Professional Documents
Culture Documents
-1-S
·1?
. ~--------~---------
---Knolls---- KAPL-1904
---Atomic Power-- AEC Research and
Develop:men t Report
---Lahorato~y ------
Correlation of Experimental
Data with Theory for
Perforated Plates with a
•
Triangula~ Hole Array
t
L. Deagle
January 7, 1958
Operated for the
- - - United States Atomic - -
- -Energy Commission by..----
~-GENERAL f) ELECTRIC==
\'· I
- - - - - - - - - :·-- --
DISCLAIMER
KAPL-1904
UC-34, Physics and Mathematics
( TID-L~500, l!.tth Edition) _
L, Deagle
January 7, 1958
'2iJ!? ~~ Jan.ZI9S9
AUthorized Classifier Date
UNCLASSIFIED
UNCLASSIFIED
LEGAL NOTICE
,,
UNCLASSIFIED KAPL-1904
ii
UNCLASSIFIED
KAPL-1904
UC-34, P.hysics and Mathematics
, . . ( TID-4500 ~ 14th E;di tion)
Internal Distribution No. of Copies
AEC, SNROO 1
. Anderson, JD 1
Boiko, A 1
Bunke, EWD 1
Cooper, WE 1
,· Davidson, DE. 1
~agle, L 18
fucumen:t · Li bra.J::·y 4
Fritz, RJ l
George, CW 1
Glasser, TH 1
Hoe·, RJ : · . 1 ..
Mains, RM 1
'
Miller, DR 1 . :· .'
Pravda, MF 1
SAR File 1 ..~.J.:
Schmitt, PM 1
Sears, Y.J 1 .. ':'·.· .
Shriver, : PR 1.
v/estmoreiand, JC 1
Wojcieszak, Rf l
EXternal Distribution
Aberdeen Proving Ground J ..
Aerojet-Gener~l Corporation 1
Aerojet-qener?-1, San Ramon (I00-880 ), 1
AFPR, Boeing; Seattle l
AFPR, Lockheed, Marietta 2
Air Fore~ Spec·ial We?pons Center 2
ANP Project. Office, Convair, fort Worth 2
Alco Products; Inc. l
Argonne qancer: Research Hospital l·
Argonne National Laboratory 10
Armed For:ces Special Weapons Project, Washington l
Armed Services Technical Information Agency 5
Army ~al~istic. Nassile Agency 1
Army Chemical Center · ·. ·· · 4
Army Signal Research and Development Laboratory 1
Atomic Bomb Casualty Commission l
~·· ....
UNCLASSIFIED
iii
····-··:--·-··-··--·--·-;·-:·-·:-·~-·.-r:-"!"~---:::.,.....-.~~-~-·-·.··--~----; -··~ ·:·· ., . ~-- .......... ;;- ,, ... ·-··"· --~·.·.··.- ....~ ... -..,.. ........ ~~-~., ... ~ .....---·---.-. ··- --- --·:- .. ·-·
U.NCLASSIFIED
. 1} . .
. UNCLASSIFIED
iv
~ ..
., __ .
UNCLJl.SSIFIED
KAPL-1904
UNCLASSIFIED
v ..
UNCLASSIFIED
KAPL-1904
Total 651
t.
UNCLASSIFIED
vi
,,
KAPL-1904
CONTENTS
ACKNOWLEDGMENT X
ABSTRACT xi
INTRODUCTION .• 1
TEST PERFORMED • .. 4
Flat Pert;orated Al\llllinurn Plate Tested for Structural
Effici.ency 4
The 3/8 Scale MJdel of the Pressure Vessel Head Tested for a
Nuclear Reactor 13
REVIEW OF OTHER DATA • 19
J. P. Duncan's Paper "The structural Efficiency of Tube Plates·
·· for Heat Exchangers" • 19
"Photoelastic Investigation of Perforated Plates, An
11
vii
. " .
.KAPL-1904
ILLUSTRATIONS
viii
,,
KAPL-1904
No~. Title
2.3 Aluminum Plate Str·ain Gage fute. - "Bottom" Surface • • 45
24 Geometry and Loading of Outer Annulus - .3/8 Afudel 47
25 The .3/8 Model Deflection Gage Data . 49
26 The .3/8 Modei Strain Gage Data 50
27 Ratio of FquiyalEmt ligament· \-lidti1 to Minimum
.;· I.igamen~ ·Width· in Bending: and Tension . 5.3
28 Equivalent Elastic Constants . 54
29 Stress MUltiplier . .. • 54
. ,,
TABLES
No.
·- .Title
·.1 •'.
ix
,,
ACKNCMLEIX}JviENT .
KAPL-1904
X
'P
ABSTRACT
.KAPL-1904
xi
.,
L. Deagle
INTRODUCTION
The purpose of this paper was to investigate the adequacy and range
of applicability of the equivalent solid plate analogy for.the calculation
of stresses arid deflections in perforated plates, as developed by G. Horvay.
This objectiy~ was to be met by comparing measured values of stresses and
qeflections with the calculated values obtained from Horvay 1 s theory. The
published li~erature, unpublished reports, and unpublished test results,
such as the t~sts conducted by and for the Knolls Atomic Fbwer Laboratory,
were considered as a source of test information to be used to meet the
stated objective. The results of unpublished tests were solicited from
twenty-one manufacturers and users of pr.essure vessels.
The problem of determining stresses and deflections in a perforated_
)
plate is fundamental to the design of all pressure containing vessels \.fhich
require multiple openings such as are found in the tube sheets of heat
exchangers an~ the heads of nuclear reactor pressure vessels. The work
of the designer is greatly simplified if the various parameters such as
plate thickness, outside diameter, hole spacing, and hole size can be
readily manipulated without recourse to empirical data. The theory
developed by Horvay permits analytical treatment of the perforated plate
by reducing it to an equi'x,aleut solid plate of' the same geometry. The
deflection is calculated by replacing Young 1 s modulus E and Poisson's
ratio :~ with fictitious elastic constants E* and ~* that are based upon
the hole size-and spacing in the perforated plate. SimilarlY, the stress
is calculated as if it were a solid plate and corrected by means of a
stress multiplier, based on the hole size and spacing. These factors, as
derived by Horvay, B.l"e preserrLed in Figures 28 and 29, Appendix D, as a
function of the ratio of the minimum ligament width to hole spacing.
