You are on page 1of 14

paper: elliott et al

Can precast concrete structures be


designed as semi-rigid frames?
Part 1 – The experimental evidence

Synopsis connections Kim S. Elliott


This paper is part 1 of two papers dealing with the semi-rigid β= column effective length factor BTech, PhD, CEng,
behaviour of beam-to-column connections in precast concrete ∆= second order deflection MICE
framed structures. In Part 1, the results of 28 full scale tests are γm γf = partial safety factors for materials and loads School of Civil
presented. In Part 2, equations are presented to predict φ= relative beam-to-column rotation Engineering, University
moment–rotation behaviour, together with a method to φE = relative rotation at beam rotation limit of Nottingham, UK
calculate the effective length factors for columns in semi-rigid φR = relative rotation capacity of a simply supported beam
sway frames. φU = relative rotation at ultimate test moment MU Gwynne
Connections were made using proprietary pinned jointed λ= axial stiffness in connector
billet, cleat and welded plate connectors, to form cruciform
Davies
assemblages subjected to sway and gravity load. They were not Abbreviations BSc, PhD, CEng,
purposefully intended as moment resisting. In some cases W= welded plate beam-to-column connector MIStructE, MICE
200mm deep hollow cored slabs and high tensile reinforcement B= billet beam-to-column connector School of Civil
completed the connections. Although the moment of SC = single cleat connector Engineering, University
of Nottingham, UK
resistance of the connections varied from 11 to 239kNm, and DC = double cleat connector
flexural stiffness from 200 to 27 000kNm/radian, when the test RHS = rolled hollow steel section
results are normalised with respect to the capacity of the beams HSFG = high strength friction grip bolt Marcelo
there is remarkable correlation leading to several useful C= characteristic 28 day concrete cube strength (MPa Ferreira
parametric equations. Designers are therefore able to calculate units only)
MEng, MSc, PhD,
the moment and stiffness values which provide the input data R= mild steel, grade 250
CEng
in the frame stability analysis described in Part 2. T= high tensile grade >460
University of Sao Paulo,
OPC = Ordinary Portland cement
Brazil
Notation hcu = precast concrete hollow cored floor unit (or slab)
Formerly Post-Doctoral
As = area of tie steel Researcher, Dept of Civil
Asd = area of dowel (billet connector) Present design methods Engineering, University
Ec = concrete Young’s modulus Precast concrete multi-storey skeletal structures are used for of Nottingham, UK
Es = steel Young’s modulus commercial offices, car parks and communal buildings, generally
Fs = tensile or shear force in bars or connector of between two and 20 storeys.They are designed and constructed Halil Gorgun
H= horizontal sway load as beam and column frameworks, carrying simply supported
I= second moment of area in flexurally cracked section precast concrete floors and staircases,and the external façade.The BSc, MSc, PhD
S= rotational stiffness = M/φ precast floors act as horizontal wind diaphragms and are fully tied Lecturer, Dicle
University, Turkey
Si = initial tangent stiffness to individual elements to ensure structural continuity in case of
Formerly Research
SE = connector secant stiffness at limiting beam rotation an abnormal loading event, or to prevent progressive collapse.
Assistant, Dept of Civil
SRC = connector secant stiffness at moment MRC Fig 1 shows a skeletal frame in which the columns are contin- Engineering, University
KsE = normalised connection stiffness = SEL/4EI uous, often manufactured in one piece up to 25m in height. The of Nottingham, UK
L= span of beam beam and floor slab connections are concealed or recessed within
M= applied test bending moment the dimension of the precast elements. Fig 2 shows a construction
ME = beam end moment at limiting beam rotation method where the connections, in this case concrete corbels, are
A. A. Mahdi
MEd = design beam end moment at limiting beam rotation exposed in single storey or multi storey columns. Part of this BSc, MSc, PhD, MICE
(with safety factors) structure (left of picture) is constructed using single storey Hyder Consulting, UK
MR = moment of resistance of beam columns supporting beams that are connected away from the Formerly Research
MRC = moment of resistance of connector column head. In most cases outlined the connections are classified Assistant, Dept of Civil
Engineering, University
MU = test ultimate moment as pinned joints, and thus the frames are designed as pinned of Nottingham, UK
P= applied column load, applied beam load jointed structures.
a= lever arm to applied load Not all precast frames are designed as pinned-jointed.
au = second order column sway deflection Connections deliberately made continuous must be capable of Received: 07/02
b= breadth of section resisting moments and shear forces due to imposed gravity loads Accepted: 02/03
d= effective depth to rebar and reversals of sway loads. Negative (sagging) moment continu- Keywords: Precast
concrete, Semi rigid,
dc = effective depth to connector ity must also be provided to counter the effects of relaxation of Frames, Beams,
fcu = compressive strength of concrete in beams and Columns, Connections,
columns Design methods,
fcui = compressive strength of infill mortar/grout/concrete in Experimental data,
Calculations
connectors
fy = yield stress of reinforcement
h= depth of section, storey height
n= number of columns in a frame
pw = yield strength of weld
pq = shear strength of dowel Fig 1.
w= uniformly distributed load Precast concrete
x= depth to neutral axis in beam skeletal frame with
xc = depth to neutral axis in connector hidden beam-to-
z= lever arm in beam column connections
zc = lever arm in connector (Courtesy Trent Concrete
α= column to beam flexural stiffness ratio with rigid Ltd, UK)

