The University of Manitoba

A Model Study for Infl uence Surfaces

of Curved Elastic Isotropic BrÍdge Decks

by

G. L. Asoka J. De Si'lva

A Thesis
Submitted to the Faculty of Graduate Studies.
in Partial Fulfi'llment of the Requirements for the Degree
of Doctor of Philosophy

Department of Civil Engineering

l¡linnipeg, Manitoba
0ctober, 1 974
 A MODEL STUDY FOR INFLUENCE SURFACES

OF CURVED ELASTIC ISOTROPIC BRIDGE DECKS

by

G. L. Asoka J. De Silva

A dissertation subrnitted to the Faculty of Gra<tuate Studics. ol'

the u¡riversity of Marritobl in purtiul firltillment ol'the rcquiremcnts
of the tleglee of

DOCTOR OF PHILOSOPI.IY
@ 1975

Pernrission has becn glantctl to thc LIBR^Ry OF¡ TilU tJNIVUR-

slrY ot" MANlrollA to lcntl or sell copics t¡f this dissert¿¡tion, to
the NATIONAL LIBRAItY Ot' CANADA to r¡licrofilnl this
dissertation and to lend or setl copics of the t'ilnr, antl uNtvERsl'ry
MICROFILMS to publish llr lbstract of this dissertation.

Tlte uuthor reserves other publication rights, und neittrcr thc

dissertutio¡r ¡ror extensive cxtructs tionr it uray be printetl or othcr-
wise reproduced without thc arrthor's writtc¡r ¡rernrissiorr.
i .:li

iSIUMMARY

This dissertation dea'ls with the deveropment of a simple

but precise experimental technìque that can be used effectively in
model analysis as a meansof studying the more complicated structural
systems. The inadequacy of the analytical methods is discussed,
indicating how these deficiencies point to the importance of considering
structural model analysis and testing as a design tool. After a
brief review of the experimental techniques available for model

analysis, a detailed description of the technique used in this work

is presented. The combination of versatitity and simp'licity offered
by this technique makes it suitable for use even in remote fiêl,d
design offic es. A unique curvature meter was developed as a part
of the work and details of its design are presented. This design has

many advantages over those used by other investigators. One of the

main advantages of the writer's design is that a twisting curvature
associated with the two orthogonal bending curvatures can be obtained
without rotating the meter. The need for rotation of the meter through
450 is inconvenient and time consuming and is required in all other
curvatures reported in the 'literature.
After a brief discussion on the necessity of the use of
curved bridge decks Ín modern highway systems, the results of tests
on three aluminium models of curved bridge decks are described and
used to show the effectiveness of the experimental technique. Two

-1
 ::Ì:i':ì.:.,

of the models chosen represent the two extremes of the simply supported
curved slab bridges that are popular in practice. The thÍrd model Ís
used to study the effect of an intermediate support on one of the
simply supported slabs mentioned above.
The concept of influence surfaces,for s'lab-o or slab-like
structures is reviewed. Non-dimensional curvature influence surfaces
are prepared from the model results. A brief description is given
of how these influence surface could be used as aids in bridge designing
as well as in checking existing curved slab bridges for abnormally
heavy moving loads.

validity of the appf ication of model results to proto-

The

type slabs has been discussed. The limits and the method of incor-
porating any difference in between the Poisson's ratio of the model
material and that of the prototype material are also included. The
mode'l results are compared with the exact solution results obtained
from the computer programme developed and are found to be in good
agreement.

The influence surfaces so prepared for the curved slabs

are compared with each other and with the influence surfaces that
have been published for rectangular bridge decks and are discussed.
Finally, thirty-five influence charts are presented in Appendix II
to be used as a permanant record.

- lt -
 NOT,ATTONS

The followìng 'i.ncludes the notatfons generally used thrsughout this work.

Occational deyiatlons are deftned where they occur tn the text,

C,, CR,, CeR Calibration constants

D Plate stiffness
E Modulus of elasticity
H HeiEht of the curvature meter
Ie, IR Non-dimensional bendi'ng curvature Ìnfluence values
IeR, InO Non-dimenstonal twisting curyature inû'luence values
K Curvature
Ko' KR, Ke+45o Bending curvatures
KOR, KRu Twisting curvatures
L Gauge 1 ength of the curyature meter
LIU Span of the model
LT M'icrometer carriage gauge length
Mo' MR, Mu*45o BendinE moments per unit length
MeR, MRe Twistjng noments per unÍt length
P Load

R Orthogonal axis in radial direction

Rl Radius of curyature
f, fe, fR, fo+45" Relative deflections
h Plate thÍckness
I CIpthsEoñal¡ a*te in tangenttal drtrection
p Fotssonrs Ratio

iii
 ACKNO1^lLEDGMENTS

The writer wishes to express his thanks to his supervisor, ,,.'.:,',,,.,,.

Dr. A. M. Lansdown, for the help, guidance and encouragement offered
by him during the writer's four year research prograrnme at Manitoba
University. ..;.
,,
,.,,',:t,,, ,

He also would like to express R.

his thanks to Dr. K. .':r:' ':

. .:: -i-r.:.: ì.t ::

Mclachlan, for his invaluable advice and assistance, particularly
-.
.:.1:,:,,

during the experimental work. His superior knowledge in instrumenta-

tion and electronics is sincerely appreciated.
He like to express his thanks to the Civil
also would
EngÍneering Laboratory staff, particularly Mr. E. Lemke and Mr. V.
Nowose'lski for their assistance in constructinq the curvature meter

and the model s.

The writer wishes to offer his special thanks to his wife

for her heìp in proof reading, tabulating test results and her to1er-
t;¡j., .,:,:.
1 ¡ a'.

ance during his many disappearances, particularly during testíng and ¡.::,,'¡,.',:,'..

the preparation of the manuscript. ''"

:
'' .; __.
''
':.j.
, :.. __-

During his four years of research at Manítoba Universityo

the writer has been sponsored by the Commonwea'lth Scholarship and

Fellowship Division of the Association of Commonwealth Universitíes. .r,,¡,,:;,,,,,.,,

For their assistance he is most grateful.

'': : '';r'

-iv-
 TABLE OF CONTENTS

Page

SUMMARY 'l

NOTATION. 'iii
ACKN0I^ILEDGMENT. iv
TABLE OF CONTENTS VV

LIST OF FIGURES . ". .viìi

CHAPTER

I INTRODUCTION.

2 ELASTIC PLATE EQUATIONS AND REVIEI^I OF SOME FREQUENTLY

USED EXPERIMENTAL METHODS IN MODEL ANALYSIS 9
2.1 Elastic Plate Equations. 9
2.2 Review of Some Frequently Used Experimental
Methods in Analysis.
Model 1l
2.2.1 Moire Method 1l
2.2.2 Strain Gauge Technique. l3
?. ?.2.3 Curvature Meter Method. 15
2.3 Ap,pt ì car¡iahì or¿ Ëqasef
:. "",. . å; '-"
cËÞqatåcrfiêorj, tôlCIö.ricþête . I 8a
.

DESIGN OF CURVATURE METER . t9

3..| Rationale Behind the Design. 19
3.2 Design of Main Body and Legs of Curvature Meter. 24
3.3 Curvature Meter Mounting Device. 28
3.4 Electronic Equipment Associated with the Curva-
ture Meter 30
3.5 Edge Curvature Meter 33

4 MODEL AND TEST EQUIPMENT. 35

4.1 Construction of Models 35
4.2 The Pre-Loading. 36
4.3 The Knife Edges and the Testing Arrangement. 39
4.4 Loading Mechanism. 42

EXPERIMENTAL PROCEDURE. 45
5..| Cal ibration of Curvatur.e Meter . ,. 45
5.2 Procedure for 0btaining Curvature Data 52
5.3 Material Properties and Calibration of Loading
Systems. à:. . 55
CHAPTER Page

6 OBSERVATIONS FROM CALIBRATION AND SYMMETRY TESTS 56

6.1 Calibration Constants for the Three Transducers 56
6.2 Checks on Experimental Results. 59

PRESENTATION OF TEST RESULTS 64

7.1 Dimensionless Curvature Influence Coefficients. . 64
7.2 EvaluatÍon of Influence 68
(a) UnÍformly Distributed Load (U.D.L. ). 68
(b) lnlheel Loads. . . 70
7,3 Comelation of Model Results to Actual Structure. 72

ANALYSIS OF RESULTS. 76
8.1 Comparison of
Inf1uence Surfaces. 76
(a) The Center Reference Points (A) of Model I
(t5o opening anole) and Model 2 (45' open-
ingangle) .! 77
(al) Io Inf'luence Surface Fígures I-1 and
IYI3 77
(aZ) Ip Influence Surface Figures I-2 and
ïrl3 . , .. 79
(aS) Igp Inf'luence Surface Figures I-3 and
I-t5 . 81
(b) Reference Points (B) for Model I and Model ä 8l
16 ïnfluence Surface FÍsures I-4 and I-16. 8t
(c) Reference Points (C) for Model I and Model 2 82
16 Influence Surface FÍgures I-5 and I-17 . 82
(d) Rðference Points (D) foi Model I and the
Continuous Model . ,, . 82
(¿l) Ig Influence Surface Figures I-6, I-18,
and I-29. 82
(dz) Ip Influence Surface Figures I-7, I-19,
- and I-30. 83
(dq) I6p Influence Surface Figures I-8, ï-20
" añä I-31. .. 83
(e) Reference Points (E) for Model l, Model 2 and
Continuous Mode'l . B4
16 Influence Figures I-9, I-21 and I-32 84
(f) Reference Points (F) for All Three Models 84
16 Influence Surfaces Figures I-.|0, I-22, and
84
(g) Reference Points (G) for All Three Model s 84
Ig Influence Surface Figures I-ll, I-23 and
ï -34. 84
:.:..:.-.:.ì::i._.::_.,:,.,..:..::;..1.:.i:a.:.,:.;...-.....:.:.:...1..1.:.:..i:..:...:::...-. ...-.::..:.:::

CHAPTER Page

(h) Reference Points (f) tor Model 1, Model 2

and Continuous Mode'l . 85
I" Influence Fiqures I-12. I-24 and I-35. 85
(i) Rëference Pointõ (A), (B)'and (c) on the
Middle Support of the Continuous Model. 85
In Influence Surface Figures I-25, I-27 and
II28 and lgp Influence Surface Figure I-26, 85

CONCLUSIONS AND SUGGESTIONS FOR FURTHER RESEARCH. . 87

REFERENCES 92
Appendix I. 97
I . I The l,rlri ter's Inverse Approach. 97
(a) Error Due to Tilt of the Meter. 97
(b) Error Due to the AdditÍonal Rotation of
the Meter 98
(c) Error Due to Shiftíng of Outer Leq. 100
1.2 The Effect of the Gauge Length . . . 102
(A) Curvature Meter at the Center of the Beam . 'l 03
(B) Curvature Meter at the Quarter Point of the
Beam. .. 104
Appendix II . , 106
2.1 Tables of Influence Coefficients 107
2,2 Influence Surfaces , . 123
 LIST OF FIGURES

P.age

Figure 2.1 Moment DÌagran 10

tigure 2,2 Curvature Meter Principle 11

Figure 2.3 Position of the Deflection Measuring 17

Transducer

Figure 3.la Curvature Meter 20

Figure 3.lb DetaÍts of the Curvature l4eter Base 20,

Figure 3;2 Real Case 22

Figure 3.3 Hypothetrìcal Case 22

Figure 3.4 Imaginary Leg F 23

Figure 3.5 Tilt of o+45' Axis 23

Figure 3.6 Curvature Meter Legs 27

Figure 3.7 Alignment of the Legs on the Base 27

of the Curvature Meter
Figure 3.8 The Curvature Meter l4ounting Device - 29

Fi gure 3. 9 The Resersed Mounting 29

Figure 3.10 The Schematic Diagram of the Electronic 32

Equipment Arrangement

Eigure 3.ll Fixing Clamp 34

Figure 4.1 Lay Out of Models 37

Figure 4.2a Loading Schedule 38

Figure 4.2b Locating of Punch Marks 3B

Figure 4.3 The Pre-Loadlng 40

Ffgure 4.4 Knife Edge Support 40

Figure 4.5 Test Set Up 41

vtll
 Page

Fi gure 4. 6 Loadi ngiulechantsu 43

Fi 4.7
gure Cantil wer System 43

Figure 5.1 Plate Subjected to fure Bending 46

Figure 5.2 Ca'libration Set Up 49

Fìgure 5.3 Posittye and Negatl've Curyature 53

F'igure 6.'l Typical Cal ibratlon Graph Showi.ng 57

Ins tgnî fi cant Hysterests
Figure 7.1 Local Curvatures ln Model and 74
Actual Structure

]X
 LIST OF TABLES

Fage

Table 6.i Test results of some of the symmetrlc 61

l oad poi¡nts fot the reference poi:nts
A, D and Dr.
 CIIAPTEB I

INTRODUCTION

Wjth the lncreastng demands for the use of statical'ly in-

dreterminate structures, especÌa11y for nodern h'ighway systems,

refi'ned and more accurate methods of analysls of these structures

are required. In applying existing str.uctural theori.es to the
analySÍs of structureSoquîte often the desiEn englneer ts requi.red
to incorporate simp'lÌfying assumptÍons with regard to the deformed
shape and structural behaviour. Even with the finite element tech-
nique which is considered to be the most powerful analytical tool at
the present t'ime, one has to work wlth a structure that corresponds
to presumed deformations or stress dÌstributîons. Usually it is
d'ifficult to v'isualize and to predi:ct the behaviou,f which is the
governing factor of the accuracy, that leads to optimum designs,

Simplifying assumptions always tend to be taken on the conservative

side for safety, and the economy of the structure depends on how
conservative they are. it is always essentñáj to verify the efficiency
as weJl as the reliabil ity, 'þefore an ana'lytical method is made avail-
abl e for the use of others,

One may suggest checking the reliabîlity by applyÍng the select-

ed ana'lytical methodto a proh'lem, tvhere the exact solution is avaj'lable,

In nost practlcal cases, however, numeri'ca1 solution of these equations
utiltzing a dîgîtal computer is the only practîcable method.
 -2-

Such a method of solution will be refert^ed to as the rrexactil nethod

in this thesis
Also, in general, the problefls that haye exact solutions
are not complex. Therefore, in solvtnE these problems, eyen by the
so-called approxirnate analyti'cal methods, it nay not be necessary
to make severe assumptlons as rìn a complex problem, Hence, this
method of checklng will not verify the relrìabÌllty of an analytical
method tru1y. In addîtjon these exact ana'lyses are really only a

better approximatîon to the exact theory because the practical

accuracy depends on the convergence of the hyperbolic and trig-
nometric series employed. Although one finds perfect, rapid
conyergence for deflections, in the vÍcfnity of the load the
convergence becomes slow for bending nonents and very slow for

.f,.u, (1 ). The series become dq\ergent l'n the vi'cinity of a con-

centrated load [2), u, mentioned in Chapter VII.
The numèrical solutions to all these powerful approximate
analytica'l methods, including finite element, fÍnjte strip, folded
plate analysis etc., that require a matrlìx formulation are just not
attainable without the aid of dîgital computers. At the same time

an analytical method is only as good as the data fed in at the

beginning since no computer FleögremeeWtltfreoppee't the poor
assunptions made in the analysl's,
of the htghly sensitiye êlectronic
ù.Iith the development
apparatus, the method of mode'l än,e&y9*saËe,G0m$s.ifäî0añSwer to the
above-menti.oned shortcomlngs of th.e analytical methods. Many
 -3-

investigators particularly in Europe(3'O'5,6) have shown the reliability

and scope of model analysis and testing as an indispensqble design tool,
particuìarly for structures of complex shapes and boundary conditions.

The testing of models not only demonstrates the true structural

behaviour under load, but also provides the stress and strain indirectly
through the deformations such as deflections, rotations, e'longations
and shortenings which can be measured on the model.

The main objective of this dissertation is to demonstrate

experimentally, how model analysis and testing can be used as a

precise design tool in analysing structures of any shape and of any

complex boundary conditions. Also its ís to build up a simple,

purpose

but precise experimental technique which is efficÍent, but unsophisti-

cated, hence could be used even in a remote field design offíce.
In Chapter II, some of the experimental methods that are
frequentìy used in the model analysis, are briefly discussed a'long
with a detailed description of the technique used in this work. '

The technique of direct measurement of re'lative deflections by

curvature meters was used in this project. As a part of the project,

a unique curvature meter was designed. The writer's design dÍffers
in many ways from the curvature meters that have been used by other
investigatorr(5'7'8'9) so far. This is discussed in chapter III.
Minor but important design details such as the design of legs, the
procedure to obtain comect center points for the legs and the
transducer holes and the method of mounting the meter onto the
 -4-

model, that had to be considered in order to achieve a high leve'l

of precision, are also included in Chapter III. The writer's inverse
approach of checking the meter anaìytically on an assumed worst case
(a defìection equal to half the thickness of the plate which is
considered to be the limit for small deflectÍon theory) to obtain
a quantitative idea of each individual error that he feels cou'ld

be present in the actual testirg, is included ín Appendix l.l.

This is dó-called the "inverse approach" because initiat'ly, the
significance of each error was studíed indívidually. Then, where

appropríate, they were considered separately to obtain an optimum

design, instead of checking the quality of the meter in a direct

way (ìike most investigators have done), by using ít on a situation
with a known solution. The danger inherent in the latter rnethod is

that, by coincidence one may use a rare situation where the errors
cance'l out each other to give excellent results, which may not be

the case in the actual testing.

The effect of the gauge I ength of the meter on the measured
values and the size of the model are discussed in Chapter VII. An
approximate idea of this effect was obtained analytically and the
details are included in Appendix 1.2.
In the type of curvature meter designed, three displacement
transducers were used on two orthogonal and one 45o axes to find the
curvatures in those three directions. The curvature meter was

calibrated along each axis, using a strain gauge testing beam

 -5-

apparatus. This apparatus also works on the same principle as the

curvature meter. Details of the calibrations are included in
Chapter V.

With the type of device described above, curvature on'ly in

one direction can be measured when the reference point is situated
on or close to a of the model. Hence another "edge meter"
boundary

was designed to be used in these situations. This meter too is dis-

cussed in Chapter III though it was not used in the testing due to
poor performance in one direction.

To demonstrate the sui tabi'lity of model analysis, mode'ls

of curved bridge decks were chosen. Curved bridges are required
particularly in highly developed urban and suburban areasn where
many railroad level crossings and highway intersections are being

replaced by overpasses, underpasses and interchanges. Even in

rural areas, a'lignment and site conditions may be such as to make

it to locate a curve on the bridge rather than on

more economicaì

the approaches. As far as the writer knows, until a few years

back curved slab bridges Ì^,ere anaïysed using the simple beam

theory. Then came the more refined analyses for curved slabs,
such as finite strip method('10), curved strip metto¿(lI) (or
curved folded plate analysis) and discrete energy method(12).
Basically all of these fal'l into the same approximate but
conservative group of analytícal methods discussed earlier. The
writer feels that, comparatively very little research work has been
carried out on curved brÍdges, although they have become increasingly
 -6-

comnpn in modern highway systems as a result of smooth traffic

flow requirements.
For straight of various boundary
and skew bridges

conditions, span/width ratîos and opening angles, one can find an

ample amount of published design data(6'l3ol4'15'16'17). They
are usually presented in a form that is more useful to the
designer, namely "influence surfaces". But, so far no attempt has
been made to proVide the des'igner with this type of design data

for curved bridges.

Therefore as a second objective, information on curved
bridge decks was obtained from the study of models and is presented
in a form, that rnay be useful as aids to the bridge designer. Such

information was not previously available.

