I

l l
l
'l'
l
A

STUDIES FROM
THE SCHOOL OF CIVIL ENGINEERING

THE DESIGN OF FLAT SLAB STRUCTURES

- AN HISTORICAL SURVEY

K. A. FAULKES

UNICIV REPORT No. R-129 APRIL, 1974

UNIVERSITY OF NEW SOUTH WALES, KENSINGTON
N.S.W., AUSTRALIA 2033

ISBN O 85841 081 8

 THE DESIGN OF FLAT SLAB STRUCTURES -

AN HISTORICAL SURVEY

by

K.A. Faulkes.*

SUMMARY
This paper traces the development of design methods
for flat slab structures, and considers the current state
of knowledge relating to the design of such structures for
both vertical and lateral loads.

* Senior Lecturer, Department of Structural Engineering,

The University of New South Wales, Australia.
 CONTENTS

Page No.
1. INTRODUCTION 1
2. EARLY HISTORY 1905-1921 1
2.1 Early Construction 1
2.2 Early Theory and Design Methods 4
2.3 Early Tests 7
2.4 Nichols, 1914 7
2.5 Early Code Requirements 9
2.6 Westergaard and Slater, 1921 10

3. VERTICAL LOADING 12
3.1 Theoretical 12
3.2 Design Methods and Building Code Requirements 15
3.3 Experimental 20
4. LATERAL LOADING 23
5. STIFFNESS PROPERTIES 25
5.1 Patel (1957) 25
5.2 Brotchie and Russell (1964) 26
5.3 Carpenter (1965) 26
5.4 Copley (1966) 30
5.5 Qadeer and Stafford-Smith 32
5.6 Aalami (1972) 34
5.7 Summary of Stiffness Properties 35
6. CONCLUSION 37
6.1 Vertical Loading 37
6.2 Lateral Loading 38

ACKNOWLEDGEMENT
REFERENCES
FIGURES
 - 1 -

1. INTRODUCTION

This paper examines published work on elastic bending moments in

flat slab structures. It traces the development of design methods and cons-
iders the current state of knowledge of flat slab behaviour.

Section 2 describes the rather stormy beginnings of flat slab

construction in the first two decades of the twentieth century. The early
controversies are of considerable historical interest and their effects may
still be discerned in the provisions of today's building codes.

Investigations into the bending moments produced by vertical loads

are considered rather briefly in Section 3, since this area has been well
surveyed by other investigators. Published work which is directed towards
stiffness properties and design for lateral loads is examined in more detail
in Sections 4 and 5.

2. EARLY HISTORY 1905 - 1921

2.1 Early Construction

The advent of reinforced concrete allowed the extension of structural

design from one-dimensional to two-dimensional elements. The development did
not come easily. One of the pioneers of flat slab construction, Robert Maillart,
wrote later:

"With these one-dimensional elements: joists, columns and beams,

the engineer was accustomed to calculate and build his structures, in such
a manner that they were everything in the world for him and any other
possibility lay far beyond his reach ..... This was the situation when reinforced
concrete emerged, but at first nothing was changed: it was laid as if it
were steel or wood, girders spanned from wal 1 to wall and from column to
column. At right angles to these main girders came secondary beams and the
space between would'be filled in with a slab without however its being
comprehended as a special constructional element. On the contrary they made
haste to divide it up into strips, those strips could then be calculated as
beams in the normal manner." (Ref. Bill, 1969).

A similar point was well made by Sozen and Siess (1963).

"Just as the first motor cars were built to look like horse-drawn
carriages, the first reinforced concrete systems were conceived in the image
of traditional types. In a timber structure, the planks carried the load to
the joists, the joists to the girders, and the girders to the columns; so
 - 2 -

must they in a reinforced concrete structure. Hence, the flat slab had to be
invented rather than developed as one of the obvious appli~Rtions of reinforced
concrete."

Credit for the invention goes first to the American engineer

t' A.P. Turner. Writing in the "Engineering News" in 1905, Turner introduced
his 'mushroom slab' thus:

"There is ample room for all to improve radically on present methods

of design and computation which are .... following after the manner of structural
iron work.

"Thus far experimental investigation has been confined almost

exclusively to simple beams and slabs reinforced in one direction only .....
That concrete lends itself readily to reinforcement in all directions should
lead the practical constructor, as far as may be, to so reinforce his work that
the deformation from a strain in one direction may be offset in part by a
force in another direction.

"Enclosed herewith (Fig. 3) a study along this line, the idea being to
avoid the expensive forms for beams to secure a neat and unbroken ceiling line
together with a considerable economy of material without sacrifice of strength."
(Ref. Turner, 1905).

The "study" referred to was probably the first structural drawing

of a flat slab ever published. It is reproduced here as Fig. 1 and shows the
four-way reinforcement system which was to characterise Turner's flat slab designs.

The first flat slab structure to be built was the five-storey

C.A. Bovey-Johnson building, constructed by Turner in Minneapolis, Minnesota
in 1906. Building Department permission was obtained for it only as an
experimental structure subject to a load test, in which the floor was required
to withstand an imposed load equal to almost three times the design load
without deflecting more than 5/8" at the centre of any panel. Two adjacent
panels were loaded; the maximum deflection was only¼", easily satisfying the
requirement.
This building was the first of many. In 1914 Turner was able to
speak of "having introduced the mushroom system in about $200,000,000 worth
of buildings and bridges, of spans from 12' to 50', in the past seven years .... "
(Ref. Eddy, 1914) and of his experience "acquired in the design and construction
of from 1000 to 2000 structures of this type." (Ref. Nichols 1914).
 - 3 -

The rapidity with which the flat slab system was accepted in the
U.S.A. is emphasised by the fact that it was usetl fu1· 809, uf all lJuildings
designed for loads of 100 psf or more in the period 1906-1913. The acceptance
was not unqualified. Turner wrote in 1914: "The conservative business man
who advances the money, as the writer has found by experience, would usually
like a bond, which may amount to anywhere from $5000 to $100,000, to assure
him that the structure when completed will come up to the guaranty." The
"guaranty" involved tests of load carrying capacity and maximum deflection.
Since the engineering profession was quite unable to agree on a method for
flat slab design, and since the cost of a flat slab was known to vary markedly
depending on which engineer designed it, the business man was undoubtedly just-
ified in his conservatism.

During this period the Swiss engineer, Robert Maillart, was pioneering
flat slab construction in Europe. Maillart's first step was to conduct experi-
ments on large scale models. He wrote: "Of course it is a problem that is hardly
solvable by calculation and so only tests with models and the measuring of
executed buildings can lead to a more certain goal. "(Ref. Bill 1969, p .165).

Figures 2 and 3 show the first flat slab structures to be built for
experimental purposes. They were constructed in 1908 in the workyard of
Maillart and Cie in Zurich "in order to first obtain a foothold concerning
the constructive possibilities of the reinforced concrete slab." The first
test slab, shown in the foreground of Figure 2, was pin-supported at the four
corners, and was quickly found to be unsuitable. The second structure, shown
in Figure 3, consisted of a nine-bay slab rigidly connected to its columns via
capitals, and "proved of such stiffness, even with point loads in single fields.
that the practical value of the system was proved." This structure was claimed
by Maillart to be the first flat slab to use a two-way system of reinforcement,
rather than the four-way pattern favoured by Turner.

This test.convinced Maillart of the practicability of the flat slab

system. "The problem now was how could it be constructed and dimensioned.
The purely theoretical way appeared to be inaccessible ...... In order to achiev,
an experimental basis for the problem, and to acquire a sufficient basis for
practice, the same firm erected a large structure with nine fields each of four
meters length. The slab was only 8 cm. thick to assure the greatest possible
elasticity." This test structure, built in 1910, is shown in Figure 4. A
single concentrated load of 1000 Kg was applied at various points on a one-
meter grid; the deflection curves for a variety of loadings were obtained
from the test readings by superposition. A number of simply supported beams
 - 4 -

were also made with the same thickness and reinforcement as the test structure.
"The influence of the single loaus coulu now !Je juugeu siuce Llte ueflecliuu
curve of the slab could be compared with the deflection of the beams under the
corresponding known bending moments." Although it seems unlikely that much
reliable information on moments or stresses could be obtained in this way, the
tests gave Maillart sufficient confidence to proceed with commercial construction
of flat slab buildings.

The first of these buildings was the Lagerhaus-Gesellschaft building,

constructed in Zurich in 1910. In addition to using two-way rather than four-
way reinforcement. Maillart' s system differed from Turner's in employing a
curved column capital (see Fig. 5) so that "the column fuses into the floor
slab corresponding to the play of forces." No information has been found on
the methods by which Maillart determined the slab thickness and reinforcement
quantities required.

2.2 Early Theory and Design Methods

The emergence of the flat slab system in the United States aroused
intense controversy among structural engineers. Sozen and Siess (1963) comment:
"For the structural engineer, plate action was an entirely new concept. The
'crossing beam analogy' thinking of the slab as two perpendicular beams each
carrying a certain proportion of the load in relation to their stiffnesses,
helped only to foster the still existing illusion that only part of the load
need be carried in a given direction. Grashof's work had already been used
by the mechanical engineers in boiler plate problems. However, this work was
represented in American engineering literature either simply as formulas
without any derivations or as a basis for arriving at questionable conclusions."

The "questionable conclusions" usually involved a misunderstanding

of what Mensch called "the mystic influence of Poisson's ratio" (Ref. Eddy,
1914, discussion) .. The chief authority for this school was Professor Eddy
of Minnesota, who developed an elaborate theory of slab action from the basic
assumption that the lateral strain due to Poisson's effect must be accompanied
by a corresponding lateral stress. His theory naturally came into conflict
with the laws of statics but it was popular among some of the "commercial
designers", notably C.A.P. Turner, since it had the effect of appreciably
reducing the negative reinforcement required over the columns.

Some engineers (then as now) doubted whether a structural system as

novel and complex as the flat slab really came within the gamut of the laws
of equilibrium. Professor Eddy wrote of "the essential divergence of the
 - 5 -

correct theory of slab action from that of beam action in which latter case
, , , , , tho moment cf the applied .fo1·ces ls equal t6 the moment of the internal
resistance, which is not true of slabs." (Ref. Nichols, 1914, discussion).

Eddy's theories did not go unchallenged. E. Godfrey complained

that "In asking the Profession to accept this mushroom theory, the author
(Eddy) is asking us to consider the laws of matter suspended for this particular
type, or else that there are special laws that apply to this special combination
of materials." (Ref. Eddy, 1914, discussion).

The practical effects of the controversy were highlighted by Angus

B. McMillan in 1910. In an article in the Engineering News (Ref. McMillan,
1910), he described six methods then in use for the design of flat slabs.
The methods fell into two broad classes: those based on (or misconstrued
from) Grashof's theory, and those using one of the variations of the
'cantilever method'. In the latter method a rectangular or circular section of
slab surrounding the column head and bounded by the inflection lines, usually
assumed at about one-fifth of the span from the column centre-lines, was
assumed to act as a cantilever, supporting the loads on its surface and a
concentrated line load at the perimeter which supported the rest of the slab.

McMillan applied each of the six methods to the design of a 20ft x

20 ft interior panel of a flat slab carrying 200 psf live load, and tabulated
the resulting slab thickness and reinforcement quantities per panel. His
table is reproduced below. Methods 1 and 2 used the cantilever approach,
methods 3, 4, 5 and 6 were deduced from Grashof's work.

