until
they reached implosion depth. The
violence of the implosion was suffi-
cient to be heard aboard ship by
means of a hydrophone hung in-the
water below the ship. Thus, the time
of the implosion was known, the wire
angle and the amount of wire necessary
to reach implosion depth was noted.
The depth of implosion was then
derived. This.depth was then converted
to critical pressure.

The technique of mitigating equipment damage (from the violent
implosion of void specimens) by almost completely filling the test specimen
with water brought up the following questions:

1. Was sufficient void space being provided to compensate for the
volume reduction of the glass pipe—end closure system at high

pressures?

2. Did this technique have any effect on the critical pressure of the

specimens tested?

9 10 in

2 3 4 5 i ie , :
Deep Ocean Laboratory °* oineering Dw.

Figure 6. Damage to aluminum end-closure plates resulting from the
implosion of a 4-inch-I|D, 6-inch-long glass pipe at 16,900 psi.

In order to answer these questions, specimens of different sizes were
provided with end-closure plates, which were drilled and tapped to receive
a high-pressure tubing connection. The specimens were then assembled,
filled completely with water, and the high-pressure connection brought
through the pressure vessel end closure (Figure 13). This line was then
connected to a graduated cylinder so that the fluid expelled from the
specimen assembly in response to increasing external pressure could be
measured. Figure 14 illustrates this setup. A graphic plot of the volume
change versus pressure for three 4-inch-!D x 6-inch-long glass specimen
assemblies is shown in Figure 15.

Experimental data from testing 4-inch-|D x 6-inch-long glass pipe
in the manner just described showed the following:

1. The net reduction of volume of the glass pipe, under an external
hydrostatic pressure of 20,000 psi, was approximately 40 ml or 3% of the
volume at atmospheric pressure. (The void space normally left in specimens
tested for critical pressure amounted to about 10% to 15%.)

2. The critical pressure of glass pipes partially filled with water did
not vary significantly from those filled with air at atmospheric pressure.

‘sajejd aiunsojo-pua winulwinye “Ajquuasse uawwioads $o uolesedaid Bulinp sabuely
01 adid ssej6 pajeoo-saqqns-auoaijis yo Ajquassy “g einbi4 adid ssej6 0} saqqns auoaljis ALY $o uoNeoddy ‘7 ainbi4

Figure 9. Appearance of glass-to-metal contact area and RTV silicone rubber
seal before the tie bolts were torqued.

Based on this information it was concluded that the practice of
partially filling the specimens with water did not affect the test results as
long as a minimum of 10% to 15% of the total internal volume of the test
glass specimen was filled with air at atmospheric pressure when the speci-
men was assembled.

The test results, presented in Table 2, indicate that ‘‘off-the-shelf”’
glass pipe of the type tested will provide reliable, transparent, nonmagnetic
instrument or light housings for a single submersion to any depth to be
found in the ocean, provided the appropriate diameter and length are
chosen. Relatively large housings (6-inch-|D x 18 inches long) are useful
for at least one cycle to depths of 5,000 feet with a safety factor of 2
when used in a system with simple 6060-T6 aluminum end plates.

Phase Il: Gasket System Tests

The second phase of the study consisted of testing a number of
different glass pipe-to-end closure bearing and sealing systems. The objective
was to develop an end-closure system capable of withstanding the maximum
critical pressure and satisfactory for repeated exposure to external hydro-
static pressure (pressure cycling service).

10

Figure 10. Use of torque wrench to avoid damage to glass pipe through
uneven loading of glass flange area during assembly.

11

Figure 11. Vented glass pipe specimen assembly attached to head of
18-inch pressure vessel (left).

In order to standardize the test procedure and to eliminate as many
variables as possible, all tests in this phase were conducted with 4-inch-ID x
6-inch-long Pyrex pipe with flanged ends. The glassware was procured
through normal commercial glassware supply channels and was used in the
“as received” condition.

Five specimens each of the following simple glass pipe-to-end closure
plate combinations were tested to determine critical pressure using the same
procedures as described under the Phase | tests.

Aluminum 6061-16 3.63 inches thick
Stainless steel type 316 2.58 inches thick
Brass (naval) 2.58 inches thick
Titanium (Ti-6AI-4V) 1.82 inches thick
Phenolic resin—glass fiber laminate 4.40 inches thick

Figure 16 shows typical samples of the end closures used.

The variation in end-closure plate thickness from material to material
isa result of the relative strengths of the materials and the calculated thick-
ness required to resist failure at an operating pressure of 20,000 psi.

Table 3 shows the results from short-term critical pressure tests with
the various end-closure materials.

2

1

; Deep Ocean Ei vineening Di :

Figure 12. Air-filled 4-inch-ID x 12-inch-long glass pipe assembly before
and after implosion at 10,400 psi. The white, granular material
is all that remains of the glass pipe following implosion.

13

vent tube

9-inch-ID
pressure vessel

Figure 13. Typical water-filled specimen assembly showing connection through
the end closure to the pressure vessel head and the atmosphere.

Figure 14. Typical setup used to measure the water displaced from and detect leaks in a
test assembly during pressurization. The displaced water is collected and
measured in the graduated cylinder at intervals of 1,000-psi pressure increase.

14

66-1 anbojeyed siour|||-suamo 44

00Z-3d anbojejyes ssejH Buluson ,,
*payoeid Ajaianas diam | |e ‘papojdwi! auoyy §
*pajsal JoN =
“Aj YBijs Payoes9 auO ‘papo|jdusi auoyy 4
“pajley BUON x

(€] (v] (v]

OSv'eé | E€E6b COOLS O08'r 0098 O0z'Z | OS9'8L 00002 00691 | ++
[v] (v] {v]

ooo'e | 89Z'p OOLL OE 0066 O0SZ'6 | O09V'9 002'6 o0S'z

XBW) YOU |-p

xalAg Yyou|-g

(Z] [€] [v]
ool'r | SZ48'9 0069 o0sS8'9 OOZ'LL OS0'8 g000'0Z

xasAg YOU|-p

{v] [v7] [v7]

; ; ‘ : ; : ; 3 c xaiAg YoUl-
oos'é | O8&S O0SS O0ES OO 006;S.- 029 o0c Zt OO” 9L

{v] [€]

- fs é 5 : rE ‘ F xalAg YOU -Z
OO 8 GLL6 00S 6 000 6 006 8 OS9 tL OSZrl OO9VL

{v] {v] xalAd
OOv'6L | GES'ZL OO6'ZL OOr'ZLL OOr'8L g 000'0z YoU |-Z/L-L

[v7]
+000'0Z

t
BAY xe Ul Bay xey Ul BAY xeyy Ul, Bay xey| Ul Bay xe Ul
adig jo adA}
Jajaweig apisu|

—}o yiGua7 Jo} (Isd) ainssasg (eo

1 eXOAg YOU I> |

("pajsa} suauloads Jo Jaquinu juaseidai— [|] ‘ajdwexa 10}—ssaquunu paja49e1q)

sa]e|g PUW WINUILUN|Y 91-1909 YIM edid sse|5 pebue|4 $o ainssesg jeaig W4a|-WOYs °Z ajqeL

is

“aunssaid jo aseajai

ay} UO Jede |Ja} Pue payxoesJD UaUUIOads BUG “JJede |/e} JOU
PIP 1Nq payxoesD Ajasanas asa SUBWIDads OM] “WIJ ApIsu! BU}
JJo payjeds diyo absey e ydaoxe Bulyoe419 OU UaLUIDads AUD 104

*JASSOA AY} WO1} PEAOWAI UBYM JJede aed pue paxde19
Ajaianes a4am Suauloads aaiy| “papojduu! uauuloads auQ

"JASSAA BY] WWOI, PSAOWWS4 UBM JJede awed pue
payxoe19 Ajajanas alam suawioeds om “papojduu! uawioeds 8uQ

*pasea|al

sem ainssaid ay} UaYM Yede {ja} pue PayxdeID AjasaAas 819M
1Nq epojdui! oO} payley suawoads OM | “UO!eZIUNssoid Bulinp
‘Ajaniyoedsas ‘OG’ pL Pue OOG'ZL 1e Papo|du! suawisads om |

“Buljse} Jaype Uatuloeds
jeoidAj © smous G ainbi4 “peraijei sem ainssaid |1jUN yea] JO
apo|dwi you pip ing Uo!Nezinsseid BuLunp paxodeso suaUIoeds | |\7

+000' 02 +000' 02

+000'0Z

+000'0Z

+000'0Z

+000'0Z +000'0Z

+000'0Z

+000'0Z

ayeulwUe| Jaqiy ssejH
poyeubaidui-ulsad 91jouayd

sseiq |PAEN

Av-l\W9-1L day

wuniuey |

OLE adAj |aals ssajule1sS

WNUIWN|e Q1-L909

BOUBUOJIJAg [EANIONIIS UO sayO,)

BAY

(sd) ainssaig jeoIW9

sjeliajey snole, $O SaINsO|D PUA YIN paisa, UaU\
suaul9ads adig sse}9 buo7-Yyou|-g x GI-You|-p $O aunssaig JedI}UD Wa] -WoOYS “E a)qe,

jeLayeyy 22} g-PUy

16

External Hydrostatic Pressure (psi)