Many model tests particularly in North America have

been used just to study a particular unfavourable case or cases of

loading, as an aid of the design of the structure or iust in
verifying the design under "worst load"(18't9'20). Such tests

provide only quaìitative information of the effect of other

"heavy loads" that may cross the bridge from time to time. Hence
these model tests are not of much use, particu'larly to a bridge
engiilrleer, who has been called upon to check existing bridges for
the passage of abnormally heavy loads. Indivisible obiects such

as transformers, electrical generators, pre-cast highway bridge

beams and turbines are few examples of the so-called "heavy'loads",
that are being transferred from one part of the country to the
 -7 -

other. The 800 Megawatt generators for Bruce Nuclear Power Station
in Ontario is a current example of heavy indîvìsible loads currently
being built for canadu(ss¡. Out of the entire turbine-generator
set, the stator core and the windings alone weigh about 240 tons.
There will be a total of four of them at Bruce Nuclear Power Station.
The use of influence surfaces is an answer to the above-

mentioned timitations of the model tests. These influence surfaces

for two dimensional slab-like structures are analogous to the

influence lines of one dimensional framed structures. An influence
surface prepared for any action (moment, curvature, twist, shear, etc. )
at a particular reference point on a slab, provides the designer
with qualitative information of the considered action at the reference
point due to a load or any pattern of loading at any ptace on the slab.
such surfaces have been used first by H. M. hlestergaara(2). A rarge
number of such surfaces have al ready been prepared for standard
cases of structures, such as rectangular and skew slabs of single
span and also of continuous spans(6'B''13'14'.|5'16'17 '21'22'23) .

If a set of influence surfaces for the structure under consideration

are available, all that the designer has to do is to draw the loading
pattern on tracing paper to the same scale as the influence surfaces
and move it about on each corresponding surface, until the critical
location of the load is obtained. This may require more than one
trial whereas a person familiar with influence surfaces will spot
critical iust by glancing through the set of influence surfaces.
cases
The method of evaluation of these surfaces is discussed in Chapter VII.
 -8-

Thlrty-fíve Ínf'luence surfaces were prepared from the

writer's model test results. They are presented in Appendix 2,
in the form of contour plans which seems to be the most useful form
of presentation for the assessment of the effect of heavy'loads on
bridges. All these surfaces are plotted in terms of nondimensiona'l

curvatures. The advantage of so doing, ís discussed in Chapter VII.

The investigator has to use his own judgement in selecting the
reference points for influence surfaces. The points considered Ín
this project are discussed in Chapter IV.
Initially two model curved bridge decks made out of high
strength aircraft alumin,tum were studied experimentally. The model
sizes were chosen such that they represented the two extremes of
the simp'ly supported curved slab brÍdges, popular in practice.
A computer programme was deve'loped for the exact analysis solution
involving an infinite Fourier series. Model resu'lts were compared
with the "exact" solution results and found to be in close agreement.

A two span continuous model deck was formed by locating

a radia'l knífe edge at the center of the span of the second model
tested. The effect of presence of the intermediate support was

studied experimentally and compared with the single span model resu'lts.
No exact so]ution is available for the continuous case.
 CHAPTER II

ELASTIC PLATE EQUATIONS AND REVIEl^l OF SOME FREQUTNTLY USED

EXPERIMENTAL METHODS IN MODEL ANALYSIS

2.1 Elastic Plate Equations

The general elastic plate theory is well established and

has been known for years. Derivations and the limitations on the
applications of elastic plate equations can be found in any standard
text book on elastic plates(Eg' 1'24). However, the equations are

given below since they are necessary for an understanding of how

the experiments described later could be used'in the analysis of

plates. Attention is a'lso drawn to two of the limitations, to
the app'lication of these equations since they are crucial factors
in model analysis.
In a pìate element, stressed in the e'lastiìc range,
the following relationships exist.

MR=D[Kn*uKo] (2.1 )

Mo=D[Ko* Kn] (2.2)

and

MRe = -Mu* = D(ì - u) KRo (2.3)

where, MR, MU and M*U are moments O:rr unlt length of the plate AS

in Fisure 2.i l0; ** ana rc, - I a2r¡ I

shown on pase
] [= P [= *ffi -ææ)
rqpresent the curvatures in the R and 0 directions produced by the

bendins momenrs M* and Mu; and olu

{= þH å#] i, ü,. rwisrins
 - l0 _

M96

Fi qure 2. I l4onrent DÍ aqram

Terms Shown all Posítive.
 - 1t

curvature due to the torsional moment MOU.

D='12(1Eh3 -I
- u')
is the plate stiffness per unit width where E is the modulus of
elasticity, u is the Poisson's ratio of the material and h is the

thickness of the elastic pìate.

Two limitations mentíoned above are:
1) The model s;tåb must be thin comparedto the other dimensions, for
the shear deformation to remain negligible in relation to the
flexural deformation;
2) The deflection of the model must be sma'll compared to the thickness,
in order that the stresses corresponding to the stretching of the
middle surface of the p'late are negligible in comparison with
i the bending stresses.

2.2 Review of Some Frequently Used Experimental Methods In

2.2.1 Moi re Method

The Moire fringe technique is nou, freqently used in model

studies where the stress and strain can be derived directly from
displ acements.
of reflecting material or coated with a
A model made

reflecting surface is set up vertically,'in front of an i'llumjnated

cylindrr'cal screen, consisting of black and white paralleì lines of equa'l

width. The model is loaded from the back to obtain unobstructed

 -12-

fringes. Images of the screen reflected by the model ín the'loaded

and unloaded states are photographed through a hoìe in the center
of the screen. Both states are photographed on the same negatÍve.

Moire'interference fringe patterns appear on the photographic

plate since when the mode'l is loaded, the reflected array of 'lines

deform following the surface displacements induced in the model

and generally do not coincide with the lines reflected by the
unloaded model. These interference fringes represent contours
of constant slope normal to the screen line direction. By rotating
the screen, the slope ín any direction can be obtained. Numerical

or graphÍca1 intergration and differentiation of the slope curves

give the deflections and the curvatures in the plate respectively.
Hence the bending and twisting moments at any location can be

found makíng use of the relationship between curvatures and

moments and knowing the flexural rigidity of the model as shown
in Equations 2.1 ,2.2 and 2.3. For more detaíls, the writer refers
the reader to F. K. Ligtenberg's paper on Moiry'method(25).
is not suitable if very high accuracy is
This method
required since there is always a possibility of human error in
finding the fringe centers and p'lotting the slope curves. This
technique becomes less precise when the model has a shape other
than a rectangle or a square, especially a curved model, such as

the one considered in this work. In a curved bridge slab, Ít is

more appropriate to present curvatures, moments, etc. using the
 - l3 -

polar coordinate system. it is convenient to use curved and

Hence

radial grid lines with a curved model. since the black and white
,:,:
screen lines are parallel to one of the axes of an orthogonal ,,,.,,,

system, the fringe patterns that appear on a photographic plate

represent contours of constant slope in a directÍon parallel
or perpendicular to only one radíal grid line at a time '.'.,

Therefore, one has to employ the conditions of equitibrium (Mohr's

::.: .:,
circle) to obtain radial and tangential curvatures at grid points
on the other radial grid lines. The other alternative is the
expensive method of taking two photographs with the screen lines
'l
paraì e'l and perpendicul ar to each radial grid I i ne. From hi s earl i er
experience on the application of Moire technique to mode'ls of
curved bridge decks, the wrÍter is aware of the amount of work
and the errors involved in finding a slope aìong a curved grid
line. The above method becomes re'latively expensive if high
accuracy is needea(8). :t.:: :

,:':jtìit.'

2.2.2 Strain Gauge Technique .,,:,.,::

:....: .::
The straín gauge technique is certainly the most widely
used technique in model analysis and has been known for many years.
Details of the technique can be found in any book on experimental
,,:,,,,,.
(Eg' 26),
stress analysis therefore will not be given here. Moments '.:::.

are calculated from the surface straíns of the model that are
obtained from the strain gauges affixed to the model. The

magnitude of the reading and therefore its accuracy increases with

the distance of the strain gauge from the middle surface of the slab. ,1,',
 -14-

For this reason relatively thick slabs are advantageous for strain
measurements. At the same time one has to keep in mind the limita-
tions imposed by the elastic theory on the thickness of the slab,
as mentioned in section 2..| above.
l^lith strain gauges, high precision can be obtained with
a little care. The development of modern e'lectronic equipment

has made the use of strain gauges in model analysís much easier in
recent years. Typical of such high precision strain gauge study
is the work of Yeginobal iQ7) where he used electric strain gauges
to study a three-span continuous skew slab model to verify his
theoretical results. But a tota'l of 20,000 strain readings were

required to comp'lete the experiment. In 1961 Rüsch and Hergen.öd.r(6)

also made use of strain gauge technique to study models of skew slabs

to produce 174 contour charts of influence surfaces for bending and

twisting moments. A test of equilibrium showed a discrepancy of not
more than 4 percent in the final results.
In most of the places on the modelr €xGêpt at fnee edgeso

interior supports and exterior supports, the gauges to be

have

app'lied in the form of "3-legged rosettes". There is a difficulty

in finding the stress peak over the reference point (even if the
gauge length is sufficiently sma'll) if of individual
rosettes made

strain gauges are used, since these gauges will have to be app'lied
at some distance from the theoretical reference point. This
drawback can be avoided by using more expensive commercia'l ly made
 - l5 -

rosettes fabricated so that alì the individual gauges coincide at

one point. Hence with this type of rosette, it is possible
common

to have all the three gauges dîrect'ly over the reference point.
For better accuracy, it is preferable to have the same number of
gauges on both sides of the slab to seperate axial from bending
and for temperature compensation. Therefore a large number of gauges

are required for a detailed mode'l study of any structure. These

high costs and the amount of labour involved become a significant

drawback for the extensive use of strain gauges in detailed
investigations.

2.2.3 Curvature Meter Method

As mentioned in the first chapter, in model analysis

stresses and strains are not measured directly, but calculated

indirectly from the deformations that are measured on the model

and the elastic constants of the model material. Since the

characterjstics deformations produced by bending are curvatures,
an obvious way of determining experimentally the moments in a

model slab is by the use of curvature measurements. The relation-

qhips that exist between curvatures and moments have already been

shown in Equations 2 .1, 2¿2 and 2.3, The curvature meter technique

is based on the calculation of curvature from the measurement of

relative deflections of points over a predetermined base length.

These curvature meters have been used in optics for

a long time in curvature measurementso under the name spherometers.
 -1
. .._:: J:r'. ..:..- .....- t, - -.'.: . : .::-.: . -.:..':.'1.'":.. .1. :,,.:;..::: . r: ;::. .::: .1.:':.:'ì : ì i:':i:^r:lr:.:,:.;,: :":::t. ,:Ìì.:.:;. ::: ; :::jr'i..:: r:::,1

- 't6 -

Simpìe curvature meters were employed in model tests, as ear'ly as

',ron?a). Andra(5'9), R. Krieger(5), H. tnleigl
tnt. H. ù,reise ß'29)
/n\ "rQ9),
and A. Mehmel t"' made improvements to the curvature meter to bring

it up to its present advanced form. of the investigators

Some

employed the method of mechanical addition of deflections in order

to include the Poisson's effect while others preferred the addition
to be done electrical'ly. The difficulties of machining for the

case of mechanical addition has made the rnethod of electrical

addition more popular. In recent years, the advancement of highly
sensitive electronic apparatus has made it even more suitable.
It was not necessary to employ any of the above-mentioned
methods of addition with the work described in the dissertation since
it was decided to present the results in the form of curvatures, and
hence they are, independent of the Poisson's effect ín the plate.

The advantage of so doing is discussed in Chapter VII as mentioned

previ ously.

The basic principle of the curvature meter is that of

measuring the relative deflections at three positions on the surface
of the slab. In Fi gure 2.2 (page l7)'4, B and C are three equatly
spaced collinear points on a deflected surface having a local radius
of curvature, R,. If this spacing (L) is smal1 compared to R,, the

curvature from A to C can be assumed to be constant. In other words,

the three points A, B and C can be assumed to lie on a circular
segment of radius R,. Based on these assumptions, the
relationship between the measured rise f over the chord length AC
 -17 -

R,

./,

I c
---- -A lvl
|- | _1
iB
lf
Ltz Vz
Figure 2.2 Curvature Meter Príncip1e.

Figure ?:3 Position of the Deflection

Meosuring Tronsducer.
 - 18 -

(as indicated in Figure 2.2, page l7) and the radius of curvature
R is derived as follows:

(ll"*(R,-f)2=R?
t2
or î +R3-2Rf +f2=R,2
where zRf = Lz ! 4fz
t2
giving R, = L- i-4f2

since f is very small compared to 1,4f2 becomes neglígible in

comparison to 12. Hence,

12
R,= ãr
curvarurer=l=T
r\r ¡2 e.4)

The displacement measured is 2f for the curvature meter used in the

work since deflection measuring transducer was fíxed at the outer
point as shown in Figure 2.3 on page 17. Letting 2f be equal to the
registered value f , then

K=4f (2;5)
L2

Equation 2.5 is the basic expression used for bending curvature ,,,.

in the meters of the type used in this investigation.

 - l8a -

2.3 Application of Elastic Plate Theory to Concrete

Many investigators(.|8'.l9,27,31 ) validity of
huu. shown the
the use of elastic plate theory in predicting the behavÍour of reinforced
concrete structuresl Elastic behaviour of a structure can be obtained
from a study of elastic model also. The results obtained from an
elastic model study are limited in value to the same extent as the
results obtained from an accurate mathematical study based on the theory
of elasticity. Thus, we may state that the findings derived from a
well conceived elastic model are no better than but as good as, those
from a properly executed study of a mathematical model. Knowels and
Hugginr(ls) f.or their model study on the behaviour of a three-span 600

skew bridge with stiffened edges concluded that tests on a model made

from a reasonabl¡l isotropic material, such as aluminum, can provide the

necessary information for the design of a reinforced concrete strusture
under both service and ultimate load conditions.
 1-!:'-. .: ìl-::-,i;;.::a'::.:i,::::
::1 "

ÇHAPTEB III

DESIqN OF CUßVATURE METER

3.1 Ratiopa'le Behllhg thê DeSlgn

In ssne recent types of curvature meters, the meter
has to be rotated through 4bo wl'th respect to one of the nain direc-
tions consideredo in order to fÍnd the twisting moment associated
w:i:th the two bending moments in those main directions. The meter

developed for the present work does not suffer from this disadvan-
tage and a twisting nonent at any poínt can be found without rota-
ting the meter. To avoid this extra rotation, an additional third
transducer was placed on an axis 45o to the mal'n axis o as shown
in the Figure 3.lb on page 20. ThÍs transducer measures the rela-
tive deflection between poÍnts 0 and E in the dìagram, In Figure
3..|'ô, AOc and BOD form the gauge lengths of the meter in iê and R

dî rection respecti ve1y.

For each meter positrìon three deflection readings are re-
corded. From these def'lection readÍngs, curvatures Kro K* and
Ko+
s" at a po'int can be calcu'lated and hence the bending moments
Me,MR and the twisting moment Mu* can be obtaîned, wíthout adjusting
the meter. From Equations 2,.| and 2,2 it îs clear that once K0 and
K* are obtained, the ca]culatfon of Mg and M* i.s straightforward.
But the calcujation of fluR is not as easy since Kon is not known

drlrectly. But from equilibrium for curyatures in a plate (Mohrrs

 Figure 3.1a Curvature Meter.

1.5"

GAUGE LENGTH IN E DIRECTION

Fi gure 3.1b Details of the Curvature Meter Base.

 -21

circle), it is known that the following relationsh'ip exists.

Ke+45" - Kr*l t5"
KoR=T '

HencerMOn is expressed in a dìfferent form, from that of Equation 2.3

namely MoR = +-s) (Ku*4uo

- Ko*r35') (3.1)

However, only Ke+45o can be obtained from the readings of the meter
as designed. Thereforeo equiljbrium is used again to cal cul ate

Ke*135o. It is known that

Ko * KR = Kg+45o *Kó*j
35"

hence, Ke+l
35o = Ke * Kp - K6+45e

Substituting back into Equation 3.1 gìves

ïu*= %')[*u*45' - (Ko * KR - rr*ouo)l

Ke
= D(t - u)(Ke+45ö - 1 Kn,,1

From the Equation 2.5

.K^=! 4f0
oL2
and
o
''R -=4fR
;"
L'

Unfortunately the expression for KO+4bo js not as straightforward as

the above tvro. This is because the required relatjve deflection between
 -22-

the points 0 and E (Figure 3.1b, page 20) on the axis 0+45" cannot
be obtained from the third transducer read'ing f6+45o alone r.rithout
knowing the tilt of that axís 0+450. This tilt of the axis 0+450
cannot be determined as easily as that for the other two axes since
along the former, the curvature meter is supported only on one leq,
i.e. the common center leg. To make this clear hypothetica'l and real
exqÍnpleE qfe constdered qlong ttre tof R exi.s).
! Axts.

o ì:---'
Fiqure 3;2 Hypgthetical Cqse Fi gure 3 .3 Real Cqs.S

As shovrn in Fìgure 3.2 lf only poÌnt c gets def'lected and point A

remains undeflected, the curvature meter will remain level. Then the

registered reading of the transducer wil'l be the true def'lection of

C relative to 0. 0n the other hand, if points A and C both get de-
flected as shovrn in Figure 3.3, as occurring in the general case, the
curvature meter will be tilted due to the uplift of the bottom tip
of the outer leg. Here the registered reading therefore r^rill not be
the true def'lection of C relative to 0. To find the true relative
deflection, the amount of uplift of the bottom tip of the outer leg
has to be deducted from the registered reading. This upìift is equal
 -23-

to one half of the registered reading of the transducer and hence the
true deflection is half the registered value, for both axes R and 0.
In the case of the axis 0+450, this upìift, which gives
the tiltof the axis can not be determined as easiìy, due to the
absence of an outer leg. If r,¡e imagine an outer leq F on the axis

6+45" of length equal to that of the other three fixed leqs A, 0 and
B (as shourn in Figures 3.4 and 3.5 be'lorv), then the uplift of the

/' | ,-t

\-(
'.í:? /, E ;T
t\ B ,' sl
+l
o

Figure 3.4 Imag'inqry Leg F

tigu¡e 3.5 Tilt of,gt459 nris
bottom tip of the imaginary ìeg F wÍl1 be defined by the uplifts of
the legs A and B, since the'tilt (inclination) of the solid base carry-
ing these legs depends on those two uplifts. tfplift of F is the
average of those of is midv,ray betrveen those two po'ints.
A and B since F
As pointed out previously the uplifts of A and B are equa'l to one
half the registered readings on the respective transducers opoosite
to them. Therefore the true deflection of E relative to 0 is the
reading registered on thethird transducer less one-fourth of the sum
of the other two readings. 0f course the signs must be taken jnto
account.
 -24-

Hence, from Equation 2.4

tu t*
Ko*45 ,= (#ulz (ro+45o - ï ,

16
t'i t*
= (re*4bo - ,
L,

Substitut'ing back into MUO

fo*fR\ 4 (fo+fR)
Mur=D(l-u)t 16
4 t'L2
1.1
U -
-L. e+4so z

(tu+ f*) (fo * fn)

- u) É [
48 -
..,
= D(t ro*4sn -

?
- r) (r,
Ë[
= D(l fsa45,. - + fo)1
B

The expressions that were used with the curvature meter designed,
in order to calculate bending and tvristing curvatures are given below

4f^
K^=
oL' :u (A)

4fo
'"p -
v_r-\ (B)

'-ã-
v -- 16 tI t þ -B3 (t, + f*)J
''oR ¡z 'g+45 (c)

3.2 Design of Main Body and Legs of Curvature Meter

It is preferab'le to make the curvature device as liqht as

poss i bl e, in order to to its selfweight a

keep the defjection due
mi nimum. Although the initial curvature due to this selfr^reight has
 -25-

no effect on the ft'na1 ì"esults, since the whoJe system can be "zeroedrt

electront'caì1y after applying the meter to the model o a Iight meter

lvill help tn a11owt'ng a greater portlton of the total allowable deflec-
tion (for the small deflectÌon theory) to be obtarìned only from
external loadl'ng after the electrl'cal "zerotng", A detalled discus-
sîon of thÍs îs included ìn Chapter V.
A piece of alumlni.um, 2u x 2r x 3/81', was used as a base for
the legs and the housîng for the transducers. The centers of the legs
and the transducer holes (Figure 3.1b, page 20) were located using a

universal mi I I ing machi.ne 04axinat V-l0, Canadf an Edel staal Limited) ,

The carriage of this machine couid be.moved în two orthogonal directions,

with a precision of'l/1000 of an inch, using the attached micrometers,
It was ascertaîned that there tntas some back lash in the horizontal
moyement of the machine carriage, Therefore, care had to be exer-

cised in locating the centers of the'legs and the transducer holes,

The transducer holes were bored out to obtain a close snooth fit"
Hence, it was possrìble to moye the transducers smoothly, easily and

quickly in the yertical direction to obtaín the zero posit'ion. Poly-

ethylene screws were used to lock the transducer in place [Figure 3..la,
page 20).

