STRENGTH OF GLASS

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP89B00487R000300650004-8
• Release Decision: RIFPUB
• Original Classification: K
• Document Page Count: 13
• Document Creation Date: December 22, 2016
• Document Release Date: August 22, 2011
• Sequence Number: 4
• Publication Date: May 1, 1955
• Content Type: REPORT

Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
STRENGTH OF (ASS
Prepared fur: SPO 21 433
Referonce s SPO 23970
PROJECTOR DIVISION
THE +' I?-II 1ER CORPOR!TXON NOf IWAAIX, CONNECTICUT
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
Introduction
The fo3lot report is a compilation of a near of interim reports concerning
tests performed on plate glass to determine 3 resistance to temperature and
psure stresses. The first part is concerned with the resistance to temperature
only and to a combination of pressure and tewrperz Lure. The succeeding parts are
concerned with pressure tests only.
PART I
1.0 GENERAL
A -.umber of tests were conducted on the mechanical propeeUes of plate glass diNcs
insofar as their resistance to temperature and pressure wore cancernc .
Two series of tests were run; a temperature test in the absence of pressure and a
combined temperature and pressure test. Briefly. these tests were conducted. ay
follower
1.1 TE T ATURE TEST
A glass disc was imply mounted at its edges by clasping it between two . alu Laing
plates. with 110" ring gaskets. The 20" xring gaskets were coated lightly with sil.i.-
cone grease. The diameter of the supporting edge Maas .1/2" less than the diwmter
of the glass disc sample. Torque oar the bolts was set at 5 i mch-pounds. The bolts
ite6 8-32 machine' screws
The plate supporting the sample was the cover over a well insulated box containing
:L--n
an electric hot plate of 660 watt capacityy. The surface of the hot plate was
distant from the ,mss surface. An 18" household. type fan was permitted to blow
a stream of air across the "cold" face.of the. test sample, The surface temperature
of the glass was measured at four points with four constantan-copper thermocouples.
Precautions were taken that the true surface temper .turye and not that of the sur-
rounding air layer eras measured. The four points were near the edge of the sample
on hot and cold sides, and at the center of the sample at hold amid. cold sides, With
the test set up the heat was applied, and time and, temperatures recorded. The t -
perature was permitted to increase until rupture occurred. 8e0 Table I.
1.2 PRESSURE AND T 1PERATURE TEST
In these tests. the glass sample was supported as described above except that the
alarms m support plate was the cover of a pressure vessel. Water used as the
pressure medium. All air was removed from. the system by suitable venting so that
an of the glass surface was in contact with the water. , A standard pressure of
10 psi gauge ' (accuracy + 1 psi) was used throughout all teats. Bolt torque on the
sample support was .5-peuudv4
The surface of the glass exposed to the atmosphere was heated in the same fashion
as described for the temperature tests Heres. of course, no circulating fan was
used. Four thermocouples mere used to measure- center and edge temperatures, sec
Table.
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
1.3
SUNMM OF. TEMP! RAT1JRE TESTS
Seven inch diameter plate glass samples ofvarious thicknesses were tested to
destruction. The support diameter was a.". No record of the glass composition
is available but standard soda-'lime glass is assumed. AD samples had polished
edges and a minixmim of surface scratches and flaws (i.e, no scratches apparent
to eye). After clamping and prior to test, the camples were tested for strain
in a polariscopo. No initial strain due to clamping or glass conditioned was
detectable in any of the samples tested,
1. Buxom OF P wSURET IE' TESTS
After the test setup completed, a 10 psi pressure was established before the
application of heat. This pressure was then euitained throughout the application
of gradually increasing heat. Eight inch diameter glass discs with polished edges
were used. They were supported on a 7,n diameter. A variety of thicknesses were
tested. Glass composition is assumed to be soda-lime. No polariscope strain
tests prior to testing were possible.
1.5 CONCLUSION
It would 'seem that resistance to fracture due to thermal stresses increases
with thickness for stead, temperature conditions. Although the supports were
designed to represent simple supports, there is probably some constraint leading
to a support which is quasi--constraining in nature. Polishing edges to reduce
sources of local stresses (nicks? scratches) is worthwhile. The average (for
design) tensile strength of plate glas at rapture is approximately 6,500 psi
as concluded from these tests. No olaim.is made to represent those figures as
true sampling. For this reason, no percent accuracy of data is indicated.