Design Method Slab Steel Amount of Reinforcement

Thickness Stress per panel
(in.) (psi) (lb)

1. Cantilever 8 16,000 2,189

2. Turneaure & 1,931
12 16,000
Maurer
3. Grashof 8 16,000 784
4. Mensch 8 16,000 2,120
5. Turner*(a) 8 16,000 549
(b) 8 13,000 718
6. McMillan 8 16,000 1,084

*Turner used and recommended a steel stress of 13,000 psi

(a) was included for purposes of comparison.
 - 6 -

The table indicates that the quantity of reinforcement used could

Val')' by iB much as 400% depending on which design method was used. Turner I s
method gave the smallest quantity of all; small wonder that his clients
demanded bonds dependant on satisfactory performance of their buildings in
load tests.

The differences between design methods were considerable; the

financial stakes for some of the designers were huge. (Turner claimed to have
used his system in two hundred million dollars' worth of structures over a
seven year period.) The debate between the 'commercial' designers and their
more cautious brethren was bitter and public. The discussion of a paper by
Professor Eddy highlighted the attitudes of the opposing factions (Ref. Eddy,
1914).
Godfrey complained that "the commercial designers succeed in scaling
down the bending moments to a point which independent practising engineers
fear to attempt ..... The theory of flat slabs is hard for any disinterested
engineer to accept." L.J. Mensch wrote of "A riotous licence of figuring by
most of the advocates of flat slab construction ..... The writer begs not to
be misunderstood as being an opponent of flat slab construction ... He protests,
however, when engineers and contractors (who, as a rule, have all to gain and
nothing to lose) represent by versatile agents that flat slab construction.
of designs as advocated by Mr. Eddy, are as good as, or better than, girder
constructions which actually have a factor of safety of four or more .... No
wonder that they can show to unsuspecting architects and owners a great
saving over all other designs, and being able to mention a great number of
examples of buildings which did not fall down (having a factor of safety of
about 2) are believed to be by the owners and architects very wizards in the
art of reinforced concrete construction."

C.A.P. Turner, on the other hand. was able to reply by invoking

the prestige of a large successful practice: "Having introduced the mushroom
system in about $200,000,000 worth of buildings and bridges, of spans from
12 to 50 ft, in the past seven years, the writer may state that more testing
has been_done and larger bonds written guaranteeing its strength than for any
other kind of concrete construction ..... when it is considered that it has
been put up in Australia, India, the West Indies, throughout Canada and the
United States, the record of achievement must have behind it something more
than mistaken ideas."
 - 7 -

2. 3 Early Tests

The weight of evidence was on Turner's side.

Flat slab buildings, no matter now designed, passed their load

tests with ease. Furthermore, evidence was accumulating from more sophist-
icated tests in which reinforcement strains were measured.

The first such test was made in 1910 by A.R. Lord, who measured
reinforcement strains on a flat slab floor of the Deere and Webber building
in Minneapolis. The 9-3/16" thick slab was designed for a live load of
225 psf, and.in the load test eight adjacent interior panels were subjected
to an imposed load of 350 psf. The maximum measured negative and positive
moment stresses were 24 and 10.4 Ksi respectively.

This was the first of several similar tests on flat slab buildings
constructed in the U.S.A. in the period 1910-1920. The tests have been summa-
rised in a thesis by. D.S. Hatcher at the Un-i' versi,t.y
. of Illinois (Ref. Hatcher,
Sozen & Siess, 1961). In all cases uniform load wa·s imposed on a number of
panels (varying from one to nine) of a large multi-panel floor. Steel strains
were measured using extensometers with gauge lengths varying from 8" to 15",
and the usual straight-line reinforced concrete formula, M = f A jd, was used
s s
to calculate the bending moments from the strains. In all tests the strains
measured at working loads, and therefore the mOJDents deduced from them, were
found to be small.

2.4 Nichols, 1914

The flat slab debate appeared to be deadlocked. Turner, who claimed

to base his calculations on Grashof's theory, designed his flat slab panels
for a total negative moment over interior columns of WL/50. McMillan, also
using Grashof' s theory, concluded that the correct figure for this moment was
WL/25. The proponents of the cantilever method doubled the figure again, and
used steel quantities four times greater than Turner's. Yet Turner's slabs
stood, and tests indicated that steel stresses at working loads were wel,
within safe limits. Perhaps the only point of agreement among the disputants
was that flat slab analysis was complex and "hardly solvable by calculation' .

In 1914 a young Boston engineer, J.R. Nichols, broke the deadlock b:

showing that the total moment for one special but commonly occurring case could
be calculated from the equations of statics.
 - 8 -

Nichols considered an interior panel of a regular flat slab floor

with square panels extending indefinitely in both directions, supported on
circular column capitals and uniformly loaded (see Figure 6). The segment
shown in heavy outline, bounded by sides A,B,C,D and E, was taken as a free
body.
Due to symmetry, no shears or twisting moments exist on faces B,C,
D and E. There are bending moments on all faces, and shears and twisting
moments on face A. The assumption is made that the shear forces on the curved
face A are uniformly distributed.

The total vertical load on the free body is 0.25w(L2 - nr~), which
must also be the value of the total shear on face A. The resultant of the
shear forces, assumed uniformly distributed around A, acts at a distance
from the centre of the column equal to 2V2r /TT.
0
Taking moments of all vertical forces about XX gives:

wL 3
Mxx = 16 (1 - 2.55k + 2.67k 3 ) where k = r 0 /L
For equilibrium, this must be equal to the sum of the bending
moments on faces C and B, plus the components normal to XX of the moments
on the curved face A.

In his conclusion to the discussion on his paper, Nichols suggested

a simpler approximation to the above formula. Extending the free body across
the full width of the panel, the approximate expression for total moment is

= WL (l _ ~~)2
8 3 L

where·W is the total load on the panel, and c the diameter of the column
cap. The error involved in the approximation is less than 1% for values of
c/L smaller than 0.3.

Obviously, Nichols' formula did not enable the moment at any point,
or across any section, to be determined; but it did provide the first sound
criterion against which designs could be checked. Nichols himself commented:
"The nature of the limitations imposed by statics is best shown by an illust-
ration. If we are told that three stones weigh 6 lb. this does not estabiish
the weight of any one stone. but it does ensure that the heaviest stone weighs
at least 2 lb."

Nichols had done little more than apply the equations of statics to a
flat plate panel, yet his paper provoked a furious discussion, five times as
long as the original paper.
 - 9 -

A.W. Buel complained that "the author's reasoning seems to be

deductive, which since the time of Lord Francis Bacon, has not been in favour
for scientific investigation." L.J. Mensch thought that Nichols' paper l'will
not advance our knowledge of the design of such floors, as he assumes a
certain relation of positive and negative moments, and fails to prove that
they may exist."

C.A.P. Turner, leaning confidently on his experience "acquired in

the design and construction of from 1000 to 2000 structures of this type,
dismissed Nichols' thesis as "mere algebraic deductions which the author has
based on certain assumptions. These assumptions and deductions by Mr. Nichols
appear to involve the most unique combination of multifarious absurdities
imaginable from either a logical, practical or theoretical standpoint. At
the very outset he assumes the illogical proposition that the mechanics of a
slab and the mechanics of a beam are identically the same." He invoked the
authority of Professor Eddy, who "clearly defines the external moment of
forces acting on a slab or beam as apparent moments."

Eddy joined the attack with an exposition of his flat slab theory,
which, as noted earlier, was based on an erroneous treatment of Poisson's
ratio effects. He further compounded the error in a tortuous attempt to
justify the use of a Poisson's ratio value of 0.5 for reinforced concrete.
This had the effect of halving the magnitude of the calculated negative
moments over the columns, an achievement which suited Turner and appeared
to give closer agreement with reinforcement stresses measured in building
tests. The agreement was in fact fortuitous, and it was perhaps unfortunate
that Eddy chose to congratulate himself in this discussion on "having brought
a rational theory to a somewhat satisfactory degree of perfection."

The forces arraigned against the application of Newton's laws to

flat slabs must have seemed formidable. G.S. Binckley. in discussing a paper
by Eddy in the same.year, wrote: "The crushing weight of practical experience
under which C.A.P. Turner, M.Am.Soc.C.E., flattened out Mr. Nichols' purely
theoretical paper tends to induce caution in others."

2.5 Early Code Requirements

The available empirical evidence appeared to support Turner's

'experience'. His slabs used much less steel than Nichols' analysis would
have required, yet they performed satisfactorily. Furthermore steel strains
had been measured in several buildings, and the total panel moments deduced
from them were in all cases much smaller than the value of M0 calculated from
 - 10 -

Nichols' formula, For example, tests on six flat slab floors yielded the
following values of totai moment:

Purdue test slab J 0.59 M

0
Purdue test slab S 0,74 M
0
Western Newspaper Union 0.72 M
0
Sanitary Can building 0.30 M
0
Shonk Building 0.38 M
0
Be 11 St. Warehouse 0.40 M
0

The engineering profession was thus faced with a conflict between

Newton's laws and apparently overwhelming test evidence in favour of
"suspending the laws of matter" for flat slabs. Unable to ignore either
Nichols' logic or the test figures, it adopted the form of Nichols' equation,
but arbitrarily reduced the magnitude of the total moment. The first Joint
Committee decided that flat slabs should be designed for a total moment of

= 0.107WL(l - ¾f )2
representing 85% of the static moment. The 1920 A.C.I. Code rejected the
conditions of equilibrium even more decisively with;

2 C 2
M
0
= 0.09WL(l - 3 L)

thereby authorising a disregard of statics which persists in building codes

to this dav.

2.6 Westergaard and Slater, 1921

Nichols' paper had shown how the total moment in an interior panel
could·be calculated. It gave no information on the distribution of moments
within a panel, and it did not explain the low stresses measured in building
tests. These problems were resolved when Westergaard and Slater (1921)
published a comprehensive paper which aimed "to present information which
correlates the results of tests of a fairly large number of slab structures
with the results of analysis, so that the report may aid in the formulation
of building regulations for slabs."

The first section of the paper, written by Westergaard, dealt with

the analysis of homogeneous elastic plates. It commenced with an historical
summary of the development of plate theory, covering 42 references from Euler
(1766) to Nielsen (1920), followed by a derivation of the plate equation,
 - 11 -

+ +

and a comprehensive discussion of bending moments in rectangular plates

supported on four sides.

Westergaard then considered a typical interior square panel in a

regular flat slab structure of infinite extent, supported on rigid circular
columns, and uniformly loaded. A solution was achieved by first using results
obtained by Nielsen from finite difference analyses of slabs on point supports,
so that the combined solution gave zero slopes and deflections around the
peripheries of the columns.

These analyses yielded the first reliable information on the

distribution of elastic moments in a flat slab panel. The results for c/L
ratios varying from O. 0 to O. 3 were given in two figures which are reproduced
here as Figures 7 and 8. It was found that the proportion of the total
static moment taken by the various parts of the panel did not vary significantly
with varying c/L ratio. Average values were:

Column Strip Middle Strip Total

Negative Moment 48% 17%
Positive Moment 21% 14% 35%

These percentages are very close to those used in today's building codes.

