Title no.

85-553

Comparison of Analysis Procedures for Two-way slabs

@ffi

by Mary Theresa Cano and Richard E. Klingner

Two-way reinforced concrete slabs act with columns and walls to

Jorm structural systems Jor resisting graviry and loteral loads. Current analysis approaches for such systems usually involve finite elements or equivalent frames (effective beam widths or equivalent
Jrame properties). Each approach has advantages and disadvantages.
As currently used, neither is completely suitoble for anolyzing twoway slgQ systems under combined gravity and lateral loads.
Thi*development, advantdges, and limitatiorc of each approach arg.*
discussed, with emphosis on the equivalent frame approach. AkY
equivalenl frame analysis method is proposed that iyvolves explicit
modeling of attached lronsverse torsional membersQateral deflections calculated by various slab onalysis methods aie compared with
publishedlgcerimental results Jor a multistory slab system under loteral loads.'Slab moments calculated by variow slab analysis methods
ore compared with each.o\her for idealized flat-plate ond two-way
slab-on-beam structurei;lihe explicit transyerse torsionol member
method is found to give good results for drifts and slab-biam actibrls
and is recommended for analysis and design of tio-way slab systems
under combined gravity and lateral loads.

Keywords: concrete slabs; lateral pressure; reinforced concrete; structural analysis; structural design; two-way slabs.

A common structural engineering problem is the design of two-way, reinforced concrete slab systems-flat
plates, flat slabs, and two-way slabs on beams. Starting

with a trial slab thickness based on deflection or

punching shear considerations, the designer must provide for satisfactory strength and stiffness under combinations of gravity and lateral loads. This rbquires that
actions within the slab system be computed.
Designers differ as to the best procedures for this.l
Available approaches include those involving finite ele-

ments and those involving equivalent frames. These

approaches can produce widely differing results, and
each has advantages and disadvantages. Thus, it is
sometimes

difficult for a designer to

select an appro-

priate analysis method and interpret the results for de-

OBJECTIVES AND SCOPE

The objectives of this paper are:
l. To review available analysis approaches for twoway slabs.
ACI Structural Journal

November-December 1988

2. To discuss co_mpulel:aided m_ejhoels based on each

approach.
3. To compare _numerr_qa! les,qllq from each method.
4. To recommend analysis methods for two-way slab
systems.

This study concerns two-way slab systems of reinforced concrete under gravity and lateral loads. It is
limited to analysis methods that are readily adaptable
to computer-aided solutions. Methods based on the
equivalent frame concept are emphasized; yield-line
methods2 and strip design methods3 are not covered.
Discussion is confined to obtaining design moments;
steel placement and minimum reinforcement requirements are not addressed.
ANALYSIS APPROACHES FOR TWO.WAY SLAB
SYSTEMS
Behavior of two-way slab systems under gravity and
lateral loads is complex. Unlike planar frames, in which
beam moments are transferred directly to columns, slab
moments are transferred indirectly, due to the.t_ots,ion"g!
fleailihly-,of the slab. Also, slab moments from gravity
Ioads can "leak" from loaded to unloaded spans; this
must be accounted foi in inatviii. fhJ need to model
torsional flexibility and moment leakag6. has given rise
to two main analysis approaches for two-way slabs:

those involving finite elements, and those involving

equivalent frames (using effective slab widths or equivalent frame properties).

Finite element approach

Slab behavioral modes can be modeled directly using
finite element methods, typically involving plate bending elements.a,s Because many elements are usually required to achieve.good results, finite element approaches are expensive for large structures. Also, the
_ Received-Sept. 10, 1987, and reviewed under Institute publication policies.
Copyright @ 1988, American Concrete Institute. All righti reserved, iricluding
the making of co-pies unless.permission is obtained from the copyright proprietors. Pertinent discussion will be published in the September-Octbbei ISES ACt
Structural Journal if re'ceived by May l, 1989.

597

ACI member Mary Theresa Cano receiyed a BS in Architecturul Engineering

and an MS in Civil Engineering from the IJniversity of Texas at Austin. Ms.
Cano has worked as a design engineer, and is currently an engineer wilh the
Bridge Division of the Texos State Department of Highways and public Trans_

portation, Austin.

ACI member Richard E. Klingner

is an associate professor o! civil engineering,

The University of Texas at Austin. He is a member of ACI Committees 531,
Concrete Masonry Struclures; 349, Concrete Nuclear Structwes; and joint ACIASCE Committee 442, Response of Concrete Buildings to Loterol Forces.

applicability of linear elastic analysis is questionable

when calculated slab stresses exceed cracking values.
For these reasons, direct use of finite element approaches is not discussed further here.
Equivalent frame approach
To reduce the complexity and cost often associated
with finite element analyses, the equivalent frame approach can be used indirectly to compute equivalent
beam widths or equivalent frame properties. In this approach, a three-dimensional slab structure is idealized
as two independent sets of parallel planar frames,
crossing each other (usually at right angles). This general classification should not be confused with the specific analysis method known as the ACI equivalent
frame method, to be discussed later in this paper.
Two familiar examples of the equivalent frame approach are the effective beam width and the transverse
torqional member procedures.. -

..-_

.ljfegqjve b-eam-width procedurA.;-The effective

beam-width procedure was developed for analyzing
two-way slab systems subjected tojglgt4l,lggdq,and has
been used primarily for flat slabs and fla-t plates. This
method incorporates the effects of slab torsional flexibility, but,aet _BgirrgnifQkece. an errectiv; ;id;h- factor cy is obtained such that a slab of effective width culr,
subjected to uniform support rotation 0, would have a
total support moment equal to that of the original slab
(width lr, varying 0). Once the effective beam width is

determined, a conventional planar frame analysis is

carried out.