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
WAIiLL i
'Thickness
Temperature Difference Between
Hot and Cold Surfaces (Through
Thickness) ?F. at Time of Fracture
Near
YAX.
Elapsed
Time to
r, Mature
Appr=.
Rate Of
Te Inc.
OF
(Start at
. Strew,
,
(Inch
Cep A
T
OF
Aran.
Rom
N to (1) )..,
,o61
56
52
65
267
24.6
33
6,200
71
108
108
315
35.5
7
10,300
.133
82
78
262
3342
4
9,750
.133 .
751
172
(3 scratches)
NO
34
5
1
170
452
5
5
000
16
.133
7
,
3
6
.,
242
206
NO FR9CTURR
1
67
,.. S
9
700
29
.
4
,
.21x2
210
202
220
(3 scratches)
N
MGM
.385
298
O
572
.1--S
9
28,00
.570
-0--
NO I1ACTURF
252
12
24j,000
Remarks
Unpolished edges
fracture started at
edge, :LO iu o-# bolt
torque. Green color..
Polished edge. 4 in.
# bolt torque. Green
color. Fracture
Started at edge.
Unpolished edge,
lVacture started at
edge. 10 in.?# bolt`.
torque. Water vh te.
No f"rac ure. Po1i she
edgee.
No f3racture. 5 in.-
# bolt torque.
Green color.
No featiure. 10 in.-
# bolt torque, Water
tab te.'Edge temp.
2960 P4.
No fracture.
Polished edge
No fraotut'e. 10 In.-
# bolt torque. Water
ite..Edge temp.
* Z. 131.9? F.
No fyraotura. 1004
bolt torque. Green
color
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
TABLE I (continued)
Note (1
Commuted from foxmu1e
S stress in
c'C * 82 x 10-7P C? thei. . oxp ansion
0 temp* difference in ?C (fr test data, rte.)
E Y ng'e modulus - iO.5 x 3 psi
14 Poisson's ratio = 0.22
Ref: Corning Blass and Roark pg. 322.0 p raph 5i.
also Morey pg. 315.
Theoretically, a simply supported disc should nearer rupture due to a steady
state, uniform temperature g adient across its this ss. The above for ulao
is for fixed, edges.
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
TABLE ii:
Tr eratuxo difference
Between Hot and Cold
Sizfaces ('! vai 1 Thick-
ness) 3 at T? of
Pressure
Th.dmess psi Near
-: ~ ..a.. ......
. Max.
OM.
Ce
Approx.
Elapsed 'Rate of Computed
Time to Temp. Nom. Store
Fracture Increase pi (tensile)
-Tin. Bea. , .n.. Note (2)
.070 7 0 0 0
70
.135 8 0 0 0 70 0
P-41300 Pressure applied g3'adu
within .1 min. Chen coder.
Fracture etarted at in-
ternal i' aw (bubble)
7, 50 Gradually, applied preecnre
in 1 sin. period. Water
wh ite?. Bad eeratch 1*"
long on lour preen re aide.
Failed at scratch.
.252 10 69 99 99 230 39 4 22,700 Wat -white.
NO FRACTURE
10 95 260 160 268
?.~- 4 l?A30 No fracture. Water whiteo
Max. edge temp. 174? F.
106 1w00 286 42.5 5 21,230 sed aide. Fracture
started at edge; ~t
edge touched support.
. edge temp. 261?F.
?~?-- 3.h 18 Both atdee tested (br re.
versing) making two runs.
More'seaerre ruu tabulation
here, No fracture. Max.
edge temp. 222,5?F (2/80
from cold race)..
NO FUCTUR
.501 10 142 283 187 380
{both
alAes
tested)
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
TABLE II (contiz ied)
Note (2
(k acted from formula for simple support, mechanic4 loading
(of irk$ pg. 188 paragraph 1)
Max* e t - (314,1)
6-err
(at center)
s stress, tensile (on surface) away from load - i.e* hot surface)
in W Tr1tr`
w - unit load psi
r - radius of support
M = 1 - reciprocal of Poisson's ratio a 4,55 for glass
1
t = thioknesa in inohee
Stressea due to to er.ature computed as before (Nott 1) . Tabulated stress
is algebraic sum of mechanical and temperature stresses.