In addition to this work on uniformly loaded interior square panels,

Westergaard also obtained some limited information on the effect of uneven
panel loading, and on moments in oblong and exterior panels.

In the next section of the paper, Slater examined the relationship

between reinforcement stresses and bending moments in beams and slabs. He
was able to show th~ inadequacy of the assumption, made in previous analyses
of flat slab tests, that moments could be obtained from measured steel
strains by simply substituting in the straight-line formula M; EE A jd.
s s s
Slater commented:

"In a cracked beam the stresses at the cracks may approach the
computed stress, but between the cracks the concrete assists so greatly in
carrying the stresses that the average measured unit-deformation over the
gauge length is likely to be considerably less than the maximum unit-deform-
ation, especially at the lower loads. It is possible also that even at the
section where a crack occurs a portion of the moment may be resisted by the
tensile stresses in the concrete."
 - 12 -

Slater investigated the relationship between measured steel strains

am! lHmtllug momemLs by testing 84 beams with va.rying steel pereenta.geS. He
compared the steel stresses indicated by measured steel strains with the
stresses calculated from f M/A jd, and found that the measured stresses
=
s s
were significantly smaller than the calculated values for all loads up to
the ultimate. With low steel percentages and low loads, the measured stress
could be as little as one-quarter of the predicted value at working loads.

These tests exposed the major reason for the differences between
the static moments and those deduced from steel strains in the 1910-1920
building tests. All of the strain measurements were made with long gauge
length extensometers; steel percentages in the slabs were low; concrete
tensile stresses were therefore very significant. There were other contributing
reasons - the effect of adjacent unloaded panels, the neglecting of twisting
moments around the large column capitals of these buildings - but errors due
to these causes would have been small compared to those introduced by
ignoring concrete tension. Slater's work brought about the major reconciliation
between Turner's flat slabs and Newton I s laws.

The 1921 paper by Westergaard and Slater was a monumental work

which remains perhaps the most important contribution yet made to knowledge
of moments in slabs under vertical loading. It marked the end of the beginning
of the search for a practical solution to the flat slab problem.

3. VERTICAL LOADING

3.1 Theoretical

3.1.1 Interior bays of Uniformly Loaded Floors

If a flat slab floor has a regular array of rectangular bays in

both directions, is uniformly loaded over its entire area, and is so large
that for practical purposes an interior bay may be considered to be surrounded
on all four sides by an infinite number of identical bays, then the two
centre-lines of the interior bay, and the four column centre-lines that bound
it, are all lines of symmetry on which no shears or twisting moments exist.
The resulting simplification of the boundary conditions makes this a special
case for which the complexities of flat slab analysis are greatly reduced.

As discussed in the previous section. this case was first considered

by Nichols (1914) who derived an expression for the total static moment in
the bay, and by Westergaard and Slater (1921), who calculated the distribution
of bending moments within an interior bay of a flat slab structure supported
 - 13 -

on circular column capitals. Westergaard and Slater dealt mainly with square
bays, although they u!JLairrnd some limited itiftil.'ffi/:ltiofl oh 1noffients in oblong
bays, with aspect ratios up to 1:1½.

An approximate solution for square interior bays supported on square

column capitals was published by Woinowsky-Krieger (1954), who used a complex
variable method in conjunction with conformal mapping, originally developed
by Muschelisvili for two-dimensional and torsional problems in elasticity.
The method was laborious, and its use was made practicable only by the high
degree of symmetry obtaining in this special case. The paper presented
graphically the bending moment distributions calculated for a slab with
column side: bay span ratio c/L = 0.2.

Gupta and Vaughan (1967) extended Woinowsky-Krieger's work to

include rectangular bay slabs supported on rectangular columns. The theoret-
ical solution was checked against experimental results obtained from a small
scale multi-bay Perspex model. Curvatures in the interior bay of the model
were measured by the moire fringe method, and showed good agreement with
calculated values.

3.1.2 General Theoretical Solutions

An ingenious mathematical solution was published by Brotchie (1957

and 1959), who observed that the elastic analysis of flat plate structures
was made complex by the interaction of loads and reactions. The loads and
reactions individually were simple in form or could be divided into simple
components; the complexities of analysis could therefore be reduced by
considering the components one at a time, and combining them by superposition.

This was achieved by supporting the slab on an imaginary liquid,

ancl. applying each load or reaction component separately. The reaction
provided by the liquid was proportional to the deflection of the slab and could
be either positive or negative, since the liquid was assumed to remain in
contact with the slab at all times.

Polar coordinates were used, and values of radial and tangential

moments were tabulated for (i) point loads and (ii) a uniform radial moment
applied round a circle of small radius. By combining the two cases, and
by numerical integration from the point loading case, the effect of any tyPe
of load on a flat slab supported on circular columns could in principle be
determined. However, the amount of work involved made the use of the method
rather unattractive except for very simple cases.
 - 14 -

Russell (1966) used Brotchie's method to produce graphs of the

distribution of moments in interior bays urn.lei- u11ifui-111 load, fd:t' dbldtil(

bays with aspect ratios varying from 1: 1 to 1: 3.

Brotchie (1963) later modified classical plate theory to examine

the effect on bending moments of the in-plane or arching forces resulting
from the extension of the middle surface of a slab due to flexural cracking.
He concluded that if the boundaries of an interior panel were fully restrained,
the in-plane forces could reduce the total moment by 17-20%. but that in a
flat plate structure with no marginal beams, or with all bays loaded, the
restraint largely disappeared. and the full static moment given by Nichol's
expression should be used.

A numerical procedure for the analysis of rectangular plates supported

on rigid columns of rectangular cross-section was developed by Ang. (1959) at
the University of Illinois. He used Newmark's plate analogue to enable
finite difference equations to be applied to the analysis of plates. In
order to reduce the amount of computer storage required, Ang developed a
distribution procedure which allowed continuous plates to be analysed by the
interconnection of single panel solutions. The procedure was based on the
Hardy Cross moment distribution technique for beams and frames, and commenced
with the "fixed edge moments" and "fixed edge reactions" at all edges of all
panels resulting from the loads applied on the panels. It consisted of
(i) successively balancing moments at each joint common to two panels, around
all joints in all edges, while maintaining continuity at the joints, and
repeating the moment-balancing process cyclically until satisfactory conver-
gence was obtained; (ii) similarly balancing one vertical reaction at a time
at each joint common to two panels, while maintaining continuity, and repeating
till the unbalanced reactions at all joints were reduced to acceptably low
values; (iii) since the equilibrium of moments attained in step (i) was
unbalanced by step (ii), (i) had to be repeated, thereby re-establishing
equilibrium of moments but unbalancing the vertical reactions again ... and so
on. Clearly many cycles of computer calculations were required for the solution
of general plate problems by this method. However, in the case of continuous
plates in which all panels were identical, the balancing of the vertical
reactions did not affect moment equilibrium at the joints, and the solution
was obtained much more rapidly. Ang's work was restricted to this case.

Woodring and Siess (1968) combined Ang's distribution technique

with a procedure proposed by Newmark (1941) in order to obtain influence
surfaces for continuous plates supported on rigid rectangular columns.
 - 15 -

Several charts showing influence surfaces for'a 3 x 3 square bay structure

were included in this paper.

3. 2 Design Methods and Building Code Requirements

3,2.1 Introduction

The Americans were the original pioneers of flat slab construction,

and have ever since led the way in devising and improving flat slab design
methods. Australian and British Codes of Practice have consistently followed
A.C.I, Code provisions in all essential points. Therefore only the A.C.I,
Code wil 1 be discussed here.

3.2.2 The Empirical Method

An outline has been given in Section 2.5 of the circumstances which

led to the adoption in the 1920 A.C,I. Code of a provision allowing flat
slab panels to be designed for a total moment 28% smaller than the static
moment. This Code also provided that, for slabs without drop panels, the
minimum percentages of the total moment to be resisted by the various sections
of the slab should be:
Negative Moment Positive Moment
Column Strip 40 18
Middle Strip 10 12

The remaining 20 9, was left for the designer to distribute "as required
by the physical details and dimensions of the particular design employed,"

The 1928 A.C.I. Code left the value of the total moment unchanged
but revised the distribution percentages, following Westergaard's work, to:

Negative Moment Positive Moment

Column Strip 46 22
Middle Strip 16 16

The distribution coefficients, and the expression for total moment,

remained unchanged in A.C.I. Codes until 1971, except that the 1963 Code
increased the design value of the total moment by up to 15% for values of
c/L less than 0.15.

The design of flat slabs by the use of the A.C.I. formula for total
moment and coefficients for distributing that moment, was referred to as
"Design by Moment Coefficients", or "Design by Empirical Method". The 1928
Code introduced a provision limiting the use of this method to structures
 - 16 -

similar to those which had provided the test data by which it was justified,
i.e. to flat slab floors contalnlng "a Sel'les uf slabs uf apprulllmaLely
uniform size arranged in three or more rows of panels in each direction, and
in which the ratio of length to width of panel does not exceed 1.33." Since
many flat slab structures fell outside these limitations, a design method
which could be more generally applied was clearly needed.

3.2.3 Equivalent Frame Methods

In 1929 the committee charged with formulating the reinforced

concrete section of the Uniform Building Code, California edition, set up a
sub-committee to investigate the possibility of treating the flat slab and
its supporting columns as a series of elastic frames. The sub-committee's
report, published later by Dewell and Hammill (1938}, led to the inclusion
of an "Elastic Frame" method of flat slab analysis in the 1933 Californian
edition of the Uniform Building Code.

The method adopted was very similar to equivalent frame methods in

use today. A full panel width of slab constituted the "beam" of the bent.
Column-slab joints were considered to be rigid. Columns were assumed to have
points of contraflexure at their mid-heights. The positive and negative
moments calculated from the frame analysis were distributed between column
and middle strips in the same proportion as was specified for the Empirical
Method.

One of the problems faced by the sub-committee was that, since the
elastic frame analysis conformed to the conditions of equilibrium, it led to
considerably larger moments than the Empirical Method, which accounted for
only 72% of the static moment. This inconsistency was removed by the simple
expedient of reducing by 40% the negative moments calculated in the frame
analysis.

An equivalent frame method for flat slabs was introduced into the
A.C.I. Code in 1941. It followed Dewell and Hammill's method closely, the
main differences being that columns were considered to be fixed at their
remote ends, and that the negative moment reduction was achieved by specifying
that for design purposes the maximum negative moment should be taken as that
obtaining at a specified distance from the column centre-line, this distance
being devised so that the resulting design total moment for interior panels
under uniform load was closely equal to that given by the Empirical Method
formula. The specification of the critical section for negative moments was
modified in the 1956 Code, but the effect remained substantially the same.
 - 17 -

Corley, Sozen and Siess (1961) compared the moments calculated from
the 1956 A,G,I, equivalent frame aualysis With some known elastic solutions.
It was found that, generally, the equivalent frame method gave values of
positive moment which were too low, and values of negative moment (before
reduction to critical sections) which were too high. After reduction the
negative moments could be either high or low depending on the dimensions of
panel and column. Usually the negative moments after reduction tended to be
too low, except at exterior columns, where they were too high. In some
cases the sum of the positive moment and the average of the reduced negative
moments could be smaller than the value of the total panel moment given by
the Empirical Method formula.