Effective beam widths so derived depend on the assumed stiffnesses of the columns and of the beam-column connection regions. Typical of the results of such
methods are the effective widths obtained by Khan and
Sbarounis.6 Though strictly applicable only to slabs
with boundary conditions and cracking consistent with
the assumptions of their original derivations, such results are qften used for a wide-r4!ge__glcases.
-Tronsverse

tirs|onat

mem_uei

iiiiii,ii.*rn" t urt-

verse torsionai mem6er pro..dur. was developed following extensive testing of two-way slabs.T-e Those por-

tions of the slab attached to the columns and transto the direction of the span (plus the transverse
beams, if any), are assumed to act as transverse torsional members, transferring moments from slabs to
columns. These transverse members are assumed rigid
except in torsion. Moment transfer is treated as occurring directly over the column width c, and along the
torsional members. The rotational stiffness of the joint
verse

598

is determined as a function of the torsional stiffnesses

of the transverse members on each sid-e of the joint and

of the flexural stiffnesses of the columns above and below the joint.
DESIGN METHODS BASED ON THE
TRANSVERSE TORSIONAL MEMBER
PROCEDURE
The transverse torsional member procedure accounts
both for slab torsional flexibility
"rrd -o.n"n'i*ffi["
and hai bbtir indorpoi6lefi nto'seneiai
i[eCi]iC d.esiin
methods. Two of these are the ACI equivalent frame

method (ACI EFM),,o and the extended equivalent

frame method.rr-r3 A new method, termed the explicit
transverse torsional member method,ta is also presented. In all such methods, member actions are computed, distributed to column and middle strips, and
then used for slab design.
AC! equivatent frame method
The ACI EFM'o first requires thar the building be
idealized as a series of equivalent planar frames (Fig.
1). The actual three-dimensional frame is assumed to be
composed o{_{gb:!ggm; (horizontal elements with
flexural stiffness I(,) supported on an assemblage of
columns (vertical elements with flexural stiffness K")
ana qransyerse torsional members (horizontal elements
with torsional stiffn6ss K).' ?iie'equivalent planar frame
is composed of slab-beams (horizontal elements with
flexural stiffness .&',) supported by equivalent columns
(vertical elements with flexural stiffness _r(,", defined as
follows)

(1/K*): (t/DK) + (t/K,)

(t)

This notation conforms to that of ACI3l8-83.'o The

flexibility of the equivalent column is the sum of the
flexibilities of the actual columns and attached transverse torsional members. Required member stiffnesses
K,, K", and K, are defined as follows.to
Torsional member stiffness K,-The transverse torsional member concept was first proposed by Corley.T
Moment transferred frorn slab to column was originally assumed to be uniformly distributed across the
width of the slab. Jirsa later proposed8 a triangular
moment variation (maximum intensity over the column, zero at each edge of the equivalent frame). The
corresponding torsional stiffness K, was then obtained
by approximate procedurese

K,:

DgEC/lrU

- @r/l)13

A)

For slabs with beams, K, lBq.(2)l is increased by the

factor I"u/1,, using the notation of ACI 318-83.t0 Eq. (2)
for K, was intended to apply to cracked slabse and was
calibrated using the results of gravity-load tests.e,15'r6
Column stiffness &-K" is independent of K, and is
calculated conventionally, using the actual column mo-

ment of inertia between the slabs, and an infinite moment of inertia within the slabs.
ACI Structural Journal

November-December 1g88

lzl

/z

lz//2

(a) Definition

-n_

of equivalent frame-plan

vrcw

(b) Members of three-dimensionol structure, Detait A

K""
(c) Members of equivalent frame, Detail A

Fig. |-Member con/igurations assumed in ACI equivalent frame method

Slab stiffness K.-Slab stiffness is calculated conventionally, including the effects of column capitals and
drop panels. The moment of inertia of that portion between the ienter of column and face of column,
bracket, or capital is then incieased by ihe factor l/(l
- cr/lr)2, both to match test results and to account for
the increased flexural stiffness of the slab-column con-

nection region.

For flat pldtes under uniformly distributed gravity

loads, slab moments calculated by the ACI EFM were
found to differ from measured values by at most 15
percent at interior columns, but by much more than

at,qlleri,or columns.e For.two-way slabs on beams,

i-t,hiq
drscrepancies-of l0 to 20 percent were observed

at some
locations.e Despite discrepancies in the distribution
of

moments, the

ACI EFM provides sufficient flexural

strength to resist the total factored static moment for

each equivalent frame span.
.

_ _F.g_B,a dgsigner,s viewpoint, the ACI EFM has three

-ilisadvantages:
First, rtime-consuming computations are required for
member stiffnesses K,, K,, K", and K"".

ACI Structural Journal

November_December 19gg

Second, ES. (2) (for the equivalent column stiffness

K*) was developed for gravityJoad analyses only. The
ACI EFM can be applied correctly to lateral-load cases
only if the slab-beam stiffnesses K, are reduced for the
effects of cracking.to

Third, the ACI EFM is strictly applicable only to

single-story substructures. The stiffnesses K"" of the

equivalent columns above and below a joint depend on

the stiffnesses of the attached torsional members framing into the joint and also on the stiffnesses of the columns above and below the joint. Based on such calculations, a single stiffness (K*), is assigned to the equivalent column at a given level. In applying the ACI EFM
to multistory structures, however, analogous calculations are performed for each level; on the next higher
level, the same equivalent column can have a different
stiffness (rQr. This discrepancy can be avoided by applying the ACI EFM to single-story substructures only.

Extended equivalent frame methods

In

rqspgqse

to these disadvantages, the ACI EFM

of

was extended by Vanderbilt and others, in the form

599

(a) Extended equivalent column

method (three-dimensional
method)

(c) Extended equivalent staL

method (three-dimensional
model)

DISTR IBUTE

DISTRIBUTE KI

*ffic

[t,*Ktr]

IG, I

-[^-+ftJ

l-__&__l

LEF;
(b) Extended equivalent column
method using nrneuz (two-di-

mensional method)

(d) Extended equivalent slab

method using nrn.e,tta (two-dimensional model)

FLEXURAL
MEMBER
(e) Special member
used in EFRAME

TORSIONAL

MEMBERS

Fig. 2-Extended equvalent frame methods (Vonderbilttr)

the extended equivalent column K"" and. extended

equivalent slab K* methoals.il-r3
The extended equivalent column method tFig. 2(a)l
uses conventional beam elements (no attached torsional
members at ends). Column elements incorporate the
flexural flexibilities of the columns, plus the torsional
flexibilities of the attached torsional members [Fig.
2(e)1, distributed to column elements above and below
the joint in proportion to the flexural stiffnesses of
those columns[Fig. 2(b)].
Tfre equivaient slab method IFig. 2(c)l uses conven_
tional column elements (no attached torsional members
at ends). Slab-beam elements incorporate the flexural
flexibility of the beams and slabs, plus the torsional

flexibilities of the attached torsional members [Fig.

to slab-beam elements on each siOe of
the joint in proportion to the flexural stiffnesses of
2(e)1, distributed

those slab-beams [Fig. 2(d)].