20,000 | Ti Py The 6061-T6 aluminum and the
iw aay 4
oastenesien i y eeratoyite resin aimpreginatce glass fiber
18,000 imploded at Revi al laminate materials performed the best
18,000 psi —=< |n under a single pressurization to implo-
16,000 wo sion. However, the cost of both the
EL. a
a basic material and the machining of the
Yo
14,000 A glass laminate end closures is much
f greater than that of the aluminum end
12,000 i closures. For this reason the 6061-T6
ij aluminum is considered the best choice
10,000 if based on these tests.
iY The second series of five speci-
8,000 L Note: — mens of each glass pipe—end closure
di Internal volume ; 2 :
1,200 ml at | combination was tested for cyclical
6,000 (evecsal f atmospheric pressure | performance. Cyclical tests started
YW | | with 10 cycles from 0 psi to 5,000 psi.
4,000 Hal bacon | Specimen assemblies were then dis-
LM O 94-7 specimen assembled and examined, and the
2,000 Me ls OQ 948specimen | mext cycling pressure was selected on
M/ © 107-2specimen | the basis of the just completed test.
0 / | | Subsequent cycling tests at successively
) 10 20 30 40 50

Accumulated Volume Change (ml)

Figure 15. Graphic plot of the volume of

water displaced from three
water-filled, 4-inch-ID x 6-inch-
long glass pipe assemblies
subjected to increasing hydro-
static pressure.

lower pressures were conducted as
indicated in Table 4. The pressuriza-
tion rate was held at 1,000 psi per
minute in all tests.

Based on these tests, naval
brass appears to be the best choice
for limited cycling service under this
set of conditions.

A third series of experiments was designed to evaluate the effectiveness
of various ‘‘gasket'’ (sealing and bearing) systems designed to interpose a ‘‘soft”’
bearing and sealing element between the end of the glass pipe and the type 316
stainless steel end-closure plate selected for these tests; effectiveness was gaged
by the increase in the cyclical depth range of the glass pipe under test.

The gasket systems were designed to serve three functions. (1) to
provide a watertight seal between the glass pipe and end-closure disc at both
low and high pressures; (2) to eliminate stress concentrations due to point
contact between the somewhat uneven glass bearing surface and the steel
end-closure material; (3) to provide a ‘’mobile”’ surface on which the glass
pipe end flange could move as the pipe diameter decreased in response to
external pressure.

lv

Aluminum 3 i :
6061-16 a 4

Stainless Steel

-m6

Nonmetaitic j
aiagg enolle-Resin .
~Fiber Laminate oe ~

Brass = Titanium

Naval Brass Ti-ai-ay_

Figure 16. Typical end closures used to evaluate the effect of end-closure
material on short-term critical pressure of 4-inch-ID x 6-inch-
long glass pipe with flanged ends.

Since the purpose of this series of experiments was to enhance the
usefulness of the glass pipe in terms of numbers of use cycles and depth
limits, an arbitrary minimum acceptable short-term collapse pressure of
3,000 psi was adopted. Systems which failed at pressures less than this
were not further tested to determine what their maximum depth rating
would be for cyclical service.

The gasket systems tested were as follows:

1. Soft (60 durometer) 3/32-inch cross section O-rings were placed
in the grooves in the glass pipe flanges, lubricated with silicone grease and
then compressed between the glass pipe flanges and the steel end plates by
tensioning the tie rods holding the assembly together. (See Figure 17.)

In each test assembly the O-ring acted as a wedge and sheared off
the external flange lip from the pipe (Figures 18 and 19). This occurred
during the first cycle to 3,000 psi for each pipe. Thus the maximum useful
depth rating for 3/32-inch O-rings is not known. Substitution of thinner
O-rings may eliminate the wedge action, as was shown subsequently in the
successful light housing design (Figures 33 and 34).

18

"S1s9} d1nssaid 91/99 40 saluas yoea 10} pasn asam sajejd Pua paysiuljas pue uauiloads adid ssej6 mau y :a10N

{0Z] °388} BuljDAd isd-Qo0'¢E ut
pasn se Ajquuasse Uawuioads
awes ‘sse|6 0) abewep on

(Isd QOO'E 1e Sa]9A9 QZ uel)

[S] ‘sse|6 0) abewep oy
(isd QO0'E 3e Saj9A9 G_ :ue}d)

[8] ‘spua 4yy0q ul Buljjeds
AWOS Pue $yx9e49 BUY 'a[DAD
yiybie BuLinp payiey yuejeas

(Isd QO’ Je Saj9A9 Q| uel)

[L] ‘a6uejy aroge yout |
}noge spua Yio ul punoue
Aem au} |} $49e49 'a/9A9 puo
-9aS Jalye paxea) pue payxde419
(isd QOG'E 12 saj9A9 G_:ue}d)

[SG] ‘abewep sayjo ou ‘apisut
Bul|jeds aul, awos ‘anoos6 Bul
-Q pue Wid apisul uaamjaq
spua y}0g ul s49e49 auly Asaf,
(isd OOG'E 1e sajDA0 G :ue} qd)

[Z] ‘punose Aem

ay} || YBnosyy payxoes9 ssej6
‘aj9A9 pully} Bulunp ainssaid
Bulseasoul sapun payeaq
(isd QOG'E 1e sajDA0 G :ue| dq)

{L] ‘aj9A9 puosas

Bulunp asnssaid Bulseasout
Japun isd QQ6'Z 1e papojdw|
(isd Q00'p 3e saj9A9 G :ue|g)

[6] ‘aj9A9 YOL

Bulinp asnssaid Burseasoul
Japun isd QOL't 1e papojdiw|
(isd QOG'p 3 SajDA9 QL ‘ue|g)

(G] ‘abewep say,0 40 Bulyjeds
Ou ‘aA00J6 Buls-C 0} jae
-4ed spua Yy0q ul Sy¥9e19 aul
(isd Q00'p 12 Saj9A0 G :ue|q)

{S]
‘abewep Jayy0 ou /an00j6 Bui

-O Pue Wi apisul uzamjaq
pua auo ul $49e19 aul} Asaf
(isd OOO'p Ye SajDA9 G :ue|g)

[G] ‘abewep sayj0 ou
‘Wd JEBU BPISUI UO 49e19 BUC)
(isd QOG'p 32 saj9A9 G :ue}q)

—

[L] ‘a}9A0 auo
Jajje paxea| 'syoe19 Aueyy
(isd QOG'p 1 sajDA0 G :ue}q)

[S] ‘ssej6 0) abewep oy
(isd QQG'p 3€ SajDA0 G :ue}g)

(G] ‘abewep

4Jay}O OU ‘apisul Buljyjeds
awos 'an0os6 Buls-Q 0} jale
“Jed spua y}og ul $49e49 aul 4
(isd 00G'p 12 sajA0 Gg :ue}g)

[S] ‘eBewep sayj0 ou ‘anoos6
Buls-CQ pue WH apisul Uaamjaq
pua auo Ul 49e49 auly Ala),
(isd 00G'p 1 SajDA0 G :ue|q)

[1] ‘a)9A9 puosas

Bulinp ainssaid Bulseaioul
Japun isd OOO’ 1e papojdwy
(isd QOG'p 1€ SajDA9 G :ue}d)

S}[NSaY pue ue|g Isa]

{L] ‘aj9A9 auo
Jaye payea| 'sydes9 Aueyy
(isd QO00'G 12 $aj9A9 G -ue}g)

[S] ‘ssej6 0) abewep oy
(isd QQ0'g 1e Sa}9A9 G :ue}g)

[G] ‘ebewep

4ayio Ou ‘apisu! Buljjeds auly
aWos '38A00I6 Buls-C 0} ja}Je
-Jed ‘spua yi0g Ul $49e19 aul 4
(isd QQ0'G 12 saj9A9 G :ue}d)

{L] ‘punose

Aem au} |j2 yBnosy} paxoe19
pue eale aje)d-pua wo}
payebedoid 4919 !aj9A9 puo
-0as Ja}Je payea| pue paxoe19
(Is QQ0'G 12 Saj9A0 G_:ue|g)

{S] ‘ebewep

4ay}0 OU ‘WI apIsUl 4O Jo
940g diyd Buoj-youl-z auQ
(18d QQ0'G 1e Saj9A9 Gg :ue}g)

a}eUIWe| Jaq} ssej6
pajeubasdui!-uisas 91joUuayg

Sseiq |eAeNy

AvlW9-!1 adAy
winiuey |

OLE adAy
|9ays ssajuleys

WAUILWN|E g 1-909

=I

[euaiey aie] g-pug |

((pa}a]dWOd sajdA9 4s} $o Jaquinu juasaidai— [S] ‘a|dwexa 10}—suaquinu pajayoe1g)

Buipeo7 aunsseig 1e\SOIPAH JeoIj9AQ Jeusa1x3 JapuN sjeluareyy snoie, yo
SA4NSO|Q pu YIM Suawioads adig sse}s Bu07-Youy-g x G]-You|-p yo soUeWO}IEg “p alqe |,

19

flange section 2. Thin (0.023-inch-thick)
Sanaa 8 gaskets made of fiber-reinforced neo-
Pyrex glass pipe prene (Fairprene 5722A), lubricated
with silicone grease, were placed
between the glass pipe end flanges
compressed
O-ring 4-1/2- and the steel end plates and then
Lae DSS as compressed by tensioning the tie
inch section x
rods holding the assembly together.
A watertight seal was provided by

WO" stainless steel applying room temperature vulcan-

pyeesi6 izing (RTV) silicone rubber to the
joint between the end closure and
Figure 17. Experimental sealing and pipe flange after compressing the
bearing system utilizing seal. (See Figure 20.)

an O-ring interposed
between the glass pipe
flange and the metal
end-closure plate.