In order to in curvatures, the

measure the minute changes

legs of the curyature meter should be pressed tlghtly agatnst Ùhe

surface but at the same time, its abi.lity;to adjust itself to the
change in t'gauge length" due to ctrcular arc defonnati'on of the plate

should not be prevented. If the outer leEs of the curvature meter

were rîgid and solidly fixed to the base, obviously
 -26-

they would offer some resistance to the deformation of the surface.

Therefore, localized stress concentrations would be formed around
the tips of these legs urhich could have a signifÍcant effect on the
local final deformed shape of the plate. An ansh,er to this is the
use of legs with free'ly movable bottom tips.
Both outer 1 egs of the meter were made of tlo parts n connected
together at the middle by a high tensile leaf spring as shor¡rn in
Figure 3.6 on page 27. The bottom part of each leq had a conical
tip of 609 and was hardened. The center'leg was macle of one solid
piece and this too had a hardened, 60o conical tip.
The outer legs were fixed on to the base as Índicated in
Figure 3.7 on page I,7. of care was taken to ensure
A great dea'l

that the plane of each leaf spring uJas orthogonal to the respective
meter axis through it. Each leaf spring offered stiffness in the
direction along the pil,ane of the spring, and little resistance in the
direction normal to jts p'lane. Therefore, the bottom part of each

outer leg was allowed to in the direction of its corres-

moveo on'l-y

ponding meter axis. SÍnce the meter axes u,ere orthogonal to each

other, any rotation of the meter about a vertical axis was prevented.

To make sure that the distances between the tips of the

outer legs and the center 1eg remajned equal to half the gauge length,
the tips of all the three legs were placed in prec'ise1y located 90e

conical shaped punch marks on the model.

 -27-

I
TOP PART

LEAF SPRING LEAF SPRING

BOTTOM PART

Two Views of an Outer Leg. Center Leq.

Figure 3.6 Curvature Meter Leqs.

9l
EI
vrl

,l
Ël
ltl
ol
..,i
zl
<l
dl

OF LEAF SPRING

Figure 3.7 A]ignment of the Leqs on the Base of the Curvature Meter.
 -28-

3.3 Curvature [vleter Mounting Device

Due to the eccentric weights of the three transducers, it
was not possible to support the curvature meter on its own three legs
alone. Therefore, a method that would ensure stability and firm contact
between the legs and the surface, without introducinq significant
restraints, to support the meter on the model.
was required

A very efficient techníque which had not been used in earlier

work, was employed to meet this desion requirement. The force re-
quired to hold the curvature meter against the model was obtained
from the action of a "string and spring" arrangement as shown in
Figure 3.8. Fírst, a nylon fishing line, knotted at one end was

passed through a hole of diameter 0.04" made jn the base at the center
of gravity of the meter. Then i't was taken through a hole (aiameter
0.04") made'in the model, right beneath the center of gravity of the
meter. After that it was passed through a helical spring. This
spring was guided by a guide pin made of aluminium, with a restraining
collar at the bottom. The purpose of the restraining collar was to

compress the spring against the model plate. To obtain the force
required, the line was tensioned and anchored against the inner wall
of the guide pin with a set screw (Fiqure 3.8, 29), wíth the
page

spring ín compression in between the plate and the collar. The tension
in the Iine could be varied by varying the compression in the spring.
It was tested and ascertained that, within reasonable limits, the
amount of tension in the line was not an influential factor on the
 SPRING

ET SCREW

Figure 3.8 The Curvature l,leter l'lounting Device.

Figure 3.9 The Reversecl Mountínq.

 -30-

resul ts.
l¡lhen it was required to load the model from underneath,
cl oser to the reference point, the loading stem interfered with the
guide pin. Then the "string and spring" arrangement had to be re-
versed as shown in Figure 3.9 on page 29.
3.4 Electronic Equipment Associated with the CúrVature Meter

earlier, the principle of the curvature meter

As mentioned

was based on the measurement of relatjve deflection over a gauge

length. Since these gauge lengths were kept as small as possible

consistent with the requ'ired sensitivity, the magnitude of these
relative deflections was generally in the neighbourhood of several
microns. At the end of each gauge length of the meter, an inductive
displacement transducer of the type l,{IE (Messrs. Hottinger Messtechnik,
Drarmstat) was instalted to measure these minute relative deflections.

of a probe connected
The transducer, type hllE, consisted

to a ferrite core wfrich could be moved axially in the bore of a coil

assembly; placed in the cy1Índrical transducer housing. The coil
assembly was made of two coils which were adjacent to one another on

a common axis. These coils were electrica'l1y connected in series

to form two branches of a l¡lheastone Bridge circuit. The t{heastone

Bridge could be balanced by moving the femite core, so that its

center coincjded with the mid po'int of the transducer. Any axial
displacement from this position resulted jn a change of inductance
in the two coils, causing a disbalance in the previously balanced
 -31

Wheastone brr'dge, The signal that would be ayatlable as a resuit

of this disÞalance, after amp'lificatÌono could be read on the meter
of a carrier frequency anplr'fier, for tht's purpose an anplîfier of
the type Kt¡lSlT-5 fMessrs. Flottinger Messtechnf ko Darmstadt) was used

in thls work. The naxlmun sensÍtivity of the meter KhtS/T=S is

2 X 10-8 metre.
To connect sinuJtaneously the three transducers used jn
the curvature meter to measurl'ng brÌdge KhlS/T-s, a switch box was
constructed. This was capable of connectÍng one transducer directly
to meter KI'JS/T-S for one swi'tch posi'tlon, Then for" the other swjtch
position it connected a frequency mîxîng apparatus S A II (Messrs.

Hottinger lt4esstechnick, Dannstat) to the meter KhlS/T-s. The other

two transducers were connected drirectly to the ìnput of the m'ixing
apparatus S A II (same manufacturer as the above equipment), The

mixing apparatus S A II is simply a mixer, which has to be used always

in conjunùtûionwri¡tlbh KblS/T-s. In using the mixer S A II two seperate
electric sîgnals could be sunmed or subtracted usjng the appropriate
sign settings. The incoming signals could be mixed to different pro-
portions using the two proportionality settings.
In'this experiment the mixer S A II was used to feed only
one tránsducer lmpulse at a timeo tntolthe measuring bri.dge KhlS/T-S.
Thlìs was achieyed by settîng oneof the proportÍonalty knobs to zero
whîle the other was set to read 100 percent. Hence, with the aid of
the switch box and the mixer S A II, it was possíble to read the
 -_I.*.,,

SWITCH
I
(,
N)
I

TRANSDUCERS
(1,ì^ ¿ã1-r'

sA ll- KWS/ T-5

FÍgure 3.10 The Schematic Diasram of the Electroníc Equipment Ärrangement.

 -33-

Probe deflections of all the three transducers on the ïreasuring bridge

Kt{S/T-s one at a tÍrne, wltthout disturblng the meter, the leads or the
connections. A schemati-c di.agram of this arrangement is shown in
Figure 3.10 on page 32.

3.5 Edge Cúrvature Meter

As nentioned prevt'ous'ly, the curvature meter dl'scussed above

could not be used to measure all the three curvatures occuring on

or close to a boundary, especîal1y at a free edge, A specÍal fixing

clamp shown i- Figure 3.1'l on page 23 was designed to carry the
curvature meter and to support it on an edge, such that the tîp of
the center 1eg of the meter was riqht over the reference point [Figure
3.'|J, 34). The conical bottom of the cylindrìca'l hole in the
page

clamp was made to coincide with the tip of the center ìeg when jnserted.

In machining the c1amp, care was taken to make sure that the points
A, B and C were solinear.
The tightness of clarnping was found to have an effect on

the results. This effect was found to be more significant on the reading
fo+4s. than on the other two, f, and f*. This was because the magnitude
of fg..45" u/as very to the other two. The resulting in-
1ow compared

accuracies caused this type of edge meter to be discanded. Instead

the curvature K, in the o directjon lvas measured at the free edges using
the ordinary currature meter as shown jn Figure 4.26 (page 38), The
other two curratures K* and K*, were obtained from the measured K, value
usring the plate theory for free edge boundary condi:tîons.
 - 34-

CURVATURE METER
FIXING SCREW HOLES

Figure 3.1 | F ixing Clomp

 -35-

CHAPTER IV

MODEL AND TEST EOUIPI4ENT

4.1 Construction of Models

Out of the few model materials that have already been shown

suitable for simulating the flexural behaviour of reinforced and

prestressed concrete structures (5' 9, 21, 27' 29), air-craft alum-
inium a11oy (ALCAN 2024-T-3) was chosen to construct the models.
Its high moduìus of elastÍcity (E = .|0.38 x lO6 psi) enabled the
achievement of adequate accuracy of measurements, with the equipment
available, without requ'iring deformations so larse as to alter the
fundamental geometry of the model slab. it was ensured that
Hence

the small deflection theory of plates remained valjd. In addition,

its nature for easy fabrication to the required curved form was also
considered to be important. Poisson's ratio (u) was 0,322 for the
aluminÍum alloy used and therefore higher than that for concrete.
The effect of the difference in Poisson's ratios and the method of
correcting the model results for varying Poisson's ratio, are dis-
cussed in Chapter VII.
The thickness of aluminium plate used was 0.1?5". The thick-
ness chosen was such that the plate could be considered thin compared

to the dimensions on plane. However this, at the zones of application

of the'load, where the "load spread" takes p1ace, is not true. Thin-
 -36-

ness of the model assisted in correlatinq the model results to those

of actual structure around these zones as discussed in Chapter VII.
It also facilitated the measuring procedure because measurable deflec-
tions could be achieved with small values of applied load.
The layout of the models is shown in Figure 4.1, on page 37.
As mentioned earlier, the model dimensÍons were chosen to represent
curved slab bridges popular ín practice. Along each radial boundar.y,
a set of holes of diameter 0.04" were drilled through the model pìate
to pass the strings to hang the "pre-'loads" that will be discussed
later in this chapter.
The reference points investigated are indicated in
Fígure 4.'l on paoe 37. The reference points at the free boundaries

were located as close as possible to the free edge. It was possible

to locate all the "edge" po'ints at distances less than one half the
thickness of the plate from the edges. Around each reference point,
three 90'o conical shaped holes were punched (Figure 4.bon page 3-8 )

to position the meter in place. As mentioned earlier, great care

was taken to ensure that the distances from the center 1eg hole (i.e.
the reference po'int) to the two outer 1eg holes were equa'l to one

half the gauge length of the meter. For each meter portion, a hole
of diameter 0.04" was drilled through the model plate at the appro-
priate place to mount the device as explained earlier. The loading

schedule used is shown în Figure 4.laon page 38.

4.2 The Pre-Loading

Due to the fact that the self weiqht of the aluminium plate
to\
or
Model 'l - '15" Openino Anql e.

l4odel 2 - 45" Opening Anqle.

c
s H
he Conti nuous ModeJ 0l = 0?. = ?.2,5"

Figure 4.1 Layout of Models.

 l4odel I and l4odel

at ìt

. The Continuous Model.

Fiqure 4.2a Loadinq Schedule.

I

Figure 4.2b Location of Punch Marks.

 -39-

r,'las not great enough t0 ensure fir¡n contact between the modeJ Pl ate

and the kntfe edge supports, sorne pressure was necessary to hold the
model plate down, Thi.s was achi.eyed by'hangìng peights fron strtngs
which were anchored to the upper surface of the rnodel through tiny
cones sîtttng at the edEeo as shown tn Flgure 4.3 on page 40. In
addition these pre-loads prevented any possible uplifts of the corners
of thlmode'l due to loading. The questîon may arîse as to whether
these loads would affect the results because they cause some initial
curvature in thermodel. Howeyero this curvature was very sma'll and its
effect was elÌminated by electronîcal'ly I'zeroingrr the whole system after
all the pre-1oads were applied. This princlple rls explained in Chapter V.

4.3 The Knife Edges and the Testing Arrangement

The knife edges were made of solid steel. As shown in
Figure 4.4 on page 40, slots were cut along the sharp edge of each

knife edge, A set of holes of d'iameter 0.04r' were dril'led through

them, to pass the strings carrying the pre-loads. These slots were
cut to avoid the difficultÌes in alîgnÍng the holes through the sharp
tips of the knife edges. The s'lots were machined to widths of
dimensions less than one half the thickness of the nlate to ensure

that the loca'llìzed stresses formed in the p'late aroung the slots due
to suspended pre-1oads were insli'gnificant.

The test of two large steel channels champed

bed consjsted

across two leyeled tables. Two double angles made of stngle ang'les
fixed back to back were clamped across these two channels, In each
 Figure 4.3 The Pre-Loadinq.

lott

T ÏJi i¡/VIJ
iiiiiiiiiiiiiiiiii
: ll

rl
l¡
1!
iii¡
i! l
tl
h
rl iiiiiiitiiii
1t'

Fiqure 4.4 Knife Edoe Support.

 -42-

of them, a gap was provided between the two single angles in order
to pass the strjngs camying the pre-loads. The two knife edges
were fixed to the double angles. This test set up js shown jn Fiqure

4.5, page 4'l .

4.4 Loadinq Mechanism

A vert'ical steel shaft with a ball transfer at the bottom

was employed to load the model plate. This loadjng shaft was guided
through two teflon bushings placed insjde a cylindrical housing. As
shown jn Figure 4.6 on page 43 the cyfindrical hous'ing was machjned

as a part of a carriage that was desiqned to slide on a horizontal

beam. The ends of this beam were fixed on to two vert'ical posts hav'inq
magnet'ic bases. Another vertical frame v¡as f ixed on top of the
previously mentioned horizontal beam. In the horizontal beam of the
top frame, a number of holes were bored at pre-determined spacings to
fix an induct'ive displacement transducer of the type TDCDT-100, Hewlett
Packard. The purpose of this transducer was to obtain the deflection
directly under the load. Acircular steel disc was fixed to the top
of thelloading shaft on which the loads were placed. Solid steel
u,ashers with a'c'ircular hole in the center to fit the shaft were used

as weights. The loading shaft could be held at any vertical pos'ition

using two long vertical Screws p'laced against the bottom surface of
the steel disc (F'igure 4.6). Hence the ball transfer of the loading
shaft could be made to be just in contact with the plate before loading.
To obtain the deflection djrectly under the load, the bottom end of
Fiqure 4.6 Loadinc Laechanísm.

Fiqure 4.7 Cantilever System.

 -44-

the transducer probe was kept on the loading shaft after it was adjusted

to the above mentioned position. This transducer was connected to the

Hewlett Packard Data Acquisition System machine to find the corres-
ponding deflection.

As mentioned previous'ly, at points close to the reference

points,it was required to load the model from underneath. A cantilever
system similar to that of a simp'le balance was employed to achieve
this purpose. As shown in Figure 4.7 on page 43, a vertical steel
needle with a ball transfer at the end was employed to load the model
from underneath. This needle was guided through a steel tube fixed
to an end of the horizontal balance beam. The needle could be adjusted
vertically to ensure that the horizontal beam remained horizontal
during loading and the deflecting of the plate. The purpose of this
was to avoid the load being applied at an inclination to the vertical.
To minimize the possible effect due to the horizontal frictional force
that could exist between the ball and the model p1ate, the lenqth
D of the needle (Figure 4.7) was kept as short as possible. A tiny
bit of grease on the ball transfer also helped to reduce the above
mentioned friction to a negìigib'le amount. When loading from below,

instead of turning the knife edges to the top of the model, a technique
of preloading the model was used and the procedures jnvolved are given
in Chapter V, page 55.
 -45-

CHAPTEB Y

EXFERIMENTAL PROCEDURE

5.'l Calibratfon of Curvature Meter

Initlally, the performance of the curyaturerneter discussed
în chapter III was tested on a plate that was subjected to pure
bending. As shown in Figure 5,lo pâgê 46 a 6il x 6" square plate
made of alunl'nium al1oy (RlCnn 2024 T-3) was supported at the oppo-
site corners A and c fron underneath and at the corner B from the top,
The curvature meter v\,as attached to the plate such that the two main
axes of the meter were a'ligned aiong the two diagonals of the p'late.
Then the plate was loaded at the corner D. As expectedn the two

transducers on the mai.n axes registered readings of equal magnitudes

but of opposite signs indicatrlng antic'lastic curvatures along AC and
BD. The third transducer registered a zero reading indicating that
the ptate remained undeflected along the lînes x and y (Figure 5.1,
page 46) as expected in the case of pure bending. In addition the
twr'sting curvature calculated from Equation tcJ on page 24 using
the three transducer readings was zeroì whîch again satisfied the
criterion of pure bending. Further, this confîrmed the validity
of Equation IDI that was developed in chapter III. This experiment
was repeated for tWo of three loadi.ng cases. For example the trans-
ducer readings obtaÍned for a particular load were +0.00lggn -0.0020
and 0.0 for the two main axis transducers and the third transducer
Figure 5..| Plate Sub.jected to Pure Bendinq.
 -47 -

!"espectiyely" This experi.ment v{as repeated with the devtce rotated

through '|80' to be certatn of the consistenÇy of the curyature meter.
The same readlngs were obtaîned wtth the signs reversed. Then the
curvature meter was rotated through 45" such that the main axes of the
meter Wereat 45. to the dlagonals of the p1ate, 1.e., in X and y
directions. As anticlpated, the transducers on the main axes registered
zero readfnEs whrìle thethird transducer regr'stered a non-zero reading.
The curvature along the diagonal of the p'late was calculated using Equa-

tìon ICI developed for Ke+4s. in chapter III, page 24. This value was
found to be equal to the curvature obtained for the same diagonal for
earlier meter set up using ühe more straîghtforward Equation iAl. Hence,
the validity of Equat'ion lClwas confirmed agaîn. The curvature meter was
rotated through .|80" from this posit'ion. The same readings were obtained
again with the sign re:versed. Then the curvature meter was adjusted
such that the main axes were l5o from the diagonals, All the three
transducers registered non zero readings, From these readrlngs curvatures
along the dîagonals were ca'lculated using the equilibrìum for curyatures
in a plate (Mohr's c'irc1e). They were equal to those obtained
With earlier set ups. Unfortunately, the above described r;experi-
fi¡cnt could not be considered as a genera'l proiôf to show the

validity of Equation ICI developed for 0+45o axis of the meter. This
was due to the fact that in all the three meter set ups the neter

axis e+45o remained level as in the simple case considered in Figure

3.2 on page 22. Hence the term (t, + f*) in Equation ICI¿a!,Wpys
 *48-

turned out to be zero unlltke În the general case shor¡ln i:n Ftgure 3,3,
page 22, 0n the other hand, this experinent dÌd not produce dat¿
inconsjstent with the equattons deye]oped for the meten axis o+45o in
Chapter III.
As mentioned earlier, the curvature meter was calibrated
along each axis using a straìn gauge testjng beam apparatus, (H.
Tinsley and Co, Ltd,). this apparatus, a circular curvature
With
could be obtained ìn the center portîon of the beam by fixing the
two adjustable roller supports A and B equî-distant from the respec-
tÌve screw jacks R and L provided at the ends of the beam, (Figure
5.2a page 49). This cîrcular curvature cou'ld be calculated using
the relative deflection neasurement obrained fron a mîcrometer nounted
at a center of a special carriage. This carriage had been designed
to work on the same principle as the curvature neter. The gauge
'length between the two fixed legs at the ends of the carriage was
much larger than the gauge length of the curvature meter. The rela-
tive deflection of the micrometero therefore, were
measurements

greater in magnitude than those of the curvature neter transducers.