Both methods reduce the slab system to a planar

frame which is then analyzed conventionally. Both in_
clude the effects of slab torsional flexibility under lat_

eral and gravity loads. Because the equivalent slab

600

method cannot reproduce the effects of moment leak_

age under gravity loads, it should be used for cases in_
volving lateral loads only.tt
While removing some of the disadvantages of the
ACI EFM, these methods require a special computer

programrT to handle the equivalent beam and slab

ele_
ments. Also, hand computation is required to distrib_

ute the torsional member flexibilities to the columns

above and below a joint, or to slab-beams on either side
of a joint.

Explicit transverse torsional member method

To eliminate these drawbacks, a modification of the
preceding models, termed the explicit transverse
tor_
sional member method,ra is proposed. As shown in Fig.

3, conventional columns are connected indirectly by

two conventional slab-beam elements, each with half
the stiffness of the actual slab_beam. The indirect con_
ggUqll, made using explicit transverse Gi"fi;;"r_
bers, permits the mo-deling of moment feaEag_e qllgell
qq_ql9b-loilio-riitnixuiliji.wnilJtneresurtinsf rameis
;onpi#;; thiils-n6t a ierious comptication. Speciat_
ACI Structural Journal / November-December 19gg

purpose programs such as ETAqs,18 available in micro_

computer as well as mainfiame versions, are widely

used for analyzing three-dimensional structures. Because the transverse torsional members are present only
in the analytical model, their lengths are arbitrary, provided that their torsional stiffnesses are consistently defined, as explained later.
The explicit transverse torsional member method has
several advantages. Structural modeling is simple and
direct, requiring very few hand computations. Also,
computed member actions in the slab-beams and transverse torsional members can be used directly for design
of slabs and spandrels, respectively. Finally, this
method can even be used for true three-dimensional
analysis of slab systems under combined gravity and
lateral loads. Two sets of equivalent frames, each running parallel to one of the building,s two principal plan
orientations, can be combined to form a single threedimensional model. This single model can be used to
calculate actions in all members (slabs, columns, and
spandrels) under as many combinations of gravity and
lateral loads as desired.

COMPUTER APPLICATION OF SLAB ANALYSIS

METHODS
Computer application of the preceding methods requires that each equivalent frame's geometry, material
characteristics, and member properties be defined. For
all except the explicit transverse torsional member
method, a planar frame is used. The secant modulus of
concrete is usually computed according to ACI 31883.t0 Procedures for calculating member properties are
summarized in Tables I through 3 and are discussed in
the following.

EXPLICIT
TORSIONAL
MEMBERS

Fig. 3-Three-dimensional model of equivolent frame

using explicit transyerse torsional member methodta
Effective width method
For computer analysis, column stiffnesses are computed conventionally and beam stiffnesses are computed using the effective width /r. Cracking, if present,
should generally be accounted for separately. Most effective width methods do not address this isiue. In this
faper, additional stiffness reduction factors of 0.33 and
0.70 were used for flat plates and two-way slabs on
beams, respectively.

ACI equivatent frame method

For computer analysis, the equivalent stiffness K"" of
each column is computed from the joints at each end of
the column. The column inertia is set so that the column's rotational end stiffness coefficient 4EJ"/h" equals
the average of the two Ku" values. Because r("" is strictly
the stiffness of a joint rather than a column, this pro-

Table 1-Modeling idealizations used in computer.aided analysis methods

for two.way stabsof member
Slab analysis
method

Effective width
method (Khan and

Columns

Modeled with
effective width

Sbarounis6)

factor

ACI equivalent

Modeled normally

frame method8

Attached torsional

SIab-beams

members

Modeled normally

Not modeled

Column stiffness is combined with attached torsional member stiffness to

give equivalent column
stiffness. If column stiffness differs as determined
from each end, average

Attached torsional member stiffness is combined

with column stiffness to

with ends incorporating

torsional flexibilities of

Special column element

Torsional stiffness of attached torsional member

is modeled using special

bers

torsionally flexible ends

cu

give equivalent column

stiffness

properties are used

Extended equivalent
column method

Modeled normally

(Vanderbilt,,)

attached torsional mem-

Extended equivalent
slab method

column elements with

Special beam
elements with ends

(Vanderbilt")

incorporating
torsional fl exibilities
of attached torsional

Explicit transverse
torsional member

Modeled normally

members

Modeled normally

method'4

ACI Structural Journal

November-December

988

Modeled normally

Table

2-computation of member properties for analyses of flat plates

GravityJoad analvsis
Method

Effective width (Khan

Uncracked

LateralJoad analysis
Uncracked

Cracked

Should not

and Sbarounisu)

Should not be

be used*

used*

Extended equivalent
column method

May be used

Should not be

(Vanderbilt")

Extended equivalent
slab method
(Vanderbiltrr)

Explicit transverse
torsional member
methodr4

Il=1"

Cracked

May be used

I!

--

ot

May be used
Ij -- 0.33 a lul

r|

May be used

May be used

usedl

Il=Io

be used"

Should not be
used"'t

May be used

Il=Io

Il, = 0.33 I'

May be used

May be used

May be used

May be used

Should not

Il=ro

K,' =

K,

r; -- 0.33 rb
Ki = 0.33 K,

ri,

0.33 Io

May be used

Il=Io

I;

--

0.33 r"

K: -- 0.33 K,

K,' = K,

*Does not exhibit moment leakage under gravity loads.

Ia
= 0.44 for flat plate example of this paper.
I Not recommended for gravityJoad analyses using
[ < Iu.