Although the theory, derivations, results (including the curves in
the appendix}, and applications to various perforated plate problems are
thoroughly treated by Horvay 1 ' 2 ' 10 "" 14 ' it is .believed' that a brief. statement
of the theory :and derivation should be included here.
KAPL-1904
1
2
r--L ·I
L r~r:n:J~r (bl
-•Cl[]dl} (c)
M ( ~)
~· .M (d)
KAPL-1904
• 3
The stress mul ti.plier . 1 ~ 2 used to correct the solid plate stress·
was obtained by considering a hexagona~ arr~y of beams subjected to in~.
plane perpendicular boundary tractions apd shear loads, as· sho'vm in
Figure 1, (a)~ The stresses in the ligE!Jllents w~re obtained from the stress
distribution on the boundaries of the ~exagonal array so th~t the maximum
str~ss in a l~gament was expressed as a function of this stress distribution,
which corres:ponded to the principal stresses in a .solid· pl~te and the
orientation of tP,e ligamel1t with respect to these pr1.ncipal stresses.
·nn~s, when the ratio of solid plate principal stresses is unity (isotropic
tension or compression), the maximum stress in .the perforated plate occurs
at the minimum li~ament width and is equal to the solid plate stress ·
multiplied by the ratio of the hole spacing to the minimum· ligament width.
When the ratio of the principal stresses is other than unity (this ratio
can vary from_l to -l)j the maximum stress is in~reased over that for the
isotropic case and is found to vary with the orientation of the lig.arilent
with respect' to tl1e larger of the principal stresses; the l:i.gament nearest.
to an angle of 45 deg with the larger·_ of the principal stresses then has :
the highest stress. The highest stress in a given plate J,s obtained when
the ratio of the principal stresses is equal to ~,1 (pure shear). Thus,
the stresses in the ligaments of a perforated plate are ·shown to be a
function of ligament width, ratio of principal stresses, and the orientation
of the ligament with respect to these "solid plaj,e" principal stresses~
The inc:J;"ease ip. stre~s caused by the change in the ratio of principal
stre.~ses from· unity is th? result of bending moments of the type shown
in Figure l(b), which'·produce additional stresses away from the center of
the ligament. Th.us, the theory predicts that .~he maximum stress will
occur in the ~enter of the ligament only for an isotropic solid ple.te
stress distr~bution and is shifted away from the center for other distri~
bu:tions. The curves in Figure ~9, Appendi~ DJ are plotted for ratios of
the principal stresses of 1, 0, and ~1 for ~ligament oriented at 45 deg
with the m~~mum principal stress.
The pr:l,ncipal restrictions and assumptions aside from those normally
associated with solid plates are:
1. The plate should have a large numbP-r of holes· (several ·
·rows}.
' 2. The holes must be arranged in an equilateral triangular
array.
J. The pl~te is loaded i~=plane.
4· Rein,fq:r-cernent of the plate by tubes inserted in the holes
is n.egligi.ble •.
5. Use of the constants E* and ~*is restricted to p~ates
whos? hoJ,.e spacing is smaller than the plate thickness •
•
NOTE:. The·roregoing restrictions are Horvay 9 s.
KAPirl904
4 •
TESTS PERFORMED
Flat Perforated AlUminum Plate Tested for structural Efficiency
The purpose of this test was to determine the feasibility of extending
. the perforated plate theory, developed by Horvay, to plates \..rhose thick-
ness was large in comparison to the radius of the plate and in which the
number of penetrations was relatively small. TI1is test was made at the
Knolls Atomic Power Laboratory, Schenectady, New York, and was undertaken
as a preliminary step to the design of the closure head for the reactor
pressure vessel for a submarine. With the foregoing in mind, the important
features of the geometry, the number and size and skew angle of penetrations,
were proportioned so Llu::1t the plate becamr:- ::m ......1/4-scale model of the head
intended for the full-size vesseL The ske\-1 angle 1:;; roxmed by the inter-
RP.ction of the penetration centerline anu.a normal to the plate surface;
this angle lies in a plane normal to a radius of the·plate.
The. test pie.ce shown in Figure 2 was machined from 61ST Alurrd.num.
The nineteen penetrations in the plate; 2.185-in. diam, are arranged· in a
triangular array with a maximum skew angle of 9 deg 52 .min in the outer.
· holes. This skewing of the penetr13.tions, dictated by nuclear r'equiremerits
of the core and external requirements for mechanical clearance, resulted
in ligamentsuoF different sizes on the ~wo surfaces of the plate. A one-
inch thick web: was left ].n the center of the plate to provide a pressure
seal for the ~ydraulic oil which is used to apply the load to the plate.
The plate thickness in the perforated portion was five inches; this
dimension was chosen as being scaled to the order of magnitude of the head
that would q·e used on the full size ves~::elo Since the test pl.e,ce was so
thick, e.Jum:i.num. W?:S chosen for its low modulus so that deflections,. which
.:are measurable with dial.micrometers, could be obtained without huving to
use excessively high loads.
The alUminum plate was clamped into a heavy steel fixture for the
duration of·: these tests, as shown in Figure J. The cavity beneath the
plate was filled with hydraulic oil and the plate was clamped against a
gasket to form a seal agairu:;t the pressure. Pr·essurization was accomplished
by means of a hand pump and the hydraulic pressure was mesasured \-lith a ·
Bourdon-tyPe pressure gage. Both strains and deflections were measured
during the test. The general arrallgement of gages is shown in Figure 2.