Hence, the micrometer sensít'ivity of l0-5 metre was quite adequate
for calibration of the currature meter. For accurate measurements of
deflections the micrometer shouJd not exert any pressure on the beam

while recording the readings. For thÍs purpose the carriage u,as pro-

vided wîth two termi'nals which were connected to an electric buzzer actuated
 (b)

Figure 5" 2 Cal Íbration Set up.

 -50-

by a high frequency current. Thus, the slightest contact between the

micrometer and the test beam would at once be indicated bv the buzzer
sound.

The calibration set-up js shown ín Figure 5.2b, page.49.

A narrow strip of alumìnium of lenqth 48", of width 2" and of thick-
ness 0.125u was used as the testing beam instead of the Bery'l1ium

Copper beam supplied urith the apparatus. This strip was cut from
the same sheet of aluminium that was used to make the models in
order to attain high level of precision. The micrometer carriaqe
v{as placed on the center portion of the beam such that the two

fixed'legs were on the center line of the beam. The curvature meter

was mounted on the at a predetermined point between the legs of

beam

the carriage withits axjs to be calibrated lyino along the center

line of the beam. More accurate displacement measurements on the
measuring bridge could be obtained if the transducer probe was c'lose
to its at the unloaded state. This was
mechanical zero position
obtained by moving the transducer housin,o vertically until the measur-
ing bridge meter needle was close to zero while all the electrical
settings were at middle positjons. The meter needle was brought to
exact zero positÍon electrica'lly and the measuring bridge was then
balanced for phase and amplitude. The measuring amplifier K.hl. S/T-5

used could be set to various measuring range sensitivites such that

even a fractiona'l part of the nominal I mm,i range of disp'lacement of

the transducer probe indicated a full deflection on the recordinq

 - 5'l

meter. The measuring bridge meter was then calibrated such that the
needle indicated a full deflection in the meter urhen the maximum
displacement comesponding to the selected sensitivity range was

sjmulated electrically. The micrometer reading was recorded at this

unloaded state and then the beam was deflected by moving the two

screw jacks downwards. The micrometer readings and the measuring

bridge meter readjngs \ivere recorded at various stages of loading.

For each sensitivity range selected these measurements were carried

out for two continuous cycles of beam deflection obtained by load-

ing, unloading, reloading and re-unloading. This was to obtain an

idea of the significance of the hysteresis of the total system

and to check whether the meter was capable of reproducing the results

faithfully. The data that were ob,tained from these calibration tests
are discussed jn the next chapter.
The question arose as to whether the calibration obtained

for a particular ini.tial balance position of the probe close to its

mechanical zero position would be valid if the initial balance posi-
tion of the probe during testing was not the same, although in both
cases the measuring bridge cou'ld be set for "initial zero" using
its electronic sett'ings. Thereforen the calibration procedure was

repeated for a number of off set posítÍons of the probe on both sides
of its mechanical zero position. For each calibratíon, the pos'itions
of the appropriate "coarse" electronic setting and the "fine" elec-
tronic setting of KWS/T-5 (or S A II) were recorded. However, it
 -52-

not possi.ble to record the positi.on of the+Textra f.tneil settÍng,

tnlas

A]l three meter axes were callbrated tn three sensitiytrty ranges

that showed deflectlons of 0.01 mlo 0.005 and 0,002 nm at full
range, In calibratÍng the o+45a axls of the meter, aTl the three
transducer readrìngs hrere recorded and used in Equatfon ICI [page 24).
The three transducers used on the three curyature meter axes rilere
marked with different numbers l'n order to nake sure that ln testing
they were used on the sane axes on whtch they were callbrated.

5.2 Procedure for gbtaining Curvature Eata

The curvature meter was mounted on the model at the point
under investigation such that the middle leg of the meter coincided
with the reference poi'nt. The three transducer housr'ngsif,üÏien',.\¡,rene

locked onto the curvature meter such that the probes were close
to their mechanical "zero" positions while the measuring brídge
electrical settings were at their middle posÍtions. The purpose of
this adjustment was two-fold. Fjrst, as mentionddeêarrl{¡êu.,the sen_
sitivity of the displacement transducer r^las at its peak when its
probe was at this middle position. secondly, this adjustment gave a

one millimetpe range above and below the I'zero" positiono thus allowing
the probe to record both positive curyature and negative curyature,
(Figure 5.3, page 53). The measurd"rtrEg, bridge meter :needle v\,as brought
to the zero positt'on for circuit, The neasuring bridge
each transducer
was then balanced for phase and ampli'tude for each

transducer cl'rcuit, l¡lhen the brfdge was balancedo it

 -53-

Positive Curvature

Negative Curvature

Fiqure 5.3
 -54-

was assumed to have "zeroed" the initial conditions of the model.

From then on, the meter registered the deflections corresponding to
change in curvature due to any additjonal load aop'lied to the slab.
Therefore, the earljer mentioned initjal curvatures produced by the
pre-loading and the weiqht of the curvature device did not have any

bearing on the curvature recorded due to the load. The measuring

bridge was calibrated internal'ly as described earlier in all the three

selected sensitivity ranges. The model was then loaded at the required
place (the "load point") and the three transducer readings vuere re-
corded. This was repeated three times, startincl with a different
transducer every time to ensure that the readÍnqs were not'influenced
by factors such as the transducer circuit transfer suritch box, connect-
ing cables, etc. After the first model was complete'ly tested this was

carried out on'ly as an r;occasional check. Every time urhen the load
was lifüdd it was checked to see whether the measurjnq bridge remained

balanced for all the three circuits. The reading of the transducer
type TDCDT-100 (HEIIILETT PACKARD) was also recorded to make sure that
the selected load did not produce deflection large enouqh to violate
the small deflection theory for elastic plates,
In the case of first model, the symmetry of the measured
values for the three reference points on the center radial line urere
checked by loading the model at load points on both sides of this

symmetric radial l'ine. For the other two models thjs symmetry check

was done occasiona'lly. In addition, gn the first model an extra

 ,... .:.: .: ' :: .::

55-

synflmetrlìc reference poÍnt [Figure 4.1 , page 37) was exanined

to check the synmetry.

In order to 'load the nodel fron undenneatho ît is usual
to change the knife edge set-up, Instead of thls complicated proce-
dure the fact that the curvature meter rneasures only the change in
curyature was used, First, the model was loaded from above at a
number of points close to both supports with reasonablv heavy Joads.
This was to ensure that the plate trras pressed agalnst the supports
when it was loaded from underneath. Then the meter vvas zeroed for
initial conditions and the load was applfed from below usíng the
cantilever system descrîbed in Chapter IV, The same experimental
procedure described in the prevÍous paragraph was then followed.
5.3 Material Properties and Calibration cif Loading Systems

AÊ,,tension test was carried out to obtain the modulus of

elasticity and the poisson's ratio of the model material, The load-
ing mechanism used to load from above was calibrated using a preci-
sion balance capable of readÌng down to the nearest l/100th of a gm.
It was ascertained that even the small friction that existed
between the shaft and the teflon bushing could be rendered negligible
by tapping lightly on the shaft after loading. This procedure was

followed at all times.

 -56-

CHAPTER VI

OBSERVATIONS FROM CALIBRATION AND SYI4METRY TESTS

6.1 Calibration Constants for the Three Transducers

The circular curvature produced by the strain gauge testing

beam apparatus was measured in in the previous

two ways as described
chapter. Then they were compared with each other ín order to find a
calibration constant "Cu" (or C* or Cr*45.). Instead of calculating
a number of "CU" 's at different levels of Toading and taking the
average value, it was decided to plot the curvature measured by the

testing apparåtus (using Equation 2.4 on page'lB ) vs the curvature

measured by the meter (using Equat'ion [A] on page 24). In each case
the slope of the best straight line through these calibration po'ints
was taken as the constant "Cu" (or CO or C6a45o). The Calcomp line
plotter computÉù progranüne "AVLIN" that is available at the computer

center of Universjty of Manitoba was used to plot the calibration

data. The slopes of these straight lines were obtained as out puts
of the programme. Excellent linearity and reproducibility of the
measured values indicated the ftrigh degree of reliabifity of the measur-

ing brÍdge apparatus and the transducers. l,{hen the data for loading,
unloading, re-loading and re-unloading were plotted on the same sheet,
it was obvious that the hysteresies in the system was insignificant.
From these calibration graphs (Figure 6.1, page 57) the reproducible
nature of the curvature meter was very evident.
 o
ol
ó
tl
I
I
xo
=P
trl
ct
(9

=o
f-1\
frö
t-
c¡

ÊE
E,O
f
v,
LOADING
ul
-Â UNLOADING
= o
-o
---'--æ RE. LOADIN G
H3
Ð ætr RE - UNLOADING
k
É,
3ç
o

o
(\¡
ó

o.oo o.o2 0.o4 0.06 0.o8 oto

CURVATURE MEASURED BY METER X IO-I

TYPICAL CALIBRATI ON GRAPH SHOWING

INSIGNIFICANT HYSTERESIS
Fi g. 6'l
 :":,:..:.:,. î:..:r-r:.:.r .t.::-:- ;.,:r._r 'a : ,,: il::'J.:;1.: .i:tl::_,tt..::':..-ì j.. ,:j..:1.-;.

-58-

Each meter axis was calibrated in three sensitivity ranges.

There was a s'light difference Ín "C" values between the three sensi-
tivity of the measuring bridge for all the three curvature
ranges
meter axes. The effect of offsetting the transducer probe from its
mechanical zero position was found to be insignificant on the cali-
bration constants CU and C0*45o in all the three sens'itivity ranges.
The constant CO was affected by offsetting the transducer probe and
this was found to be very significant ín the 0.002 mm (tne f¡ighest)
sensitivity range. In the cases of 0.0.| mm and 0.005 mm sensitivity
ranges this effect was found to be significant onTy when the probe
was offset considerably from the mechanical zero posÍtion. However,

this did not have any bearÍng on the testÍng since the positions of
electronic balance settings of KhIS/T-S (.or S A II) were recorded during
each cal ibration.

A different procedure that was used in plottinq the cali-

bratíon curve for the 0+45" axis of the meter because of its compli-
eated expressíon for curvature is given below.

The circular curvature calculated from the testing beam

Qtr
apparatus = Ë\T where L, is the gauge length of the carriage meter.
\ LT/
The same curvature calculated on 0+45o axis of the meter
fo+45" - å (fe * fR) ]. f, f* can be
#r, I
= The readings and
corrected sínce C, and CO are known.

q 2 - IãL
8F 16 r fo+45o 3
I
(cofe + c*fp) 1
 ,59-

re+4bo = å,h,' * $ (core + c*rp)

Ler F'r =
å,h,,*$(coro+c*rp)
Hence F* vs f0*45o was plotted to find Co*45o. Co*45o ít
really a calibratÍon for the third transducer alone. Hence, in correct-
ing the curvature measured on the e+45o axis, a'l'l three calibration
constants Ce, CR and Cg+45o wêI"ê used ín the Equat'ion [C].
The calibration constants obtaÍned are qíven below"

Ful I Deflection
of the SensÍtivity c^
U
cR ce+4s"
Range.

0.01 mm 1.002 0.992 0. 995

0.005 mm t.0ll 0.998 0.991

0.002 rm 1.008 0.999 0.987

6,2 Chggks .on ,Experimentat. .Rqq.t{J tS

As mentioned in the previous chapter in the case of the

fírst model tested, the model s'lab was loaded at loadino grid poínts
on both s'ides of the synrmetric radíal line ín studying the three
reference points A, B and C. Since these reference points were on

the symmetric radial 'lÍne, this provided a comprehensive check for

syrmetry. For all the three reference points A, B and C, the trans-
ducer values fo U
to be remarkably close for symmetric load
were found
points [e.g. , g, 9'lh, h' / etc. (Figurne 4.ä, page 38 )]. The maximum
 -60=

difference obtatned was.muc[ less than one tralf of the snallest read-
able diy'i'sr'on [0.0002 m m) of the 0,0] .n.n scale used. At the reference
point A, the registered transducer values of f* were snall compared to the
orthogonal axr's values fr, Some discrepancy could be notecl for these
small values of f, for sJ¡.r.metric loadfng. l-lowevero on'ly in four
instances these dîfferences s,ere more than the smallest readable divi-

} of the 0,005 m m scal e used. The oby jous sol ution

cil"sn¡{0n00il" iiì}ûrì

appeared to be the use':of hi'gher loads so as to obtar'n higher readings

for f^K that could be read on a less sensitÌye scale than the one al-
ready used. However, ìt was not possible to use hÍgher loads sjnce
the small deflectîon theory of plates would haye been violated. The

alternate solution, for the above probìem are discussed in Chapter IX,
The third transducer readings f6+45o obtained at reference point A
were entíre1y different from each other for symmetric load points.

However, when the twisting curvatures KeR at A were calculated using

fo, fR u¡d f9+4b" in Equatjon ICI for syrmetric load poìnts, remark-
ab'ly close values v,,ere obtaÍned with opposite signs. This was an
excelient demonstration of the highly precise performance of the
curvature meter. It a'lso proved that synmetric boundary conditions
had been obtaíned from the model'ling technique emp'loyed. These

transducer readings are listed in Table 6..] on page 61. As expected

zero values were obtained for Kr, at reference point A when the model

was loaded at points on the s¡rnrnetric radial lÍne, All the three
transducer readi.ngs obtalned for reference polnt D for all the load
 _61

Table 6.1 Some of the Symmetric Load Points for the Reference Point A.

Load Points Transducer Readings

U R: 0+t45o

a 0. 00705 -0.00296 0. 001 52

b 0. 00850 +0.0a221 0.00236
c 0. 00576 -0. 00090 0. 00250
d 0. 00958 -0.00232 0.00272
e 0. 00740 -0.00300 0.001 65
f 0. 00600 -0.00225 0.001 I 0
f' 0.00603 -0.00225 0. 001 73
g 0.00683 -0. 00'| 64 0. 001 65
g' 0.00683 -0.001 75 0.00221
h 0.00755 -0. 001 05 0. 00239
h' 0.00755 -0.001 05 0. 00249
'l 0. 00760 -0. 001 78 0. 00235
i' 0.00762 -0.001 81 0. 001 98
j 0.00660 -0. 00240 0.00.l 7g
j' 0. 00664 -0.00232 0. 00.| 40
k 0.00603 -0.00208 0. 001 02
k' 0.00609 -0.002ï 6 0.001 94
I 0.00558 -0. 001 45 0. 00] 23
ll 0. 00560 -0.00.| 44 0.001 88
m 0. 00536 -0.00t 11 0.001 38
m' 0.00536 -0.00.l 00 0.001 83
n 0.00532 -0. 001 29 0.001 89
nt 0.00534 -0.001 30 0.001 01

o 0. 00535 -0.001 79 0. 001 g8

ot 0. 00542 -0. 001 80 0. 00081
 -62-

Some of the Readings at Symmetric Reference Points D and D' *

Transducer Readings at the Transducer ReadÍngs at the

Load Reference Point D Reference Point D'
Poi nts â g+4S"
R 0+450 0 R

a 0. 00403 -0. 001 42 0. 001 42 0.00401 -0.001 38 n 001 41

b 0.00467 -0.00122 0. 001 52 0. 00464 -0.00122 0 00't54

c 0.00540 -0. 001 I 5 0.001 44 0. 00538 -0.001 I 6 0 00142
d 0. 00534 -0. 001 28 0.001 04 0. 00532 -0.00] 28 0 001 04
e 0.00453 -0.001 50 0. 00054 0.00452 -0. 001 50 0 00052
f 0. 0051 4 -0.00200 0. 001 55 0. 00295 -0.00093 0 00il8
f' o. 002e5 -0.00098 0. 001 20 0. 0051 I -0. 00'| 96 0 001 58
g 0. 00585 -0.001 36 0. 001 95 0.00342 -0.00095 0 001 20
g' 0.00341 -0. 00096 0.001 20 0. 0057 9 -0.00'| 34 0 00200
h 0. 00660 -0.00070 0. 00205 0.00400 -0.001 00 0 001.l9
h' 0. 00400 -0.001 00 0.001 20 0. 00653 -0. 00070 0 0021 0
'l 0.00648 0. 00t 53
-0.001 40 0. 00328 -0.00088 0 00063
i' 0.00322 -0.00085 0. 00068 0.00645 -0. 001 40 0 001 55
j 0. 00564 -0.00202 0. 00076 0. 00330 -0.001 03 0 00041
j' 0. 00330 -0. 001 03 0. 00042 0. 00564 -0. 00202 0 00078

At the reference point D' the curvature meter is located with a 90o
rotation (clockwíse) from that at D. The load.ing points a, b, c, d
and e are loeated on the symmetrica'l radial line. Hence for the
above mentioned loading points identical transducer read'ings are
obtained for the reference points D and D'as shown above.
 -63-

points were in good agreement with the corresponding values obtained

for the symmetric reference point D' as shown in Table 6.I on page 6l .
In the other two models tested, only occassional load sym-
metry tests were performed.
 CHAPTER VII

PRESENTATION OF TEST RESULTS

7.1 Dimensionless Curvature Influence Coefficíents

The moments M* M* and M*U could be obtained from Equation
2.1 , 2.2 and 2.3 on page 9 using the curvatures calculated from
test results. Influence surfaces presented in terms of these moments
would not be of much direct use to the designer since they were cal-
culated for a certain load P and for a plate material of certain
Poisson's ratÍo p. Thereforeo it was decided to present the results
in the form of non-dimensional curvature influence surfaces as shown

below. This method would help in attaining a better accuracy in

correlating the model results to prototype slabs made of materials
of different Poisson's ratios. The method of obtainíng this correc-
tion for Poisson's ratio is given below. Howevero some discrepancies

would still be present and they are discussed later in thjs chapter.
I0, I* and Ir* are the non-dl'mensional influence coefficients corres-
ponding to the curvatures Kr, K* and KU* where
I^e = -Þ--
DKo
(7.1)

I,r
KP=
DKR
(7.2)

IoR = (l - u)D Ko,R

(7.3)
 -65-

These curvature ínfluence coefficients can then be used to find the

moments of prototype slabs of the same shape but of different Poisson's
ratios, in the following vJay. If the Poisson's ratio of the proto-
type material 's up and the concentra.ted load ir Pp, then:

M3 = Pp (rË * uo rf;) (7.4)

M[ = Ppftil*uor[1 (7.5)

rË* = pp ,rËn (7.6)

where I[, Ifr and IIO are the mode] curvature coefficients. This can
be shown in the follor^ring uray:

Prototype curvature/unjt load = B#.ffi

where Dm = model stiffness

Km = model curvature for a load Pm

Pm = model load
Op = prototype stiffness
Therfore, for a prototype load PO

''p-ry.Ë#=ËËrwr
v-

But Im = Tffi
Pm

Hence Kp = lL
Dp'I*
 -66-

rherefore M[ = Dp (r! + uo rfl)

-= DprBËtä*uo$frrf;t

Pp(rä*uolfl)

Euqations 7.4 and 7.5 are simplified forms of more comp'lex expressions

that have been developed by S. Timoshenko and S. Krtege¡(24) to modìfy

the bending moments in slabs for a change jn Poisson's ratio from u*

to up. The expressions given by Timoshenko and Krieger are:

M3 =
$_ f tt - uruo) r'r[ + (uo - u*) uff
'm
1

ufl =; 1rz
tm
t (l - u*uo) Nfl + (uo - u*) u[ 1

It can readi'ly be shown that the above equations are not valid r'¡hen
applied to the slabs with boundary condÌ:tions such as free edges or
cases of elastic supports--where u is implied in those boundary condi-
tions. For examp'le, at a free edge, the use of the above equations
for conversion of moments from u, to uO Vields an expression
^l
MpR - ur) N[ J ror the transverse moment M*.
t,ç t (uo Cl early

this expression is not zero although the transverse moments at a free

edge must be zero for any Poisson's ratio. However, it has been shown

that, althouqh the use of the above equatÍons leads to some errors
near a free edge or an elastic support, it does not influence the
 -67-

moments elsewhere on the s1ab,

(24' 27' 30, 31, 32).
A detailed
investigation of the effect of change of poisson's ratio is beyond

the scope of this dissertation. However, the writer has carríed out
a survey of literature available on this subj..t (15, 24,27,30,31,32)
and thereforewith the aid of those he offers the following comments.
(31)
RobÍnron in his investjgation of centrally loaded square
slabs found that a reduction of Poísson's ratio from 0.30 to 0.lb
resulted 'in a considerable reduction Ín both spanwise and transverse
moments under the load point. The effect of this on span-wise moment

at the center of a free edge was found to be negligib'le. He ascer-

tained similar results for a reduction from 0.3 to zero. In the
reference (30) it is mentioned that BaTas and Hanusku (32) huu.
observed that for a uniform loading, an increase in poisson's ratio

from zero to 1/3 resulted in an increase of deflectis'n at the center

of a free edge. For a concentrated load near the free edge, an increase
of both deflection ahd spanwise moment was observed. yeginobol i (27)
from his studies on continuous skewed plates commented that the
deflections and moment increased rryith Íncrease in Poisson's ratio.
The above comments illustrate only a few examp'les of the research
that has been done on this subject.
The writer concludes that in correlating the model results

to the prototype, drV difference in Poisson's ratio betureen the model

and the prototype should be considered. sínce the use of EquatÍons
7.4 and 7.5 leads to some errors near a free edge or an elastic support,
 -68-

at those places it appears safer to use the mode'l results directly

in the design without any coversion if the model test has been carried
out on a material of hîgher Poissonrs ratio. than the prototype(2.|).
The reason for the above recommendation is that the model deflections
and moments will be higher than those of a material of lower Poisson's

rati o.