Computation of member properties for analyses of two.way

slabs on- beams
Table 3

Method
Effective width

(ACI T-beam width'o)

Extended equivalent
column method

(Vanderbilt")

Extended equivalent
slab method
(Vanderbiltrr)

Explicit transverse
torsional member
method'a

GravityJoad analysis

Lateral-load analysis

Uncracked

Cracked

Should not be

Should not be
used*

Il=Io

Il

May be used

May be used

May be used

used*

May be used

Il -- Io
Ii =L

May be used

Ii=1"

May be used

Il=Io
Ki __ K,

r!=L

Uncracked

Il = I*,'
I! = 0.75 I,

Should not be

II :

l--L

May be used

Il
I!

IO

May be used

used"'5

Il=Io

May be used

May be used

Il : I"r,t
Ki = o.33 K,
| = 0.'1s L

Cracked

May be used

= I,nf

Iq|

0.75

I"

Should not be
useds

May be used

Il -- Iu
Ki=K,
I::1"

Il = I"tt.t
Kl = 0.33 K,
I! = 0.75 I,

*Does not exhibit moment leakage under gravity loads.

I I"nt : 0,42 Ib for
two-way slab on beami example of this paper,
I I*, : 0.70 Io for two-way slab on beams example
of this paper.
r Not recommended for graviryJoad analyses using
.{j < 1r.-

cedure cannot be applied consistently to multistory

structures. Other required member properties are the
beam area

A,

length

Z, shear area Au, and moment of

inertia 1u. Beam properties are based on the gross section, using the full slab width /r. Effects of cracking
need not be considered explicitly for gravity loads, because the formula for K"" was calibrated using cracked
test specimens.T,e However, this may not be sufficient
for a gravity-loaded slab previously cracked by lateral
loads. For lateral load cases, slab-beam cracking should
be considered by reducing the slab-beam's moment of
inertia by a reduction factor (usually 0.25 to 0.33 for

flat slabs or flat plates'o).

the slab-beam stiffness should be multiplied by a reduction factor of 0.33 for flat slabs or flat plates.r0'rr

Extended equivalent slab methodir

As noted, this model should not be used for gravity
loads.tr Gross member properties are used as in the
preceding. Transverse torsional member stiffnesses are
distributed to special torsional end members on each
side of a joint in proportion to relative slab-beam stiffnesses. When lateral loads are involved, the slab-beam
stiffness should be multiplied by a reduction factor of
0.33 for flat slabs or flat plates.to,tt

Explicit transverse torsional member method

Gross member properties are used for slab_beams
and columns. Area, moment of inertia, and shear area
are calculated conventionally. For computer input, the
torsional stiffness ,I of the transverse torsional mem_

Extended equivalent column methodll

For gravity loads, gross member properties are used
as in the preceding. Transverse torsional member stiffnesses are distributed to special torsional end members
above and below each joint in proportion to relative
column stiffnesses. This process can be carried out automatically by the computer program EFRAME.T'7 Reduction of slab-beam stiffnesses for'gravity-load cases
is not recommendedlt because it results in erroneous
moments at exterior columns. For lateral load cases,

. Using an arbitrary length

602

ACI Structural Journal

bers is calculated by the following procedure

K, = D9Ec/lrll-(cr/tr)13
[Eq. (2),

Z for the torsional

repeated]
members,

November_December lggg

2'- 9"

l'-9"
Tv p.
1

Meosured

v
I

Iv

Comouled
EFE'WIDTH
(o: O.5l)
EXPI-ICIT.

Kec, ETABS
Kes, ETABS

FLOOR 7' x 6'-0" x t7a"

(

Fig.

4-NRC

model used

for drift

typicol

o.oo5

comporisons

J;G/L

(3)

L/G

(4)

K,;

Effects of cracking need not be considered explicitly

for
gravity loads, because the ACI expression for
K, is con_
sistent with some cracking.T,e However, this may
not be
sufficient for a gravity-loaded slab previously cracked
by lateral loads. When Iateral loads are present, sla6_
beam cracking should be considerea Uy muttiptying
the
slab-beam's moment of inertia by. a reduction factor
of
0.33 for flat slabs or flat plates.'o

Comparison of different methods: Lateral drift

calculation
To compare the accuracy and ease of use of the
preceding methods, each was used to
compute the de_
flections of a small-scale, multistory flat plate test
specimen,re,2o cited also by Vanderbilt.i
This specimen,
tested under the auspices of the Canadian
National Re_
yargh Council (NRC), is referred to here as the NRC

Model, and is shown in Fig. 4. Measured lateral

deflec_
tions of a transverse frame of the NRC model were
compared with those computed by the following
four

methods:
a. Effective width method (Khan and Sbarounis)6

b. Explicit transverse torsional member methodra

c..

Extended equivalent column method (Vanderbilt)lr

-

d, Extended equivalent slab method (Vanderbill;rr

Because cracking was not observed in the NRC
Model,2o all member properties were calculated neglect_
ing cracking. Since the computer program EFRAMET,T?

ACI Structural Journal

,ors

Fig. S-Loteral drift of transverse frome, NRC model

Therefore

J.

DRTFT (in)

the torsional stiffness ,I, of each is then calculated

that

K,i:

0.o

November-December lggg

was not available, the extended equivalent column

and
extended equivalent slab methods were implemented
using the ETABS program. Calculated results,o using
ETABS were within 5 percent of published results,
ob_
tained by Vanderbilt using EFRAME,,T for the'same
structure. Shearing deformations were neglected
throughout. Computation of member propertie-s is de_
scribed in detail in Appendix B of Reference 14.
The results, shown in Fig. 5, indicate that for this
example the e ff ectivg-.yi"*
'ffililtTfiiriee[r. -g-re1,! o d gi ves very .accur ate
lalelal dri_fls,
other rnlit
giving very
slmllar answers, all overestimate drift by"ar,
as much as 20
percent. In evaluating these results, it should
be noted
that only the effective width method was developed as_

suming an uncracked slab. A typical slab structuie

would be more likely to have some cracking, and its
lateral drifts would be closer to the values ;ffiffi; ;;
either of the equivalent frame methods. ihe effective
widih method and the explicit transverse torsional
member method were judged much easier to implement
than either of the extended equivalent frame methods.
While this observation probably would have changed
slightly had the EFRAME program been available, it is
advantageous for a method to require only standard

analysis programs.