The reason for the high initial negative moments given by the
equivalent frame method lay in the Code assumption that the slab-column joint
was infinitely stiff. Even if the column itself were infinitely stiff the
slab on either side of the column would undergo curvature. The error was
magnified at exterior columns because the Code in effect assigned infinite
torsional rigidity to the edge beams, by assuming the equivalent beam to be
infinitely stiff within the limits of the column.

To improve the accuracy of the equivalent frame analysis for vertical

loads, Corley proposed the following changes in the methods of calculating
the stiffness of the ersatz members:

The stiffness of the slab within the slab-column joint, instead of

being assumed infinite, should be taken as that of a slab twice as deep as
the real slab.

The stiffness of an interior column in a flat plate structure

should be taken as infinite within the joint. For flat slab structures with
column capitals, the 1/EI diagram for the column should be assumed to vary
linearly from zero at the mid-depth of the slab, to the 1/EI value for the
column at the base of the capital.

At exterior columns, the edge beam-column should be considered as a

single element whose average stiffness ¾c was to be calculated thus:

= where

Kbc = stiffness of the beam (or slab) - column combination

m = a distributed torque applied along the axis of the edge member

1
 - 18 -

= rotation of end of column due to bending in column

= average rotation, due to twisting, of beam with respect to column.

In calculating et, the torque applied by the slab was to be assumed

to be linearly distributed along the edge member. Where there was no edge
beam, a portion of the slab equal to the width of the column was to be con-
sidered to offer torsional resistance.

Moments computed by the proposed method were compared with moments

measured in tests on five models, and in almost every case showed better
agreement with the measured moments than did moments calculated using the
1956 A.C.I. equivalent frame method.

3.2.4 The 1971 A.C.I. Code

The 1971 A.C.I. Code incorporated significant changes in flat slab

design rules, based largely on work carried out at the University of Illinois
in the 1960 1 s.

Two design methods were still specified: the equivalent frame method,
and the "direct design method", which was based on the former "empirical
method", but had less restrictive dimensional limitations so that it could
be applied to a greater range of structures; the panel length:width limitation
was extended from 1.33 to 2.0, and the ratio of successive spans from 1.2
to 1.33.
The formula for total panel moment, which had endured basically
unchanged for fifty years, and which had been much criticised for its fail-
ure to account for the full static moment, was abandoned in favour of;

M
0 where

w = applied unif~rm load per unit area

L = span transverse to direction for which moments are being calculated
2
L = clear span between column faces
n

If the column reaction were assumed to be concentrated at the corners

of the column, the total moment for an interior panel would be;

= w (L L 2
8 2 n
where:
 - 19 -

c1 = column dimension in'direction of span

c2 = column dimension in transverse direction

The code formula is an approximation to this expression, and is accurate to

within 5% for values of c /L smaller than 0.20.
1 1
Concentrating the reaction at the column corners is the most
unconservative assumption that can be made in deriving the total moment.
Perhaps the code committee, like its predecessor in 1921, was unwilling to
go too far in one step beyond existing engineering practice. Using the more
conservative assumption of shear forces distributed uniformly around the
column perimeter the expression for total moment would be;

M (1 - +
0

For square columns and square bays with c/L = 0.25 this expression
gives a value of M0 about 12% greater than the code formula. The difference
decreases with column size, and since the truth probably lies somewhere between
the two extreme assumptions about the distribution of shears around the column,
the error involved in the code formula should be small for modern flat plate
structures.

The new code stipulated that for internal panels 65% of the total
moment should be assigned to the negative moments, rather than the previous
code's 62%. For exterior panels the distribution was governed by a formula
which took into account the stiffness of the "equivalent column" in a manner
similar to that proposed by Corley. The division of the positive and negative
moments, into column strips and middle strips was essentially unchanged for
interior panels, but for exterior panels without edge beams 100% of the
negative moment was allocated to the column strip, on the basis of analytical
investigations into the elastic distribution of moments in flat slab floors.

The equivalent frame method in the 1971 Code followed the principles
of Carley's proposals, which were outlined in the previous section. The most
important change made was to extend to interior columns also Carley's device
of calculating column stiffness on the basis of the effective stiffness of a
slab-column combination in which the strip of slab between the columns could
rotate relative to the columns. This was found to be necessary in order to
allow more accurately for the effect of pattern loads on positive moments.

The critical section for negative moments was now taken as the face
of the column. Moments were distributed between column and middle strips in
the same proportions as for the direct design method.
 - 20 -

The revisions featured in the 1971 A.C.I. Code have made the design
of flat slah5 rather more complex, but there is no tluu!JL that design moments
calculated under the new provisions reflect much more accurately the elastic
distribution of moments for slabs under vertical loading.

3.3 Experimental

Any attempt to determine experimentally the distribution of bending

moments in continuous flat slab structures requires a major outlay of time,
money and laboratory resources, and few such attempts have been made. The
experimental work which has been published has been aimed chiefly at checking
the adequacy of design methods given in building codes.

Bowen and Shaffer (1955) developed an optical method called "photo-

reflective stress analysis", or the "Presan method", for measuring curvatures
in an acrylic sheet model. The method, although time-consuming and expensive,
could be applied generally, but the results given in the report applied only
to a square interior bay supported on circular columns with a c/L ratio of
0.063. These results showed general agreement with the distribution of moments
given in the 1951 A.C.I. Code; the biggest difference found was that the
measured column strip negative moments constituted just over SO% of the
total panel moment, compared with the code's 46%.

Higgins and Lin (1956) reported tests on a flat slab model consisting
of a 3/8" cast aluminium sheet with six 17" square panels arranged three by
two. Air pressure was used to apply uniform load to one panel at a time, and
readings from individual tests were superposed to give results for all panels
loaded. The authors' main conclusion was that the positive moments at midspan
were consistently higher than A.C.I. (318-51) moments, and that the code
positive moments should be multiplied by factors varying from 1.4 for column
strips in exterior panels, to 3.4 for middle strips in interior panels. However,
it should be observed that the A.C.I. analysis assumed that the remote ends of
the columns were fixed, whereas from the published diagram of the model it
seems that its columns were closer to a pinned than to a fixed condition.
This would increase the positive moments in the slab; the increase would tend
to accumulate as individual test results were superposed to obtain results
for multiple panel loading.

A major experimental investigation was carried out at the University

of Illinois in the period 1956-1963. (Ref: Sozen and Siess (1963); Hatcher,
Sozen and Siess (1961); Hatcher, Sozen and Siess (1965) .) The investigation
was initiated by the Joint ACI-ASCE Committee on Design of Reinforced Concrete
 - 21 -

Slabs with the aim of obtaining information on the strength and behaviour of
various types of multiple-panel floor slabs. Five test structun,, were made;
of which the first two, the flat plate and the flat slab, will be considered
here. Each structure comprised nine 5ft square panels arranged three by
three and was designed in accordance with the empirical method of the 1956
A.C.I. Code. They were model structures, in the sense that the designs were
for structures four times as large, with 20ft x 20 ft bays, and all dimensions
were scaled down by multiplying by 0.25. Coarse sand was used as aggregate
in the 1.75" thick slabs, and the reinforcement was cut from 1/8" square bars,
annealed to give stress-strain characteristics similar to intermediate grade
reinforcing bars and specially rusted to improve the bond performance. The
flat plate was designed as a typical apartment building floor with a live load
of 40 psf and an imposed dead load of 30 psf for partitions and finishes,
giving a total design load of 155 psf. The flat slab was designed for 'light
storage' loading of 200 psf plus dead weight 85psf giving 285 psf design load.
Edge beams were incorporated in both structures, and the flat slab had 10"
thick drop panels and column capitals at all except the corner columns.

In the design load tests the flat plate was hardly cracked, whereas
the flat slab was cracked extensively both top and bottom. Steel stresses
were generally about 4-5 ksi in the flat plate and 16-18 ksi in the flat slab.
The differences in behaviour of the two structures at design load arose from
the fact that the design load for the flat slab was about double that for the
flat plate, although both used the same slab thickness. Clearly a large
portion of the moment in the flat plate at this load was still being carried
by concrete tension in the largely uncracked slab. It should however be
noted that the modulus of rupture of the small aggregate concrete used was in
excess of 600 psi, much greater than would be expected in normal structural
concrete.

At a load of 225 psf on the flat plate, 45% higher than the design
load, localised peak stresses up to 31 ksi were recorded at the interior
columns, but steel stresses generally were only 15-18 ksi, still below the
design working stress of 20 ksi. Evidently even at this load a considerable
proportion of the moment in the flat plate was still being carried by concrete
tension.

Slab moments were deduced from measured steel strains by means of a

moment:steel strain relationship obtained from tests on concrete beams with
very low percentages of reinforcement. The A.C.I. design moments for the
interior panel were naturally concluded to be low, since the Code provided
for only 72% of statics. Otherwise the tests showed reasonable agreement
 - 22 -

between measured and A.C.I. moments except in the vicinity of the spandrel
l.H1a111s. Design moments were lower thall measured moments at the first interior
column line and much higher at the exterior column line.

As a check on size effects in the University of Illinois quarter-

scale models, a companion flat plate structure to 3/4 scale, with 15ft square
panels, was tested by the Portland Cement Association Research and Development
Laboratories in Skokie, Illinois (Ref: Guralnick and LaFraugh (1963)).

The behaviour of this structure under load was very similar to that
of the University of Illinois flat plate. At working loads the plate was
largely uncracked. In the centre panel the flexural rigidity appeared to
remain that of the gross section up to a load of about 255 psf. In no case
did the maximum steel stress at working load exceed half the design allowable
stress of 20 ksi.

Deflections were proportionally a little higher than in the smaller

model, presumably because the modulus of rupture of the concrete was lower.
The P. C. A. structure also had more cracks than the ¼ scale model, which was
to be expected since the bond properties· of the 12" deformed reinforcing bars
used must have been much superior to those of the plain rusted 1/8" square
bars used at the University of Illinois.

The distributions of moments at working loads were very similar to

those found in the¼ scale model. The authors observed that the measured
stresses in the negative reinforcement were much higher than those in the
positive steel, and that positive moment cracking was slight relative to
negative moment cracking, indicating a case for placing more of the total panel
reinforcement in the column strips over interior columns.

It is interesting that the authors judged it still necessary, in

1963, to open their concluding remarks with the following statement:
"In view of the agreement found between measured moments at service
load and the computed total static moments, it is confirmed that the equations
of statics are applicable to flat slab structures, as they obviously must be."

Following these one-quarter scale and three-quarter scale concrete

model tests, R.C. Elstner (1970) tested al/14th scale methyl-methacrylate
plastic model of the same structure, in order to explore the usefulness of
small-scale elastic models as a means of studying flat slab systems. The
agreement between the results from this model and those from the University
of Illinois tests was generally good, the moments measured on the plastic model
being on average about 5% on the low side. Scaled deflections also agreed
fairly well with the larger model.
 - 23 -

4. LATERAL LOADING

Sefore 1960, no significant attempt had been made to assess the

behaviour of flat plate buildings under the action of lateral loads. Blakey
(1962) wrote:

"Flat plate structures are not usually used in tall slender buildings
because it is believed that their lateral rigidity is inadequate unless special
stiffening frames or shear walls are provided. Whether this belief is accepted
or not, the occasion will arise when it is necessary to calculate the behaviour
under lateral load of an unstiffened flat plate structure, and it has been
found that this may also be done on the basis of elastic frame analysis. In
this analysis the 'beam' in the frame is taken to be a strip of the slab equal
to the width of the shear head if steel columns with grillage connections are
used, or a strip equal in width t6 the column plus three times the thickness of
the slab if concrete columns are used without shear heads. These strip widths
are chosen to provide the best estimate of lateral deflection.