This bearing—sealing system
worked reliably on each of three test
assemblies cycled 10 times to 3,000
psi, and on one specimen cycled 10
times to 3,000 psi and then 10 cycles
to 4,000 psi. Two others failed on the 15th and 18th cycles to 4,000 psi
after being cycled 10 times to 3,000 psi. This system is considered useful
for limited cyclical service with 4-inch-diameter flanged glass pipes to depths
of 6,000 feet. The same gasket system was found in previous studies"? to
perform successfully in cyclic tests on glass domes to depths of 40,000 feet.

3. The gaskets in this case consisted of 1/8-inch-thick polyimide
resin (Vespel SP-1) washers fitted into grooves in the steel end plates. These
washers were of sufficient width that the glass pipe did not touch the steel
end plates. During assembly the watertight seal was provided by applying
RTV silicone rubber to the area between the washer and the glass pipe.

(See Figure 21.)

Of the two test assemblies cycled, one failed through cracking during
the first cycle at 6,000 psi after 10 cycles at 3,000 psi, 10 cycles at 4,000 psi,
and 10 cycles at 5,000 psi. The second test assembly completed 10 cycles
each at 3,000 and 4,000 psi, cracking on the first cycle at 5,000 psi. This
system is considered useful for limited cyclical service with 4-inch flanged
glass pipes to 8,000 feet.

4. Composite gaskets were made by soft-soldering 1/8-inch-thick
copper washers to 1/8-inch-thick steel washers and then facing off the flat
surfaces parallel and providing them with a 32-rms finish. The steel faces
were then coated with a 0.0025-inch-thick molybdenum disulfide

20

Figure 18. Four-inch-ID x 6-inch-long glass pipe utilizing an O-ring
interposed between the glass pipe flange and the metal
end-closure plate after exposure to 3,000 psi external
hydrostatic pressure for one cycle.

21

Figure 19. Closeup of failure area of glass pipe tested with O-ring gasket
between glass pipe and flange and metal end-closure plate.

22

<n flange section
: of 4-inch-ID x

RTV 6-inch-long
silicone Pyrex glass
rubber pipe

2q _ /_— 0.025-inch-thick
fiber-reinforced
neoprene (Fairprene
5722A) washer with
silicone grease
+r Stainless

steel

type
316

Figure 20. Experimental sealing and
bearing system utilizing a
fiber-reinforced neoprene
washer between the glass
pipe end flange and the
metal end-closure plate.

flange section
of 4-inch-!D x
=~ 6-inch-long
1 Pyrex glass
void pipe

RTV silicone

rubber (approx 2 1/8-inch-
0.015-inch eS thick, 3/4-
thick) Ne inch-wide
senosceesoesce Vespel
gasket

Figure 21. Experimental sealing and
bearing system utilizing a
Vespel washer between the
glass pipe end flange and
the metal end-closure plate.

dry lubricant material. Assembly was
effected by placing the molybdenum-
disulfide-coated face of a washer on
the steel end plate, placing the glass
pipe flange (coated with RTV silicone
rubber) in contact with the copper
face, then the second washer on the
opposite RTV silicone rubber coated
glass pipe flange (copper side down),
and finally placing the second steel
end plate on top of the washer. The
assembly was then compressed by
tightening the tie rods. The joints
between the washers and end plates
were then sealed with RTV silicone
rubber. (See Figure 22.)

The purpose behind this system
(and those described in paragraphs 5,
6, 7, and 8) was to provide a bearing
surface (copper in this case) which
would be soft enough to accommodate
irregularities in the glass pipe end
flanges and thus effectively eliminate
stress concentrations due to point
loading at those areas. Figure 23
shows the relative indentations in
the aluminum (see paragraph 6),
copper, and lead (see paragraph 5)
bearing surfaces which resulted from
one pressure cycle to 3,000 psi. These
indentations give an indication of the
degree to which these materials
“accommodated” the configuration
of the glass bearing surface. In
addition, it was hoped that the glass
and washer assembly would respond
to changing external pressure as a
unit when changing diameter due to
the effects of such external hydro-
static pressure. The molybdenum
disulfide coating on the washer—end
closure contact area was provided to

23

surface of
glass lapped flange section
to remove of 4-inch-ID x

irregularities 6-inch-long
Pyrex glass

pipe

tO

1/8-inch soft copper

RTV Ne
ra (solder, used to join
silicone
washers)
rubber

; j_ 1/8-inch mild steel
— molybdenum
disulfide

ty «Stainless
steel 316

Figure 22. Experimental sealing and
bearing system utilizing a
composite copper—steel
washer between the glass
pipe end flange and the

metal end-closure plate.

permit the washer to slide on the end
closure as the diameters changed with
pressure. If this system performed as
hoped, it would not only reduce the
stress concentrations in the glass—seal
bearing area but would also largely
eliminate tension cracking of the glass
pipe end flanges caused by the glass
dragging’ on the end closure as it
attempted to change diameter in
response to changing external pres-
sure. Neither this system nor the
similarly designed lead bearing washer
system described below performed
satisfactorily at 3,000 psi. Examina-
tion of the failed specimens showed
many small cracks which originated

in the glass—metal bearing area and

copper—steel

: aluminum §

Jead—steel

Figure 23. Aluminum, composite copper—steel, and composite lead—steel
washers showing the relative depth of indentation resulting
from pressurization of experimental sealing and bearing system
assemblies to one cycle at 3,000 psi external hydrostatic pressure.

24

propagated up into the pipe wall. While this system very likely is satisfactory
for cyclical service at some pressure less than 3,000 psi, it was not tested at
lower pressures and its maximum useful depth rating is not known.

5. Composite gaskets were made by cementing 1/8-inch-thick steel
washers to 1/8-inch-thick lead washers with epoxy cement (see Figure 24).
The composite gaskets were then treated and assembled to the glass pipe
end closures in the same manner as described above for the copper—steel
gaskets.

While the lead bearing system provided the most ‘‘compliant” of
the glass-to-metal bearing systems (see Figure 25), it did not produce the
desired critical pressure for the system. All specimens tested failed during
the first cycle to 3,000 psi. While this system very likely is satisfactory for
cyclical service at some pressure less than 3,000 psi, it was not tested at
lower pressure, and its maximum useful depth range is not known.

6. The gaskets in this test assembly consisted of 1/4-inch-thick
washers machined from 6061-T6 aluminum and finished to a 32-rms sur-
face finish. A 0.0025-inch-thick molybdenum disulfide coating was then
applied to one side of each washer. Figure 26 shows the component parts
of the specimen assembly. In the final assembly, the bare sides of the
aluminum washers were sealed to the glass pipe end flanges with RT V
silicone rubber and the molybdenum-disulfide-coated sides were placed
in contact with the steel end-closure discs. The assembly was compressed

by tightening the tie bolts, and the
Hanne cectione? joints between the washers and steel
+ finch-ID x 6-inch-long_ end plates were then sealed with RIV
Pyrex glass pipe
silicone rubber (Figure 27).

surface lapped
to remove
irregularities

1/8-inch-thick lead

(epoxy used to join Figure 28 shows the aluminum
ae ite I bearing ring after 10 cycles to 3,000 psi
rubber sd elena followed by 10 cycles to 4,000 psi. The
aN Bere resulting indentation is very difficult to
detect. One specimen assembly survived
two 10-cycle tests successively to 3,000
a eae and 4,000 psi. A second specimen
failed after one cycle to 3,000 psi and
Figure 24. Experimental sealing and a third specimen survived three 10-
bearing system utilizing a cycle tests successively to 3,000,
composite lead—steel washer 4,000, and 5,000 psi and six cycles
between the glass pipe end to 6,000 psi.

flange and the metal end-
closure plate.

ZS)

Figure 25. Composite lead—steel washer following external pressurization of test
assembly to 3,000 psi. The lead has plastically conformed to the con-
figuration of the glass pipe end flange and its integral O-ring groove.

Figure 26. Component parts of an experimental test assembly used to test
the seal provided by an aluminum washer interposed between
the glass pipe end flanges and the metal end-closure plates.

26

Figure 27. Assembled experimental testing system for aluminum seal
ready for insertion in a pressure vessel.

Figure 28. Experimental aluminum washer after being externally pressurized
in a test assembly to 3,000 psi. The indentations resulting from
the glass pipe pressing on the aluminum are just barely visible.