The strains were measured with.Baldwin A-19 electrical resistance strain
gages of 1/8-in. gage length; stresses were then computed from these
strain gage readings. The strain gages were mounted on the.exposed surface
only, at the narrowest part of three ligaments, i20 deg apart around the
center hole·. :Three gages were mounted across each of the three ligaments;
one gage in the center and one at each edge of the ligament. The deflections
were measur.ed by 0. OOOl~in. -dial micrometers of which eight were mounted •
at four different radi;i.. The .scribed circl.es show the locations of the
dial.'g~es, (see Fi~e 2).
KAPL-1904
FIGURE 2. Al~inum Plate with Top Side Up
KS-12003
Unclassified
6
KAPL-1904
7
KAPL-1904
8
----
0 .0 10
----
</)~,-
--'<ll
~
~ Si
t; g
.008 --
TOP--
~
.006t---BOTTOM.__
- BOTTOM
........
..................
..............
...._, ·~
,,
-MEASURED
-- CALCULATED
r--~ ~
~!::
:t ~ .004 ~ ~ ',
0
0
-.......
~~
r-,
.00? ...............
~'
I~
0 2 :3 4 5 6 7
~ r-...
8 9
r - FIAniiiS FHOM ~ liN.I KS·IIH)~5
UNCLASSIFIED
The tes t ing of the model consisted of two separate tests; (l) the
strains and deflections were measured on the "top" surface with the largest
ligament uppermost in the test fixture, and (2) another in whicll the measure-
ments were made on the "bottom" surface wit h the minimum ligament width
uppermost. This reversal of the plate in the fixture resulted in a change
in the loading of the plate . This change in the actual loading caused what
appe8-rs to be an anomaly in that the strains measured on the "top" surface
were larg er than those mectsur ed on the ''bottom" surface while Lhe ueflections
measured for the "t op" were smaller than those measured for the ''bottom"
surface. The explanation for this : s as f ol l O\.,rs .
When the measurements were made on the "top" surface,the pressure in
the lower half of the nineteen holes produced an unbo.lanced force on the
s o-called solid ouLe1· annulus, aa chown in FieurP. 18 Appendix B, (see
Figure 3 also) . This introduced a moment that opposed the rotation caused
by the t ransverse pressure load while at the saute Llme an in-plane tensile
str ess was produc ed. Thus, it was possible for the deflection to be reduced
but to have Bt.rRi ns on the t ensile surface remain relatively high while
the strains on , the compression surface would undergo a mru·ked reduction.
Should this force be counterbalanced, an increase in deflection would be
expected with a possible decrease in the tensile stresses. This counter-
balancing actually takes place when the test piece is reversed in the
f ixture to allow the testing of the ''bot tom" surfctee since there is now
fluid around the outer periphery of what is now the lower half of the plate
rel ative to the fixture .
KAPL-1904
9
In addition, the posit ion of the neutral axis of the plate is now
such that the horizontal discontinuity force (T i n t he analysis) which
results from the eccentricity of the neutral axis of t he perforated
portion with r espect to t he center of gravity (CG ) of the solid outer
annulus is now compress i ve and causes a further reducti on in the tensile
stresses but adds to the compressive stresses. Unfortunatel y, no gages
were on the under side of the plate t o verif y these deductions. F~wever,
the same relative variation of s t r ess and defl ection was prod ':!ed analyt-
i cally when the effects mentioned were taken into account; therefore, the
foregoing is judged to be a valid explanat i on of a phenomenon which, when
it fi rst appeared, was thought to be an error in r ecording the data but,
when subsequently checked, it was found to be val id inf ormation.
The average stress across the narrowest part of the ligament around
the central hol e and the maximum ef lections, as axrived at experimentally
and analytically, are given in Ts.b . c lo
K.APL-1904
10
"To12" Surface
A* B* . c* Average
Edge of Ligament 5. 96 7. 57 6.21 6.59
Center of LieamP.nt 5. 51. 5.29 5-96 5.60
Edge of ligament 6. 94 6. 94 6-94 6.94
"Bottom" Surface
Edge of Ligament 5. 37 4· 70 5.37 5.13
Center of Ligament 4. 81 4· 60 4·70 4· 70
Edge of Ligament 5. 30 4· 70 5.42 5.14
ligaments across which the stress is nearly uniform and, therefore, does
not make allowance for Lhe ~tress concentration efft:H.:t ol>RP.rvcd in the
test except for the approximation which was made by extrapolating the
calculated curves in Figure 29, .Appendix D, to the limiting stre.s s concen-
tration for a hole in an infinite plate . The comparison of test · values
with theory is shown in Table 1 on p . 9 for average values of stress
across the ligament since this is all that the theory attempts to do. The
significance of Table 2 (above) is the variation shown in the stress across
the ligament which the designer would have to account for by other means
when the ligaments are no longer slender.
KAPL-1904
FIGUPE 5. The 3/8 Model Test Setup - Strain Gage Selector Switches
KS-18663
Unclassified
FIGURE 6. The 3/8 l{odel Test Setup - Support St:and and Pressurizing Pump
KS-18664
Unclassified
1.3
For static load- carrying capacity, only the average stress across
the ligament is of interest; however, the stress concentration effect at
the edge of the hole becomes important for fatigue considerations.
In general, i t was concluded from these tests that it was permissible
to extend Horvay 1 s theory to the plate tested or to . one of similar propor-
tions which has the same type (skewed) and number of penetrations . Because
of the numerous assumptions other than those in the perforated plate theory,
it is not possible to assign quantitative value to the error that may
exist when this theory is used; however, since the over-all error was found
acceptable, it may be deduced that the error introduced by Horvay 1 s approach
to the perforated plate problem is small even for this extr eme case which
barely fulfills some of the assumptions on which Horvay derived his constants
and stress multipliers.
The .3/8-Scale MOdel of the Pressure Vessel Head Tested for a Nuclear
Reactor
The measurements of stress and deflection made on the pressure vessel
head were part of a larger test program in which a .3/8-scale model, in steel,
of the reactor pressure vessel, closure, and pipe nozzles were t ested for
pressure stresses, distortions, and pipe reaction stresses. This test work
was done at Combustion Engi neering Company, Chattanooga, Tenness ee n
The purpose of this test was to provide experimental evidence of the
structural adequacy of the design of a reactor pressure vessel and to
provide a means of checking the analytical procedures used in the design
before fabrication of the full size vessel was completed .