7.2 Evaluation of Influence Surfaces

(a) Uniformly Distributed Load (U.D.L. )
As mentioned earlier an influence surface drawn for a parti-
cular reference point prnesents the influence of a unit load any-
where on the slab at that point. Therefore, the influence of any
load is measured by the product of the magnitude of the load and the
corresponding ordinate of the influence surface. The distributed
load can be considered as closely packed concentrated loads.
If p is the'loading per unit area and i is the influence
ordinate of curvature under the load p, the curvature created at the
reference point by this loading is:
dI = (p i) RdedR (7 .7)
If one replaces arc length Re by T, then small arc length Rde becomes

dT. Therefore, Equation (7.7) becomes

dI = (p i) dT dR

The total curvature due to uniformly distributed load over the whole slab
is given by the summation of each of these curvatures dI.
I=fdI=.f(pi)drde (7. B)
 -68 o-

A (u,v)

erence point A ond lood point P wilh

R ef
coordinotes (urv) ond lgrp¡ respeclively.

Figure 7. lo
 69-

The influence ordinate i depends on four variab'les, nameìy u, V, 0, and R.

The reference point location is denoted by u and v. The load location
is given by R and o as shown in Figure 7.1a (Page 68a). Therefore,
Equation 7.8 becomes

I(u,v)=lI (R,e) i(u,v,o,R) dTdR (7.9)

If one rep'laces the acutal dimensions R and T by dimensionless coordinates
u=.Ïj- o=TR r-T
"=g
¡=I
,l=î
where I is an arbitrary 'length:
then dT = lda dR = ]db
Hence Equati on 7.9 becomes

I (u, v) = l' ,, p (a, b) i (c, do ô, b) da db (7.10)

This evaluation formula shows an important general feature of influence
surfaces. That is they are independent of the absolute p'late dimensions
and that only the shape of the plate is important.
In general, uniformly distributed loads are constant. There-
fore, Equation (7..|0) becomes
I (u, v) = p.l2 t¡ i (c, d, ô, b) da db (7.11 )

The double integra'l in Equation (7.'ll) can be interpreted geometrically

as the volume of the influence surface over the loaded area. This
geometric meaning is the key for the evaluation since, in general, it is

not possible to determine the exact influence function i in order

to perform the double integration in Equation (7.11) mathematically.
l,rlhen the contour plan of an inf]uence surface is given, it
: siirmpfl,e +prdbl.bm to fi nd the i nfl uence vol ume" f rom p1 ane
i'soã "

projected areas of the contours or from the section through the con-
 -70-

tours as shown by A. Pucher (13). These calculations are made in a

manner identical to e4rthwork volume computations, usinc Simpson's

rul e'where appropriate.

Although any I in Equation
arbitrary value could be used for
(l.ll ) in thjs work the width of curved bridqe deck is used for l.
(b) Wheel Loads

In designing highway bridges, it is essential to study the

effect of a truck load urhen Ít travels across the bridge. The magni-

tudeof the curvatures depends on the position of the tuuck on the

bridqe. The critÍcal posítion of the truck is usually determined
by a trial and error procedure. The wheel spacinq is drawn on a piece
of tracing paper to the same scale as the influence diagram and then
it is moved about on the influence surface until the critical location
of the truck is obtained. lrlith a little practice, one can approxi-
mate to this position very closely by inspection, noting that in
general, one wheel load should be on top of, or as close as possible
to the reference point. All the wheel loads can be considered as

concentrated loads except that wheel load,,which is over the reference

point or very close to it. In th'is case the load distrjbution should
be considered, since at or close to the reference point the Ínflu-
ence surface changes rapidly. 0ften a load distribution apg:le, of 45o
/1?\
is used \'-".
In the treatment of other rvheel loads it has been assumed
that they are concentrated loads. But íf the contact area is large,
 -71

consideration of load dÍstribution is necessary, as discussed in

greater detail ín Reference ('13). An example of the use of influence
surfaces in truck load analysis is given ín Reference (¡g) by A. M.
Lansdown.

ïnfluence surfaces are useful to the bridge designer not

on'ly in assessing the effect of abnormally heavy loads on àn* exist-
ing bridge, but also assist in the design of bridges in many ways.

The critical moment locations can be obtaÍned quickìy if a set of

ínfluence surfaces for the structure to be designed are available.
They also become useful in laying out the reÍnforcement for the struc-

ture. ldhen the pertinent influence surfaces for the particular struc-
ture to be designed are not available, the designer can employ a set
of ínfluence surfaces representing a structure that is closely similar
to that under desÍgn. The hypothetical structure so chosen can assÍst

him to amive at the final structure. For example, when a bridge

engineer is called upon to design a brÍdge, the tota'l span and the
approximate width will be the only definite properties of the bridge
that he will at hand. The plane geometry of the bridge of course
have

depends on the traffic requirements at the particu1ar location. If the

influence surfaces are avai'lable for various shapes of structures of,

different span to width ratios, the engineer can select a structure
that appears to be the most suítable for his problem. Then he can
start to build up his particular structure using the critical moments
in the selected influence surface as the primary design moments.
For instance, if the designer has selected a horizontal'ly curved
 -72-

bridge deck, at once he ulill notice from the influence surfaces that
the critical locations are the mid points of the two free edges

instead of the mid span. Hence, it r,rou'ld indicate to him that edge
beams are likely called for. Similarly, in the case of a skew

slab, the influence surfaces will indicate the obtuse corner as the
critical location that should be consídered in detail. From an in-
spection of influence surfaces alone, with a little experience one
may be able to pick a hypothetical structure that will be close to
the final structure.

7.3 Correlation of Model Results to Actual Structure

The gauge length of the curvature meter controls the accur-
acy of the measured values and therefore Índirectly governs the chojce
of size of the model. This stems from the curvature meter's measur-
ing only an average curvature over the base'length L of the meter.
The ratio LlL, L, is the span of the model slab) therefore
(where

becomes a prime factor for the accuracy of the modeï resuï ts. The

assumption of constant radius of curvature within the base length L

is assisted by the selection of smaller ratios of L/Lr. An analy-

tical approximation to the effect of L/1, orìr. curVâtures is presented
in Append ix 1.2. t^leigter and lnlejse (29) ,..ommended that for prac- .-.I.

tÍcal purposes ratios of (L/L¡a) of 1/10 to l/8 is satÍsfactory.

Mehmel and Weir. (8),n.ntioned that the error that could result from
a ratio of 1/6 was less than one percent in the mid span region.
The ratio used jn this investigation was l/18 wjth respect to the
 -73-

center spans of the simply supported model slabs, whereas it was

l/9 with respect to the continous slab. The ratio was very close to
117 in the transverse direction with respect to the width of both

model sl abs:

Any curvature meter does not accurately measure the curva-

ture at or in the close vicinity of a concentrated load where the
curvature of the deformed plate changes rapidly. This Ínaccuracy
in curvature rneasurement is due to íts peak value being rounded
(flattened off) as a result of the curvature meter's measuring onìy
an average curvature over the base length.

Theoretícally the curvature under a perfectly concentrated

load ís a singular function which reaches an infinite value directly
under the loadpoint. In reality, however, perfectly concentrated
loads do not exist. As explained by l^festeguur¿ (2), the load tends
to spread over a definite contact area. For analytical punposes

the "concentrated load" is distr¡i:buted over an even greater area

than the actual contact area. The further distribution is usually
considered by projectÍng it on to the neutral plane alons 45o lines
as shown in Figure 7.1 on page 74. such a spreading of the load
reduces the infÍnite curvature under the load to a finite value.
The greater the distributjon of the "concentrated load", thesless sharp
the at the load locatjon. The contact area
maximum peak becomes

between the ball transfer of the loading shaft and the model slab

is much smaller than that of a real wheel load and the actual slab.
 PROTOTYPE
MODEL SLAB

a\
/\,/1
rl
tt

'f;

J:

\ ii:
\.:

t0
t ¡, I

t
¡e
\Þ ': ':
t,
t; I

E FFECT OF THICKNESS OF SLA B

\\
I
¡
¡

CURVATURE
UNDER LOAD
'a - - ---2

CURVATURE Lw
UNDER WHEEL
LOCAL CURVATURE IN LOCAL CURVATURE IN ACTUAL
M ODEL. STRUCTURE.
Fig. 7. I
 -75-

Hence, the load spread area on the model is less than that on the
actual slab as shown Figure 7.'l on paoe 74. Therefore, the
in
curvature due to a unit load on the model at the 'load point is qreater
than that occuring on the actual slab due to a unit wheel load.
A'lthough the curvature meter measures an average curvature, the

difference between the measured curvature and the practical curva-

ture on the actual slab tends to be neg'ligib]y small for practical
purposes, wíth the model tending to give higher (conservative) va1ue.
 CHAPTER VIII

ANALYSIS OF RESULTS

8.1 of Influence Surfaces

Comparison

In general, the contours of the influence surfaces follow

a definite pattern for the comesponding reference points in all
the three models (15'opening anoìe, 45o opening angle and continuous)
except for the three reference points on the ínterior support of the
continuous model. The spanwise O-axis of the curvature meter was
found to be more sensitive to the location of the load than the
other two axes R and 0+45". Hence, as it can be seen in Appendix II
the contours of the IU Ínfluence surfaces are more closely spaced
than those of I* and IU*, The I, influence coefficients are higher
in magnitude compared to the corresponding IO and IU* coefficients
at all the loading grid points in all the models.
Ïn the transverse curvature influence surfaces IR (Figures
I-2 and l;il4) for the reference points: A and D (Figure 4.1, page 37)
of the single span models, 'large areas of anticlastic curvatures
are noticeable unlike in the case of
transverse influence surfaces
for rectangu'lar slabs and skew slabs as given by Pucher (13) and
/ Â\
Rusch\'/ respectively. This anticlastic curvature occurs because

the slab behaves more like a cantilever in the transverse direction

due to the fact that a considerab'le portion of the horizonta'l1y
 ..-!1..-'. ,..
'1,,1 -.l.t J f.."-,.t'

-71 -

curved slab is unsupported outsìde the chord of the outer circujar

Ë0g". ThÌs cantilevertng effect becomes less for curyed slabs sf
short spans and naryow openiìng angles that are nore stmtlar to the
case of rectangu'lar slabs. The transyerse curyature influence sur-
face Io
K- [Ftgure I-30) for the reference polnt D of the continuous
model shows the dinlnutîon of this cantilever effect. In general,
for each rnodel the maxÌmum ordÌnates of Iu surfaces for the
edge reference points B,CÇn E, F, G, H (F.lgure 4,1, page 37) are
greater than those for the corresponding A and D, the center and the
quarter span reference points respectïve1y. The ratios of the corres-
ponding Ie vajues for the edge reference poÍnts and center reference
^^j,^e^ t^ ^ 'B
þtjiiitÈ (E:ìj: þ , t0 , 'o , t0 ) are considerably greater than
'C 'E 'F
T.
le iÃ-
-e ie rD-
-0 io iD
-0 -ete
those ratios in the case of a rectangu'lar slub (131)*. The above
comparison shows that the free curved edges must be considered jn
detail in analysing and designing curved slabs.

(a) The center Reference Points (A) of Model I (.|5î. opening angle)
and Model 2 (45' opening angle)
(a.,)
- t- Io Influence Surface Figures I-1 and I-i3
u

The genera'l shape of the infJuence contours and the maximum

influence ordinates for bsth cases are practically the same, except
that in model 2 the contours start to open out towards the outer

The rectangular slabs considered for comparltson were of span to

width ratîos of 2 and 3. The models tested had span to width ratio
of 2.7 .
 -78-

edge. This occurs because as the opening ang'le is increased, the

outer edge becomes more flexible and inner edge becomes more stiff
even though both models have the same center span length. The above

statement is clear from the fact that the influence ordinate at the
mid point of the inner edge is reduced from a vaìue of 0.605 for
model I to a value of 0.538 for mode'l 2,while the ordinate at the

mid point of the outer edge is increased from 0.690 for model I to
0.804 for model 2. The influence ordinate at a similar point on a
rectangular slab of 3 to I span/wídth ratio (close to that of the
modelis) is 0.696 (.|3) Although the maximum ordinates are practical'ly
the same, a close comparison reveals that the area covered by each

contour of particular influence ordinate value is increased as the

openíng angle is increased. Therefore, the total Ínfluence volume

under the influence surface fór the reference point A of model 2 is

greater than that of model l. From the contour patterns Ít can be
seen that at regions close to the inner edge the contours of parti-

cular influence values that are obtained for model 2 with a load
close to the mid radial Tine occur for a load further away from
in the case of model 1 . The above shows that
the mid radial I ine
the inner edge of model I of sma'ller opening angle is less stiff
than that of model 2.
the influence surface I-1 and I-13 are compared with
Ì¡lhen
(13)
Pucher's influence chart 19 to. the center of a cantilever
plate strip and with Molkenthin's influence chart 13 (14) for the
 -79-

center of a simply supported rectangular slab, it can be seen cl ear'ly

that the ínner edge becomes stiff and starts to act more líke a

fixed edge.

(uZ) I¡ Influence Surface Figures I-2 and I-.l4

of the contours for both cases appear to be

The shape

similar to each other and to those that have been published for the

rectangular slabs
(t3). One striking difference between these in-
fluence surfaces and those for the rectangular slabs is that a sma'ller
area is covered by the positive contours of these influence surfaces
for the curved slabs compared to the other surfaces. The reason
for the presence of more negative contours is that in the transverse
direction the slab acts more like a cantilever as a result of the
middle strip of the curved stab being stiffer than that of a rec-
tangular slab. This extra stiffness is due to the supports being
aligned in a'radìal direction instead of beinq paralle.l to each
other and also due to the reason mentioned earlier in this chapter.
To illustrate the above statement, an example of a plate strip having
non-parallel supports is considered.
 -80-

From the above shown moment vectors, it can be seen that the tw'isting
and the bend'ing momentat the support in the X direction are not
zero unlike in the case of parallel supports. This illustrates
the presence of twistíng due to radial'ly aligned supports.
If we consider a vtrtua&, displacement ô the total energy
in the system can be given by the following equation.

Pô = XEB+XEf

But if there is no twistjng as in the case of paral'le1 supportso the

above equation will have only one üerm (í.e. rEr) on the i,.ight hand
side too. for ,:the above equatÍon to be true, the magni-
Therefore,
tude of P has to be less. The above examp'le illustrates the increase
in stiffness due to radially aligned supports. Pucher's influence
chart l8 \''/ for the transverse curvature at the center of a c'lamped
l'r ?\

plate stríp is useful for illustration of the above mentioned behav-

iour of the curved p'late. This cantílevering action is also assisted
by a dÍstribution of twisting curvature across the slab espec,ia'l.ly
close to the outer edge even though their influence ord'inates are
low. Some evidence of this cantilever effect can be sèen '-1c r
in the influence surface Figures I-2 and I-14.
The anticlastic curvature influence ordinate at the mid

point of the outer edge íncreases as the opening angle is increased.

 -81

(u¡) Ir* Influence Surface Fiqures I-3 and I-15

The influence surface Figures I-3 and I-15 are practically

the same. The low twisting influence coefficients compared to the

two bending influence coefficients at the reference point A indi-

cate that the load carrying capacity at this point relies mainly on
the bending restraint of the slab. However, as mentioned earlier,
a distribution of this twisting curvature can influence the stress
distríbution in the transverse direction. It also indicates that
the principaì curvatures and their directions remain close to those
of I^UKand I,, as the load moves away from the symmetric radial line.
From the two influence surface diagrams I-3 and I-15 it can be seen

that the twisting curvature at the center point of the slab is not
influenced by the opening angle unlike the bending curvatures Io
and I..
K

(b) Reference Points (B) for Model I and Model 2

I o I nf
'l
uence Surface F'i gures I -4 and I -l 6

There is no considerab'le difference between these two

influence surfaces and those that have been published for a similar
point on a rectangular sluÞ (t3' 14). However, when compared with
each other, the stiffeningL.of the inner edge as the opening ang'le

Ís increased can be seen. In the case of model 2 the influence

ordinate starts to change to low values rapid'ly as the load is moved
towards the inner edge unlike in model l. A similar effect can be
 :::-.:'l:r.:. : r;., .j,r,:;t.: ', ¡,,i ¡,r:,ar, rt,.:_:¡:,,¡:r.::,.;::." r:,:,

-82-

observed in the rectangular plates as the span to width ratio is

l1
U ?
v' l4)
decreased

(c) Reference Boints (C) for Model I and Model 2

Iu Influence Surface Figures I-5 and I-.|7

flexibility of the slab at regions close to

The greater

the outer edge with increment of the opening angle is evident from
these influence surfaces (I-5 an¿ I-17) in the same way as described
above but with an opposite action taking p'lace in this case. The

influence surface for model f is simjlar to that of a rectangular

slab. In the case of model 2 the shape of the contours changes at
regions close to the outer edge. This change in the contour patterns
is due to the increased flexibility of the outer edge. Okamura and
(34)
Matsui ontained a simjlar set of contours for the mid point
of an inner edge of a curved slab of 30o opening ang'le.
(d) Reference Points (D) for Model l, Mode'l 2 and the 0.ontinuous Model

(dl) I, Inf'luence Surface Figures I-6, I-18 and I-29

The surfaces I-6, I-.I8 and I-29 appear to be similar to It
influence surfaces discussed in (a.,) above except that they are
not symmetrical due to the unsymmetrical location of the reference
point. In all the three cases maximum ordinates are reduced, of course.
When the influence surface I-29 for the continuous model is compared

with the influence surfaces I-13 and I-18 for model 2, the reduction
in max'imum influence ordinate and the 'larger spacing of contours
 -83-

indicate that a considerable reduction in bending stresses can be

obtained with the aid of an interior support. Some negative contours
of low influence values are present in the adjacent pane'l due to
continuity of the slab across the support.
(dZ) I* Inftuence Surface Figures I-7, I-.|9 and I-30
of anticlastic curvature contours are promÍnent
The areas

in the influence surface Fígures I-7 and I-.l9 for models I and 2
as in the case (ar) above. As mentioned earlier, considerable areas
of positive contours are present in the case of the continuous model
äs in the case of a rectangular slab. The short span length and the
small opening angle permit the curved slab to behave'in a manner
similar to a rectangular slab. This similarity to a rectangular
slab is the reason for the above behaviour.