Dy

This system should be suitable for capsules for cyclical service to
6,000 feet. The one failure at 3,000 psi serves to point up the fact that
the quality-control criteria applied to this type of glassware do not serve
to exclude potential rejects for this type of application.

7. \n this bearing—sealing system, two previous concepts were
combined. Aluminum washers similar to those described in No. 6 above
were fabricated. They differed in fabrication and treatment only in that
a recess was machined in one outside edge to provide an O-ring groove
when the assembly was made. (See Figure 29.) The O-ring in this groove
provided the watertight seal between the washer and the steel end-closure
disc. A second O-ring was placed in the groove in the glass pipe end flange.
This O-ring made a watertight seal between the washer and the glass pipe.
Molybdenum disulfide dry lubricant was applied to the areas of the washers
in contact with the steel end plates, and silicone grease was used to lubricate
the O-ring between the washer and the end plate, but no RTV silicone rubber

was used in this system.

This system did not work satisfactorily at 3,000 psi. Each of the
specimen assemblies tested failed from cracks originating in the O-ring
groove of the glass pipe end flange—as in the case of the first gasket system
tested, which also used an O-ring in the glass pipe flange. While this system
very likely is satisfactory for cyclical service at some pressure less than
3,000 psi, it was not tested at lower pressures and its maximum useful

depth range is not known.

Pyrex glass pipe

silicone
grease

O-ring, 5-1/4-
inch-ID, 3/32-

inch section

{ 1D, 3/32-inch section

S 1/4-inch-thick
ae 6061-T6 Al
dry molybdenum
disulfide

Figure 29. Experimental sealing and
bearing system utilizing
O-rings and an aluminum
washer interposed between
the glass pipe end flange
and the metal end-closure
plate.

+—+ 4-inch-ID x 6-inch-long

compressed O-ring 4-1/2-inch-

8. These gaskets consisted of
1/4-inch-thick 60-shore-hardness neo-
prene sheet material cut into washers
and fitted into grooves in the steel end
plates. (See Figures 30 and 31.) The
contact area between the glass pipe
flanges and the rubber washer was
lubricated with silicone grease, and the
watertight seal was made after assembly
by applying RTV silicone rubber to the
joint between the glass and rubber.

This system failed by cracking
and leaking in the transition area
between the pipe flange and the cylin-
drical pipe body after seven cycles to
3,000 psi.

28

flange section
of 4-inch-1D x

6-inch-long
Pyrex glass
pipe
RTV
silicone silicone
rubber grease

1/4-inch-
” thick soft
neoprene
rubber
gasket

steel type
316 end
plate

Figure 30. Experimental sealing and
bearing system utilizing
a 1/4-inch-thick captive
neoprene washer between
the glass pipe end flange
and the metal end-closure
plate.

threaded for
high-pressure
vent

Each specimen assembly of the
eight different systems described above
was filled with water and the specimen
vented through a high-pressure connec-
tion (Figures 31 and 32) to a sight
glass (outside the vessel) at atmospheric
pressure. This enabled the observer to
detect any minor leaks in the assembly
which might otherwise go undetected
until the specimen was removed from
the vessel (assuming it was intact) on
completion of the series of tests. Pre-
vious experience with cycling tests of
glass specimens in which the glass
cracked but did not fail violently
indicated the desirability of a leak
detector to determine precisely when
a leak occurred. This high-pressure
connection from the interior of the

Figure 31. End-closure plates used in testing the Vespel and the neoprene
washers (shown here). The high-pressure vent is utilized in the
experimental measurement of internal volume change and for
attenuation of implosion force.

29

water-filled specimen to the atmosphere
also served to permit the specimen to

high-pressure 4 : . ,

vent connection fail by cracking or to implode without

the potentially damaging shock which

results from the implosion of a void or

even partially void specimen.

Table 5 summarizes the results
of these experiments.

From the viewpoint of simplicity
and low cost, the fiber-reinforced neo-
prene gasket (system 2) is the most
satisfactory to depths of 6,000 feet.
The systems which promise to give the
8,000-foot depth ratings for cyclical
service are 3 and 6, which utilize Vespel
and aluminum gaskets, respectively, to
eliminate failure from the rubbing of
the glass pipe end flanges on the end
closure as the glass pipe changes in
diameter with changes of external
pressure.

Phase III: Prototype Housing Tests

5

Deep Ocean Lab. =< Ocean Engineering Di

In order to demonstrate typical
Figure 32. Typical method of providing applications otglass She uo Unetersse
a high-pressure vent connec. US@- two prototype housing designs
tion from the interior of a were tested, one for a 1,000-watt light
specimen assembly. and one for an instrument housing.

Light Housing. Figure 33 shows a deep-submergence light specifically
developed to utilize a glass pipe housing with conical flanges. This design
utilizes a 1-1/2-inch-ID x 6-inch-long, flanged Pyrex pipe as the transparent
section. Figure 34 shows the construction.

Operational testing of this assembly indicated a collapse pressure of
9,000 psi. It was subsequently successfully operated and pressure cycled at
2,500 psi. In view of its successful performance at 2,500 psi, it is considered
reliable for operation at 5,000 feet.

Instrument Housing. Figures 35 and 36 show a prototype instrument
housing developed at NCEL. This housing utilizes two 4-inch-ID, flanged
Pyrex glass pipe caps and a section of 4-inch-ID, flanged Pyrex glass pipe

30

6 inches long. Interposed between the glass elements are two aluminum rings
provided with fiber-reinforced neoprene washers, which provide a compliant
seat between the glass and metal components. Sealing was effected by the
use of a silicone grease, and initial compression was supplied by tensioned
tie rods and metallic pipe flanges.

This assembly was pressure tested and found to fail at an average of
2,800 psi. Subsequent pressure cycling at 1,000 psi demonstrated the use-
fulness and reliability of this system at depths to approximately 2,000 feet
in cyclical service.

MODES OF FAILURE

The specimens failed in two ways: cracking followed by implosion
and cracking without implosion. Those which cracked without implosion
probably would have imploded had pressurization been continued beyond
the 20,000-psi limit used in these tests.

Figure 33. Prototype deep-submergence light for 10,000-foot depth service
using 1-1/2-inch-1D x 6-inch-long glass pipe.

3]

62-927

eee ET
ses imran
ies RE oe tele
Siivigd ¥ AvaWaSSW DIA-HW Fs 1 (COW 2
ONISNOH did SSW19 41 : : penin ice Doras Inscie Sisi ve adid sse|6 pabue}} Buoj-youl-9

a jeoWuouv Nvi50. S20 BSS cage geet X Gy-youl-z/,-| Buisn aotnsas yidap-1004-990'01 40} 1461)

EAR a Preral rimoreeer ee aduabsauigns-daap adAj}O}01d e 40} sjieyap UOIONAISUOD “PE ainbi4

TN suave

Twelve wouraiwo50 wn | _ON aaa
1 COKE "A O24, M COO/ 1a FLOM F395)
"ved _|"(WxO) 7002 Use Wwe? 2xabnd WwOd! | 7m;
TOD |[O2Sks SWINGOD) "HO?" 079) IIe SSCID'INO| 2
WNOD _| (1906S 1 ITALE ONINZOD) “Hla Fi "JHNOTF — z Eire
TOD | 12P-OSk SOLIINNOD INIatW A NSoKW | 2 Ea
iss, LON Xon_INve- Z| 7 ial
702 Woo Fi 07 Fs owe-o| 2 2
Waa Yor t- co 'sawmsom | 2 Z
709 | LOvENOD 139905 BW, 'SN7G WOrnr> | 2 geal TIES SINISE
(opwmcrs| QTD D TaeNOS- SONZ F7/0I JID
‘porestirey| ALOINI? IS 625" COSE' IM EO 'OideS | 2 6 99 S005 SIWINIDIO. IVI -E GCE
7wo>_| FQIM_FIAIOD FIBIXFS Pla |ORE | H2 2 SNOISWIME NOMLIESr . L
TOD (25069 mvwex0s) ‘sie7en | 2 SMO?IQY S& IP 7IKKS @- Sa ons
Danl P09 Give on7 | 2 = OILOY ISIMAHUO SEIINA GIWVWIITOL ~ aa
ssvre oF "asicvor | 2 e/ SISOF satpro We IWF PP ae 21078) Ag
¥e:0°9.% Cox WNitiz7 | € | ef | CRLON SORIA drahwiw/s Cada 4 OnaTe.
P2NQee |_sR/ O90" » we OG2” "LeIS OWiacs | FNO| F7 WHS SSIINNWEE BE IP Of WILN 7 onibar Siinide Livia! Cie ——
sstoe | 27x 090 > WC O82)" Lo¥s Saas |INO| 27 JosL0L0ee JKL WO O7sN Oa eE par mere3 | Cl) ©
eoLnzi77” |3wol 27 FIIM ORLSI? SINFL! SIPEALIESIAON Pena ge doe BLD
E 20M EO = Cl ale FRM | 2 Flow
= Fe io> 9 22> Ye Tams [no aaav08 1008 @
Worn | * org aor Bawsum | 2 e/