The general arrangement of the test setup showing the test stand and
pressure vessel, electric strain gage selector swi t:ches, potentiometers,
and pressure gage for measuring the hydraulic load is shown i n Fi gure 5.
The vessel was subjected to internal pressure; the working fluid is
hydraulic oil. Pressurization was obtained from _a motor driven pump, shown
::fn Figure 6. Strai ns were mee:u:nu·ed by means of (l/4=in. ) . Baldwin electrical
resistance strain gages, and dial micrometers (O ~ bOOl in.) were us ed to
measure the deflections .
The head has nineteen skewed penetrations to which nineteen standpipes
are attached on t he upper surface. Only three of these standpipes were
made to scale for the model; these were: the center st andpipe, one inter-
mediate standpipe, and one of the outermost standpipes. The remainder of
the holes were fitted with _piJ:iPR which passed through the hules and were
welded to the head by outside fillet welds on the upper surface. The
penetrations. are closed against pressure by the standpipes so t hat t he head,
KAPL--1904
14
')
KAPL-1904
Iig. 5 3/8 Mo del Eead - Top View
KS-18662
UNCLASSIFIED
16
KAPL-1904
17
'··
·.
~ ~':- -
'·
,~,
.-28 00
' ~-.,'.., ~\
.~41:::-~. .
~
-24 000
,
"'' \
.................
r--,
',
-20
,.,.., ', ._.... \ "·
'\
~~
,,
-.... \
......
'iii
S-16000 ' \\.
"\
"\,
\ ..... ,
\
~ STRESS DIRECTION
Q:
:n
o CIRCUMFERENTIAL
-120 00 - A RADIAL
\
\ \'
\
'"'
I
•a MEASUR EO I \\ .
b
-80 oci
06 CALCULATED
\·
.. \ \
-\,\.\
-4000
o-
0 ..
....
....
. 2·-
··'
=
-·
4·
r- RADIUS FR0t.11
...
6
Surface Stress
. .. ' ' . .
. ~ . . . ' .
point midway: :Petween the ~enter and the outer edge than the curve' through
~e m~asured points; however, it is pr~bably close enough for ~ost engineering
purposes.· -The per cent of error between the maximum deflection as measUred
and the max~ deflection as calculated. is approximately ~he same as the
..) variati·on of th~ maximum stresses and could be accounted for in the unknown
•
edge concii tions.
The stresses that were measured and calculated were the maximum average
stress. across· the ligament; L e., the stress at the narrowest part of ~the
ligaments and·. not the highest stress in the 'ligament as predicted by Horvay 1 s
theory •. · It is, not possible to measure accurately this maximum stress by a
strain gage technique; to obtain this stress a photoelastic analysis would
·,.
be necessary._ However, in the model which was investigated, this effect
can be neglected since the variation in principal stresses, in the equivalent
solid plate, from the center to the outermost ligament is small. In the.
ligaments around the center hole the ratio of principal stresses is
essentially lJility, and the maximum stress is at the center of the ligaments.
KAPL-1904
18
.020
r---
~----- -- MEASURED
.018
.016 r-- ....... , ---- CALCULATED
·2
::::. .014 ~"-..... ' .....,
~
8 .01 ,
....
~
..J
.010
~ ' I'~
~ .008 ' ....,
,,
.............
~ .........
.00 6
.004
.002 ·..;; KS-18647
~ Unclassified·
2 · 4 6 e -~ IZ 14·
r- RADIUS FROM t_ (IN.)
In the outermost ligaments, according to·· theory, the maximum, stress will
have shifted from the narrowest part of the ligament and the value will be
higher than the value at the minimum ligament width; however, since the
variation in principal stress is small and hence their ratio is still close
to unity, the stress at the narrowest part of the ligament will be very
close to the maximum stress theoretically predicted. Consequently, from the·
viewpoint o·f the structural integrity of the head,- the slight increase in
stress over that which occurs at the narrowest point of the outer ligaments
is of no concern since the point of maximum stress in the head is at the ·
center ligaments. The significance of the foregoing, when an attempt. is
made to evaluate the usefulness of the perforated plate theory, is that no
direct check of this phase of the theory is possible from this test.
However, the close agreement between the radial variation of calculated
stresses and the radial variati9n of the measured stresses, as shown in
Figure 9, strongly indiQates that the plate is behaving ~s predicted in
t...:.'18CO.~y and, hence, warrants this further extension of the theory.
KAPL-1904
19
TEST PIECE
KS~l8649
Unclassified
K.APL-1904
20
,,
KS~l8650
ONE -PASS PATTERN
KAPL-1904
21
The dimemlions of the plate used in the hydrostatic test for first,
a two...:.pass anq then a one-pass pattern w:ere:
Outside diameter, ,in. 8
Diameter of perforated portion, in. 5.5.
.·:
Plate ~hickness, in. · o. 5.34
Hole pitch, -in. 7/16
Hole size, 'in. 5/32
. .
In addition, hydrostatic tests were'made on an 8=in.-OD plate whose
initial 2--in. ; thickness was progressively reduced to l. 821 in., 1. 590 in. ,
1. 266 in., 1.-104 in., and 1. 000 in. .Also, the effect of increasing hole
size for a 2-J.n.-thick plate underbydrostatic load.was investigated,.with
:the hole size:being increased from J/J2 in. to 1/8 in., to 5/32 in.