COMPARISON OF DIFFERENT ANALYSIS

METHODS: IDEALIZED FLAT.PLATE
To compare the accuracy and convenience of differ-

ent slab analysis methods, the same four methods were

used to compute slab moments in an idealized two_story

flat-plate frame:
a. Effective wldth method (Khan and Sbarounis)6

603

tll
if

As noted previously, the ef_fectjve

INTERIOR EOUIVALENT

_!en{g{ equivalent glab

Slob thickness 8"

All col. 24"x24"

Fig. 6-Three-dimensional view of ideolized two-story

flot-plate example

wid_1h and 9a
m91!q{s_. cannot model moment

leakage, and hence should not be used for gravity-load

analyses. In analyzing slab systems for combined gravity and lateral loads, the preceding two methods can
only be used if the analysis is split into two parts: either
of the preceding two methods is used for the lateralload portion of the analysis, and other methods (such
as the extended_equjvalent qolumn or the explicit transverse torsional member methods) are used for the gravityJoad portion. The results are then combined manually. This process is referred to here as a "two-model"
analysis. The other two methods (extended equivalent
column and explicit transverse torsional member methods) can be used for gravityJoad as well as lateral-load
.analyses. Using

a single model, results for different

load cases can be computed and combined automati-

cally.

b. Explicit transverse torsional member methodra

c. Extended equivalent column method (Vanderbilt)"
d. Extended equivalent slab method (Vanderbilt)"
As shown in Fig. 6, the idealized example frame has
20-ftbay widths, a uniform l2-ft. story height, and24-

in. square columns. The slab thickness of 8 in. was seIected based on the shear and moment transfer and deflection provisions of ACI 318-83,'0 assuming a dead
load of self-weight plus 15 lb/ft2, and a live load of 50
lb,/ft'?. The frame was analyzed for gravity loads and
also for lateral loads of 20lb/ft'z. Member properties
were calculated as shown in Table 2. Joints were considered ri gid, sh e aring 4eloflgalLb B!*y-e,f 9 t9g!99q94,
and member actions were computed at member faces.
Table 4

Slab moment results for gravity loading

Results for gravity loading are shown in Table 4. The
extended equivalent column method, which is.almost
identical to the ACI EFM,rt is used as the standard of
comparison for gravity loading. For gravity loading of
an uncracked structure, the effective width method and
the extended equivalent slab method give expectedly
poor results. The extended equivalent column method
and the explicit transverse torsional member method
give good results.

To u_sq q $Letj flq!:p_!el_e-q94_elfor ggavity g! ygll.as

laleral loading, the slabs shquld be qraq[ed. However,
when the slabs alone are cracked, both extended equiv-

Comparison of slab moments, idealized flat plate example

Slab moments at Level 1, kip-in.

Exterior Interior
column column

Load case and analysis method

Gravity pattern loads, uncracked (1.4D
Effective width methodu
Extended equivalent column method'l

Interior
column

Exterior
column

+ l.7L)

Extended equivalent slab method"

Explicit transverse torsional mbmber method,o

Cravity pattern loads, cracked slabs

Effective width method"
Extended equivalent column method"
-Extended equivalent slab method'r

Explicit transverse torsional member method'o

1629

r699

1675

850

1980

t'176

lt97

665

762

854

1995

1790

661
292

683

6?5

866

700

t46

257

299.

881

7t3

882

2041

1787

3'7

-37
_40

553

Gravity pattern loads, crack slabs, and torsional

members

Explicit transverse torsional member

method,ra

Ki :

0.33 K,

Lateral loads, cracked slabs

Effective width methodu

Extended equivalent column methodl
Extended equivalent slab method'l
Explicit transverse torsional member
method,la Ki = K,

Lateral loads, cracked slabs

- 3'1

-50
-49
-49

45
43

37

40
40

44

-40
-39

31

-23

23

39

-37

-45

37

-43

50
49
49

-31

-16

-44

torsional

members

Explicit transverse torsional member

method,'o

K: :

0.33 K,

Combined loads 0.75 (1.4D + l.7L + l.7W)

Two-model method (Extended column
uncracked + effective width cracked)
Extended equivalent column method, cracked
Explicit transverse torsional member
method,'o K,' : 0.33 K,

604

-38
52'1

550

t268

1396

882
571

t446
t534

123'7

1316
1365

1318

ACI Structural Journal

1420
t35't
14'71

6s9

982
647

November-December 1988

alent frame methods and the explicit transverse tor_

sional member method give erroneous results fo1 grav;
ity-load cas-es. As shown in Table 4, decreasinl the
slab-beamstif f nes5_i4qgagqstherl"b-";;*;;;:

p o it s, lelh er
hAqtqigAfqll eA _4,rt, ;r a u e ex
p6ctEd-ihe-rea;n-iilahE is ifiaTiri"
.;iffi*
t.

IO x
4

l2 x 2O girders

lnterior equivolent
'

"*

conliiictrid to the torsional members rather than to the

slab-beams. When the latter are made more flexible,
the
increased relative stiffness of the torsiorrt-;;;;;;$ii
causes support moments to increase.il,ra
The solution to this problem is to include the effects
of cracking in the transverse torsional members as well
as the slabs. This modification is easy to carry
out with
the explicit transverse torsional *.-L., method. Table
4 shows the results for Ki : 0.33 K,.
-Bggu-lts are close

to those of the extended equivalent cotrrr--metnoO

:' ' Cuncraakeil

case)

Slab moment results for laterat loading

Lateral loading results are also shown in Table 4.
When slabs alone are cracked, all four meth.ods give
similar results. When transverse torsional members are
also crackedln the case of the explicit transverse tor_

sional member method, moments are decreased slightly.

Slab moment resutts for combined loading

(gravity plus laterat)
Combined loading results are also shown in Table 4.
Using the "two-model', procedure, gravity_load mo_

ments calculated using the extended equivalent column

method (uncracked members) are combined with lat_
eral-load moments calculated using the extended equiv8
alen!_ qla!_me1ho,{ (crackbd slabs). Using tfre t,single_

model"'procedure, gravity- and iateral_ltad moments

are calculated using the extended equivalent column
method, as well as the explicit transverse torsional
member method. Cracking is taken into account for
the

slab-beams in each case.(Ii = 0.33 1r). Because the

ex_
tended equivalent column rirettrOa does not permit

easy

incorporation of cracking of the torsional members,

ii

2O spondrels

frome

l.2ao" +

2+o"

l.