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
PART II
2.0 PRESSURE, TESTS vs. THICKNESS
A series of pressure tests were made on various thicknesses of 8" O.D. plate
glass discs using the pressure test set up described in Part I. Three thick.
nesses of glass were used; 1116", 2/81' and i/!1". Prior to testing the blanks
were inspected for strain under a polariscope and found to be free of strain.
They were then clamped in the pressure tester under 5 inch-.pounds of bolt torque.
Care was observed to eliminate all air pockets. Pressure eras then gedual.3y
applied until the glass broke. In almost all ca"s,s the fracture started at
a surface scratch which had been noted and identified prior to the test. All
discs had polished edges.
Table III is a plot of glass thickness vs.. the stress in psi at time of fracture
as computed from the stress fornnile for discs which is:
S 3W,,,, (3m+l)
(at center)
where: W - WIT r2 = total load
at = reciprocal of Poisson's ratio = .54
t = thickness in inches
The mean thicknesses tested were .0638" which broke at an average pressure of
14.2 psi; .132" which broke at an average pressure of II psi, and .2530 which
broke at an average pressure of 25.7 psi. The support diameter was 77" OD. It
will be noted that the allowable stress decreases with an increase in thickness
of glass. Normally, one would expect the curve to be a straight line having a
fixed value of stress for ax thickness. A theoretical investigation is shown
in Part III to determine the reason for this effect.
PART III
3.0 T TICAL AMMG
rrey r r.A yM ~1 ^
The following covers an analogy between the mechanical properties of glass and
reinforced concrete. Design data on reinforced concrete is well known. In ad-
dition. its properties.are almilar to those of glass in that concrete has con-
siderable strength in compression but none at an in tension. Glass in turn has
considerable strength in compression and some strength in tension. .
Standard reinforced concrete design requires that the neutral axis of the support
member (beam or slab) be determined first. This location is dependent upon the
ultimate fibre strength of the concrete in compression and the ultimate fibre
strength in tension of the steel reinforcing bars, and in addition the ratio be-
tween the modulus elasticity of the steel and the modulus elasticity of the concrete.
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
PART ZXI (coat)
For purposes of this discussions the standard nomenclature for concrete design
will be used. The basic equations area.
k fc
=
fs+fc
(1)
(2)
For our purposes.% we will apply reinforced concrete design data for a beam as if the
beam were made of glass. It has been well-established in literature that the tensile
strength of glass is a limiting strength. Since the beam is made of . glass, the
material is homogenous and the value 'n' Is therefore unity. Further, it is assimaed
that the ultimate strength in compression of the glass is a constant and the ultimate
strength in tension is a constant, thus the value of tk' will be constant. Nor then
the moment in the lower half of a uniformly loaded beam is
R As fsjd (14)
where 3 = l-k (5)
3
f$ m ultimate strength in tension
N ' ' is the area of the glass which is in tension. This in turn will. be proportional
ton certain skin thickness (value undetermined) multiplied by the width of the beam
W. It is this skin thickness which is assumed to vary with the total thickness of
the glass. The mexim.un moment in a simply supported rectangular beam of length 'I' is
Hmax = 1/8 Wi
where 'Wt is the total load, uniformly distributed. If
(6)
we equate It and 6 we have
W== Kld (7)
where 'Klt is a constant dependent upon the shape of the began
and type of load. In the case of a glass beam 'K11 might vary as f d' varies o3* be
constant- while 'dT is raised to some power. This All be shown later. In the case
of a reinforced concrete beam, Kl will remain constant.
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
PART III (coat)
For a rectangular (cross section) steel beam similar equations may be applied.
For a uniformly loaded simply supported beam of steels the maxiimirn moment is given
by equation 6.. Further the value of the moment is also given in terms of the beam
shape and its Ultimate strength by
rr=SI
(8)
where c 2d #rl bd5
Equating equations 6 and 8 tae have
W =d2 (9)
where K2 is determined by beam configuration and typo of load.
it should now be noted that in the case of the concrete desig the load varies as the
first power of the beam thickness, and in the case of an all steel beam the load varies
as the square of the beam thickness.