i::uncan cbmputed deflections of the one-pass plate for concentric
loading and hydrostatic loading by using the fictitious constants E* and
~* (used by Horvay). for the perforated portion. In his procedure the
slope_ and deflection of the perforated portion were matched with the solid
OUter annUlUS -Of the plate. These formulaS aJ."e derived from TimOSkenkoV S
Plates and Shells and are given in the appendix to Duncan 1 s paper. The
'Co;npar:i~on···;r.;:~·a'lculated · stress and deflection of simply supported undrilled
plates with measured values made it possible to estimate the edge restraint
caused by fricti<:m at the support. By applying this correction to calcu-
lations of deflection for the perforated plates, Duncan obtained a·quantita-
tive estimate·-:of the error caused by using the fictitious constants in the
"equivalent plate constant" ·analysis. The calculated and measured values
of-central deflection agree within 15% for the 0.5J4~in.-thick plate under
hydrostatic load~ The error between calculated values for central deflec-
tion and measured values was 13% for the plate subjected to concentric
loading. In qoih cases the calculated deflections were larger than the
measured defle·ctions. Duncan concluded that "the tests on one-pass plates
·confirmed with fair accuracy the proposal that Horvayvs thick plate theory
might be applled to the· central drilled area··· .• "
. .
funcan points.out in his reply to the discussion of his paper that
"the essential1y. experimental study had been instituted to examine a . ·
particular case arising in practice, and had been extended as far as
opport\mi ty and equipment wou,ld allow to broaden its scope and usefulness
to designers.·: It was, therefore, not always well suited to arbitrate some
of the finer .theoretical points raised." One such point that was neither
prov~d nor disproved was the method proposed, by Horvay, for computing the
maximum stress in the ligament.
KAPL--1904
22
KS-18651
Unclassified
The solid plate was used to provide measured values of solid plate principal
·~·
stresses for comparison with calculated ~alues of.principal stresses. The
ratio of solid plate principal stresses is essential, ·in the theory, to
the prediction of maximum ligament stresses. Curves which show the theore-
tical and measured solid plate stresses were drawn and very good agreement
was obtained except, as was expected, at the point of application of the
load. The maximum ligament stress was determined from the boundary fringe
order at eachof·the holes (1 to 6); the location of these .holes may be
seen in Figure 13. The point of maximum stress was found in the inclined
ligaments at a distance ~/L = 0.2 from the center of the ligament; ~ .v~iable
distanGe alo~ the ligament with .the ·origin at the center and L is .
the ligament length measured along the center line of the ligament. The
following tables show a comparison of theoretical ligament stress with
observed stresses for solid plate stresses obtained experimentally and
analytically;·
Horvay's
Experimental Theory
1}. &
Hole p~ xp X
.E.. )[
1 13.7 -4.5 -0.33 78 75
2 14·4 -4.5 -O.Jl 90 73
3 16.3 -4·5 -0.28 90 71
It 20.2 -3.6 -0.18 92 66
5 29.6 -0.4 -0.01 72 60
6 52\3 9.0 0.17 48 52
Tables 3, a and b are data for ·the perforated plate subjected to two
diametral loads. See p. 55 for list of· symbols. KAPL-1904
.. ' . ........
24 .. ~,.
Horvay's
ExJ2erimental Theor;y
B 8
Hole _P_ ~ X p p
Horvay' s
Experimental Theory
"'cr· "'(J
Hole ___.£_ ....2S.l2-- X ...E._ L_,
1. 7.1 7.1 1 151 18
2 9.7 3·7 0.38 133. 42
3 13.5 . 0 0 98 60
Ll. HLl -2.6 -0.14 91 66
5 25.5 -4.2 -0.16 83 67
6 37.8. 0 0 61 60
Table 4, a and b,are data for the perforated plate -subjected to three equal
radial loads.
where & = maximum ligament stress:
p·= largest principal solid plate stress, and
x = ratio Of principal solid plate stresses.
Rules 1 to 6 are counted from tha center radially ontwflrd,
Data in foregoing tables show that fairly good agreement between theory
and experiment is obtained for the case of two diametral loads although the
theory underestimates the stress in holes 2 .through 5 by -25%. For the
case of three radial loads no agreement at all is obtained at tlle l!E:mte!·
(800% error) while fairly good agreement (15% error) is obtain~d at the outer
·hole. Guernsey attributes this discrepancy, in the central holes, to the .,.
presence of pure bending which Horvay' s theory neglects.
KAPL-1904 •
,;··
25
CONCLUSIONS
Based on-the foregoing tests, with regard to the general validity of
the equivalent- plate constant theor.Y and specifically to t~e usefulness
of the theory:~s developed in detail by G. Horvay, these conclusions are
made:
l. The perforated lattice does, in fact, behaves as an
equivalent plate of reduced stiffness \vhen it is sub-
j.ect~d to -transverse loading.
2. The ~esults, as obtained from the Horvay 1 s theory, are
sufficiently accurate for engineering applications for
a range of plate sizes extending from thin _plates
(radius-to-thickness ratio 7. 5 to 1 for Duncan's tube
she~t~} to thick plates (radius-to-thickness ratio
1.6 to 1 for the reactor model head) with a variation
in the number of penetrations ranging from the large
number in Duncan's tube sheet tests down to the rela-
tively small number (19 holes) in the reactor model
head (J/8 ·scale model).
These conclusions apply only for Horvay' s ·"thick plate 11 theory, which
requires the hole spacing to be smaller than the plate_thickness; the
assumption of the derivation_being that the plate was thin; i.e., the
radius was large compared With the plate thickness. r-.)'o conclusions regarding
Horvay's "thin plate" theory (hole spacing larger than the plate thickness)
can be drawn from these tests. Also n,othing· can be said about the utility
df the method in the range, explicitly excluded by Horvay, for which
2.<.!.<1,
2 2R
KAPL-1904
26
RECOMMENDATIONS
Despite the ~ncouraging results which were obtained in correlating
the theory with test results, considerable work on perforated flat piates
subjected to transverse loading still remains to be uone. With 1•cga.rd
to the triangular hole array' thel·e ls .still unccl•tainty' f:rnrn the ;.,
experimental standpoint, concerning the variation and location of the
maximum stress in the ligament that are caused uy Lln::: uff.set bending vThich
results from the difference in magnitude of the principal stresses and
from the ligament orientation with respect to these stresses.