2ao"

Slory height = l2'

Slob lhickness = 6"
All columns, 16"x

16"

Fig. 7-Plan view of idealized two-story, two_woy slab

with
beams example

of 6 in. was selected based on the deflection provisions

of ACI 318-83,'0 assuming a dead load of self_weight
plus 15 lb/ft2, and a live load of SO lb/ftr. The frame
was analyzed for gravity loads, afid also for lateral
Ioads of 20 lb/ftr. Joints were considered rigid, shear_

ing deformations were neglected, and member actions

were computed at member faces.
The frames were analyzed using both two_model and

single-model procedures. For the effective width

method, widths of slab-beam members weiil as ildfined

bt tfie T-beam width provisions of ACI 3t8-g3 (Refer_
ence 10). Effects of slab-beam cracking were accounted

for using an average effective moment of inertia, equal

in this case to abofi 0.42 times the gross inertia.,. For
the qlh.qr-fngtg4s, slab-beams and transverse torsional
members were as defined by the ACI equivalent frame
method.10 Effects of slab-beam cracking were ac_
counted for using an average effective moment of in_
ertia, equal in this case to about 0.70 times the gross
inertia.ta Column stiffnesses were calculated using
the
gross inertia, multiplied by a reduction factor
equal to
0.75 (Reference l0). procedures for calculating mem_
ber properties are summarized, in Table 3.

gives erroneous results for the combined load

case. The

explicit transverse torsional riremlier method, on the

other hand, gives results close to those of the two_

model method.

COMPARISON OF DIFFERENT ANALYSIS

METHODS: tDEAL|ZED TWO.WAY SLAB
ON BEAMS
To compare the accuracy and convenience of differ_
ent slab analysis methods, the four methods were used
to compute slab-beam moments in an idealized two_
story frame made up of two-way slabs on beams:
a. Effective width method (ACI effective width for

T-beams).Io

b. Explicit transverse torsional member method.ra

Extended equivalent column method (Vander_

- .-c.

bilt).'r
d. Extended equivalent slab method (Vanderbilt).il
As shown in Fig. 7, the idealized frame has two_way
on beams, 20-ft spans, a uniform l2_ft story
.sl1b1
height, and l6-in. sqrar. columns. The
slab thickness
ACI Structural Journal / November_December
lggg

Slab.beam momenl results for gravity loading

These results are shown in Table 5. The extended
equivalent column method (uncracked case), almost
identical to the ACI EFM,I is used as the standard for
comparison for gravity loading. For gravity loading of
the uncracked structure, the extended equivalent slab
method gives expectedly poor results, ,hil" th. ,*_
tended equivalent column method and the explicit
transverse torsional member method give almost iden_
tical results.
When the slab-beams alone were cracked, the ACI
effective width method and the extended equivalent
slab method were much less accurate than the other
methods. Both the,extended equivalent column method
and the explicit transverse torsional member method
gave slightly high results, although not as far off
as
those of the preceding flat-plate example. As before,
this problem was resolved by using thi explicit trans_
verse torsional member method with cracked torsional
members as well as slab-beams. Results are very close
605

ol srab-bearn moments, ideatized two.way

I1l9J-_-comparison
Deams example
Slab-beam moments at Level

Load case and analysis method

Gravity pattern loads, uncracked (1.4D + lr7L\

ACI effective T-beam width meihod,o.r4 '
Extended equivalent column methodri
Extended equivalent slab method',
Explicit transverse torsional member method,n

Gravity pattern loads, cracked slabs

ACI effective T-beam width method,o',.

Extended equivalent column method,,

Extended equivalent slab method,,
Explicit transverse torsional member method14

Gravity pattern loads, cracked slabs

members

Exterior Int.rioffi
column column
1457 196'7
948 20s7
1083 1584
952 20',12

513
895

1591

1908

805

99s

2065
t'709
2080

605
889

103

1002

l,

srab on

kip-in.

column

column

814
881

8'74

torsional

Explicit transverse torsional member

method,'o

K,' =

0.70 K,

Lateral loads, cracked slab-beams + columns

ACI effective T-beam width methodr0.,a
Extended equivalent column method,,
Extended equivalent slab method',

Explicit transverse torsional member

method,'a

Ki :

K,

962
- 105
-164
- 163

2089
100
133

129

896

-95 95
- l0l l0l
- 104 104
- 10t 101

-164

133

163

131

-99

99

606

1643

l3r6

1506

583
559

r683

- 100
- 133
- 129
- 133

105

t64
163
164

Lateral loads, cracked slabs + columns +

torsional members
Explicit transverse torsional member
method,'a

Ki = 0.7O K,

Combined loads 0.75 (1.4D + t.7L + l.7Wl

Two-model method (extended column
unoacked + effective width cracked)
Extended equivalent column method,'ciacked
Explicit transverse torsional member
method,'o K,'
0,70 K,

to those given by the extended equivalent column

method (uncracked case).

Slab-beam moment results for laterat loading

These are also given in Table 5. All methods ixcept
the ACI effective width method give satisfactory results. Unlike the preceding flat-plate example, torsional
member cracking using the explicit transverse torsional
member method does not significantly decrease slabbeam moments under lateral load.

Slab.beam moment results for combined loading

(gravity plus lateral)
These results are also shown in Table 5. Using the
"two-model" procedure, gravity-load moments calculated using the extended equivalent column method (no
cracking) are combined with lateral-load moments calc u I at ed u s i n g t h e ex.t e n e-{ e qqi_vp!_e_n!
{
- Clab _ir1g!Lq*4'j
(cracked slab-beams and coiumns). Using tha ;;sin;i;-

model" procedure, gravity- and lateral-load moments

are calculated using the extended equivalent column
method, as well as the explicit transverse torsional
member method. Slab-beam and column cracking are
considered identicalty in both cases, and torsional
member cracking is considered in the Iatter method. As
shown in Table 5, all three methods give acceptable results. The single-model procedures are much more convenient.