(dS) Ir* Influence Surface Figures I-8, I-20 and I-3.|

The influence surface Figures ï-8 and I-20 are similar to
each other. The twisting at quarter points also seem
curvatures
to be small compared to lr's at those poínts as ín the case of the
center reference poínt. The genera'l pattern of contours of Fiqure
I-3'| for the contin¡rous model is similar to IeR influence surfaces
I-3 and I-15 for the center poÍnt of model I and 2, although they are
unsymmetrical due to the presence of the midd'le suoport. The contours
on Figures I-8 and I-20 are also unsymmetrical as the reference
points are located away from the symmetric radial line. From Figures
 -:':;:-:,::Ì-ì:-ìi,;

-84-

I-8 and I-20 it can be notíced the amount of negative twisting curva-
ture contours for the reference point D is decreased with the increase
of the opening angle.

(e) R.efsren-ce Points {.E). fo.r M.odeJ_'I.; J4gdel.¿ 3nj thg Continuou.s Mö-qe]

IU Infl uence Figur:es I-9, I-21 .and I-32

The Figures I-9, I-21 and I-32 appear to be similar to those

discussed in (b) above except that the contours are unsymmetrical.
In Figures I-9 and I-21 this again occurs because of unsymmetric

reference point locations. In Figure I-32 the unsyrrmetry is due to

the presence of a míddle support. The support is also the cause

for the negative contours that appear in the adiacent pane1.
(f) Re-feJ-eqc.e.Joinls (F) for All Thfee Mo-{êlg

Iu Influence Surface Figur:es I-10,.1-22.and I=33

The foregoing remarks apply equa'l1y to the comparison of

all the three influence surface contours Figures I-10, I'22 and I-33.
Again as in the case of the earlier discussed influence surface
I-17 for the reference point C of mode'l 2 the pattern of contours
for T-22 also dÍffers from those of I-10 and I-33.
(g) Relgrglcg SojnF (e) fo.r Al.l. T.hree Mods:|.s..
Iu Influence Sut:face FÍqures I-11, T-23,.and I-34
From Figures I-1.l, T-23 and I-34 it can be stated that
the stresses at these corners are significant only when the load Ís
near those corners. Also, it can be noticed that they are not in-
 -85-

fluenced signif,ícant'ly by the other factors such as opening anqle

and number of spans.

(h) Reference Points (H) for All Three Models

I, ïnfluence Figures I-12, I-24 and I-35
The foregoÍng remarks are true for the influence surface
Figures I-12, I-24 and I-35 for the reference point H also.
(i) Reference Points (A), (B) and (C) on the Middlè Support of the
Continuous Model

I, ïnfluence Surface Figures I-25, I-27 and I-28 and Ir* Influence
Surface Figure I-26
The ínfluence surface Figures I-25, I-27, I-28 and I-26
represent the curvatures I, and Ir* at the interior support. As one
would expect, the negative contours are predominant in the case of
Iu's at A, B and C. There is no irregularity in Iu* surface Figure
I-26 presented. The on'ly difference in form between these surfaces
and those that have been published for moments and curvatures at
supports involve a question of slab thickness. The influence sur-
faces presented in this work have been calculated for a slab of real
thickness and hence the contours die down to a zero value at the
reference point. Since the publíshed influence surfaces are obtained
for an infinitely thin slab, the contours merge in to a maximum
value at the reference poínt. This question of infinitely thin slab
in terms of real slab created an interesting problem for Kurata and
 -86-

ct\
\'''.
I
and Hatano The model results were compared with the "exact"
solution results obtained from the computer programme developed a'long
the línes of that of A. Coull and C. Das (36), andi,wé,.ne::,foundi'to be in
good agreement. The differences for IU values at the reference points
A and D were not more than 2.5 per cent. At the edge reference
points B, C, E, F, G and H the differences for I, vaìues were in
the region of 4-5 per cent. ',lnlowever, these differences are accep-
table since the model results were obtained for practical edge

reference points close to the edges whereas the theoretical results

were calculated for theoretical reference points exact'ly on the edges.
The differences for the other curvature values IO and Ir* were in

the range of 3-5 per cent. For a very few readínqs the differences
were as high as 6-8 per cent, but these were ljmited to the cases
where the recorded transducer readÍngs were of small magnitudes.

This consistency between the test and analytical results enhances

the confidence ín both the ana'lytical and the test results.

 ':l;-:i-:.:,:::':,.:l:,..ì::']:,,:.:ì:..|.-.r.:.:::.:.:'.':.''.;:':'.1'.'¡.1.::..,..'','.

-87 -

CHAPTER IX

CONCLUSIONS AND SUGGESTIONS FOR FURTHER RESEARCH

The writer concludes that the material Dresented in this

dissertation provides adequate evidence to show the reliability
and accuracy of the prediction of structural behaviour by means of
model,analysÍs. A simple but precise experimental technique that can

be used effectively in model analysis has been presented. The adapted

method does not require an advanced knowledge of mathematics that is

usually required ín solving fourth order partial-differentia'l plate
equations nor does it require the aid of electronic computers as

in the case of anaìysis. The use of this experimental

numerical
method is not límited to cases of simpte (or simpljfied) boundary

conditions as in the case considered in this work. It is more use-

ful in analysing slabs of any shape, particularly of comp'lex boundary
conditions which defy the accurate analysis by mathematics: 'j I
Versatile but unsophisticated nature of thís experimental method

makes it possible to be used even in a remote field design office.

The simplicity of the apparatus required J.s the main advantage

of this method. However, in using this techni-que one has to remember

the compromises that are required to make between the gauge length
of the curvature metero the span and the thickness of the model and
the magnitude of the joads so as to obtain reliable results.
 j ::';.:::::i:.;::i:" ;r.'.

-88-

validity of the app'lÍcation of model results to proto-

The

type slabs has been discussed. Thellimits and the method of incor-
porating any change in Poisson's ratio between the model material
and the prototype materÍal have also been included and have been

shown to pose no major difficu'lties. The model results have been

nresented in the forms of contour influence surfaces useful as a
permanent record which can be used as aids in bridqe designing as

well as in checking ex'isting curved slab bridges for abnormally

heavy loads, that may cross the bridge from time to t'ime.
Most of the influence surfaces prepared illustrate the

substantially greater flexibility of the longer outer edge compared

to that of the inner edge. It is shown that the inner edge becomes
stiff and starts to act more I ike a fixed edge with the increase of
the opening angle. The presence of large areas of anticlastic
curvatures is in behaviour of these curved
a notable difference
slabs compared to rectangular slabs or skewed slabs. These suggest
that a well distributed torsional action is present in curved slabs
unlike.in the case. of a rectangular slab. However, this prediction
could only be conformed by further detailed studíes at more locations
closer to the outer edge and to the inner edge. In general, the
maximum at the free edge of the slab. The consider-
curvatures occur
able reduction in maximum curvature at the center and the larger
spacing of the contours due to the presence of an interior support
provide some valuable data to the designer. The influence surfaces
 :,: :,. : :.. 1.. 1...:.j ::..:

-89-

at A, B and C for the support moments indicate that there is a con-

siderable difference among these negative moments even for a uni-
formly distributed load over both spans. This unevenly distributed
negative moments at the interior support in turn indicate that the
reactions on the supports are not evenly distributed. This leads
to a bearing problem which has to be studied from the measurement
of support reactions.
A highly precise experimental technique that could be used
to check the reliabjlity of various analytical approaches has been

presented. This experÍment also could be used to find out the limits
of the approximations that has to be incorporated in using these
analytical methods in order to obtain data accurate enough for practi-
cal purposes. For instance in the case of a finite element programme,
the shape and the size of the mesh pattern that is required for an
optimum design could be found very easily and quickly using this
model technique as a basis of comparison.

A study of the influence of number of bearings and the

spacing of them on the bending and torsional curvatures is recommended

for further studies following. It is desirable

along with the to
study further the effect of change of opening ang'le for sing'le span,

two-span and three-span curved bridges. The influence surfaces of

the above mentionednsü{rüG}tüfesiAd:on,g: wiithhehe. otthe,rl a:lféady";publ ished

surfaces will provr'de invaluable design data for the bridge engineer
to have in hand when he is cal'led upon to design bridges. However,
'' 1. . - .,. 4 t . r.,\t. . :, : :. : - ì.r. ì -.: -t..'.t. :...'.: . . : í.:'. :.: ; : i: -: : : ; _: ."1

-90-

the writer recommends the following improvements to the technique

used above.
(1) An automatic influence surface p'lotter should be

emp'loyed to cut down the time required for the ÈedÍous process of
plotting influence surfaces by hand. At the same time jt would help
in selecting better reference points (more critical) during the
experiment itself since the surfaces for a'lready considered points
are i:n hand.

(2) A thick slab and hence a'larger model that allows the

use of heavier loads should be used, so as to obtain considerab'ly

high readings in all the three transducers. A material of high
modulus of elastic'ity'is suitable for larger models since in a
stiffer model the deflections due to its self-weight and other dead
loads are small. Hence, a greater portion of the allowable deflec-
tjons could be obtained from the external loading a1one. Glasso

micro-concrete, sand filled epoxy are few examp'les of the suitable

model materia:ls.
(3)
of spring loaded transducers are recommended
The use

since with these type of transducers the curvature meter could be

attached to the underside of the,model. Hence, the load could be
applied from top even at locat'ions close to the reference point.
The curvature meter so designed is a simp'le but precise
apparatus that is
for the influence surface p'lotting.
most suitable

In turn the avaÍlability of influence surfaces for variety of standard

 -91

shapes of slabs is a valuable aÍd to the des'igner for pre'liminary

des'ign or for special l oad analysi s .

 REFERENCES

(l) Jaeger, L. G., Pergamon Press (lOe+¡, "Elementary Theory of

tlastic Pl ates".
(2) hlestergaard, H. M., "Computation of Stresses in Bridge Slabs
due to hlheel Loads", Publ ic Roads, Volume II, No. 'l o March, 1930.
(3) Rowe, R. E. and Base, G. D., "Model Ana'lysis and Testing as a

Design Tool", Proc. I. C. E., V. 33, Feb., 1966.

(4) Rowe, R. E. and Best, B. C., "The Ljse of Model Ana'lysis and

Testing in Bridge Desiqn", Preliminary Publication, Seventh

Congress of the I. A. B. S. E., Rio de Janeiro,1964, p. ll5-121.
(5) Andrä, hf., Leonhardt, F., and Krieger, R. o "Vereinfachtes
Verfahren zur lvlessung von l4omenteneinfl ussflächen bei Pl atten",
Der baningenieur, V. 33, n. ll. Nov. 1958. p, 4A7-414,
(6) Rüsch, H. and Hergenröder, A. "Einflusfelder der Momente

Schiefwinkliger Platten", Third edition 1969 with an English

translation, hlerner-Ver'lag-Düsseldorf, l9 pp. and 174 charts.
(7) Mehmel, A. and Weise, H., "A Contribution on the Structural
Behaviour of Skew Slabs Using a Model", English Translation,
N. R. C. of Canada, Technical Translation 1128.
(B) Mehmel , A. and l,{eise, H. , "Model InvestigatÍon of Skew Sl abs

on Elastically Yielding Point Supportsl', Cement and Concrete

Association Library TranslaùionyrNo . 123, 1965.
 -93-

(9) F., "Influence of the Spacing of the

Andrä, l,rl. and Leonhardt,
Bearings on Bending Moments and Reactions in Sing'le-Span Skew
Slabs", Cement and Concrete Association Library Translation
No. 99, 'l960.
(10) Cheung, Y. K. "The Analysis of Cylindrica'l Orthotropic Curved

Bridge Decks", International Association for Bridqe and Struc-

tural Engineering Publications, Vol . 29, Part II, 1969, pp 41-52.
(ll) Meyer, C. and Scordelís, A. C., "Ana'lysis of Curved Folded Plate

Structures", Journal of the Structural DivísÍon, ASCE, Vol 97,

No. ST-10, Proc. Paper 8434,Oct, '1971, pp 2489-2480.
(12) Buragohain, D. N., "Discrete Analysis of Cylindrica'l 0rthotropic
Curved Bridge Decks", International Association for Bridge and
Structural Engineering Publ ications, Vol 32, Part l, 1972,
pp 37-47.
(13) Pucher, Adolf, "Influence Surfaces of Orthogonal Anistropic
Plates", Springer-Verlao, Vienna and New York, .|964 (3rd ed.) '.l.t:-.:.i--
:.::_ _...
'. :. ::.
35 pp. and 93 charts.
'

(14) Molkenthih, A.n "Influence Surfaces of Two-Span Continuous Plates

with Free Longitudinal Edqes", Sprineer-Verlag, Heidelberq and
.165
New York 1971, 44 pp. and charts.
(l5) 0'lsen, H. and Reinitzhuber, F. "Die Zweiseitig Gelagerte P1atte",
Ver'lag W. Erust & Sohn, Berlino 1959, (3rd ed.) ll3 pp.

(16) Krug, S. and Stein, P., "lnfluence Surfaces of 0rthogonal Aniso-

tropic Plates'J. Sprinoer-Verìag, Ber'lin, 1963,33 pp. and 193
charts.
 -94-

(17) Homberg, H. and MarX, W. R., "schiefe Stäbe und platten", hlerner

Verlag, Düsseldorf,, 1958, 324 pp.

(18) Aneja, I. K. and Ro1'l, F., "Experimental and Analytical Investj-
gation of a Horizontally Curved Box-Beam Highway Bridge Model",
Second International Symposium on Concrete Bridqe DesÍgn, ACI

Publ ication SP-26.

(.|9) L. C. and Huggins, M. l,,l.o "Model Study of a 60-
Knowles, l,{.
Degree Three-Span Skewed Bridge", Department of Civil Engineering,

University of Toronto, March .|965, (0. J. H. R. p. Rep 35).

(20) Aneja, I. K. and R011, F., "Model Tests of Box-Beam Highway

Bridge with Can.ti"levered Deck Slabs", Conference Preprint No.

395 A S C E, Transportation Engineering Conference, Philadelphiao

Pa., Oct.17-21, 1966.

(21) Kurata, M. and Hatano, S. "The Influence Surface for the Indeter-
ninate Cl amping Moment of Sl ab-Bridge-Type Sker^red Pl atesT,
I. A. B. S. E. Publications, V01,24,.l964, p. l0l-'|12.
(22) Chen, T. Y., Siess, C. P. and Neu¡nark, N. M., "Studies in Slab
and Beam Híghway Bridges, Part VI. Moments in Sìmply Supported
Skew l-Beam Brídges", University of Illinois Engineering, Exp.
Sta., Bull, 439, Jan. 1957, 72 pp.
(23) Kawai, T. and Thürlimann, 8., "Influence Surfaces for Moments

in Slabs Continuous over Flexible Cross Beams'.1, I. A. B. S. E.

PublicatÍons, Vo1, 17, 1957, p 1.|7-138.
 -95-

(24) Timoshenko, S. and l,{oinoh/sky-Krieger, S., "Theory of Plates

and She'|1s", McGrar¡r-Hi1'l , New York, 1959.

(25) Ligtenberg, F. K., "The Moire [rlethod", Proceedinqs of the

S. E. S.4., Vol.12 No.2,.|955, p. 83-98.
(26) Holister, G. S., "Experimental Stress Ana'lysis',, Cambridge,
University Press 1967, 322 pp.
(27) Yeginoba'li, 4., "Continuous Skewed STabs", Ohio State Univ.
Eng '9. Exp. Sta . , Bu'l . 178, Nov. I 959.
(28) Schmidt, E., "Modellversuche zur Bemessung von Baukonstrucktioneu",
Schweizerische Bangeitung, lg4g.
(29) l,rleigler, H. and hleÍse, H., "Modellstatisches Verfahren zur
Aufuahine von Einflussflachen von Platten", Beton und StahTbetoubau,
May, 1 959.
(30) Kennedy, J. B. and Tamberg, K. G., "Problems of Skew in Concrete

Bridge DesÍ9n", Dept. of Highwayso 0ntarío, Report No. RR 144,

March I 969.
(31) Robinson, K. E., "The Effect of Skew on the Behaviour of Simply
Supported Bridge 51abs", Cement and Concrete Assoc., Tech, Rep.

TRB/271, Ju1y, 1957,

(32) Balas, J. and Hanuskao 4., "Der Einfluss der Querdehuungszahl
auf den Spannungszustand einer 45o schiefen Platte".
(33) Lansdor,vn, A. M., "The Use of lnfluence Surfaces in Assessing

Moments and Stresses in Short Span Bridges Subjected to Abnor-

maily Heavy Loads", Trans. E. I. C. V., May, J966.

 -96-

(34) Okamura, H. and Matsui; K., "Automatic Recording EquÍpments

for the Model Analysís of the Influence Surfaces of Plate",
, ., 1

Proc. of the Symp. on the New Ideas in Structural Design, J. S. C. E., ', ,'.-,,

Dec. , I 963.

(35) Engineering Digest, Vo1. 20, No. 5, May 1974, p.7.

(36) Coull, A. and Das, P. C., "Analysis of Curved Bridge Decks", ;:,: ,

:'i:i'
Proc. I. C. E., Vol . 37, May 1 967, p. 75-85
,:,::-t:.. :
':':.'._--:l_.
 APPENDIX I

l.l Thehlriter's Inverse Approach

A qualitative idea of each indivídual error, that the
writer feels could be present in the actual testing, Ís obtained by
checking the curvature meter analytically on an assumed worst case
as mentioned ín Chapter I. This so called "inverse approach" is
given below.
A deflection equal to the half the thickness of the plate
is considered as the worst case. The required relative deflection
f is also assumed to be equal to the maximum a'l'lowable deflection
although for the considered urorst case, a relative deflection between
any two points within the plate will always be much less.

(a) Error Due to Tilt of the Meter

Fig. A Fig. B
 -98-

F'rom the above Figure A, it can be seen that

sln v-
r_f
uz
For the "worst case" conditions L=1.5", f=0.0625"

and hence, sin o-- 0.0625

r.5 x I = 0.0833
oft
'6 = 4" 48'
If the additiona'l rotation due to the of the outer'leg
movement

(tfiat is shown in the next paraqraph to be equal to second order

quantity of O) is neglected, as shown in Figure B above the trans-
ducer will register a reading Ah' instead of the true reading 2Ah
due to the rotation ó.

2Lh
=
^h' cos ó

2Lh
cos 4o 48'

2Lh
(0. ee64e)

2^h (1.00352)

Therefore, maximum possible error due to tilt of the meter is 0.35

per cent.

(b) Error Due to the Additional Rotation of the Meter

 -99-

Let o be the addít'ional rotation of the meter due to the movement

of the'leg as a result of e'longation (or contraction) of the surface

of the plate.

*L* = H9'
4
but
LX-.| Hz + ¡z
2t
Hence, 0 ,M L,
-2 = H9'

Since ,ry + L,
ct

It is shown that the addÍtional rotation o is a second order quantity

of 0. e can be calculated as follows,

elongation e = .f 2
and € = Kh/Z .:.:...,::
,::a,..a,:,,t,a,::,,

ulhere, K = i+ f = Q.062s,', L = 'l.s,, ,.,,.:.,,,.,,,.

and h = 0.'125"

hence € - 8Ì<-0'9925 0.125

(1'5)2 x" 2

= 0.01 39

1Ã
ande=0.0139x+ 2

= 0.0.| 04
 - t00 -

Þ=H0
-2
Si nce fl = I.0"

and Q= 0.0208 radians

Thereforer o ( 02 < 0.000433 radians

oF, < 0.0250 r; ..t ..ì. ::

ït can be seen that a is insígnificant compared to ô.

(c) Emor Due to Shifting of 0uter Leg

Àh'
2^H

O+o(

o
I
ïf the bottom tip of the outer leg.sits on the wall of the conical
shaped punch mark instead of being at the vertex as shov,rn in the
above figure, it wíIl affect the transducer reading. The magnitude
of the emor and the correct size of the punch mark can be obtaíned
as follows. For convenience the outer 1eg is consÍdered as made of
one solid piece. The angle subtended by the chord length at the
center of radiai is assumed to be 2o and the angle of the conical
shaped punch mark is taken as 2a.
 I0l
Therefore,

sin 0_ L/2 cos

R
0

ôh = lsÍn[gO-(o+o)]
'l cos (o + o)

From the above figure,

l= 2LH
sin ö sin ['180 - (0 + 0 + o)]
_
oF' 1| _ 2ÂHsinó
- sfn(o+o+0)

hence, ôh - t a)
2¡H sÍn o cos (o
For the worst case cond.ition
sin(6+0+o)
¡2
ft = ft:,
õT
f = 0.0625, | = 1.5, 0 = 4" 48'

giving sino = Ll}cosQx9!