Also, the variation of stress across the ligament as the ligament
becomes thick is uncertain. In addition, there are no published test
results for plates with perforations such that the ratio of the hole
· spacing to plate thickness is less than unity. The writer knows of only
one set of tests that was conducted with plates in which the hole spacing
was very much larger tJ:lan .the plate thickness. These tests were conducted
by the General Electric Company 1 s ·Aircraft Nuclear Propulsion project at
the Evendale Plant. The writer has seen some of the preliminary test
results and there appeared to be fali:·ly good agrccmont vli th theory; how-
ever, no details a:L"e ava:tlablc. To I'esolve the uncertainties concerning
the stress distribution in flat plates with a triangular hole array,
the following test prog'ram is suggested:
l. Make four plastic models of perforated plates for a photo-
elastic study of the stress distribution under transverse
loar:l.
2. The models should have hole diameters such that the four
plates have ratios of minimum ligament width to hole
sp~~ing of Q.~, 0"2, 0.1, and 0.05, respectively.
J. The plate diameters should be ~24 to J6 in.
4· The plate thickness should be such that the initial ratio
of plate radius to thickness is 1. 5 to· 1; · subsequently the
thickness could be· reduced in four or.. more ·steps· until.the
thinnest plate practical was produced.
,.
KAPlr-l9Q4
27 .
KAPL-1904
<
. '•
THIS PAGE
WAS INTENTIONALLY
LEFT BLANK
29
APPENDIX A
... . . . . .
ANALYSIS OF· .PERFORATED PLATE WITH SKEWED PEm'ORATIONS AND SOLID
_OUTER ANNULUS .
'
A perforated plate with skewed holes is subjeqt.ed to a uniform pressure
load and·a. restraining edge load as shown in Figure 14.
The radius "b" to the OD of the perforated portion is arbitrarily
chosen to give the best approximation to the outer boundary of the hole
pattern.
........,,-----a ____,_.-t
p
,,
FIGURE 14. Idealized Perforated Plate
The skeweq.hole array causes the neutral axis of the plate to. be
displaced :Qy an amount '.'e" from the geome.tric centerline of the plate~
That portion of the plate outside the radius "b" is treated as a
solid ring.·.
The ·perforated portion inside the radius "b" is treated as a solid plate
by using Horvay's equivalent elastic constants for Young's modulus and
Poisso!'\ 1 s .ratio.
The stress and deflection of the perforated.portion (to be referred to
·as part· "B") is determined by ·the applied load and the restraint caused by
the ring ·(to .be called part "A"). See Appendix E for the .list of symbols.
KAPL-1904
30
.r-
.
I··R
, .. c
I b
;r
·.
-
T
.
w ... ~ -- I
·-:-
ri, +e' M
" :t___ .....E~·
- - t
..
l
ai ...
Q •
l__,_ ~ ~4 fj .
4 ••
Q
c,
then
($ee Reference 19), (1)
KAPL-1904
3.1
(5) .
-
.K
A bc W
; (7)
therefore,
·,eA(e+e 1 ) + KA.[ P ·- pg (~) - 1;- wJ
T (8)
KA +· KB
where
(9)
. ,KAPL-1904
)2
r~-t
Qonsider plate "B" and let 6B Rotation of "B" at .rI = b (Reference 18)
then
(11)
w11ere
E'*t:3
D* = . (12)
12(1-1-1*2 )
therefore,
y
KAPL-1904
33
Mia~ _
--
· EI
.b
(1+~-t·*)D*
2
[pb .
--M
. 8
J (14)
(16)
M _ a 1 (KA+KB) + bCA(e+e 1 )
2
. ; {{cB P~ 2
CB [~1 (KA+KB) + bCA(e+e 1 ) 2J + bCA(KA+KB) .
____:k(e+e 1- )
.[ K ..... m~ ,;.....
c w + [. 91· -· KA( e+e 1
- )J
b p + pg '[a+c
-- --,--,A
,·J
.{ . KA + KB al . · · . . ~A + KB · ~1 · . 2al
(19)
+ p{a..;b)
KAPL-1904
34
00
k:JOO
A~LJO
A . .
and
t
e = Y - 2 (21)
therefore,
(.22) .
The equivalent elastic constants and stress mill tiplier are 'lli:U:H:H.l uu
the rat~o of the hole spacing to the·minimum ligament width as measured at
the center oi the plate; i.e., at t/~.
Bending stresses in perforated plate 11 B11 are as follo\.J'S: in a solid
circular plate the bending stress in both·the radial and tangential
direction is given by:
y
KAPL-1904
,.·
35
The bendi_ng s-tr.$ss in the plat~ "B" can be obtained by superimposing the
stress caused. py the unifol}Il pressure ·1oa9 p and the edge mom~nt· M.
The bending moments -caused by press~~e on simply ·su::ppo_rted circular
~-
Mr '"' i6 (13+!l*)(b -r
2 2 ), (see Reference 18). (24)
• (26).
Bottom surface:
KAPL-1904
·:·:.::-:.·· ...... .........................
'•,
36
2
r (1+31'*) J M}
+ + ~ }} (30)
(31)
. (32)
'.(33)
Top surface:
E
'(34)
r
[\-)t.t.nm · aurface:
(35)
KAPL-1904
. 37 .
Deflections:
' '
~The ·vertical displacement at any l'adius r is the sum· of tne displace-
ments caused by the rotation of "A" and the bending or· "B" plus the shear
deflection of "B " ·
'
where wA = displacement at r =b caused by rotation, and
/
'/
{36) .
=
(b2-r2)
(l+~*)D*
{p64 [..
-
*.
{5+~ )b 2 - r 2 (1+~*)
. .·
J - -
M }.
. 2. .
•.. (37) .
ws =2-
3 . 4G*t
p
(c2-r2); . . (38)
where
• * :. . 2h - 2h .E
G - 2R G - ZR X 2(1+~) 1 (39)
(40)
(41)
Rotations:
· The rotation, of "B" at any radius .r is given by:
e{r) = (l+~*)D*
· r ·{E.. ~b2 (3+••*). - . ~ 2.(1+~•*>]
16 L . . ... . ~ .. • .·M~.
) r
< b.·
·
(42) ..