SUMMARY
This report has focused on the following analysis
methods for two-way reinforced concrete slabs:
606

1..

t700

130'7

1324

131

163

1443 817
t418 9lt
1437 885

1509

1523

Effective width methods, exemplified by the

method of Khan and Sbarounis.6

2. Transverse torsional member methods.
a. ACI equivalent frame method.to
b. Extended equivalent slab method (Vanderbiltlr).

c. Extended equivalent column

method

(Vanderbiltrr).

d. Explicit transverse torsional member metfod.ra

The ACI equivalent frame method was developed
and calibrated for single-story substructures under
gravity loads and cannot be consistently applied to
multistory structures., In the extended equivalent column and extended equivalent slab niethods, the torsional stiffnesses K, of the transverse torsional members are distributed between adjoining columns or slabbeams, respectively. Both methods are used with a specialized computer program.rT The explicit transverse
torsional member method accounts directly for moment leakage and slab torsional flexibility. It requires
only conventional computer programs and can be specialized to either of the extended methods.

Lateral deflections computed by all methods were

checked against results obtained from an uncracked,
scale-model, flat-plate structure.re,20 While results from
the effective width method were most accurate, the
slightly greater drifts predicted by the other methods
might be more reasonable in a real structure with some
cracking. All methods were considered to give satisfactory deflection calculations.
The preceding methods were also compared with respect to accuracy and convenience in calculating mo-

ACI Structural Journal

November-December 19gB

ments for two idealized two_story structures,

one a flat
plate, and the other a two_way slab on
beams. Gravity_
load moments, lateral-load moments, and combined
gravity- and lateral-load moments were
examined.
gravity_load
moments
were
compared
.Calculated
with those from the extended equivalent column

Table 6

Recomm

ffi l,m,:is$*,s*;*?T*f

Type of twc-way slabTwo-way slab on bearns

Il

= I*r,*
= 0.7s I,
Ki = 0.70 K,

method, previously verified for gravity_load mo_

I,
r
I

I!

ments.rr For the flat_plate frame and the two_way

slab
on beams, the extended equivalent column method
and
the explicit transverse torsional member method gave
good results.

of ti

irEpo

4. Additional examples involving three_dimensional

5. Correlation with additional experimental data

on
drifts and member actions.
ci tations

c2

corresporl:?lt'JBl"

""".r

Cross-sectional constant to define torsional properties

[see
Eq. (13-7), Chapter l3J
Size of rectangular or.equivalent rectangular
column, cap_
ital, or bracket measured transverse to the direction
of the
span for which moments are being determined,
in. (Chap_

ters 11 and 13)

E
G

= Modulus of elasticity of member (concrete)

= Shearing modulus of member (concrete)
Io, Il = Moment of inertia of beam
1g.oss ani cracked, respec_
tively)

Advantages of the explicit transverse torsional

member method

f: = Moment of inertia of column (gross and cracked, respectively)

Iur, : Effective moment of inertia (based on ACI T_beam width)
f",

Because it gives reasonable results,

does not require
special computer programs, and permits

used

easy consider_

November-December 19gB

.Irpt.

structures.

*.*u.r

U.urn,

1. Wall loads on beams of two_way slabs.

2. Relation between computed spandrel torsions
and
ACI compatibility torsion provisions.
3. Simpler ways to include the effects of cracking
in
two-way slabs on beams.

iaiiiridtory--di:i{eciio;
actions
""a
were obtained using a single model
based on eittrer the
explicit transverse torsional member method,ra
or the
extended equivalent column method.n Recommended
procedures for calculating member properties
for use in
all methods are given in Tables I through 3.

ACI Structural Journal

o,

method:

Ged
rli.it 'qgjgionA;;;ue;-@@
way stabs on

umn axial loads, a single computer

model, including the
effects of cracking, can be used to calcurate
laterar
drifts and all member actions (slabs, walls,
and col_
umns). Calculated slab actions can
be assigned to col_
umn and middle strips, and slab design
cin easily be

for two-way ,tut

Needed research
Further research is needed on the following
topics
related to the expricit transverse torsionar
member

CONCLUSIONS
of their accuracy and relative simplicity, two
methods were preferable for analysis of siabs
u"ae.
combined gravity and lateral loads. For flat plates,
sat_
istaelqrlllefl ecliaa!_e-sd_nqe*u_"r_aeirsr!_tid;ffi E_
tained y{ry=1ji$-ls=ry9d"l

ation of cracking in both slab-beams and transverse

torsional members, the explicit transverse
torsional
member method is proposed as a powerful
method for
analyzing two-way slab systems under
combinations of
gravity and lateral loads. Member
stiffnesses for this
method are as recommended in Table
6.
Using this method, an entire three_dimensional
building can be analyzed at once, under combinations
of gravity and laterar loads in difierent plan
directions.
If beam end shears for the equivalent fru*er;;;;
each direction are corrected to avoid
doubling the col-

0.70 Io

completed. No hand calculations are required

for load
combinations. The explicit transverse torsional
member
method is believed to present current ACI equivalent
frame method concepts in a form that is both powerful
and convenient for design purposes.

Because

bdami;

*1"1,

Combined gravity-load and lateral_load moments

were compared with those computed by
a two_model
procedure. The extended equivalent
column method
was used to calculate gravity_load moments.
The effec_
tive width and extended equivalent slab methods
were
used for lateral-load calculations for flat plates
and
two-way slabs on beams, respectively. For the
flat_
plate case, the explicit transverse toisional
member
method (using a single model) gave results very
close to
those obtained with the two_model procedure
and was
much more convenient. For the two_way slab
on
beams, both the extended equivalent column
and the
explicit transverse torsional member methods gave
re_
sults very close to those obtained with the
two_model
procedure and were much more convenient.

vgise

#'nj':",*'ii:.'"''

I*,
!,,

for effective beam width in two_way slab on

example

=
I,u

J,

K"

K*

beams

Effective cracked moment of inertia used for

beams in twoivay slab on beams example
Gross moment of inertia of slabs or slab-beams
Torsional stiffness of fl torsional member

=
:
= Flexural stiffness of column; moment per unit rotation
(Chapter t3)
: Flexural stiffness of equivalent column (Chapter 13).