"In the analysis, relative stiffness factors may be calculated from the
uncracked gross concrete sections for the 'beams'. The 'beam' moments should
not be regarded as confined only to the width of the 'beam' strip, but may be
added to the column strip moments."

Although these comments expressed views which were probably widely

held among structural engineers at the time, no basis for them in theory or test
has been found.

Beresford (1962) published some results from lateral load tests on

an experimental lightweight concrete flat plate structure in Australia. The
structure was not designed specifically for the lateral load tests, which const-
itute& only one of several series of tests on it, and it was not well suited
for providing information on the lateral load response of typical flat plate
structures. It had three 12' x 9' bays in each direction, and consisted of a
lightweight concrete slab supported on slender steel columns, with two types of
slab-column connections; it was therefore a rather unusual type of flat plate
structure.

Lateral loads of 3000 lbs were applied simultaneously at slab level

to each of the four columns on one of the longer sides of the structure.

The main emphasis of the report on the tests was in comparing the
experimental results with those obtained from an equivalent frame analysis,
according to the S.A.A. and A.C.I. building codes, in which the slab stiffness
was calculated on the basis of the gross moment of inertia of the full panel
width.
 - 24 -

Column moments, deduced from measured steel strains, were found to

be 10% to 20% greater than those calculated. This was attributed to the fact
that the uncracked section had been used in obtaining the slab stiffness.

In the attempt to find what proportion of the slab width, assumed

uncracked, would yield the correct values of moment and deflection when used in
a conventional rigid frame analysis, three such analyses were made, with as.sumed
slab widths of 2', 6 1 and 12' (the full panel width). It was found that the
width assumed made little difference, the resulting column moments being almost
constant for the three cases; the calculated deflection increased by less than
15% as the slab width decreased from 12' to 2' . Beresford concluded that
"the width of column strip which may be assumed is not critical as far as
assessing column moments is concerned". Nevertheless he recommended that
moments due to lateral loads should be determined using the column strip width
only, and that the moments thus determined should be added to the column strip
moments due to vertical loads.

The insensitivity of the column moments to the assumed value of the

slab width reflected the fact that the steel columns used were extremely flexible
relative to the slab. The columns were constructed from two 2\" x 2l211 x ¼"
boxed angles,-having a compound moment of inertia of only 2.25in 4 • Even for
the smallest slab width used in Beresford's analysis (2') the "equivalent
beam" EI was three times the column EI. Figure 9 shows that for EI ratios
greater than two little change in moment or deflection occurs. Had the structures
been built with concrete columns, say 9" x 9" with an EI value about twenty
times as great as that for the steel posts used, the moments and deflections
would have been very sensitive to quite small changes in the assumed effective
width of slab.

An investigation into the restraint offered by the connecting slab

to coupled shear walls in a regular multi-storey apartment building was reported
by Barnard and Schwaighofer (1967). The building was effectively infinite in
length. The shear walls occupied the entire width of the building except for
the central corridor which comprised 1/9 of the building width. The longitudinal
spacing of the shear walls was 4/9 of the building width. The authors wished
to establish what width of slab could be considered effective in coupling the
shear walls together.

The problem was investigated by testing a 22 storey model, consisting

of two shear walls¼" thick and 4 11 long separated by a 111 gap, and connected
by a floor slab l/8 11 thick, 911 long and initially 4 11 wide. Strains were
measured at the two extreme fibres of one of the shear walls. Lateral loads
were applied at every second storey.
 - 25 -

No attempt appears to have been made to restrain the slope of the

slab at its longitudinal edges, so that the relationship of the model tests to
the real situation is a little uncertain.

The method used to investigate the effective slab width was to test
the model with 4" wide slabs, then cut all the slabs to a smaller width, and
test again. The smallest slab width tested was 13/16". The reasoning behind
this procedure was that "if the entire width of the floor slabs is effective
in coupling the shear walls, then any reduction in the width should lead to
an increase in the wall stresses in the lower part of the structure".

The model test results showed that the extreme fibre stresses in the
wall increases by 15% from 540 to 620 psi as the width of the slab was decreased
by 80% from 4" to 0.81". This would seem to indicate that within this range
the wall stresses were little affected by the slab width, and it is difficult
to accept the authors' conclusion that the test results "show conclusively ....
that the entire slab width is to be considered as effective in coupling the
shear walls".

In the discussion, D. Michael of Ove Arup and Partners reported that

his investigation of the same problem had shown that "the dominant dimension
determining the effective slab width is the clear opening dimension between
the walls. Up to a slab width equal to the clear opening the slab is almost
fully effective. Additional widths of slab thereafter add decreasing amounts
to the slab stiffness."

5. STIFFNESS PROPERTIES

5. 1 Patel (1957)

The first known systematic attempt to gain information on the stiff-

ness of flat plates was undertaken by M.N. Patel in a doctoral program, completed
in 1957, whose purpose was "to study the interaction between a column and a
flat slab due to an unbalanced moment at the joint caused by any external or
internal load".

The problem was investigated experimentally by means of a 4' x 3'

stainless steel model with 12" square bays and 2. 7" dia., circular columns.
The model had fifteen column positions, five in the longitudinal and three in
the transverse directions; transversely the plate extended 6" beyond each outer
longitudinal row of columns, the edges being left free. Fourteen of the column
positions were permanently fixed, and consisted of steel washers clamped to a
 - 26 -

supporting plate. The central column joint on one of the transverse edges
could be rotated !Jy luatliug au aLLached ta'.titilever arni. Un this edge deflections
were prevented but rotations were not restrained except at the fixed column
positions. The edge therefore simulated a line of symmetry in a long building
in which rotation was applied to a typical interior column joint.

The measured moment per unit rotation of this column joint was found
to be 6.2Et 3 /12 in-lb. Doubling this value, and dividing by D, gives a non-
dimensional stiffness coefficient M/D0 = 11.3, applying for an interior slab-
column joint with a c/L ratio of 0.225, or 0.20 if the circular columns used
are converted to equivalent area squares.

Plate deflections were measured on a one-inch grid, but the accuracy

of the measurements was rather poor.

In parallel with the experimental work, finite difference analyses

were carried out. In these calculations, which were apparently done manually,
all panel edges we,re assumed to be fixed, except those contiguous to the
rotated column. Agreement between analysis and experiment was poor.

5.2 Brotchie and Russell (1964)

Brotchie and Russell quoted some results of calculations on the stiff-

ness of square internal panels in flat slabs supported on circular columns and
subjected to lateral loading. Details of the method of calculation were
contained in the Part Two Supplement, Appendix 4, of the reference article.

The stiffness figures tabulated in this article are plotted here as

Figure 10. It is important to note that the stiffness referred to is that of
an interior joint in an infinite structure, when an entire transverse· TO\< of
columns is rotated simultaneously.

5. 3 Carpenter (1965_)

5.3.1 Scope and Methods

In 1965, J. Carpenter presented in a doctoral thesis the results of

an analytical and experimental study of the elastic behaviour (stiffness
properties, distribution of moments and deflections) of flat plate structures
subjected to lateral loads.

The analytical work·was limited to the case of a typical square

interior panel in a regular structure with square columns. The experimental
work concentrated on the same case although some limited information was
 - 27 -

obtained on the behaviour of the plate around a centre column on the edge of
the structure.

The structure considered was a flat plate floor consisting of four

equal square bays in each direction. It was assumed that, if a rotation were
applied to the centre column in this structure while the other 24 columns were
held fixed against rotation, the behaviour of the plate around the centre
column would be independent of the edge conditions of the structure. The
floor was considered to be an intermediate floor in a multi-storey building,
with the points of inflection in the columns at mid-height, so that the loading
could be represented by a single force in the column stubs.

The experimental work was carried out on two Perspex models, which were
identical except for the size of the column cross-sections. Model A had !2" x ½"
columns, model B had 1" x 1" columns. The slab consisted of a¼" Perspex
sheet with four 9" x 9" bays in each direction. Columns were loaded by equal
and opposite forces at the ends of 3" column stubs above and below the slab.

Both models were instrumented similarly. A profile plotter was used

to measure deflections on a 14 x 14 grid in a panel. Surface strains were
measured at about 30 locations associated with each loaded column, using linear
electric resistance strain gauges aligned parallel to one or other of the column
lines. Stresscoat brittle lacquer was used in an attempt to determine visually
the direction of the principal strains in the vicinity of the columns.

5. 3. 2 Analytical Work

An analysis of the flat plate structure was attempted only for the
case of rotation of the typical interior column. The mathematical model used
was a .fixed-edge circular plate with a radius equal to twice the column spacing.
The centre of the plate had a rigid inclusion simulating the rotated joint.
The inclusion problem was solved by an approximate method involving complex
variable theory and ·conformal mapping. The solution process produced deflections
at the eight other column positions included within the circular plate; these
deflections were reduced to zero by applying point loads of appropriate magnit-
udes at the column corners.

A "reference structure" was defined with which the various properties

of the flat plate structure could be compared. The reference structure was
identical with the A. C. I. "equivalent frame" for elastic analysis, except
that the slab and columns were not considered to be infinitely stiff within
the joints.
 - 28 -

5.3.3 Experimental Results

Analytical and experimental results were compared, for the case of

rotation of the centre column, as a guide to the effectiveness of the mathe-
matical model in simulating the behaviour of the physical model.

Stiffness

The values of M/D0 for the two models are compared with the analytical
predictions in the table below. For the purpose of this comparison the stiff-
ness of the reference structure was adjusted thus:

M
=
4EI
E =
120(1-µ 2 )
I Lt3
=
8 L t3 72
Hence M = 4 (1-µ 2 )
00

For an interior panel this value must be doubled.

EXPERIMENTAL ANALYTICAL REFERENCE

RESULTS RESULTS STRUCTURE
Model A
½" X ½" 5.38 4.95 6.96
Columns

Model B
1 11 X 1 11 6.49 6.90 6.92
Columns

The predicted stiffness was 8½% less than the measured stiffness for
Model A, and 6% greater than the measured stiffness for Model B.

In each model loading tests were performed on two central edge columns.
One column had its outside face flush with the edge of the slab; the other
stood proud of the edge by a distance equal to one half of the column dimension.

The measu~ed stiffness of the panel containing the flush column was
82% of the interior panel stiffness for Model A, 96% for Model B. The measured
stiffness of the panel containing the exposed column was 69% of the interior
panel stiffness for Model A, 90% for Model B.

Carry-Over Factor

The experimental results indicated that almost all of the carried-over

moment was taken by the first longitudinal column. The design of the experiment
did not provide for straightforward measurement of the moment in the fixed
column, and considerable difficulty was experienced in deciding a value for this
 - 29 -

moment. In fact the analytical and measured values of the carry-over factors
varied by up to 30%, and there were ohvious inconsistenciez in the experimental
values, as shown in the table below:

CARRY-OVER FACTORS

MODEL A MODEL B

Interior Col. (Anal) 0.222 0.288

Interior Col. (Meas) 0.170 0.244
Flush Exterior Col. (Meas) 0.142 0.248
Exposed Exterior Col. (Meas) 0.189 0.227

The carry-over factor was defined as the ratio of the moment produced
in the first longitudinal fixed column to that fraction of the applied moment
which tended to rotate the fixed column, this fraction being one-half for an
interior column loaded, and unity for an exterior column loaded.