= NOMS
Orr32
COKF-2] YFIHSCM @)

sete |
arb mi SOMGPDE! IEB-2v 30 ixo19m 5 (Ip) j a
ST7OH IONOTS INWUNOTS ONY SOUIIKL COX FL I |
NO ATWOAIE? FLS6E Wie FLVINCIHD KA &
£7 000F HLe3T TWNOLLYIICO IS © CL SoS Here 5 :
5 Bas (27
A! OPLUIRO FP QL LON LHd/7 2 g
OFORINENS FUHM NOILVAISOD STIOONUNOD } |
FOND PLPN/ROZAIY S) FISTS LNG! FO F717 ‘ ER
pao We OOZE !AOZ/-MOOO! MX ICAL Fev a 8 on ATOWISS TOR-HW
997 ‘SNIWNT O00'D2 “KOZIEw A SANDY OS) 204 =
‘ %e OOBE 'A 02/-M 0001 NKO FeAL "en z
ISN SNIANT COO FE “Koxede A SICH OF 207 POMEL T9ZS SERINE
LOW NIWA 2e72P SSH7D | MEM EO OF die
2LYO FL FIVIN OL SONPL SYINIL OHS doze = ) A7BWIESY
Ho Fenvaze ISH 357 (1) INTL! FIONCH LON OP 1 DNides PIS

SUSTSILIVHAD TNOMLE IFO } ao! - sorl
woe apevens i
eres see

FENOIG FO SKIP
(CG2Y NO) LEIS SWISS &

SECT
oan (OZ) FFigeCe (e))—
rove : ;
(ice Gea, (@0r-2] uve GNF (i)
ES She SUS e720 Pe

IM BR Patt INI OD
[ee xe

is

a ; |

(Se am = PE
\\ seco? 2/097 . S72r i wero? - Wa % NS
bt Ao Paha Dec Fe n0 / \

an R2-P wed
TaM! BO = 4
N ) a
A Se
Giaceared ae ; oa, Fo

Eo Dine | (esi) 26 02-2 (ose) ateonen> sr»

32

Table 5. Results of Pressure Cycling Tests on 4-Inch-!D x 6-Inch-Long Glass Pipe Assemblies Having Various Bearing—Sealing
Systems Interposed Between the Glass Pipe Flange and the Metal End Plates

(@ = completion of cycle without leak; # = failure by cracking and leaking; 4 = failure by implosion.)

Successive Pressure Cycling Tests on Same Specimen

Bearing System
Interposed Between
Glass Pipe Flange
and
Stainless Steel 316 End Plate

0 to 3,000 psi,
cycle —

0 to 4,000 psi,
cycle —

0 to 5,000 psi,
cycle —

0 to 6,000 psi,
cycle —

6) 4) (3 7 3) 3} a) i (7 8) t2ga5 O87 8 A i@ jl 2s 4 B 7 i &) io

1. Neoprene O-ring (60-durometer,
3/32-inch section)

2. Fiber-reinforced neoprene gasket
(0.023-inch thick)

3. RTV silicone rubber with Vespel eeeee0eeeee eeeoee0ee8e8e8 @ eeeeeee?e2se es
washer eeoeeeeeeeet eeeeeeeee eis

4. RTV silicone rubber with copper-

A
steel composite washer plus tema
molybdenum disulfide coating

5. RTV silicone rubber with lead- Pa
steel composite washer plus a

molybdenum disulfide coating

6. RTV silicone rubber with aluminum | @ @ @ @ @ @ @ @ @ @ eeoeeoeveeee @ ea

washer plus molybdenum disulfide eu
coating eeoeeveeeveveee|/eeeeveeeeeeel i e#eeeeeeevseejeeeeees
a
; ry
7. Neoprene O-ring—aluminum washer 5
plus molybdenum disulfide coating 7
a

8. RTV silicone rubber with thick
rubber washer
ee | ee ee

33

es rie

2 3 4 Ci ; 7 a 4
Deep Ocean Laboratory ».'=.° Ocean Engineering Division

Figure 35. Component parts for a 4-inch-|D prototype glass instrument
housing for cyclical service to a depth of 1,000 feet.

Failure by implosion produced a loud, sharp report clearly audible
up to 50 feet from the pressure vessel. High-pitched ‘‘cracking’’ noises were
generally heard prior to implosion. As implosion pressure approached, these
noises increased in amplitude and became clearly audible to personnel behind
a barricade several feet away from the pressure vessel. As the testing program
progressed, it became apparent that the inception of the cracking sounds was
not being reliably observed due to its initial low amplitude and ambient
noise. In order to observe this phenomena and positively determine when it
started, a contact microphone and an audioamplifier were procured and the
microphone securely attached to the outside of the pressure vessel. Using
this technique, it was noted that the first ‘cracking’ frequently took place
at relatively low pressures. However, it should be noted that in some cases,
even though cracking noises were observed, there was no visible damage to
the glass tubes. This cracking noise in many cases may be generated by the
glass bearing surfaces dragging and ‘‘chattering’’ on the end closure as the
end of the pipe decreased in diameter in response to the external hydrostatic
pressure.

The actual cracks observed in specimens appear to be of three
general types. First the cracking and subsequent spalling of very thin shards
in the area where the inside rim of the pipe contacts the end plate (Figure 37).
Second are the concentric shear cracks propagating from the pipe surface in
contact with the end closure (Figure 38). These cracks, in most cases, result
in the cracking off of the portion of the pipe flange which is greater in outside
diameter than the main section of the pipe, and thus unsupported. They
appear to be a result of a combination of the shearing of the glass pipe flange

15)

» Deep Ocean Lab. Ocean Eng. Div

Figure 36. Assembled prototype
4-inch-|1D glass instru-
ment housing for
cyclical service to a
depth of 1,000 feet.

from the main body of the pipe and
tension resulting from the dragging of
the pipe flange across the end plates
as the glass pipe diameter is reduced
by external pressure. Third are the
radial cracks extending from the
glass-to-metal contact area up into
the main body of the pipe. (See
Figure 39.) These cracks originate
at the ends and propagate radially
along the longitudinal axis of the
pipe. These cracks often intercon-
nect so that the pipe falls apart when
the pressure is relieved, though in
many cases the badly cracked pipe
may not leak while under increasing
or constant pressure. These cracks
appear to be the result of stress con-
centrations resulting from small
protruding irregularities on the pipe-
flange face. These irregularities on
some of the specimens are large
enough to cause the pipe to ‘’rock”’
when placed on a flat surface. These
irregularities can be eliminated by
grinding and polishing the pipe
flanges. This was not done, however,
since one objective of this study was
to evaluate off-the-shelf glassware,
not custom-finished glassware.

PREDICTION OF CRITICAL PRESSURES

Although implosion testing of glass pipes with conical flanges has
experimentally determined the critical pressures of various sizes of glass
pipes, it is also important to be able to correlate this data with some sort
of an analytical expression for the prediction of critical pressures. If the
correlation between experimental and calculated critical pressures is good,
then such an analytical expression can be used with confidence to predict
the critical pressures of glass pipe sizes that have not been tested in this

particular experimental program.

36

Figure 37. Typical example of the cracking and spalling occurring in a
4-inch-!D x 6-inch-long glass pipe with aluminum end-closure
plates externally pressurized to 20,000 psi.

“shear’’ cracks

Figure 38. Shear cracks result in the spalling off of the portion of the pipe
flange greater in diameter than the outside diameter of the main
section of the pipe (4-inch-ID x 6-inch-long pipe with titanium
end-closure plates externally pressurized to 20,000 psi).

Sy)

“radial” cracks

toe pee ee Anon ae \2
Deep Ocean Labora Ww. )\

=e Aa RSET REECE

Figure 39. A typical example of radial cracking (4-inch-1D x 6-inch-long
pipe with stainless steel 316 end-closure plates externally
pressurized to 20,000 psi).

There are many analytical expressions for the calculation of critical
pressure in cylinders with or without stiffeners. Generally, the complex
analytical expressions are better than the simple expressions for predicting
the critical pressure. Thus, at first glance, it would appear that it is much
more desirable to use complex expressions than the simple ones since the
calculated critical pressures will be much closer to the experimental ones.
Unfortunately this is true only so long as detailed measurements and spec-
ifications exist for the given test specimen, as when it is an item custom-made
to very rigid dimensional and material specifications.

For mass-produced glass pipes with conical flanges detailed specifica-
tions to close tolerances are not available because in the mass-production
fabrication process large variations in wall thickness, roundness, and quality
of glass welds exist. Because of the discrepancy that exists between the
nominal and actual dimensions of the pipes, calculated values must differ
considerably from experimental values even if the analytical expression
used in the calculations is the correct one. Because of this discrepancy,
little can be gained by going to elaborate analytical expressions. when only
the nominal pipe dimensions supplied by the manufacturer are used in the
calculations.