Also

refers to Vanderbilt,s Extended Equivalent

fiame Metfrod
(equivalent column)

(K*)' = Flexural stiffnesses of equivalent columns on

Levels I and
(K*)' 2, respectively
K^ = Refers to Vanderbilt,s extended equivalent frame method
(equivalenr slab)

K"

Flexural stiffness of slab; moment per unit

rotation

K,, K,'

Flexural stiffness of torsional members, uncracked

and
cracked respectively (refer to Chapter 13).
r(,, refers to i,i

Arbitrary length of transverse iorsional members

(Chapter t3)

torsional member

607

:
:
=

which moments are being determined, measured center-tocenter of supports (ChaPter 13)

Y 67' No' 1 I
'
Analysis for Slab Desigp." ACI Jcnxrr' hoceedings '
E' S'
Eberhardt'
C'
A'
by
Nov. 1970, pp. 875-8&l- AIso, Discussion
ProClosure'
Hoffman, Ti Huang, J. C. Jofrirr' Y' K' Hanson' and

Effective width factor

ceedings

Length of span transverse to direction of the span for

Uniform rotation at column, used in deriving effective

width factor o

CONVERSION FACTORS

ii

I ft = 0.305 m
I in. = 25.4 mm
I lb/ft2 = 4.882k9/m'
I kiP : 4'448 kN
I ksi = 6.895 MPa
I Psi = 0.006895 MPa

ir

ii
ri

ii

REFERENCES

questionnaire distribl."Design of Reinforced Concrete Slabs"'

uted by ACI-ASCE Committee 421' 1986'

Robert, and Gamble, William L" Reinforced Concrete

pp' 274-464'
Slabs, John Wiley & Sons, New York, 1980'
and Con3. Hillerborg, Arne, S/rrp Method of Design' Cement

i.iuri,

crete Association, Wexham Springs,

191

5' 256 pp'

4. Pecknold, David A.,"Slab Effective Width for Equivalent

No' 4 Apr'
Frame Analysis," ACI JouRNAL, Proceedings V' 72'
1sls, pp. tl\-tsl . Also, Discussion bv F' H' Allen' P' leP' Darvall'
n. g. Ctover, and D. A. Pecknold, Proceedings V' 72, No' 10' Oct'
1975, pp. 583-586.
ACI
S. Btias, ZiadM.,"Lateral Stiffness of Flat Plate Structures"'
JounNer, Proceedings V. 80, No. 1, Jan'-Feb' 1983, pp' 50-54'
6. Khan, Fazlur R', and Sbarounis, John A',"Interaction of Shear
Walls and Frames," Proceedings, ASCE, V' 90' ST3' Part l, June
1964, pp. 285- 335.

7. Corley, W. Gene; Sozen, Mete A.; and Siess, Chester P',"The

Equivaleni Frame Analysis for Reinforced Concrete Slabs," S/ructural Research Serres No. 218, Department of Civil Engineering,
University of Illinois, Urbana, June 1961, 168 pp'
8. Jirsa, James O.; Sozen, Mete A.; and Siess, Chester P',"Pattern
Loadings on Reinforced Concrete Floor Slabs," Proceedings, ASCE,

V.95, 3T6, June 1969, PP. l1l7-1137'

9. Corley, W. Gene, and Jirsa, James O.,"Equivalent Frame

V. 68, No. 5, May l97l' pp' 397-4OI'

10'ACICommittee3lE'..B'JilditrgCodeRequirementsforRein.
Deforced Concrete (ACI 318-83)"' American Concretelnstitute'
Code RequireBuilding
on
pp,
and
"Commeatary
111
troit, 1983,
pp'
ments for Reinforced Concrete (ACI 318{3)" 155
of Unbraced
Analysis
Frame
D.,"Equivaleot
M.
Vanderbilt,
11.
structural
Loads,"
Lateral
Reinforced concrete Buildingi for Static
Colorado
Departmenr'
Engineering
Ci;il
No.
36,
Reporl
Research
State University, Fort Collins, Colorado, July 1981' ,

iZ. Vuna.rUitt, M. Daniel,"pquivalent Frame Aoalysis for Lateral

Loads," Proceedings, ASCE, V. 105, ST10, Oct' 1979' pp' 19811980' pp'
1998. Also, Discussion by ZiadM. Elias, V. 106, ST7, July
p'
245'
1981,
Jan.
107,
STl,
167l-1672, and Closure, V.
13. Vanderbilt, M. Daniel, and Corley, W. Gene,"Frame Analysis
of Concrete Buildings," Concrete Internotionol: Design & Construction,Y.5, No. 12, Dec. 1983, pp.33-43.
14. Cano, M. T.,"Comparison of Analysis Procedures for TwoWay Slabs," MS Thesis, Depaftment of Civil Engineering, University of Texas, Austin, Aug. 1984.
l5."Reinforced Concrete Floor Slabs - Research and Design,"
Bulletin No. 20, Reinforced Concrete Research Council, American
Society of Civil Engineers, New York, 1918,209 pp.
16. Guralnick, Sidney A., and LaFraugh, Robert W.,"Laboratory
Study of a 45-foot Square Flat Plate Structure," ACI JounNel, Proceedings V.60, No.9, Sept' 1963, pp. 1107-1185.
17. "EFRAME," computer program' National Information Service
in Earthquake Engineering, University of California, Berkeley'
18. Wilson, E. L., Hollings, J. P. and Dovey, H' H',"Erars:
Three-Dimensional Analysis of Building Systems (Extended Version)," computer program' National Information Service in Earthquake Engineering, BerkeleY, 1977.
19. Zelman, Maier I.; Heidebrecht, Arthur C.; Tso, W' K'; and
Johnston, William A',"Practical Problems and Costs of Fabricating
Multi-story Models," Models for Concrete Struclures, SP-24, American Concrete Institute, Detroit, 1970' pp. 159-185'

20. Hartley, G.; Rainer, J. H; and Ward, H. S',"Static and Dy-

namic Properties of a Reinforced Concrete Building Model," Building Research No. 140, Division of Building Research, National Research Council of Canada, Ottawa' Apr. 1919,24pp'

i
l

j
:]

608

ACI Structural Journal

November-December 19BB