L2

- cos 4o 48' _
^ 8 x.l.5
0.0625
?
oF, = 0'.|66.l

S= 9o 36'

If 0, = 45o

ôh = AH sin 4" 48' cos (9o 36' + 45")

sin (4o 48' + 9o 36' + 45o)
.,^, sin 4" 48' cos 54" 36'
sin 59" 24'
= zAH . 0.05'|
 -102-

It is significant compared to the transducer

can be seen that ôh
reading 2ÂH. Thereforeo Ít is necessary to take precautions to see
that the bottom tip of the outer'leg does not get shifted on to the
wall of the conical shaped punch mark. However, this occurs very
rare'ly since both outer legs have flexible bottom parts.

1.2 Thê Effêöt of thê GauEe LênEth

An analytical approximation to the effect of gauge length

on the curvatures is obtained as follows. Let us assume that the

curvature meter is used to measure the curvature on a simple beam
subjected tõoadlÞadtPthü ùbeteenter.

The deflection profile of a simply supported beam loaded at center

is represented by:

Pç x3
= -ml
L^ ''JLr'*l
Y
n-
1

2
[* - i]'- ro J
tDl

Then I et the curvature meter be pos i t'ioned on the beam and I et

 .'J,.,, -r.. :t::,, j :..:.. ::¡..: r:,1: _:t:.:.:-:.:.i: -...-..¡ 1; ;,;'^:¡l;'.1;.),.:;..:;.,: ,.-:. -.:. 1_: 1.: : : ;:;1:i
" : ;,1

- 103 -

PL:
Y. = beam deflection at the center leg = È
Yo = beam deflection at the outer leg
Yt = beam deflection at the transducer

(A) Suppose the curvature meter is used to measure the curvature

at the center of the beam. In thÍs case to employ the curvature
meter principle, only Y. and Yo are required.
For a gauge length of Lr/20
L^ L_ 191-
to the outer leg from end A =
Distance
Ì - Aà = *=
l or L^2 l gL-
{'0 = -P.-rLr-.þ13- Þ x-3 1
EI '12' 40 ' 16 40 r
since f = Y.-Yo
|3 'tor 3

l-f r 3 + 1 rl9-L"ì3
r ì7 r40 )/ -'"t
- -m*,'l
Er 4ã
3
PL^
=È [o.00007682 ]

Hence, K. =
Ï+
t
= f.-^ }!*1[0.00007682]
tr
:' ',,,
¡-Stz
\m/
pL^ .j.,..,.. ,,
0.2458 =ï:
LI

PL^
at the center from beam theory = 0.25
¡1å .
But the true curvature
,. : i..;;.:
The difference between the two curvature values is -l.67 per cent. ,,'..
LL
SimiTarly for gauge lengths ot ñ ana 5*!- , differences of -3.33
per cent and -6.66 per cent are obtained respectively.

* The square bracket term in the Equation D is neglected when x < Ls/Z,
 - 104 -

(B) Suppose the curvature meter is used to measure the curvature

at the quarter point.
L^
For a gauge length of *

I
I
I
u.i
evl

LS I
-s 91,
Distance to the outer leg from end A= T- 40
=
aõ'-
1,,
Ya= - P- fs-'
EI'12x64-. r *Þl
4'
I

1 I PL,,

and
%ã"T1*
'll
t. +b (ab-:)' SX
3 ot
L^ 9L^
ft= 'c -Y'o = I'EI
Y D
r- 768
5l
4õ- ¡
3
PL^

Ti [o.ool2l]

Distance to the transducer from endA

L'+b- llLs
=
=
440 40

s'imilarlV, ¡, = Yt-Y.
1,, Lrt ]
= -hrì"r#r.- '16 x ]t.t¡ +U-
40 768
 a,'_ ,.1

- 105 -

= H[o.oor3r7]
hence, eI _ 0.00121 PLs L,
EI 4õ
_
-S
0,0484 PL 2

and EI

A
"2
_
-5
0.00'H 3t 7 PL^3
/,,

_
EI

0.045269 PL- 2
l40
-S
EÏ

oo3l 32 pls 2 L,
therefore, 2f = (ol - oz) Ls/40 =
tJ.
*
EÏ 40

ol*1.
g1 v] ng
8f'
- E- I
= 4xoroo3.l..3.?
(Ls/20) 2
. Er
*
t.ç.

40
o_l:
= 0.12528
EI

PL,
But the true curvature at the quarter point from beam -rI=
theory 0.125;-
length of 's .
I
Therefore, the difference is 0.24 per cent for a gauge
Similarly, differences of 0.0032 per cent and zero per cent are
n
tl
obtained for gauge lengths of11 anolrespectÍvely.
r0 5
 ':.:-:,:..- r.: :_ :::. : :::.

APPENDIX ÏI
 107 -
2.1 Tables of Influence Coefficients.
(a) Influence Coefficients Io - Model I (15" Opening Angle)

Load
Reference Points 0n l5o Opening Ang'le Model

)o i nts A B c n E F G H

a 0 690 1 555 0.553 0. 383 0.387 0 356 0.087 0.094

â1 I '186
à2 l 017
'l
â3 085
â4 0 B6'l

ô5 0 Bt9
â6 0 io, 0 747
d7 0 630 0.541
b 0 675 0 839 0. 598 0.361 0.37 9 0 357 0.081 0.091
c 0 814 0 0.670 0. 350 0.359 0 36'l 0.080 0.086
C1 0 730 :o'
C2 0 764
C3 0 750
C4 0 695

C5 0 670
C5 0 627

C7 0 608

Cg 0 610
C9 0 608 o.) sq
cto 0 517 0.43.|
ct t 0 492 0.41 I
d 0 639 0 uro o.ss+ 0.346 0.333 0 365 0.077 0.075
e 0 605 0 1 .475 0.352 0. 304 0 359 0.074 0.069
e1 1.193
ê2 1.030
.l.088
ê3

eq 0.841
e5 0.82]
€5 0.629 0 523

( Conti nued )
 - 108 -

Load
Reference Points 0n l5o 0pening Ang'le Model

Poi nts A B c D E F G H

f 0. 584 0. 64ì 0 499 0.489 0.549 0 396 0.1l4 0.1t4

g 0.542 0.61 4 0 529 0.453 0.522 0 407 0.1 12 0.t13
h 0.498 0.539 0 567 0.428 0.453 0 45.| 0.t06 0.1t0
l 0. 503 0.469 0 621 0.433 0.387 0 5il 0.095 0.098
j 0.517 0.426 0 630 0.436 0.337 0 526 0.086 0.087
k 0.393 0. 384 0 365 0.493 1.327 0 0.156 0.1-22
k1 0.712 :u'
k2 0.679
k3 0.4942 0.608
k,,T
ks I .000
k5 0.863
k7
k8 0.600
ke 0.588
kro o. gge 0. 51
'l

krt 0.354
l o. so+ o. sgz 0 i,, 0.4944 0.679 0 t* o.lsz o.iz+
m 0.347 0.364 0 0.61 0 0.461 0 0.'t 30 0.138
lll 1
:" 0.48] :,,
fll2 0.460
m3 0.463
llì r, 0.459 0
Ìll 5 0.508 !"
lll5 0.533
tttT 0.558
lll6 0. 555
m9 0. 523
mto 0.472

( Conti nued )
 - 109 -

Load
Reference Points 0n l5o Opening Angle Model

?o i nts
A B C D E F G H

lllt t 0 3BB
frtz 0 371

Illt g 0 378
lïl rr1r, 0 372 0
ffits 0 246 :
llltO 0 240
n 0.344 0. 334 0. 384 0 476 0.361 0 652 0.'l 07 0.1 39
o 0. 35t 0. 304 0.379 0 0.305 1 235 0.091 0.130
O1 1" 0 656
o2 0 659
O3 0 876
O4 0 781
O5 0 925
05 0 841

O7 0 621
Og 0 558
O9 0 344
p 0.ì95 0.185 0. 189 0.294 0.361 0 207 0 320 0.085
q 0. 183 0. t84 0. 189 0.270 0.338 0 232 0 232 0.096
r 'l
0. 76 0.174 0.1 90 0.244 0.277 0 272 0 t3t 0.t39
'l
s 0. 73 0.168 0. 188 0. 251 0.220 0 325 0 082 0.221
t 0.17 4 0..l 58 0.180 0.263 0.'183 0 349 0 063 0.297
u 0.150 0.166 n '105 0 926 0.045
V 0.1 32 0.1 59 0 116 0 205 0.054
.l34
w 0.116 0. 0 l3t 0 087 0.087
X 0.123 0.1 09 0 154 0 049 0.t93
v 0.132 0.094 0 158 0 035 0.922
f' 0.28t 0.272 0 266 0 050 0.075
g' 0.264 0.268 0 265 0 062 0. 069

( Conti nued )
 - 110 -

Load
Reference Points 0n l5o Opening Angle Mode'l

Poi nts c
A B D E F G H

h' 0.259 0 262 0.263 0.062 0.060

i' 0.254 0 252 0.260 0.059 0.057
j' 0.255 0 237 0.249 0.056 0.052
k' 0. I 826 0 185 0.'l 78 0.043 0. 049
I' 0.177 0 179 0.173 0.042 0.046
m' 0.'l 70 0 174 0.'l 70 0.040 0. 043
nt 0.t69 0 067 0.168 0.0393 0.036
ot 0.'l 68 0 162 0..|65 0.0386 0.032
p' 0. 094 0 088 0.091
q' 0.0884 0 087 0.089
r' 0. 0852 0 086 0.084
S' 0. 0852 0 083 0.082
t' 0.0848 0 081 0.080
 - llt
(b) Influence Coefficients I* - Mode'l I ('15" Opening Angte)

Reference Reference Reference

Load
Poi nts Load
Poi nts Poi nts
Load

Po i nts )o i nts Po i nts

A D A D A D

cl -0.293 -0 .129 m3 -0.048 i' -0.063

a6 -0. 'l 62 lïl r,
I
-0. I 02 j' -0.076
b -0. '182 -0 .090 lll 5 +0.033 k' -0. 055
c +0.I 58 -0 .071 m6 -0.054 I' -0.049
ct -0.1 1 0 filT +0.009 mt -0.043
c2 -0.047 llì 6 +0.0.| 3 nl -0.046
ca -0.041 frì 9 -0.041 o' -0.049
Cr.I -0.099 mto -0.049 p' -0. 026
c5 0.0 ffitt -0.035 q' -0. 025
c6 -0.094 lfltz +0.01 7 rl -0.024
c7 -0. 031 ffit g -0.022 sl -0.0247
c8 -0.082 flìtq -0.076 t' -0.0251
ca -0.1 4t ffits -0.026
cto -0.097 -0.077 Into -0. 02.|
cit -0.089 -0.067 n -0.090 -0.095
d -0.'l s7 -0. 079 o -0.122 -0.172
e -0.243 -0.1 1 1 p -0.069 -0.1 1 9
f -0. 23] -0. I 82 q -0.052 -0. 052
g -0.142 -0.1 01, r -0.044 -0.006
h -0. 07 5 -0. 043 S -0.046 -0.044
1 -0. I 23 -0.087 t -0.060 -0. I 00
j -0. 200 -0.1 50 u -0.059
k -0. I 45 -0.21 'l \, -0. 026
k3 -0.1 28 w -0.007
kro -0. 093 X -0.021
t -0 100 -0. I l3 v -0.049
m -0 079 +0. I 0t f' -0.089
llì 1 -0. 058 g' -0.071
m2 -0.01'l h' -0.062
 -112-
(c) Influence Coefficients IÊ - Model 2 (45" OpenÍng Angle)

Reference Reference Reference

Po i nts Poi nts Poi nts
Load Load Load

Po i nts Po i nts Po i nts

A D A D A D

a 0.0 :0 . 139 Ill 3 +0. 028 i' +0.033 +0.028

a6 +0.0344 lI '.
+0. 046 j' +0.048 +0.072
b 0.0 -0 .082 m5 -0.01 5 k' -0.059 -0.077
c 0.0 -0 .020 llì 5 -'0.029 I' -0.034 -0.045
C1 0..0 n7 -0.021 mr -0.004 -0.0'14
C2 0.0 m8 -0.006 rìt +0.025 +0.01 I
C3 0.0 ffi9 +0.002 ol +0.047 r0.048
Cr, 0.0 fito -0.0.| 2 p' -0.033 -0.039
C5 +0.001 ffilt +0.004 q' -0.01 I -0.023
C5 +0.032 fntz *0.01 0 rt -0.003 =0.007
c7 +0.002 ffitg -0.023 S' +0.01 2 +0.009
Cg -0. 026 Íl tr.+ +0. 01 0 t' +0. 026 +0.024
C9 -0. 028 rlìtS +0. 002
cro +0.027 -olosg ffito -0. 0l 4
cit -0.021 +0.020 n -0.025 +0. 0'18
d 0.0 +0.040 0 -0.047 +0.048
e 0.0 +0.092 p +0.033 -0.0.| I
f +0.060 *0.140 q +0.01 I +0.001
g +0.042 -{.090 r +0.003 .0.007
h +0.003 -0.0'l 9 s -0.01 2 -0.0'| 3
l -0.033 +0.049 t -0. 026 +0.002
j -0.048 +0. 095 u .-0.001
k +0.059 =0.077 V t0.002
k3 -0.08'l W -0.003
kro -0.008 X +0.007
'l +0.034 -0.045 ¿0.002
v
m +0.004 -0. 0t 4 f' -0.060 *0.1 I 2
lTl
1 -0.062 g' -0.042 -0.065
m2 :0.018 h' -0. 003 -0.01 B
 - ll3 -

(d) Influence Coefficients Io - Model 2 (45" OpenÍng Anqle)

Load
Reference Points

Po i nts
A B c D E F G H

a 0.804 I .593 0 671 0.453 0.420 0 420 0. 'l 05 0.084

â1 1 .260
à2 I .087
a3 1 .085
a4 0.853
â5 0.824
â5 o. oga 0.7 47
ð.7 0.624 0.568
b 0.769 0.854 0 666 0.409 0.404 0 405 0.1 00 0.085
c 0.842 0. 640 0 722 0.371 0.372 0 399 0. 095 0.082
C1 0.773
C2 0.795
C3 0.7 67
C4 0.709
C5 0.706
C5 0.667
C7 0.626
Cg 0.6'14
C9 0.58.| 0 761
cro 0.563 0.468
ctt 0.515 0.429
d 0. 634 o. so+ 0.834 0.347 0.327 0 384 0. 084 0. 078
e 0. 538 0.399 1.48t 0. 332 0.268 0 371 0. 07'l 0.077
ê1 1.t51
Q2 0. 995
€3 1 .076
ê4 0.864
e5 0.815
e6 0. 637 0 513

( Conti nued)
'.-r'...'.1-'.'.::::.:.,_.'.'. - . -a- .'i... _'."- ::'

-il4-

Reference Points
Load

Poi nts n
A B D E F G H

f 0. 659 0 637 0. 600 0 569 0.582 0.468 0 133 0.'108

g 0. 599 0 6t5 0. 592 0 507 0. 545 0.467 0 127 0. 'l 06
h 0.536 0 526 0.611 0 446 0. 463 0.47 6 0
'tt6 0.105
i 0. 500 0 438 0. 634 0 428 0.378 0. 508 0 103 0.104
1
rJ 0.466 0 36t 0.622 0 399 0.302 0. 5'l t 0 085 0.102
k 0.446 0 387 0.437 0 1.383 0.4_22 0 0.114
k1 0.740 l"
k2 0.703
k3 0 552 0.631
k,, 0.876
ks 1 .021

k6 0.854
k7 0.820
k8 0. 631
ks 0. 609
krn 0 0.542
ktr a* 0.353
I o. g9, 0 i,n o.izo 0. 545 0.684 o. q¡s 0..| 64 o.irg
m 0.365 0 346 0.409 0. 683 0.473 0.47 5 0.142 0.'l 3'l
ml 0.524
m2 0.483
fll 3 0.473
I't'ì
r'. 0.455 o. sgz
flì 5 0.522
IIl5 0. 582
n7 0.613
m6 0. 598
fïl 9 0. 532 :
Il'l t O 0.468

( Conti nued )
 - il5 -

Reference Points
Load

Po i nts
A B c D E F G H

ffitt 0 418

ffitz 0 391

ffits 0 382
l'l'l r¡l r, 0 367 0 ut
frlts 0 277
ffito 0 ?54

n 0. 330 0.304 0.399 0 454 0.352 0 uto 0 112 0. 143

o 0. 3'l 0 0.254 0.366 0 3Bl 0.27 0 1 242 0 082 0.1 45
o1 0 644

o2 0 680

03 0 887

o,,I 0 782

o5 0 949

O6 0 837

o7 0 612

o8 0 572

09 0 345

p 0.228 0. 20t 0.228 0. 34t 0. 360 0 255 0 321 0.085

q 0.202 0. t87 0.217 0.297 0.334 0 262 0 237 0.094
r 0.186 0.t71 0.207 0.257 0.273 0 2BB 0 134 0.130
S 0.'l 70 0.154 0.197 0.246 0.202 0 327 0 0Bl 0.21 5
L
L 0.156 0.133 17Ã 0.226 0.t55 0 342 0 052 0.302
^
u 0.170 0.161 n 129 0 932 0. 046
V 0.1 44 0.159 0 135 0 218 0. 054
W 0.124 0.t35 0 147 0 087 0.083
X 0.ì21 0. 'l 05 0 ì56 0 047 0.172
v 0.tts 0. 078 0 155 0 025 0. 908
f' 0.333 0.306 0 323 0 075 0.062
g' 0.307 0.287 0 309 0 076 0.062

(Conti nued)
 - 116 -

Reference Points
Load

)o i nts A B c D E F G H

h' 0. 283 0.267 0.301 0.069 n 059

i' 0.264 0. 238 0. 283 0.064 0 056

j' 0.247 0.zil 0.266 0.056 0 055

k' 0.221 0.202 0.220 0.048 0 036

l' 0.208 0.19t 0. 209 0.048 0 037

mt 0.t92 0.'l 73 0.1 98 0.047 0 038

n' 0.'178 0.154 0.183 0.043 0 038

o' 0.165 0.143 0.t73 0. 037 0 040

p' 0.109 0. 097 0.'l1t 0.024 0 0204
q' 0.t0.| 0.090 0.103 0.023i 0 020
r' 0.093 0.084 0.098 0.022! 0 01 73

sl 0. 086 0. 078 0. 092 0.017 0 0t 70

t' 0.079 0.071 0.085 0.016 0 0l 66

 _117 -

(e) Influence Coeffícients IR Model 2 (45' Opening Angle)

Reference Reference Reference

Poi nts Po i nts Poi nts
Load Load Load

Poi nts A D
Po i nts
A D
Poi nts A
l'ì
U

a -0. 362 -0. t80 m3 -0.048 i' -0.077

ô5 -0. 203 lll r,
I
-0. 096 j' -0.084
b -0.225 -0.126 m5 +0.036 k' -0.080
c +0.142 -0.0908 m6 -0.080 l' -0.068
C1 -0. t 4'l m7 -0.0.| 2 mt -0.058
c2 -0.072 m8 +0.003 nl -0. 054
ca -0. 052 m9 -0. 0491 ot -0.055
Cr, -0.'t 02 ffiro +0.0485 p' -0.040
c5 -0. 0l 9 ffit t -0.046 q' -0.034
c6 -0. I 23 frtz +0. 026 rt -0. 030
C7 -0.046 fnt g -0. 023 S' -0. 028
c8 -0.090 ffit,* -0.071 t' -0.027
ca -0.'138 ffits -0.036
cto -0.1 I 3 -0. og+ ffirs -0. 020
cti - 0. 096 -0.070 n -0.0891 -0.097
d -0. t 5l -0. 091 4 o -0.il2 -0. I 56
e -0. 2'l 0 -0.il4 p -0.085 -0.'t 63
f -0.287 -0. 250 q -0.062 -0. 078
g -0.'t78 -0.1 44 r -0.049 -0.01 2
h -0.092 -0. 063 s -0.047 -0.044
I -0. t 2'l -0. 094 t -0.053 -0.09t
j -0.'175 -0. I 49 u -0.081
k -0.179 -0.283 V -0. 038
k3 ì0. 1 54 W -0.015
kto -0.1 20 X -0. 021
'l
-0. I 23-0. I 58 v -0. 045
m -0.0886 +0.099 f' -0.'t 25
ml -0.075 g' -0.'t 00
n2 -0. 003 h' -0.08]
 - ll8 -