38

In view of these considerations, it was decided for critical pressure
calculations to utilize only the nominal pipe dimensions supplied by the
manufacturer and an analytical expression which would combine simplicity
with fair accuracy. Such an analytical expression is the R. von Mises?
equation for buckling of monocoque cylinders equipped at their ends
with simple supports in the form of ring stiffeners or bulkheads (Equation 1).

e=eoc a? x
= i og | nr? ¢ 2)? - 2 rn? + |... (1)
2 2 2 2 sal H2
2 a satan 2 Qa
i= eS | Ile e tue a
2
where HM, = v2 + (1+o1p]|2 + (1-000
+
fie, = AS calf += (itl 2:0))\p: = (1-0%)(14 12),
2
eg eal
3D?
elo dle
oO Cores ae
Pos as
n? + a?
se meld
22 {L

To solve Equation 1, one must find that whole number n of buckling
lobes on the cylinder which would make the collapse pressure p a minimum
for a given mean cylinder diameter D, wall thickness t, and length between sup-
ports L. Although this analytical expression is rather simple, time consuming
and repetitious calculations must be performed before that number n of
buckling lobes is determined which makes p a minimum for a given cylinder
under hydrostatic pressure. For this reason, several approximations of
Equation 1 have been developed? (Equation 2) which in conjunction with
the expression for long unstiffened cylinders* (Equation 3) permit rapid
calculation of the collapse pressure for any cylinder.

39

The pair of equations that when substituted for R. von Mises
expression permit rapid approximation of the collapse pressure due to
buckling are:

t\ 9/2
jie 242i la eh
(1 aioe = OAS iS 1/2
3
j fae (3) i

where L = length of cylinder between stiffeners, inches
D = mean diameter of cylinder, inches
t = wall thickness, inches
E = modulus of elasticity, psi
fu = Poisson's Ratio
p

= collapse pressure, psi

After solving both equations for the dimensions of a given cylinder, the
higher collapse pressure is utilized as the correct one. However, a more
desirable solution to Equation 1 would be a set of plotted nondimensional
curves that would permit the user in the field to determine immediately
the predicted collapse pressure for a given cylinder without involved calcu-
lations. Such a plot of Equation 1 has been prepared (Figure 40) in
nondimensional form. Because they are nondimensional, the graphs can
be used to predict the buckling collapse pressure of monocoque cylinders
between simple supports regardless of cylinder composition.

A major problem encountered in the comparison of experimental
and analytical data is that test specimens and methods of test do not
exactly agree with the basic assumptions of the analytical expression. In
the experimental testing program of standard glass pipes with conical
flanges, three basic differences (see next paragraph) from the analytical
expression exist. The analytical expression is based on the assumptions
that: (1) the monocoque cylinder is of uniform wall thickness, (2) the
cylindrical radius is uniform throughout the length of the cylinder, (3) the
ends of the cylinder are simply supported, (4) the material is perfectly
elastic, and (5) the implosion of the cylinder is not initiated by failure of
the material, but by elastic instability of the cylinder.

40

aunssaud oNe]SOUpAY jeUJa} xa JAaPUN S4aUas js UBAMJa S||AYsS JeoupurjAo yo BurjyoNG Onse|3 “Op aunbi4

S

ol

10/4i6ue>

10

asdejjoo ye

sago| jo Jaquinu ——"

SOSIW “A “Y
uonenb3

oor

000't

o00'0t

000'000'L

OL * (Pd) = anssasg Bulsde|jo9 ssajuoisuawiig

“

41

Pecbaprp beep trwaprhd de)

The properties of glass pipes with conical end flanges differ
substantially from those assumed in the analytical expression. The pipes,
first of all, are not of uniform thickness and do not possess a uniform radius.
This is shown by the variation in thickness of plus or minus 20% from nomi-
nal value and the plus or minus 4% variation of external radius measured on
glass pipes tested. In addition, the conical end flanges do not give the glass
cylinder simple end support, but a support which is neither simple nor rigid,
but a cross between the two. The taper in the end flanges introduces an
additional uncertainty: the cylinder length between end supports. If the
overall length of pipe sections (including the tapered flange portion) is used
in the equation, different critical pressures will be obtained from the equa-
tion than if the length of the cylindrical portion between tapered sections
of the pipe is used. In addition to these uncertainties, there is added the
presence of a stress raiser in the form of an O-ring groove in the flange that
may cause the failure of the glass pipe at lower pressure than if the implo-
sion of the pipe took place due to elastic buckling. Only after the user of
the analytical expression understands all of these differences between the
assumptions on which the equation is based and the measurements of the
actual test specimens is he ready to intelligently apply the equation to the
calculation of critical pressure due to buckling for glass pipes with conical
flanges.

For the purpose of comparing calculated with experimental implosion
pressures, the following dimensions and properties of pipes were used for
plotting of experimental results (Figure 41):

D—mean diameter, as determined by subtracting one nominal wall
thickness from nominal external diameter specified by manu-
facturer

t—nominal wall thickness specified by manufacturer of glass pipe

E—nominal modulus of elasticity for borosilicate glass used in the
fabrication of pipe, 9.1 x 10° psi

L—nominal length of pipe, the distance between the two flat bearing
surfaces on the ends of the pipe

Comparison of experimentally determined implosion values with the
calculated curves for the same t/D ratio discloses rather good agreement
between them. The agreement becomes more remarkable when it is noted
that there are so many dimensional variations in the pipes and the flanges at
the end of the pipes do not represent the simple supports specified by the
analytical expression. Only one pipe size, the 6-inch-|D x 6-inch-long pipe,
imploded at pressures significantly lower than the calculated pressure.

43

Characteristically, this is also the only pipe configuration tested where the
distribution of cracks was distinctly different from those in other pipe con-
figurations. In the 6 x 6-inch pipe configuration, all of the cracks were radial
oriented along longitudinal axis of the pipe, while in other pipe configurations
they were mostly of the circumferential type. It would thus appear that
because of the low L/D ratio for the 6 x 6-inch pipe configuration, the stress
distribution is such that failure occurs at lower hydrostatic pressure due to
failure of material initiated by stress raisers on the bearing surfaces rather
than at the higher pressure predicted for it by elastic buckling theory.

50,000

I eal al 1] T ane etl Plein
1-inch-ID pipe
1.5-inch-ID pipe

cA
s
ie)
5 10,000 @
: ia 32 -inch-ID pipe __|
re —
®
2 >
Ae
3 ie
in 3- and 4-inch-
Legend < ID pipes
> @ 6-inch-!D pipe g .
® 4-inch-ID pipe 6-inch-ID pipe
& 3-inch-ID pipe
+ 2-inch-ID pipe
@ 1.5-inch-ID pipe
1,000 — 1 ete | ea | | l
1 2 3 4 SiG 10 100

Length/Diameter

Figure 41. Comparison of actual pipe implosion pressures with pressures
calculated on the basis of Equations 2 and 3.

In general, with only one minor exception, the plot of R. von Mises
equation has been found helpful in predicting the implosion pressure due to
buckling of glass pipes with conical flanges. Needless to say, extensive cracking
may, and in most cases does, occur prior to the act of implosion. Because of
it, the actual operational depth to which the glass housings may be repeatedly
subjected without damage is considerably less. The relationship between the
magnitude of the implosion depth and of safe operational depth can be seen
from comparison of Tables 2 and 6, which show the implosion pressures and
safe operational depths, respectively.

44

Table 6. Recommended Depth Ratings* and End-Closure Thicknesses for
Various Diameters of Flanged Pyrex Glass Pipe

One : Thickness of **
‘ ; Multiple
Length of Inside Round-Trip ives End Closures
Pipe Diameter Dive, : (in.)

é , Depth
(in.) (in.) Depth (ft) (6061-T6

(ft) Aluminum)

* These ratings reflect not only the experimental data contained in this report, but
also experimental data from NCEL TR-523, TR-559, and past experience with
reproducible failure pressures for mass-produced stock glass items.

** For a one-way dive, bare aluminum end closures may be used; for round trip or
multiple dive service nylon fiber reinforced neoprene gaskets (Fairprene 5722A)
are required.

45

SUMMARY
Phase |

Off-the-shelf flanged glass pipe can be used to provide transparent,
nonmagnetic instrument housings of very simple and inexpensive construction
for use in the ocean. Depending on the size requirements, the Pyrex glass
pipe tested in this study has a depth capability ranging from 2,000 feet to
the greatest depths in the ocean for one-time (no cycling) use.

Phase ||

1. Of the readily available materials tested, 6061-16 aluminum proved the
most satisfactory for use in a simple end plate-to-glass (without bearing
gaskets) closure system for one-time service.

2. For limited cyclical service, naval brass appears to be the best choice in
a simple end plate-to-glass (without bearing gaskets) closure system.

3. Of the various bearing gasket systems tested, fiber-reinforced neoprene
washer (Fairoprene 5722A), the Vespel washer (6061-T6), and the aluminum
washer system showed the most promise.

Phase III

1. A prototype light housing utilizing 1-1/2-inch-ID glass pipe was designed,
constructed, and tested; it was found to be useful for cyclical service at
depths to 5,000 feet.

2. A prototype instrument housing utilizing 4-inch-ID glass pipe was
designed, fabricated, and tested; it was determined to be useful for cyclical
service to depths of 2,000 feet.

CONCLUSION

Flanged glass drain pipe provides a useful, inexpensive, transparent
capsule material for enclosing lights and instruments for undersea use.