Almost all beam and plate equations relate the uniformly distributed load to the thickness in the form of the equation
W 0 K3dn (10)
Load tests on glass plate have been reported prwriously-. This data has been used to
establish the value of 'K31 and tn.. We find that for a disc of 7? support diameter
the value of K is ' 6930 and the value of t nt is 1.32o These values have a relatively
small spread when applied to any one of the 22 pressure tests made.. It is interesting
to note now that the equation for a glass structural nx mber .lies somewhere between
that for concrete and that for steel as evidenced by the exponent for the thickness rds
Three borosilicate plate glass discs were pressure tested and two rectangular boro-
silicate pieces of plate glass were tested. The values for the ultimate strength, as
computed from plate data (as given by. Roark)- fell within the clusters shown in the
plot of thickness vs. ruptured strrength. Therefore, it may be safely assumed that
the ultimate strength of glass due to mechanical loading is 5000 psi. This value is
the vertical asymptote of the curve afore mentioned. For thicknesses less than i/1",
this value increases as shown by the same cute.
PART IV
4.0 TOUGHBr D GLASS
Consideration was given to the use of toughened glass, and therefore an investigation
of the optical properties of toughened (or hardened) glass was made. There was avail-
able a 20 mm thick piece of Schott BK-7 glaze which had been specially hardened. This
type of glass was origiY used for submarine periscopes. The glass was considered
specially hardened in the sense that the large double refraction normally associated
with hardened glass had been reduced. The piece was first examined in a. polar .sco?.;o
M9,.
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
PART IV (coat)
and showed a strain of approximately 80 mil.imicrons per centimeter, A 1" aperture
collimator was set up and collimated on a reference surface. The disc of hardened
glass was then placed near the objective of the collimator and shifted back and forth
by hand. motion of the image in the collimator showed degradation of the point
source when viewed with the sample in place. In addition, notion of the sample in
the collimated beam caused displacement of the image by approximately 2.5 mils. It
was concluded$ thereefore, that hardened glass would be unsuitable as a window.
Of academic interest is Schott's comment on this glass that its chief value for
optical uses was for window material when the light was collimated and 900 incident
to its surface. Further, since the glass was hardened glass, polishing should not
exceed more than 1 mm removal of the surface of either side.
4.1 ULTIMATE GLASS STRENGTH
A mentioned in Part III, the asymptote to the thickness vs.. ultimate strength curve
lies approximately at 5000 pounds psi as the ultimate strength of glass for exT .
thickness. This figure should be used, with caution,, however, because it was based
upon test data applied to simply supported plate equations. The actual test did
not simulate a true simple support but one which lay approximately midway between
a fixed support and a simple support. For purposes of the discussion to follow,, this
type of test support will be designated as a quasi-simple support. A second plot of
thickness vs. ultimate strength was made using fixed plate formulae and the test data.
Such a curve was parallel to the original curve but shifted to the left. These two
curves then can be considered as. designating that the actual test support vas an equal
combination of simple and fixed supports, a curve of thickness vs. ultimate strength
for the quasi-s nple support should lie midway between the two aforementioned curves.
Actual plots showed that the asymptote for the illaxxed support lay at 2500 psi. This would
then place the asymptote with a quasi -simple support at 3750 psi. This figure (3750)
represents the true ultimate strength of the glass.
A theoretical approach may also be used to establish this asymptote, as followsa The
equation for the maid== strength of a disc in a fixed rapport is
4 1C t (1)
For a simple support and using a value of .22 for Poisson's value, the maxim= strength
is
a 1.61 x
If we as that the strength for a quasi si le support lies half way between the
two, then
s*- 8a+5
2
(2)
(3)
S' l.6 f+ f = 1.305 a. (IL)
2
From our experimental data we then get 8 equal 5000, S f equal 3003,. and S I equal 4050
psi. This is in fairly good agreement with the data derived from the. curves.
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
PART N (ooh)
Thus numerical values of Ss and Sf are fictitious as used here, and their use is
restricted to specific plate foxmula3. However the value, of S1 is the, true
strength of glass. Since we do not have a formula for a quasi--siinple support.
the true ultimate strength of glass is of not too great a value for this current
windx problem. For this problem, the value of the ultimate strength shmald. be
taken as 5000 psi and used with simple support plate formulae for those cases
where the actual mounting is a quasi-simple support.
4.2 S+I,AHY'
As determined from these tests, the true ultimate mechanical strength of glass
lies between 3700 to 4000 psi. This is true regardless of the gloms composition
since both borostlicate and lime glass were included in these tests.
The attached curve includes data derived from testing borosilicats plate glass
discs and rectangles*
? ,M,llo?
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8
 Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8 A 1'
ffmw
it miH
HWHI
t -
Approved For Release 2011/08/22 : CIA-RDP89B00487R000300650004-8