Plate Moments

The experimental values of plate moments were generally within about

20% of predicted values, where these were given for comparison. The distrib-
ution of moments in the immediate vicinity of the columns could not be obtained
from either the experimental results or the analysis. No information was
obtained on twisting moments.

5.3.4 Predicted Structural Behaviour of Interior Panels

In the final section of the thesis it was assumed that the derived
analytical expressions correctly predicted the behaviour of interior panels
around a loaded column. On this basis the conditions in an interior panel
were investigated for two loading cases:

1) Central Column ·only rotated, with all other columns fixed.

This case, which has already been discussed, gave information on
stiffness properties needed for frame analysis. Figure 11 compares Carpenter's
results with those published by two other research teams. In examining this
figure it should be remembered that all computed results apply to interior
panels only. The comparison with Brotchie's findings is based on Brotchie's
definition of stiffness in which an entire transverse row of columns is rotated
simultaneously. The results from Khan and Sbarounis were taken from model
tests on a single panel with a central column, with no provision for applying
moments at the transverse edges.
 - 30 -
,

The curve marked "Brotchie square" was obtained by considering as

equivalent, square and round colUmns of equal area.

In this figure 'c ' is one half of the column dimension in the
X
direction of the span 1 . All curves drawn apply to square interior panels.
X

2) Equal Rotation Applied to All Columns.

This case is of interest in giving information on the distribution

of moments in flat plate structures subjected to lateral loads.

All the work done on this case used a Poisson's Ratio of 0.35. The
results were all presented as ratios of the behaviour of the reference structure
Since the "equivalent beam" of the reference structure could not adequately
take into account the fact that the stiffness of the plate structure depended
on the column: span ratio c/L, or that the carry-over factors were quite
different from those applying to a prismatic member, the use of reference
structure served only to obscure the results, and detracted from the usefulness
of this section of the thesis.

5.4 Copley (1966)

An interesting approach to the problem of the stiffness of a typical

interior panel in a multi-storey flat plate building, when all columns are
subjected to the same rotation, was contained in a thesis by J. Copley in
1966.

Two approaches were used in this thesis:

1. In a chosen multi-storey building, the assumed effective width of

the slab was varied while column sizes and loading were kept constant, in
order.to probe how sensitive were the moments transferred between floors
and columns to change in plate stiffness;

2. The deformation characteristics of a plate under the action of an

applied moment were studied and compared with the deformation of a beam
under the same loading, in order to arrive at an effective width for the
plate. This study was limited to the case of typical interior bays in an
infinite building.

1) Equivalent Frame Analysis of Multi-Storey Building

The building chosen for analysis was twenty storeys high and two bays
wide. Columns were on 24' x 24 1 grid, and the floor to floor height was 10'0".
The plate thickness was 10". Column cross-section sizes varied from 10" x 10"
 - 31 -

at the top to 25" x 25" at the bottom of the building for exterior columns,
and from l 0" x 10" at the top to 35" x 35" at the bottom for intei-lu1· eolumns.
Design wind velocity was 90 m.p.h., resulting in a horizontal load per floor
of 5.84 kips. The design vertical live load was 100 psf.

This building was analysed by computer for five different assumed

effective plate widths, viz., full plate width, and½, 1/3, 1/6, and 1/12
of full plate width.

It was found that the changes in assumed effective plate widths

cause.cl only minor changes in the bending moments for floors above the 6th
floor. The mean results were obtained for the assumption of 1/3 of full plate
width, and this value was recommended for use for the upper floors of a
building. Copley commented:

"For the worst consideration of vertical and wind loading the moments
from widths B/2 and B/6 do not vary by more than 10% from the value for B/3.
Since this is less than the errors expected in the estimate of loading,
strength of materials, etc., this width could be taken for design purposes.

"However, for the lower six floors .... the effective plate width
is important in design and a reasonably close estimate of its value is
necessary."

2) Mathematical Analysis of a Plate

The effect of wind loading on the system was represented by an

infinite plate supported on columns each of which exerted a moment on the
plate. Initially the cross-sectional size of the columns was taken as zero.

The moment M applied by each column to the plate was replaced by a

couple consisting of two point loads (each P) acting at a distance c from the
column centre~line, as indicated in Figure 12.

Since the.building considered was regular and infinite in extent,

the lines x =0 and x = a, were lines of contraflexure where the moment and
the deflection were zero.

The analytical model therefore reduced to a rectangular plate of

infinite length with simply supported edges and loaded by a series of point
loads at a distance c from one edge.

The case of an infinitely long simply supported plate subjected to

a concentrated load at a distance c from an edge has been analysed by Timoshenko;
Copley was therefore able to obtain the solution for his analytical model by
superposition.
 - 32 -

For example, for Poisson's ratio= 0.1, B = 2a and c = 0.la, the

rotation of the plate at the column was found to be:

e = 0.431 P.a
E.t3
The slope obtained from beam theory, with the full width of the plate
assumed effective (I = 2a. t 3 /12) was:

e = 0.171 ~
E.t 3

Therefore for this case the effective plate width was obtained as 0.171/0.431 =
39.7% of the full width.

For other ratios of column spacing: span the results obtained were
as shown in Figure 13.

It can be seen that the effective width of the slab for stiffness
was fairly constant at about 40% of the span (not of the full plate width)
for ratios of column spacing: span greater than 3/4.

The effective width expressed in terms of the full plate width (B)
became very small as the ratio of full width: span increased beyond unity.
This was because the beam stiffness increased directly with plate width, but
the presence of the adjacent columns had very little effect on the stiffness
at the centre column. For columns arranged on a square grid, the plate stiff-
ness was only 1.1% greater than for the case where the column spacing was
infinite.
The effect of using a column of finite width, instead of zero width,
was then investigated. It was found that, for a square column grid, and a
column width equal to 0.la, the stiffness increased by 10% to 0.393P.a/E.t 3 •
Since·the increase in stiffness was small Copley recommended that it should
be neglected.

5.5 Qadeer and Stafford-Smith (1969)

Information relevant to the stiffness of flat plates was contained

in an article published in 1969 by Qadeer and Stafford-Smith.

This paper examined the resistance offered by a flat plate to rotation

of shear walls, with the object of determining values for the interaction
bending stiffness of the slab.

The structure analysed was an idealised regular building shown in

Figure 14. Most of the work dealt with a typical interior bay, ADCB, the slab
 - 33 -

being symmetrical and continuous across boundaries AB and CD. The object was
to find the slab stiffness for simultaneous equal rotation of all shear walls.
It was assumed that as the walls rotated, their sections at the slab remained
plane. A finite difference analysis was made for several cases with varying
values of C, W, and L (Fig, 14 (a)), The value of Poisson's ratio used in all
analyses was 0.19. The work is of interest because analyses were made using
values of the shear wall length W small enough to approximate longish columns
in a simple flat plate building.

In an attempt to obtain verification of the analytical results, a

small model was constructed. The slab was made from asbestos cement sheet,
and the walls from heavy steel plate, so that the inplane strains in the
walls would be small enough to have only a negligible effect on the measured
slab stiffness. Direct verification of the analytical results would have
required a model with provision for restraining the slope across the continuous
boundaries, Because this was difficult to achieve, the model tested had
all edges free, and an analysis was made for this condition using the finite
difference technique, The analytical and experimental results showed reason-
able agreement, and it was accordingly assumed that the same method of
analysis would give reliable solutions for the 'real' structure with continuous
edges.

Analytical Results
Stiffness

The slab stiffness was defined by the parameter k = M/D0, For the
smallest shear wall considered, W = 0. lSX, k was found to vary with the shape
of panel as shown in figure 15, For a square panel k = 4.4 approximately
for zero overhang.
Edge panels have a stiffness less than half that of interior panels
because the longitudinal edges are free except where connected to the shear
walls. Three cases of edge panels were analysed, as a result of which the
authors concluded that it would be sufficiently accurate to take the stiffness
of an edge bay as 42% of.the stiffness of an interior bay,

Effective Width

If an 'equivalent beam' be defined with stiffness EI, where

r = Yet /12, Ye being an effective width, such that the stiffness of the
3

equivalent beam is the same as the slab stiffness, it can be shown that:
 - 34 -

y
e
y= (Y = bay width)

The value of the slab stiffness parameter k was found from the
finite difference analysis. The variation of the effective width ratio Ye/Y
with shape of panel is shown in Figure 15, for the smallest shear wall size
W = 0.15X. It can be seen that the effective width for stiffness purposes was
found to be considerably smaller than the panel width. For the common case
of a square panel, the effective width was only 1/3 of the bay width; even
for the rather unusual case of span: width= 4:1, the effective width was
still only 80% of the bay width.

5.6 Aalami (1972)

Aalami considered the moment-rotation characteristics of square

plates supported on central square colwnns, and postulated that the rotational
stiffness must lie between two limits: in the upper limit the colwnn-plate
junction rotated as a rigid body; in the lower limit the colwnn was asswned
to add no bending stiffness to the plate, so that the deformation of the colwnn-
plate interface was a continuation of the deformation sufface of the plate
around the colwnn. Although the upper limit would appear to correspond nruch
more closely to the real situation, results for the lower limit only were
presented.
Three ratios of colwnn side, c, to plate size, L, were considered:
0,05, 0.10 and 0.15. Each was analysed by a finite difference procedure for
three asswned boundary conditions of the plate, shown in Figure 16. The
resulting stiffness coefficients M/D0 are tabulated below. For comparison
the stiffness coefficients calculated by Carpenter (1965) for rotation of an
interior square column joint (assumed rigid) in a continuous square bay plate
structure are also tabulated.

Case 1 Case 2 Case 3 Carpenter

Simply Supported Fixed Sway Condition Continuous

0,05 4.17 4.25 4.10 4.7

0.10 5.53 6.24 5.40 6.45
0,15 6.78 6,59 6.59

As would be expected, the results for Case 2, with all edges assumed
fixed, give the best agreement with Carpenter's figures for a continuous plate
structure.
 - 35 -

Aalami drew two conclusions from these figures, for c/L ratios up
to 0.15:
(i) that the stiffness was not sensitive to conditions at the boundaries;
(ii) that the stiffness was not significantly affected by the span of the
plate in either direction, but depended predominantly on the column
size alone.

Conclusion (i) is reasonable, although the boundary conditions

assumed did affect the stiffness values by up to 20%.
Conclusion (ii) is not valid, and could not in any case have followed
from the work presented in the paper since the same plate span was used in all
analyses. Other work indicates what would be expected a priori, that the
stiffness is a function of slab span as well as column size.

5.7 Summary of Stiffness Properties

5.7.1 Introduction

Evaluation of the published analytical and experimental data on plate

stiffness is complicated by the fact that stiffness has been defined in
different ways by different investigators, that some investigators have worked
with circular, others with square, columns, and that they have used materials
with different values of Poisson's Ratio.