RECOMMENDATIONS

A summary of recommended depth ratings and end-closure thicknesses,
based on the available experimental data and experience of the authors is given
in Table 6 for various diameters and lengths of Pyrex glass pipe with flanged
ends,

46

REFERENCES

1. Naval Civil Engineering Laboratory. Technical Report R-532: Light
housings for deep-submergence applications—Part |. Four-inch-diameter
glass flasks with conical pipe flanges, by J. D. Stachiw and K. O. Gray. Port
Hueneme, Calif., June 1967. (AD 653293)

2.——. Technical Report R-559: Light housings for deep-submergence
applications—Part I|. Miniature lights, by J. D. Stachiw and K. O. Gray.
Port Hueneme, Calif., Jan. 1968. (AD 663890)

3. David Taylor Model Basin. Report 366: The critical external pressure of
cylindrical tubes under uniform radial and axial load, by D. F. Windenburg.
Corderock, Md., Aug. 1931. (Translation of: R. von Mises. ‘’Der kritische
Aussendruck fur allseits belastete zylindrische Rohre,’’ in Stodolas Festschrift,
Zurich, 1929, pp. 418-430)

4. “Application of the energy test to the collapse of a long thin pipe under
external pressure,’ Cambridge Philosophical Society, Proceedings, vol. 6,
1888, pp. 287-292.

5. J. D. Stachiw and K. O. Gray. “Instrument capsules for deep submergence
undersea use,’’ in Proceedings of the 20th Annual Conference of the Instrument
Society of America, Los Angeles, Calif., Oct. 4-7, 1965. (Paper no. 12.1-3-65)

47

No. of
Activities

305

DISTRIBUTION LIST

Total
Copies

20

10

13

305

Defense Documentation Center

Naval Facilities Engineering Command
NAVFAC Engineering Field Divisions
Public Works Centers

Public Works Center

RDT&E Liaison Officers at NAVFAC
Engineering Field Divisions and
Construction Battalion Centers

NCEL Special Distribution List No. 9

for persons and activities interested in
reports on Deep Ocean Studies

*paqiiosap ase ubisap Bulsnoy jUaWNs}sUl

adAjo10Jd auo pue ubisap }y61| adA}0}0Nd aUG ‘paj}uUasaid Osje ase Suua}sAs 84NsO|d-pua

JO} SuOIWepUaWWODaY “suo! eoI;dde jeod1j9A9 pue di4}-PuNo4 ‘di4} AeM-au0 410} pa}uasaid

aue ejep yjidap jeuoljeiado papuawiwodas pue ainssaid asde|joD ‘Hulpeo| ainssaid jeusa}xa

a1j9A9 pue Waj}-J4OYs uapUN Papoj duu! aiaM SayoU! OE O} Sayou! g WO} Hulbues sy}bua]

ul pue sayou! g 0 you! | WoO.sy BuiBues siajawelp ul sadid ssejb pua-pabue|4 ‘s}uswinsysul

pue s}y61) Jazevuapun 4104 sbulsnoy jUasedsues} se UO!}ed1|dde 10} spua pabue}} jed!Uu0d
yum sadid ssej6 ajqeyiene AjjelosauLuod ajzenjere 0} sem Apnijs siy} JO aAljoalgo ayt

L00-Z0-LO-SLO4-A ‘| edid ssej6 pabuels jelosauuW0g—sbuisnoy 1461) aouabsalugqns-dasq "|

Palsissejour 6961 4eW sn! d gp 8LO-YdL
MIYydeIS *q ‘f pue AesID °C ‘> Aq ‘(jeul4) SANA GAONV14
TVOINOD HLIM SAdid SSV1D “Il LYWd—SNOILVOIMddV

FJONSOYSAWENS-d3aad YOS SONISNOH LHOM
Asoyesoge | BulsaauiGuz Aig jeAeN

—

*‘paqiiosap aie ubisap Bulsnoy }UauINAISUI

adAjo}01d auo pue ubisap }ybl| adA}0101d aU ‘pa}Uasaid osje aie SWa}sAs a4nsojo-pua

JO} SUONePUaUWODaY “suO!}edI;dde jeo!;9A9 pue ‘di43-puno4 ‘di4} Aem-au0 40} pajuasaid

aue ejep y}dap jeuoljesado papuatuWoda pue ainssaid asde|jog ‘Bbuipeo) ainssaid jeusayxa

d1j9A9 pue W4a}-}1OYs apUN Papojduu! a4am sayoU! OE 0} saydU! g WO} Buibues suibuUa|

ul pue sayou! 9 0} you! | Woy BuiBues ssajawelp ul Sadid ssej6 pua-pabue|4 “s}ualunij}sul

pue s} yb) wayemsapun 404 sBulsnoy juasedsues} se uol}eo1|dde 10} spua pabuely jed1u09
yim sadid ssej6 ajqejiene AjjelouaWod ajenjena 0} sem Apnijs siyi 40 aAlzoafgo ay,

LOO-L0-LO-GLOA4-A ‘| adid ssej6 pabuely jelosau0D—sbulsnoy }Ybi| auabsawiqns-daeaq ‘|,

Palsissejour) 6961 4EW snl! 'd gy BLO-UL
miyseis *q ‘f pue Aer “QO *> Aq ‘(1eUl4) SANA GAONV14
TVOINOD HLIM SAdid SSV1D “Il LYWd—SNOILVON1ddV

FJONADYAWENS d3ad YOS SONISNOH LHDI1
Asoyesoge7 Bulsaauibuz jig jeaen

|
|
|
|
|
|
|
|
|
|
|

*paqluosap aie ubisap Bulsnoy Juawinajsul

adAjo101d auo pue ubisap 1y461) adAj0101d auQ ‘pa}uasaid osje aie swWaisAs a1nso|9-pua

JO} SuOI}epuatUWODaY ‘suo!}eo1|dde jeo!j9A9 pue ‘di4}-puno4 ‘di4} Aem-auo 404 pajuasaid

ase eyep yidap jeuoijesado papuauiWodas pue ainssaid asde|jog ‘Hulpeo| ainssaid jeusaixa

d1[9A0 pue W4d}-}40YS JBPUN Papojdiu! aiaMm sayou! OE 0} SAYydU! g Ws} Bulbued sy bua]

ul pue sayou! g 0} you! | Wod4 Bulbued ssaj}awelp ul Sadid ssej6 pua-pabue|4 ‘s}uauns}sul

pue syyblj Ja}eMsapUN 410} sHulsnoy JUaJedsued} se UO!}eD1|dde 410} spua pabue|s jeod!U00
YlmM sadid ssej6 ajqejrene AjjeiosauWod ajyenjena 0} sem Apnijs S14} $0 aAiioafqo ay)

L00-Z0-LO-GLOA-A °*| adid ssej6 pafuely jerosauiWi0g—sbulsnoy }y6!| auabsauiqns-daaq “1

Paisissejouy) 696L JEN snl! d 8p BLO-HL
myiyoeys ‘q ‘Pf pue Ard "OC > Aq ‘(jeul3) SANA GAONV14
TVOINOOD HLIM SadId SSV1D “Il LYWd—SNOILVOMIdd¥

SONADYAIWANS-d330 YOs SONISNOH LHON1
Asoyesoge 7 Bulsaauibug jiAid jeaen

*paqisosap aie ubisap Bulsnoy jUawnajsul

adAyo1010 auo pue ubisap 1461) adAjO1O1d aud ‘pajuasaid osje aie swia}sAs asnsoj9-pua

JO} suoijepuawiluosay ‘*suol}ed1jdde jeo!joA0 pue ‘di43-puno4 ‘diaz Aem-auo 410} pajuasaid

ase ejep yidep jeuoljesado papuatulwodas pue ainssaid asdejjoD “Bulpeo| ainssaid jeusaixa

D1J9A0 pue W4a}-}40Ys JapUN Papo|du! aiam sayou! OE 0} seyou! g Wo. Bulbued sy}bua|

ul pue sayou! g 0} Your! 1 Wo.) BulBued siazewelp ul sadid ssej6 pua-pabue|4 ‘s}uauun4jsu!

pue s}y6lj sazemsapun 40} sbulsnoy }uasedsuej se uoljeo!|dde 410} spua pabuely jeo1uca
YIM sadid ssej6 ajqeylene Ajje!osaULW0d ajenjena 0} sem Apnjs S14} JO anlzoafqo ay |

LOO-Z0-LO-GLO4-A “1 adid ssej6 pabuel} jerosauitu0g—sBbulsnoy }Y461) aouabsawiqns-daaq *|

palsissejour) 6961 Je snii!d gy 8LO-dL
miyseys °q ‘f pue Aeid *O “> Aq ‘(1eul4) SANS GAONV14
TVOINOD HLIM SadId SSV1D ‘Il LHWd—SNOILVON1ddV

FAONADYSWENS-d34ad YO4 SONISNOH LHONM
Asojyesoge 7 BulsaauiBug jIAid jeAen

—

;
4

‘
Se
Nei
S
ike

Ae

é

Wi

Ti

mri

*paqisosap aie UBisap Bulsnoy JUauNs}sul

adAj0101d auo pue ubisap }y61| adA}O101d aud ‘pa}uasaid Osje ase Slua}sAs ainsojo-pua