. . . .
,..
r>b. (43)
\ To find the slope at any r, the first derivative of the shear deflection
must be added to the equations for rotation;. he. ,
dw
dr
;:;; - (44)
KAP!rl904· ·.
THIS PAGE
WAS INTENTIONALLY
LEFT BLANK
.39
APPENDIX B
·.
..I 1--- c
b2
·b3
...
\
R
\! I
~
T
··)M -- '
~
' t + - :
- e
t
t
~~-tti e2
. . I\ r~ t p
a, . lr
\ '
t1-.
I
,.
·H+•+.
.p Q . b4
.
. '
~ b
.
c,
• a '.
"
KAPir-1904
40
KAPL-1904
41
L'ottom 8u.r.face
c -----?1'1
f
a
• ';Bottom" Surface
TI1e geometry used in t.his computation is the same· as that for the "top"
surface 1.-ri t..h the following exceptions~
c . - 8.25 in. radius to point of support
g = 2.50 in. vertical outer surface exposed to pressure
m = 1.50 in. .vertical distance of point of support to
CG of outer annulus.
The applied forces are the same as for th·e "top" surface with the
following·except:i.on~=J:
R ..
-
p = 4·135 lb/in.-psi
. reaction at support.
KAPlr-1904
/
42
CA I,
X
•
b
1 +-
CA(e+eJ2 ,,
al KA + KB
+ p(a-b)
The remainder of the computation follows the general case, the foregoing
being the only exceptions which were brought about by the difference in
geometry and loading.
KAPL-1904
43
Experimental Data
The measured deflections and stresses were obtained from the ex-
perimental da~a plotted in Figures 20 and 22 for the "top". swface and
Figures 21 and 23 for the "bottom" surface. The deflection measurements
were duplicated at each radius for comparison and the read~ngs were
averaged to produce one.set of data. Th~ failure of the strai~ht line,
through the measured. points, to pass through z~ro is attributed to the
initial preload required to seal the gasket against the hydraulic load.
The deflections are given in the following list.
Deflections/Pressure lDad
Radius, Deflectionoz mils/;esi
Gage in. .Top Bottom
1 1.5 5.00 6.60
2 3·4 4.15 5.45
3 4·5 2.90 4.60
4 7.0 0.80 1.70
ligament* 1 2 J
** a b c a b c a b c
" -- -- --
Bottom 5.37 4· 70 5.42 4.70 4.60 4· 70 5.30 4.80 5.37
Top 6.21 ).96 6.94 6.94 5.29 7.57 6.94 5.54 5.96·
KAPL-1904
........ ·• . ~. l •••
44
,-
8 .
I.
/
~~2
-- v ./"" . . ~.L_
2
v ~r:::: ~~~
~
!--"
4
KS-18652
0
· o 200 40o soo aoo 1ooo 1200 14oo 16oo Unclassified
PRESSURE LOAD (psi)
•
10 1-
_a /
V' 2
/~v3
(/)
__.
~ ./,....x
~Vx
/ ~ v .
~ ~f/ I
--
i
__..; 4
2
..---- ____. --
~ L.-----'
1-.--
-
KS-18653
.. .. ---
200 400 600 800 I 000
PRESSURE LOAD (pSi)
1200 1400 1600 Unclassified
FIGURE 21o Altnninum Plate Deflection Data - "Bottom"· Surface
KAPL-1904.
45
7
- 10001----t---+-----' 3,4,9
"''g
6
I, 5
z 800 f----+---+-~
:::: 2
~- 8
z
~ .600
Iii
200
QOL-~~~~~--~~7=-~~~~ KS~l8654
200 4oo ~~~su:Eo~oAo~~~31 1200 1400 16oounclassified
FIGURE ~2.Q All,.lillinum Plate Strain Gage Data = "Top" Surface
•
1\.
I OOO•f----+-----1-
<1
cr
Iii
KS-1865.5: .··
0 ·200 400 600 800 1000 1200 1400 1600. Unclassi:fied: ......
PRESSURE LOAD (psi)
-~
vi KAPL-1904
The Young's modulus. for the material is E = 10 x 10 6 psi; therefore,
the Above strains when multiplied by 10 7 became the stresses in psi per psi
of pressure load. It should be recall~d that the measurements represent
two. separate tests; in both cases tension in the lig~ent was measured,
the surface with ~he thickest ligament was designated the ''top. 11 The
strains are presented in the above manner so that the strains at the same
point on one surface can be T(l.Ore easily compared with those on the_.·other
surface. In both cases there is a variation across the ligament, the
larger variation occu,rsi with the· larger ligameht.,
KAPL-1904
47
APPENDIX C
,.
rr~=b3
.C
I
I. ~ b2
\1
''
. '
...
'
RI I
'f!
. . ,
-~-,1.
I I .. .. +
e. ..
~-·
I
' j ' i' p
.rr--;::1!·,
~
g· - .· J'·
f f I.:.,.
p ...
~
:..t.'
.f •
·'
:\
I
~
o,
'c,
.j ·-.
'
~~ b, I.
Q b
.~.
" .. (J ""'
.'
K/\Pir 190 I;
48
The '
equivalent plate constants from Figure 28· are:
'
.-
E* =_0.145}j:;
' . ~* = 0.56.
T.he actual Young's modulus and .Poisson is ratio for the plate mate:i·ial
(steel) are:
E 30 X .10 6 psi, ·1-1 = 0.3
=
G = 11.5 x 10 6 psi shear·modulus.
TI1e forces applied to 4he ring are:
Q = 12,800·lb/in.
R = 19,500 lb/in • .reaction at support
p = 15,600 lb/in. pressure in 12 outer·holes
p = 2, ;>OO. 1b/in. 2 uniform pressure load
. w = 2,900 lb/in • (f~r coeffic,ient friction f'.·= 0.15).
The foregoing values are used in the general analys~s as presented in
Appendix A · to compute the equivalent solid plate stress. The stress
multiplier (Y) from Figure 29 is ·taken from the cU!'ve for t3 - +1 for·
T) = 0.24 and is Y = 4·3~
KAPir-1901~
49.