(f) Influence Coefficients lrn - Model 2 (45' Openino Angle)

Reference Reference Reference

Poi nts Load
Poi nts
Load
Poi nts
Load

Poi nts roi nts Po i nts

A D A D A D

a 0.0 +0.'l 93 rll g -0.034 n' *0.009

â5 +0.04t m9 -0. 025 ot +0.022
b 0.0 '0.129 ffito -0.037 p' -0. 056
c 0.0 +0.062 fïìti -0. 0t 7 q' -0.039
c6 +0.036 frtz -0.03.| rl ^0.022
c7 +0.005 ffitg -0.043 S' -0.006
Cg -0.023 lïì r r, -0.028 t' +0.01 I
C9 -0._024 ilts =0. 01 2l
cro *o]oo ffite -0.027
ctt =0. 0l I n -0. 01 6 r0.009
d 0.0 =0. 002 0 -0. 035 +0.022
e 0.0 +0.051 p +0.041 =0.021
f +0.076 =0. I 93 q +0.024 -0.0.| I I
g +0.051 =0. I 34 r +0.008 -0.020
h +0.009 *0.058 s -0. 007 -0.025
l -0. 026 -0.0t I t -0.0t 9 -0.007
j -0.036 +0.054 u -0.006
k +0.073 -0.1 1 0 V -0.005
k3 -0. I 21 u, -0.01 0
kro ,,0. 029 X -0.01 3
I +0.044 -0. 076 v :0.006
m +0.01 2 -0.042 f' -0. t 58
mi -0. 099 g' .-0. I 07
n2 -0.051 h' -0. 057
fll 3 -0.006 i' .0.008
[ì,,I +0.01 I j' +0. 037
m5 i0.047 k' -0.il 0
fll5 -0. 059 I' -0.076
n7 ì0. oso ml -0.042
 - ll9 -

(g) Influence Coefficients Io - Continuous ModeÏ

Load
Reference PoÍnts

Poi nts A B c D E F G H

a -0.048 -0.086 -0. 0008 +0.082 +0.095 +0.048 +0.814 +0.026

b -0. 044 -0. 053 -0.0t 6 +0.064 +0.079 +0.054 +0.205 +0.039
c -0.039 -0.025 -0. 032 +0. 054 +0.063 +0.07] +0.072 +0.063
d -0.031 -0.008 -0.061 +0.054 +0.042 +0.084 +0.030 +0. I 57
e -0.024 +0.001 -0.098 +0. 053 +0.025 +0.089 +0.01 3 +0 .7 99

f -0. 093 -0. ì 69 -0.0009 +0. I 64 +0. l87 +0.084 +0.287 +0.048
g -0.0854 -0. I 05 -0.027 +0. I 36 +0.177 +0. I 0l +0.200 +0.064
h -0.076 -0.049 -0.063 +0. I 09 +0. I 28 +0.'t48 +0.'l 02 +0.096
1 -0. 060 -0.01 7 -0.117 +0. I 09 +0.092 +0. I 88 +0.048 +0.177
j -0.046 +0.002 -0. I 92 +0.1 04 +0.048 +0. 205 +0.027 +0.257
k -0.1 23 -0.249 0.0 +0.23.| +0.374 +0. I 09 +0. I 54 +0.056
l -0.t16 -0. I 45 -0. 033 +0.212 +0.303 +0. 1 33 +0. I 35 +0.069
m -0. I 06 -0. 067 -0.083 +0. I 83 +0. I 83 +0.'|,95 +0.095 +0.097
n -0.0848 -0.01 9 -0. 1 63 +0. I 73 +0.1 0B +0.300 +0.058 +0. 1 45
0 -0. 058 +0.0047 -0. 285 +0. I 44 +0.063 +0.3 80 +0. I 3l +0. I 43
p -0.1 3t -0.326 +0.006 +0.256 +1'.042 +0.1 I 3 +0.092 +0.055
q -0. I s5 -0. I 78 -0.029 +0.258 +0.380 +0.1 50 +0.087 +0.064
r -0.127 -0. 076 -0.087 +0.391 +0.?04 +0.21 I +0.068 +0.076
s -0. 096 -0.01 I -0. I 93 +0.209 +0.1 I 6 +0.370 +0.048 +0.085
t -0. 063 +0.0066 -0.352 +0. I 58 +0.065 +0. 964 +0.029 +0.095
u -0..|'16 -0. 387 +0.009 +0.221 +0.356 +0.098 +0.0556 +0.046
V -0. 1 32 -0.'185 -0. 020 +0.202 +0. 288 +0. I 25 +0. 0559 +0.057
W -0. 'l 39 -0. 066 -0.077 +0. 1 7l +0. I 78 +0. I 83 +0.048 +0.056
X -0.098 -0.01 2 -0. 201 +0. I 63 +0. I 02 +0. 295 +0.037 +0.058
v -0. 054 +0.009 -0.406 +0.'t 33 +0.058 +0.3 55 +0.022 +0.063
z -0.073 -0.432 +0. 01 I +0. 1 21 +0.'t 62 +0.070 +0.029 +0.027
at -0.1 07 -0.157 -0.01 0 +0.0997 +0. I 57 +0.088 +0. 032 +0.029
b' -0. I 36 -4.042 :0. 048 +0.0959 +0.'l 14 +0.124 +0.030 +0.030
c' -0.078 -0.007 -0.1 64 +0.097 +0.071 +0. I 70 +0.022 +0.029

( Conti nued )
 -120-

Load
Reference Points

)oi nts
A B c D E F G H

d' -0. 029 +0.008 -0.425 +0.087 +0. 041 +0. I 88 +0.016 +0.029
'l +0.006
el -0.022 -0.41 -0.068 +0.059 +0.035 +0.01 0 +0.01 32
fl -0.049 -0.077 -0.00] -0.059 +0.065 +0.038 +0.01 6 +0.01 4
g' -0.1 1 I -0. 01 2 -0.016 -0.048 +0.050 +0.056 +0.0'l 3 +0.01 1

h' -0.039 -0.003 -0.092 -0.044 +0.031 +0.068 +0.0.l 0 +0.01 23

i' -0.008 +0.004 -0.369 -0.045 +0.020 +0.076 +0.007 +0.01 I !

jr -0.079 -0. 079 -0. 066 -0.01 B -0.01 6

kt -0.066 -0. 062 -0. 060 -0.01 5 -0.01 64

'l ,
-0.058 -0.051 -0.060 -0.01 5 -0.01 5
m' -0.052 -0.041 -0.063 -0.01 2 -0. 0l 2
nt -0. 049 -0.032 -0.069 -0.01 I -0.01 I
o' -0.0494 -0. 092 -0.082 -0. 020 -0. 0l 9
p' -0.041 -0. 076 -0.072 -0.01 I -0. 0t B
q' -0. 037 -0. 060 -0.070 -0.01 5 -0. 0l 6
r' -0. 031 7 -0.049 -0.075 -0.01 4 -0. 01 6
SI -0.0305 -0.042 -0.077 -0.01 2 -0.01 3
tl -0.056 -0.0s2 +0.0.| 3 -0.01 4s
ut -0.047 -0.048 +0.0098 -0.0t 28
vt -0.039 -0.0458 +0.008 -0.009
w' -0. 031 -0.0462 +0.008 -0.008
xt -0.029 -0.0481 -0.0077 -0. 009
 - 121

(h) Influence Coefficients I* - Continuous Model

Reference Point Reference Point

-oad Point -oad Point
D D

a -0.048 a' -0.020

b -0.01 4 b' +0.026
c +0.008 ct -0. 005
d -0. 002 d' -0.04.|
e -0.024 e' -0.036
f -0. 095 f' -0. 009
g -0.024 g' +0.008
h +0.026 h' 0.0
I - 0. 004 i' -0. 0l 9
j -0.048 j' +0.020
k -0. 1 40 k' +0. 0l 0
I -0.045 I' +0.0066
m +0. 066 mt +0.007
n -0.01 5 nt +0.008
0 - 0. 069 ot +0.0.| 4
p -0. I 59 p' +0.0.| I
q -0.059 q' +0.009
r +0. I 76 rt +0.007
s -0.024 st +0.009
¡L -0.075 tt +0.0083
u -0.'t 37 ut +0.0075
V -0. 03 9 vt +0.004
v', +0.066 W' +0.005
X -0.0'l 2 X, +0.006
v -0.065
z -0.086
 -122-

(i) Influence Coefficients IuR - Continuous Model

Reference Reference
Load
Po Í nts Load
Po i nts

Poi nts Poi nts

A D A D

a -0. 0l 5 +0.01 6 at -0.029 -0.0.| I

b -0. 0l 0 +0.009 b' -0.007 - 0. 005
c -0.003 +0. 002 ct +0.0234 +0.01 4
d +0.004 -0. 009 d' +0.029 +0.006
e +0.003 -0"014 gt -0.027 -0. 009
f -0.035 +0.039 f' -0.01 9 -0.01 I
g -0. 0t 7 +0. 025 g' -0. 008 -0.002
h -0. 005 +0.003 h' +0. 021 +0.01 I
+0. 008 .i +0.026 +0. 002
1 -0.022 '
j +0. 020 -0.022 j' +0.008
k -0.038 +0.035 k' +0.004
I -0.01 8 +0.020 l' +0.0.l 0
m +0. 00.| -0.004 mt +0. 002
n +0. 0'l 5 -0. 028 nt -0. 002
o +0.031 -0. 028 ot +0. 0l 5
p -0.048 +0.01 7 p' +0.002
q -0. 034 +0.003 q' +0.003
r -0.004 -0. 009 rt +0.002
s +0.0.l 9 -0.008 st -0. 0004
t +0.037 -0. 01 4 t' +0. 005
u -0.059 -0.002 ut +0.009
V -0.035 -0. 01 5 V' +0. 003
W -0.003 -0. 01 0 W' -0.0006
X +0.023 +0. 0l 9 xt -0.0008
v +0.04ï +0.001
z -0. 053 -0.0.l I
 -123-

2.2 INFLUENCE SURFACES

 o.69

I
J
N)
Þ
I

tn
I FIGURE I - 1

Infl-uence Surface I^(, for the Reference Point A

 - 1tro
IJ
ü)
t - 0.1
-*Q

'',:..\ .,,.',i'. .l',. ,:

: tl :: . .'..., :
;,.1:
.:: .,i:.:.
 ':

.:l
':)
'ì'
j.
- 0-294 4.,

::'
,.j

i l;:

-o---F
-o.o84 \ I
:;.
j,
:Ì
J li:
N) .t;
(¡ .rl
| ì,,

-o.243 :.;:l

FIGURE I - 2
Influence Surface I-K for the Reference Point .rl

l,:

Q = l-5o
A
u) -?t 0 = 0.1
t. -
r ì-i
.i.

.1,

.,ì
 o.063

(ii\
-0.063

\\ /¡
\\ /t
\. -ioo ./ |
\\ ii' tt ,'

I
J
f\)
oì
I

// -o.o3

-oo495 o.0495

FIGÜRE I - 3

Influence Surface IU* for the Reference Point A

A_ 15"
UJ
-=
r 0.1

|:

':
 I

f\)
\¡
I

FIGURE I : 4'
Infl-uence Surface Ia for the Reference Point B

--* p 15"
0.1
;
o.554

I
J
f\)
co
I

FIGURE I .
-_]- 5

Infl-uence Surface IO for the Reference Pofnt, C

--e 0= 15o

g = 0.1
 -:1'

i;, :

,.i

ì-:,

i'l
'ri'

,.t,
o.5
Ê,
'l // o.t5
li 0. I

::;
'..4

| .1

':.;
J
f\) :;

| ¡l

,l;
ìi.

.t.
:.
f'n
I -6 rl,,

rì

Infl-uence Surface I^o for the Reference Point ':\

D
U = 15 0
(,^)
-+ ó) (¡ = 0.1
r
r

':, .:,,. ',.

' :.
-,
:..
ì,
i]
.';:,:,.,-. .
., ..;.,:,,.)
 - o.?1"

\ì\----. -o.ot{ i

I
J
(,
O
I

-oo25 /.) +'

-'i
,i ,*il
-o.l72l

FIGURE T- 7

Infl-uence Surface In for the Reference Point

A- l5 o

-+8 ûJ
-=
r 0.1
 /l
/i
,t
/t

(Á)
J

o.o98

FIGURE I - 8

Influence Surface I'n for the Reference Point D

A_ 15 "
u)
-=
r 0.1
 lt ,

,':

:!.t
rl

O.25 O.2 O.fs O.l

(,
J

f\)
I

FIGURE

\ I
Influence Surface I0 for the Reference Point E

(,
rl I t-
--, 0
(,
r
15

0.1
0
 o.395

(,
J

(^,
I.

FIGURE I - 10

Influence Surface Iê for the Reference PoÍnt F

'-> e H
= 15o
û)
r = 0.1

:i. '.
 -.i r

,i.

":]

,:,
:.,'i

92:6 'tii

.'ì

i,:.

I i:
(á)
J

Þ
:.'
;'i.'

| :.,

AR ;lì

lG FIGURE I - 11
;ii
Inf luence Surface IO for the Reference Poínt G

--Þ 0 A- 150

tù
-=
r 0.1

'.:.:: i:";i
r;.': .11.i'.1 ..,.:; _:,.,
,:.
',.: .:: , ,t";ill. ,.t'
:1 .. .::, "
 'tì

i.'

i..

I
J
(4,
(tl
I

o.92-.z^

FIGURE I . L2
'::
Influence Surface Ia for the Reference Poínt H
:.

-*0 0 = 150
Ì,
ì,

:l
''¡ I

u)
t = 0.1
 I
J
(^)
oì
I

FIGURE I . ].3

Infl-uence Surface IA for the

Reference PoÍnt A
fl
= 45"
r = 0.3
ûJ

1:.
 -0.362

fr.Z
/-t'
.'
?-' -o.15

r_N --__-Qfzs
l/.' /
\
-op0?6
\- l' 'o,í^
I
09/
it t
I
I J

\(,
I

fn
I -]-4
FI GURE

Infl-uence Surface In for the

Reference Poínt A
0=45"

r = 0.3
üJ
 o.o79 -o.o79

__ _ -o.or
r---__--0.06 .-qOS- -
I ll rl. \ i
tt
trtt t., \---' _____O.O3 QÚí..
\*t-\*-
--
:-"]o;;r '\ \-.-' -'
---O.g¿ ---
O'
--
\-' ¡

¿
4ro2 -
--ool (^)
æ
I

o.039

, FIGUREI-]-5
Influence Surface I0n for the
Reference Pol-nt A
e=45"
üJ

t
= 0.3

. l. : .:l
t.593 1.4

I
J
(¡)
(.c,

.FIGURE I , ].6

Infl-uence Surface I^U for the

Reference Point B

U = 45"
r = 0.3
trj
 _ 0.65

t'/
66

I
J
Þ
O
I

fn

FIGURE I - 17

Influence Surface Ie for the

Reference Point C
0=45"
u,
r - 0.3
 0.593

Þ
I

tR
,::
\ '/a

'::
FIGURE I - 18

Influence Surface I^ for the .'':l

u
Reference poínt D
lil
0=45"
: = 0.3
/¡l

r
.,:

.,,'' , i¡, ,,,,,,

:'¡:
 -o.z9z

:=l:È
*o.e5
)òpzs- -o.o75
/
/ I
I I
I J
Þ
N)
I

-o.tee

\n
FIGURE I - ]-9

Influence Surface I-.t( for Èhe

Reference Poínt D

0=45"
ûJ
s 0.3
t
 ;6-15
\\
/

I
I

l
I t/i
\\ J
Þ
I il..

:i
t (,
\ I
\

o.o55

FIGURE I - 20

Influence Surface I^-

9t(
for the
Reference Point D

0=45"
g = 0.3
r
\.381

'oøg ,o:

I
J
Þ
-Þ
I

FIGURE I . 2L

Influence Surface I0 for the

Reference Poínt E
e=45o
g = 0.3
t
 ------
ouud.ou
o.1u
.

,' !.4fÞ --

o25 o.2 J
I

Þ
(tl
I

lz4z

FIGURE I - 22

Infl-uence Surface I0 for Ëhe

Reference Point F
0=45"
0 = 0.3
r
ofr?.
5 0.2

I
J
è
oì
I

0-o87

' FIGUREI-23
Influence Surface Ia for the
Reference Poiit c

= 45o

r = 0.3
ûJ
 o.t15

oou ?'
I
J
Þ
\
I

FIGURE I - 24

Influence Surface I0 for Èhe

Reference Point H
8=45"
g = 0.3
r
 I

\- -. J
Þ
oo
I

-0.06ã

FIGURE I' - 25

Infl-uence Surface Ig for the

Reference Point A
i:Ì:

0.L¿ = 0^ = 22,5"
OJ

r = 0.3
 oo59 -o.o59
tì
V
I
I

ii
I
, I
-oo5 -o.ol
\ \
l/ I
'l I ,ortt..
l/
1=
ffi
-l
\--
J

,
\ ôô
.T / f ----P'> /
\--- - -\'I 'f-- -1"

w
ï-- -1I
Y-
f o.or
-\-
-=-Y
/\
\

\
r
\

'ït,
\
t
t
\

I
\ I

o.o2
I
,
1-Tì .t
I
I
J
è
(o
I

I , c .o5 /
o03 \ tt
F-
o.o4l o.o

1n
FIGURE I - 26
Influence Surface I^-
0tf
for the
Reference Poínt A
A
9r :- Â
0^ - aa c
¿¿.)
o
L¿
u)
t = 0.3
 I

(tl
t=
I

tn
I

FIGURE I - 27

Influence Surface I0 for the

Reference Poínt B

0t = 0Z = 22.5"
(¡
r = 0.3
 I
J
(t
J
I

FIGURE I - 28

Influence Surface Iav for Èhe

Reference Point C

\e e,-0^=22.5"
L¿

oJ = 0.3
t
 -o.08
o.zbq

I
J
(tl
N)
I

IR
I
FIGURE I - 29

Influence Surface Iav for Ëhe

Reference Point D
0 . = 0^ = 22.5"
J-¿
û)
t = 0.3
 :t

-o.t59

I
J
(tl
(,
I

:1e16

FIGURE I - 30

Infl-uence Surface In for the

Reference Polnt D
0,=0^=22.5"
LZ

u) = 0.3
r
 I
J
(tl
Þ
I

FIGURE I - 31
-_-----_----
Influence Surface I^^
ul(
for the
Reference Poínt D

0.=0^=22.5"
L¿

CtJ

r = 0.3
 -o.o98
\9+?

!
J
(Jl
(tl
I

picunu r- ¡z
Influence Surface I^ forU
Lhe
Reference Point E

0.=0^=22.5"
L¿

(j
T
= 0.3
o.$4

I
J
(tr
or
I

964

FIGURE I - 33

Influence Surface Ie f or the

Reference PoínÈ F

U
L =02 = 22.5"
(¡
t = 0.3
 -.lr¡. il'.1'

o BVA

I
J
(tl
\¡
I

-o.o,2

, FIGUREI-34
Influence Surface I0 for the
Reference Point G
0.L¿ =0^ = 22.5"
u,
r - 0.3

;1. .ì
 I
J
(tl
oo
I

-o.oß
-órss

FIGURE I - 35
_-.----_
Influence Surface I^ for the
U

Reference Point H

A
u.L¿ - Â
v^ - a,
¿¿.)
Eo

ÛJ
-=
r 0.3