40} SUOI}EPUaLUWODaY “suO!}edI|dde jed1j9A9 pue di4}-puno4 ‘di4} Aem-auo 410} pa}uaseid

aue eyep y}dap jeuoljzesado papuawiwosas pue ainssaid asde|joD “Hulpeo| ainssaid jeusa}xa

o1j9A9 pue W4a}-]40Ys JapUN Papojduu! asam sayou! QE 0} SayoU! g Wd} Bulbues suibua|

ul pue sayou! 9 0} You! | Woy GulGUes SudjaLUeIp Ul sadid ssej6 pua-pabue|4 ‘s}uawiNsjsu!

pue s}y61) Jay}evuapun 4104 sbulsnoy jUasedsues} se UOI}e91|dde 10} spua pabuely jeo1u0o
yum sadid ssej6 ajqejiene Ajjeidsauuod ajenjena 0} sem Apnjs siy} JO aAizoaf[qo ay

LOO-Z0-LO-GLOA-A ‘| adid ssej6 pabuej} jerssauiLu0D—sbulsnoy }Yb1) aouabsaluqns-daeq “|

Paljissejour 6961 4eW snl 'd gp BLO-UL
miyseis °q ‘fp pue Aes °C *> Aq ‘(jeuls) SONS G3ONV14
AVOINOD HLIM Sadid SSV1D “Ill LHVd—SNOILVOIMddV

FJONADYAWENS-da4ad YOs SONISNOH LHOIM1
Asojyesoge Bursaauibug |iAig jeaen

eS Se ee es eee

“‘paqisosap ase UBisap Bulsnoy JUauNsysU!

adAj}0}01d auo pue ubisap jyb1) adAj}o}O1d aUGQ “pajUasaid Osje aie sWia}sAs ainsojo-pua

JO} SUOIepuUaWWOdaY “suo!}edi|dde jeo1j9A9 pue ‘di4}-puno4s ‘di4} Aem-auo 40} pajuasaid

aie ejep yiydap jeuoijesado papuawiwodas pue ainssaid asdejjog “Hbuipeo| ainssaid jeusia}xa

a1j9A9 pue Wia}-J10Ys 4apUN Papojdu! asam sayouU! OF 0} Sayou! g W144 Buibues syibuUa]

ul pue sayou! g 0} youl | Wosy Bulbues Siajaue!p ul! Sadid ssej6 pua-pabue|4 ‘s}uawnsjsul

pue s}y61) Jajemsapun 1045 sBulsnoy jUuasedsues} se UO!}edI|dde 104 spua pabue}s jeo1u0d
yum sadid ssej6 ajqejiene AjjeidsauiWod ajyenjera 0} sem Apnijs siy} Jo anijoa[qo ayt

LOO-Z0-LO-GLO4-A “| adid ssej6 pabuely jerosauUUWOD—sbulsnoy }Yb1| aduabsauqns-daaq “|

Paljisse|aur) 6961 Ae sniiid gy BLO-dL
miyoers *q ‘f pue Aerd *Q “> Aq ‘(leul4) SANA GSONV14
TWOINOOD HLIM SAdid SSV1D “Ill LYWd—SNOILVOM1ddV¥

SONADYSAWENS d3ad YO4 SONISNOH LHOIM
Asojesoge] Bulsaauibug |iaig jeaen

LO0-L0-L0-SLO4-A ‘I
palsissejoun

TVVSINOOD HLIM SAdid SSV19D

L00-L0-L0-SLO4-A ‘I
palyissejoun

TVSINOD HLIM SAdid SSV19

|

*paqluosap aie ubisap Buisnoy }uawnjsu! |

adAj0}04d auo pue ubisap 1y61| adAjoOJO1d aU ‘pajUasaid Osje a4e SUa}SAS a1nsojd-pua |
JOJ SUOIFEPUaWIWIODAaY “suO!}ed1|dde jed1|9A9 pue ‘di4}-puno4s ‘dis AeM-au0 40} pa}uasaid

aue ejep yidap jeuoljesiado papuawwodas pue ainssaid asdejjog “Hulpeo| ainssaid jeusa}xa |
d1J9A9 pue wWia}-J10Ys JapUN papoj dw! ajamM sayoul OE 0} saydu! g WO} Bulbued sy}bua}

ul pue sayou! g 0} You! | WO1s BulBues suajalwelp ul sadid ssej6 pua-pabue|4 ‘s}uawinsjsu! |

pue s}yBij sazeMsapuN 40} sbuisnoy JUasedsued se UO!}eD1|dde 410} spua pabue}} jeod1U09

yim sadid ssej6 ajqeylene Ajjeidiawwod ajenjera 0} sem Apnijs sy} $0 aAljoalgo ay |

|

|

|

adid ssej6 pabue}} jelouauiw0D—sbulsnoy iybl| aduabsawiqns-daaq “1

696L 4eIN snl! 'd 8b 8LO-YL
miysers “q ‘f pue Ae1d °C *> Aq ‘(jeul4) SGN3 GSONV14
“IH LYWd—SNOILVOMddV

FONADYAWENS-d3sad YOS SONISNOH LHSIT
Asoyesioge Bursaauibug jag jeaen

|
|
*paqisosap aie ubisap Bulsnoy }Uuawnaysu! |
adAjojo1d auo pue ubisap 1y61) adAj0101d auQ ‘pajuasaid osje aie swia}sAs ainsojo-pua |
JO} SUOI}EPUatLWODaY ‘suo!}zedI;dde jeo1j9A9 pue ‘di4j-puno4 ‘di4} Aem-auo 10} pajuasaid
ase ejep yjdap jeuoijesado papuawiWwodas pue ainssaid asde|jog “Hujpeo| ainssaid jeusa}xa |
21J9A0 pue Wsa}-}40Ys JaPUN Papo|dw! aiam sayou! gE 0} sayou! g WoO4; Bulbued sy}bua} |
ul pue sayou! g 02 You! | Wosy BulbuUeJ Siajatue!p Ul sedid sse|B pua-pabue|4 “s}uatUNijsU!
pue s}y61| Jazemsapun 404 sbulsnoy jUasedsueds} se UO!ed!|dde 104 spua pabuels jeo1u09 |
yim sadid ssej6 ajqeyiene Ajjelo1aWiWod ajenjena 0} sem Apnijs siyi 40 anlzoalgo ay |
|
|
|
|

adid ssej6 pabuely je;osauWW0D—sBbulsnoy 3461) auabsawiqns-daeq *)

6961 4eW snii'd gy 8L9-YL
mMiydeis ‘q ‘fp pue Aeay “OC > Aq ‘(|eUl4) SANS GSAONV14
“IL LYWd—SNOILVOINddV

JONADYAWENSd34ad YO4 SONISNOH LHOINM
Asoyesoge 7 BursaauiBug |i jeAaen

. n

Unclassified
Security Classification
DOCUMENT CONTROL DATA-R&D

Security classification of title, body of abstract and indexing annotation must be entered when the overall report ts classified)

ORIGINATING ACTIVITY (Corporate author) 2a. REPORT SECURITY CLASSIFICATION

Naval Civil Engineering Laboratory Unclassified
Port Hueneme, California 93041

REPORT TITLE

LIGHT HOUSINGS FOR DEEP-SUBMERGENCE APPLICATIONS—PART III. GLASS
PIPES WITH CONICAL FLANGED ENDS

DESCRIPTIVE NOTES (Type of report and inclusive dates)

Final; November 1966—June 1968

AUTHOR(S) (First name, middle initial, last name)

K. O. Gray and J. D. Stachiw

6. REPORT DATE 7a, TOTAL NO. OF PAGES 7b, NO. OF REFS

March 1969 48 5

CONTRACT OR GRANT NO 98, ORIGINATOR'S REPORT NUMBER(S)

| PROJECT NO Y-F015-01-07-001 TR-618

9b. OTHER REPORT NO(S) (Any other numbers that may be assigned
this report)

DISTRIBUTION STATEMENT

This document has been approved for public release and sale; its distribution is unlimited.

SUPPLEMENTARY NOTES 12. SPONSORING MILITARY ACTIVITY

Naval Facilities Engineering Command
Washington, D. C.

ABSTRACT

The objective of this study was to evaluate commercially available glass pipes with conical
flanged ends for application as transparent housings for underwater lights and instruments.
Flanged-end glass pipes in diameters ranging from 1 inch to 6 inches and in lengths ranging
from 6 inches to 36 inches were imploded under short-term and cyclic external pressure loading.
Collapse pressure and recommended operational depth data are presented for one-way trip,
round-trip, and cyclical applications. Recommendations for end-closure systems are also pre-
sented. One prototype light design and one prototype instrument housing design are described.

fe)
DD Hoe | 473 NCEE) Unclassified

S/N 0101-807-6801 Security Classification

Unclassifi

Security Classification

KEY WOROS

Light housings

Flanged glass pipes

End closures

Sealing and bearing systems

Implosion

Cyclical loading

Short-term loading

DD .2%"..1473 (sack) Unclassified

(PAGE 2) Security Classification

r

Hite