/
70.0
-- ---
60.0 ·-:-
50.0
..
.------ 4
§ ~
r--- 3
~ 40.0
~ .....-2
z ~
0
i= ~ /I
Iii
..J
l1.
~
30.0
~ ---- v /
~
/__::__
/
~
v
20.0
v
r-
10.0 ~ ~
v
/ KS-18657
Unclassified
0 500 1000 1500 2000 2500
PRESSURE LOAD (psi)
This is the ratio ot; ligament stress to solid plat.e stress at the
narrowest part of the ligament. This stress is the maximum stress in a
ligament, in· accordance vith the theory, only for thos~ ligaments around
the center hole. The variation in the principal sol~d p~ates stres~es
from the center to the oute~ ligaments is small so that . the
. ,
maximum
.
stress in the plate is the stress in the ligaments aro~nd the ·center_hole.
Since this variation in the principal stresses is small, no attempt was ·
made to fipC. the locatton and magnitude of t.he t.lleur~ti_cal maximum
stresses in the ligaments other than those around the center hole.
Therefore,-ihe value of Y = 4.3 was used tq compare cal9ulated st;esse~
with the meas~ed stresses at the center of the ligaments.
The 3/8 Model Experimental Data
Th~ measured defleotiono shown· in Figure 10 and the limeasured 11
_stresses in ,Figure 9 \vere obtained· from the experimental Q.ata in Figure 25
and 26. There_ were many more strain gages than those ~l).own (including
duplicates of the_se gages for. comparison purposes); l).owever J the data
from these strain gages are the only ones of interest for correlation ·
with calcu1~ted ligament stresses. The deflection gage readings ~ere
also dupli~ated;_ those shown were taken from along one radiusJfrom'the
center to the outer edge of t.he .head:
KAPL-1904
50
,.,
2200
J: 2000
0
~ 1800
f3
5 1600r---~r---~----~--~~~---+
~
@1400+ 2
.~ 1200-
2
b1 '100 '
Iii 4
600
200 KS-18658
3
~--~~--~~~~--~~--~~---Unclassified
0 500 1000 .1500 2000
PRESSURE LOAD (psi)
2500
...
FIGURE 26o The 3/8 Model Strain Gage Data
Deflections
Deflections
Tot.a.l Relative to :
Gage Radius z · in. Deflections! in. Gage No. 4a in ..
KAPL-1904
51
At the edge of the head, gage No. 4 me~sureq the lift of th~ head; there-
fore, defleotions relative to gage· No. 4 represent the net pending of the
heada '· .
.stress
~- .Radiusz in. · strainz iri./in. stress~ J2Si
·KAPL-1904
THIS PAGE
WAS INTENTIONALLY
· LEFT BLANK
:APPENDIX D
The following curves are takeh from the published works of. G. Horvay
and are included here becaus.e they are an integral part of the calculations
for the alu.minum perforated p;Late and the J/8 scale model head. The
symbols·used are defined as follows:
-2h
1
= mi.nimum ligament width
2h equivalent average ligament width for bending stiffness
2.n equivalent average ligament ·width for tensile s'tiffness
2R = hole spacing or pitch
·E. = Young 1 s modulus
D = flexural rigidity
E*}-
* -
J.l
Horvay's equivalent elastic constants
y = stress Qultiplier
f3 = ratio of principal solid plate stresses.
FIGURE 27. Ratio ()f Equivalent Ligament Width to. Minimum Liga-
ment Hidth, in Rending and Tension
4.0
3.0 ~
~ r--.. . t'-...Z.!L
£h
!'"--- ........ r-
2.0
-- r- ..
I . h:- ........ r-
~
2h -=t::::: --- --- .... _t:-
--- --
1.0
I.
.01 .02 .03 .04 .05 .I .2 .3 .4
-~
,5
:.:-- :::-::::
0
KS-13659
·Unclassffied
54 . /1
1.0
--
.9
.8 r- fJ-111
.7
.6 ....... ....... t'?'
.5
r-- ...... ,..... ~,
4 /
~I r'· ~'--· I'"
.3
~:/
2
/
///
/
.10
.09
/
/
/ L
/
/
.oe D /
.07
.06 D / E*
.0!1 / / E ~ ....... ..... . - ... .... --· .. ··-····
()4
v /
~
.03 / "' I
~/ /
v
.02
.v ./
v ...
-----
..
- ----
v
.. ·----- . . . ·-··
·0 I.01 .2 .3 .4 .5 .6 .7 .8 .91.0
.02 .03 .04 .05 .10
KS-18660
1l= ~ Unclassified
2R
-· --- -•OCB~
200
l\ -9-
10 0
\\ /3· -1
(J= +I CTr• CTn
50
\
,\ /3 =0 -Q- CTn=CTO
~
~1\_,, -9-
0\
20 - /3 •-1 a-1-CTJ1
0. ~ \
',',
!l.O ,--~~
..............
~ _
..........
...__ ..
2.Q I
1.0
KS-18661
Q5 1.0
.01 .0~ Unclassified
KAPL-1904 :r
55
APPENDIX E
r ...
A •
..,-
KAPlrl90L~
~.
56
t
in.·-lb
Mr = sum of external moments on outer annulus,
in.
KAPlr-1904
. 57
KAPL-1904
58
REFERENCES .
9. K .. A~· •Gardner,
"Heat-Exchanger Tube-Sheet Design, " Journal Applied
Mechanics, Vol 19, p. 159, 1952.
10. G. Horvay, "Bending of Honeycombs and Perforated Plates," Journal
Applied Mcoha.nios, V~l 19, p. 122; 1952.
11. Discussion of Items (9) and (10) Journal Applied Mechanics Vol 19.,·
P· 572 and 402, 1952~
12. G.. Horvay, "Transient Thermal stresses in Circular Disks and Cylinders, "
Tr~nsactions. AS1~ Vol 76, · p. 127, 1954.
...