In order to enable comparisons to be made, the published results

will be adjusted, where necessary, to apply to square columns, and the effect
of Poisson's ratio will be eliminated, thus:

Shape of Column
Where circular colrnnns have been used, the side dimension "c" for
square columns of equal area wi 11 be substituted; c = 0. 885x diameter of
circular column.

Poisson's Ratio (µ)

It will be assumed· that the stiffness of the slab panels is proport-

ional to the 'plate stiffness' D
3
E. t

The panel stiffnesses will be discussed in terms of the dimensionless

parameter M/D0, where Mis the total moment applied to the slab at a column,
and e is the resulting slope at the column. With the assumption above, M/D0 is
independent ofµ.
 - 36 -

All the work summarised here deals with the stiffness of interior
panels of regular multi-panel flat plate floors. No information is available
on the stiffness of edge or corner panels, apart from the limited work of
Qadeer and Smith, who analysed a central panel in an infinitely long building
one panel wide (Section 5.5).

5.7.2 Square Interior Panels

1. Stiffness defined in terms of the rotation of one interior column,

while all other columns are held against rotation.
This definition was used by Patel (1957) and Carpenter (1965).
Results from their model tests, and from Carpenter's analytical work,
are plotted in figure 1·;.

The figure indicates the trend of increasing panel stiffness with

increase in c/L ratio. However it is clear that more work is needed to
verify and extend the range of the stiffness values.

Carry-over Factors

Patel and Carpenter agreed that the only column which received a
significant carry-over moment was the first longitudinal fixed column in the
direction of the applied moment. The actual value of the carry-over factor
for moment to this column is uncertain. Carpenter's results suggested that
the factor was about¼ for a c/L ratio of 0.10, and increased with increasing
c/L.
Carpenter also stated that carry-over factors to the other fixed
columns had been found, in calculations not presented, to be about 0.035 to
the diagonal column, and -0.045 to the transverse column.

**2. Stiffness defined in terms of the rotation of an entire transverse row

of columns, adjacent rows being held fixed against rotation.

Brotchie defined stiffness in this manner, and Carpenter adapted his

analysis to this definition for comparison with Brotchie' s work.

The results of the two analyses are compared in Figure 18. Within
the range of c/L common to both there is some divergence between the trends
of the two graphs. No experimental check is available for this case.

**3, Stiffness defined in terms of equal rotation of every column.

This was the case investigated by Copley.

 - 37 -

For square panels of side L, and a column side dimension c in the

direction of the applied rotation, of O.lOL, Copley's analysis gave the
following values for interior panel stiffness:

For column width= zero, M/D0 = 5.52

For column width= 0.05L, M/D0= 6.05

It may be surmised that for a square column, width equal to O.lOL, the stiffness
would have been about
M/D0 = 6.6

This compares with a value of 6.5 for stiffness definition 1 (see

Figure 17), and 5.5 for stiffness definition 2 (see Figure 18).

5.7.3 Rectangular Interior Panels

Copley's analysis indicated that panel stiffness, for all columns

rotated, remained practically constant for any value of panel width B greater
than the panel span Lin the direction of the applied rotations.

For values of B/L greater than 0.8, the effective slab width for
stiffness in an equivalent frame analysis, was a function of the span rather
than the width of the panel (see Figure 13).

6. CONCLUSION

6.1 Vertical Loading

In the history of the development of flat slab design for uniform

vertical loading, four major milestones are discernible:

1) The derivation by Nichols in 1914 of an expression for the total static

momen~ in an interior panel;
2) The determination by Westergaard and Slater in 1921 of the distribution
of moments within a panel;
3) The devising 0£ an equivalent frame design method by Dewell and Hammill
in the early 1930's;
4) The extensive programme of experimental and analytical work commenced
at the University of Illinois in 1956, leading to the improved and ration-
alised design methods incorporated in the 1971 A.C.I. Code.

Although new analytical techniques for the determination of slab

bending moments have been devised since Westergaard and Slater published the
first solution in 1921, none has been useful as a design tool. Very few
solutions for particular cases have been published and these have almost all
 - 38 -

been only for the simplest case of an interior panel in a uniformly loaded
structure,

The design methods incorporated in current Codes are adequate for

regular structures under uniform load, such as the 3 x 3 bay structures
studied in the University of Illinois programme. Their adequacy is less
certain for partial loading, and their applicability to cases of line loads.
or concentrated loads has not been explored. Their usefulness for structures
in which, due to irregular layout or uneven loading, column-slab joint
rotations are significant, is very doubtful.

6.2 Lateral Loading

The design of flat slab structures for lateral loads to-day is
shrouded in ignorance as complete as that which marked design for vertical
loads in C.A.P. Turner's day. Usable information on stiffness properties
is virtually restricted to the limited data obtained for square interior
panels by Carpenter (1965) and Patel (1957) shown in Figure 17. No reliable
data are available on the stiffness of edge or corner panels, or on the
moments carried over to adjacent columns. Clearly much work is needed before
a rational method of design for lateral loading can be devised.

ACKNOWLEDGEMENT
The author is pleased to acknowledge the financial support
provi~ed by the Australia Research Grants Committee for a research
programme into the design of flat slab structures.
 - 39 -

7, REFERENCES

References are listed in alphabetical order of authors' names,

AALAMI, B. (1972) "Moment-Rotation Relation Between Column and Slab", Journal

of the American Concrete Institute, May, 1972, Title No. 69-27 pp, 263-269,

A.C. I, COMMITTEE 442, (1971) "Response of Buildings to Lateral Forces",

A,C, !. Committee 442 - Mark Fintel, Chairman, Journal of the American
Concrete Institute, February, 1971, pp. 81-106,

AMERICAN CONCRETE INSTITUTE ( 1971) "Building Code Requirements for Reinforced

Concrete", (A,C.I. 318-71),

ANG, A, (1959) "The Development of a Distribution Procedure for the Analysis

of Continuous Rectangular Plates". University of Illinois, Civil Engineering
Studies, Structural Research Series No. 176, May, 1959.

BARNARD, P.R. & SCHWAIGHOFER, J. (1969) "The Interaction of Shear Walls

Connected Solely Through Slabs", Proceedings of the Symposium on Tall
Buildings, University of Southampton, April, 1966.

BERESFORD, F. P, (1962) "Experimental Lightweight Flat Plate Structures Part V

- Deformation under Lateral Load". Constructional Review, Dec,, 1962, pp. 17-23,

BILL, MAX (1969) "Robert Maillart - Bridges and Constructions", Pall Mall
Press, London 1969. (Contains numerous passages from the writings of R, Maillart),

BLAKEY, F.A. (1962) "The Design of Flat Plates by Simple Analysis"

Constructional Review. November, 1962.

BLAKEY, F,A. (1963) "Australian Experiments with Flat Plates". Journal of the
American Concrete Institute, Proceedings Vol. 60 No. 4 April, 1963. pp. 515-525,

BOWEN, G. & SHAFFER; R. (1955) "Flat Slab Solved by Model Analysis" Journal of
the A,C.I,, Vol. 26, No, 6, February, 1955. pp. 553-570,

BROTCHIE, J .F. (1957) "General Method for Analysis of Flat Slabs and Plates".
Journal of the A.C.I. Vol. 29, No. 1, July, 1957. pp. 31-50.

BROTCHIE, J, F. (1959) "General Elastic Analysis of Flat Slabs and Plates".

Journal of the A,C.I., August, 1959,

BROTCHIE, J .F, (1963) "A Refined Theory for Reinforced Concrete Slabs",
Journal of the Institution of Engineers, Aust,, October-November, 1963,
 - 40 -

BROTCHIE, J.P. & RUSSELL, J.J. (1964) "Flat Plate Structures, 1. Elastic-
Plastic Analysis". Journal of the A.C.I. Vol. 61, No. 8, Aug. 1964, pp.959-995.

BRAYTON, L.F. (1910) "Methods for the Computation of Reinforced Concrete Flat
Slabs". Engineering News Vol. 64, No. 8, Augi,ist, 1910, pp. 210-211.

CARPENTER, J.E. (9165) "Flexural Characteristics of Flat Plate Floors in

Building Subjected to Lateral Loads". Ph.D. Thesis, Purdue Uni. , 1965.

COPLEY, J.C. (1966) "The Investigation of Moment Transfer Between the Columns
and Plates in Tall Flat Plate Buildings without Shear Walls". Master of
Technology Thesis, University of N.S.W., 1966.

CORLEY, W. G. & JI SRA, J. D. (1970) "Equivalent Frame Analysis for Slab Design".
Journal of the A.C.I. November, 1970. pp. 875-884.

CORLEY, W.G., SOZEN, M.A. & SIESS, C.P. (1961) "The Equivalent Frame Analysis
for Reinforced Concrete Slabs". University of Illinois, Civ. Eng. Studies,
Structural Research Series No. 218. June, 1961.

DEWELL, H.D. &HAMMILL, H.B. (1938) "Flat Slabs and Supporting Columns and
Walls Designed as Indeterminate Structural Frames". Journal of the A.C.I.,
Proceedings Vol. 34, January-February, 1938. pp. 321-343.

DIVISION OF BUILDING RESEARCH, C. S. I. R.O. (1961) "Experimental Lightweight

Flat Plate Structure". Constructional Review Jan. 1961, pp. 21-32.

EDDY, H. T. (1914) "Steel Stresses in Flat Slabs". Transactions of the

American Society of Civil Engineers. Vol. 77, pp. 1338-1453.

ELSTNER, R. C. (1966) "Tests of Elastic Models of Flat Plate and Flat Slab
Floor System". 'Models for Concrete Structure' - A.C.I., Special Publication
No. 24, Paper SP 24-12, pp. 289-320.

GUPTA, K.K. &VAUGHAN, R.C. (1967) "Bending of Uniformly Loaded Column-

Supported Flat Plates". Journal of Strain Analysis, Vol. 2, No. 2, 1967,
pp. 109 - 116.
 - 41 -

GURALNICK, S .A. & LaFRAUGH, W. (1963) "Laboratory Study of a 45-Foot Square

11
F.lat Plate Str11rt11re • ,Tournal of the A.C.L Sept, 1963, pp. 1107-118~.

HATCHER, D.S., SOZEN, M.A. & SIESS, C.P. (1961) "A Study of Tests on a Flat
Plate and A Flat Slab". Uni. of Illinois, Civ. Eng. Studies Struct. Research
Series. No. 217, July 1961.

HATCHER, D.S., SOZEN, M.A. & SIESS, C.P. (1965) "Tests of a Reinforced Concrete
Flat Plate". Proceedings of the ASCE, Vol. 91, No.ST5, Journal of Struct. Div.
October, 1965, pp. 205-231.

HUGGINS, M. W. & LIN, W. L. (1956) "Moments in Flat Slabs". Proceedings of the

ASCE, Journal of the Struct. Div., Paper 1020, July, pp. 1020-1 to 1020-30.

LORD, A.R. (1911) "A Test of a Flat Slab Floor in a Reinforced Concrete Building",
Proceedings of the National Association of Cement Users (ACI) Vol. 7, pp. 156-
179.

MACMILLAN, A. B. (1910) "AComparison of Methods of Computing the Strength of

Flat Reinforced-Concrete Plates". Eng. News. Vol. 63, No. 13, March, 1910,
pp. 364-367.

NEWMARK, N.M. (1941) "Note on Calculation of Influence Surfaces in Plates by

Use of Finite Difference Equations". Journal of Applied Mechanics, Vol. 8,
No. 2, June 1941, p. A-92.

NICHOLS, J.R. (1914) "Statical Limitations upon the Steel Requirements in

Reinforced Concrete Flat Slab Floors". Transactions of the ASCE, Paper
No. 1309, Vol. 77, pp. 1670-1736.

PATEL; M.N. (1957) "Interaction of Flat Slabs and Columns". Ph.D Thesis,
School for Advanced Studies, Michigan State Uni. of Agriculture and Applied
Science 1957.

QADEER, A. & STAFFORD.-.SMITH, B.S. (1969) "The Bending Stiffness of Slabs

Connecting Shear Walls". Journal of the ACI Title No. 66-37, June 1969,
pp.464-473.

ROZVANY &HAMPSON (1962) "Multi-Storey Frame Analysis for Vertical Loading".

Journal of the ACI, July, 1962, pp. 959-965,
 - 42 -

RUSSELL, J.J. (1966) "Analytical Data For Direct Design of Reinforcement

Distribution in Internal Panels of Flat Plate Structures". Commonwealth
Scientific & Industrial Research Organisation, Australia, Division of
Building Research, Technical Paper No. 20, 1966.

SOZEN, M.A. & SIESS, C.P. (1963) "Investigation of Multiple-Panel Reinforced

Concrete Floor Slabs". Journal of the American Concrete Institute (Proceedings
Vol. 60, No. 8) pp. 999-1027, August, 1963.

STANDARDS ASSOCIATION OF AUSTRALIA (1963) SM Code for Concrete in Buildings.

AS'CA2-1963.

TURNER, C .A.P. (1905) Discussion of "Reinforced Concrete Warehouse for North-

West Knitting Co. Mineapolis, Minnesota". Engineering News, Vol. 54, No. 15,
October 12, 1905. pp. 383-384.

WESTERGAARD, H.M. & SLATER, IV.A. (1921) "Moments and Stresses in Slabs"
Journal of the A.C.I. Vol. 17, 1921, pp. 415-538.

WOINOWSKY-KRIEGER, S. (1954) "On Bending of a Flat Slab Supported by Square-

Shaped Columns and Clamped". Journal of Applied Mechanics Vol. 21, September
1954, pp. 263-270.

WOODRING, R. E. & SIESS, C. P. (1968) "Influence Surfaces for Continuous

Plates". Proceedings of the American Society of Civil Engineers, Journal
of the Structural Division, STl, Jan. 1968, pp. 211-226.
 i<··-·-A!Jl71..._ ... . --·,-.el7!-,.. ->i, ....-AJI>!..____ ., Top View af Column Froming,

f
'
1 -
',•,;•,",'N• •
,,_.,..

....
Nlwt,

~-
FIG. 3. "MUSHROOM" SYSTEM OF CONCRETE REINFORCEMENT PROPOSED BY C. A. P.
Soctiono.l Elevation.

TURNER.

EARLIEST KNOWN FLAT SLAB DRAWINGS

ENGINEERING NEWS, OCTOBER 12, 1905.

Figure 1.
 MAILLART TEST SLABS 1908

Figure 2,

--------------

MAILLART NINE-BAY TEST FLOOR 1908

Figure 3.
MAILLART EXPERIMENTAL MUSHROOM-
SLAB BUILDING 1910.
Figure. 4.
FLAT SLAB BUILDING
FOR FEDERAL GRAIN
STOREHOUSE, ALTDORF,
SWITZERLAND, 1912,
DESIGNED AND BUILT
BY ROBERT MAILLART
Figure 5.
 L

D
I
I
Y---'

C B

~r---+--1~\
X
4r~Resultont of Load w Acting
3lt on Column Q.uadront

7 ~+++tttlls
C
w ~psf]

Mc(.........._ _........,) Ms
J --Resultant of Uniform Shears
~ Around Arc
2 r0
n:

NICHOLS FREE BODY DIAGRAM

( From Nichols 1914)

Figure 6.
 ca!,

-.10

-1:\
~ cl
~
'ls ,14
I
1 I
I
~ -.16
a/Mil

F1<:. 14, C'oEFFll'IE:,;'J's OF BEN!JlNtl !l!U~IE:'1T8 l'Elt l ':,;n· \\'wn1 J:,; .\

,:,11· IHI·: I ,n:1tJOR !'A:-;Er, ()f' A l'N!FOR~!LY Lo.\J>E(J FI.\T :,S! ..\11 \\'1n::,;
J 1t1I~:-:11'.\ ·:.. ({.\TIO 1~ Zr:Ro; .:\lo~n:XTH Acnos:s T1n: E1Jta; .\~P Tlfh
('1:,n:1t l.1,·i,:.

BENDING MOMENT DISTRIBUTIONS

FROM WESTERGAARD &SLATER
Figure 7
F.H;. ] 5.-t'oEFJ: H'lE.XT:S OF BEXDlXG }.iO:.U.!::.:'.;'TS PER L"~IT \\' l!JTH J~ ,\
::--:1.g:Al:E ];'1 l'EIUUH P~\:,;EL Of' A l'.XIFOR~LY LO.\UED 1"LA'£ SL.\.H \\°HE:'i
Pors;-;u)l'.·s H_\'fIO IS ZLRO; ~IO:,\f£XT$ A.LO.:'iG THE E.:LJGE A~1J THE: C1-:~n·:1t
LI:-iL
BENDING MOMENT DISTRIBUTIONS
FROM WESTERGAARD & SLATER

Figure 8
 3k A 8 C

25 ---1 1' I ls'9

,
9
,
9
,

MA
20 1·0

0-9
·-
~
C
I
0·8
Ms
....C I5
0·7 -
QJ C
E
0
~ 0·6 <(

Ol
....CJ
C
·- 10 Mc 0·5 C
"O 0
C
QJ
ro \ 0·4 u
~,...,

" .___
--- oA
--_.._-=-=- 0·3
-
o
Q;

QJ

5
0·2

0·1
Assumed Effective Slab Width (ft.)
1 2 3 la 5
0
1 2 3 4 5 6 7 8

Ratio EI (Slab)
EI (Column)

EQUIVALENT FRAME ANALYSIS

OF BERESFORD TEST FRAME.

Figure 9.
 SQUARE PANELS,
12 CIRCULAR COLUMNS, µ =0

10

-oeM
6

STIFFNESS GIVEN APPLIES

4 TO THE CASE OF AN ENTIRE
TRANSVERSE ROW OF COLUMNS
ROTATED SIMULTANEOUSLY.

0 0·04 0·08 0·12 0·16 0·20 0·24 0·28 0·32

C / L

BROTCHIE AND RUSSELL STIFFNESS ANALYSIS

Figure 10.
 1·1 - - - - - - , . - - - - - - - . - - - - ~ - - - ~ - - - - - .

5
(l)
1·0 7 -;;;
("t)

en 0·9
0
(l)
I.)
6 ~
I..
C 0·8
...
(l) ....
0

-a.,
( l)
0::
0 ·7

06
5-
(])
0
'-
4"2:. /
/
...-computed Results
Brotchie D efinltion
0
...., Khan, Sbarounis
/ /
(l)
0·5 / V =0·30
> 3 /
..... 0·4
0
(l)
0::
0·3 KEY - Experimental Values
2
(/) V = 0•35
(/) 0·2
Interior

-....
(l) 0
C 1
0· 1 b Flush Exterior
.....
V1
• Exposed Exterior

0 0·02 0•04 0·06 0·08 0·10

Column Size Ratio ex/I x

STIFFNESS OF SLAB ELEMENT

V = Poisson's Ratio ex : 112 Column Dimension
(From Carpenter (1965)

Figure 11.
 ~ ~ "p
I I
2a I ct- SIMPLY
SIJPf'IORT-
,. lp I ~1 lp a· a ED EDGES

I I y
I
PWJ I pWJ p.,..._x

COPLEY ANALYTICAL MODEL

Figure 12.

%
C 80
E
::J
0
u
..,_
O 60
~
0
ti)
0 Ol

..., ·-u
.C
-0
C
40
In Terms of Span - 40•14 %
·- 0
:3: 0.

....a,0 ti)

a.. 20
a,
.....(.)>
....,._
QJ
o,..____.,____.,____.,___________,
LLl
½- ¾ 1 2
Column Spacing Normal to Load
Ratio
Column Spacing in Direction of load

COPLEY EFFECTIVE WIDTH ANALYSIS

Figure 13.
 0

. . - R
'

I
- ,--•
1
.
,f_, '
-
2
8

C
--~J -
I

. A

Le-I: w .I. ~ ,I. w :l·c.l

V

(a )

f l l t l f f f l f f , , ffff1/l1//f11uu11/

( b)
Slab and Cross-Wall Structure;
a)Plan: and bl Elevation.

STRUCTURE ANALYSED BY QADEER & SMITH (1969)

Figure 14.
 - -
~ -±½0/1///2- J
f.-------- _______ l
_ ~- ~sx
O·f 5
-
2<_j
X

9 0·9

8 0·8

7 0·7 (
~
~
6 0·6 .c:
(/)
......
(/) "'O
(I/
5 0-5 :?:
........
C
~
...., 4 0·4 :;;
1./)
u
.&
.....
3 0,3 LU

2 0·2

1 0·1

0 0·2 0·4 0·6 0·8 1·0

Y/X
QADEER AND SMITH ANALYSIS

Figure 15.
 SIMPLE CLAMPED UNSUPPORTED

--1--y 11 SUP P.,.O-R Tii-S=:I=:=;, I ~ ERO ROT: Tl 0:

fu·~J u'r L.
1
I
I ' "'
a x 1 ~- a x a u 1 l__ x
j_ M, TM, L ---
TM, '~~~
-7 M,-,
I- a --l I- a ---1 I- a --I s IM p L E
SUPPORT
~
'3M,
SIMPLY CLAMPED SWAY
SUPPORTED EDGES CONDIT ION
CASE 1 CASE 2 CASE 3
Fig. 2. Square Plates on Central Columns; Dimensions and Boundary
Conditions of Three Cases Analysed.

BOUNDARY CONDITION.S USED IN ANALYSES

BY AALAMI (1972)

Figure 16.
 12
PATEL-a
( Exp.)

10

CD (Analytical)
Cl CARPENTER
' 6 (Exp.) \
~
[J

0 0·04 0·08 0·12 0·16 0·20

c/L
Note: M is the Moment Applied -to and 0 the Resulting Slope
at on Interior Column Joint when all other Joints ore
Held against Rotation.

STIFFNESS OF INTERIOR SQUARE PANELS IN REGULAR

FLAT SLAB FLOORS SUPPORTED ON SQ.UARE COLUMNS

Fi£! ure 17.

 12

10

(Analytical)
8

6
~CARPENTER
~1/ (Analytical l

0 0·08 0·16 0·24 0·3"2

c /L
Note: M is the Moment Applied to and 8 the Resulting
Rotation at each Interior Joint in an Entire
Trans1terse Row, Other Joints be"ing Held Fixed.

"BROTCHIE STIFFNESS~ OF INTERIOR SQUARE PANELS IN

REGULAR FLAT SLABS SUPPORTED ON 50.UARE COLUMNS.

Figure 18.