14|The Structural Engineer – 19 August 2003


paper: elliott et al

Fig 2. (left)
Precast concrete frame
using beam-to-column
corbel connections
and beam-to-column
head connections
(Courtesy Blatcon Ltd,
UK)
Fig 3. (right)
Precast concrete
permanent formwork
to receive reinforced
cast in situ infill to form
monolithic connection
(Courtesy New Zealand
prestress, thermal movement, creep and shrinkage. In seismic Concrete Society)
regions precast concrete elements are used to form fully rigid
connections by making reinforced cast in situ cruciform joints, as
shown in Fig 3 where the precast element is used as permanent
shuttering. Many of the inherent advantages of speed and struc- ness, the maximum bending moment at the foundation to each
tural efficiency in a precast concrete structure are lost due to the column is the sum of the sway moments, ΣHihi/n, and the P-∆
considerable quantities of in situ concrete and site placed rein- moment ΣPiau accumulated over the full height of the structure.
forcement required. Attempts to form moment resisting The sway deflection au is dependent on the effective length of the
connections using mechanical fasteners,such as steel cleats,bolted column. Unlike structural steelwork, where the rotational stiff-
brackets or welded plates, etc. have not been wholly successful, ness of beam-to-column connections influences the magnitude of
often because of a need to provide tensile resistance in the bottom au, current precast concrete design practice ignores the presence
of the beam. The result is that pinned joints at beam to column of beam-to-column connections in such a way that sway moments
connections are preferred from both a design and construction in the column may not be distributed to the beams.
point of view. BS 8110:19851 is particularly harsh on column effective length
In design, three dimensional precast structures are simplified factors in unbraced precast frames. The ratio αc of the flexural
as 2-d frames.The in-plane structural system in Fig 4a comprises stiffness of the column to that of a precast beam is taken as 10,
solely of beams and columns. Floor slabs, in the out-of-plane direc- Fig 4. resulting in an effective length factor for the column of β = 2.3. For
tion, do not contribute to the flexural stiffness or strength of the Schematic precast example, in a 10m high column of 300 × 300 mm cross section the
beams unless they are specifically designed as composite beams. skeletal frames. (a) In- second order deflection au, according to BS 8110, Part 1, eq. 32 &
Sway deflections for such a pinned and rigid plane frame are plane sway in pinned 34, is a staggering 880mm! This clearly results in an impractical
shown in Figs 4a and 4b. In the perpendicular direction in Fig 4c jointed frame. (b) In- size of columns. Although many precast concrete designers know
no frame action is possible because the precast floors are simply plane sway in rigid this method of design as being very conservative, refinement has
supported on beams. frame. (c) Out-of-plane not taken place due to a lack of data on the flexural properties of
The resistance to horizontal load and second order P-∆ deflec- torsional mode in beam-to-column connections. It is reasonable to accept that the
tions in the pinned jointed unbraced structures shown in Fig 4 is pinned jointed frame. flexural properties of a mechanical connection containing only a
provided solely by the cantilever action of the columns. Referring (d) In-plane sway in precast beam will be very limited, and indeed research by Mahdi2
to Fig 4a, assuming that n number columns are all of equal stiff- semi-rigid frame has shown this to be the case, especially where a sagging moment

4a 4b

4c 4d

19 August 2003 – The Structural Engineer|15


paper: elliott et al

Fig 5. (above) Aspects of this work have been previously published by the
Continuity tie steel authors7,8,9, and by others on items (iii) and (iv)10,11.The analytical
bars placed in the floor techniques presented in Part 2 enable calculation of connection
zone acts compositely strength and stiffness, and the effective length of columns in the
with the precast beam format of modified BS 8110 equations.
Fig 6. (right)
Types of precast Concept of semi-rigidity
concrete beam-to- The response of a frame to bending moments will be greatly influ-
column connectors. (a) enced by the rotational behaviour of the connections, in particu-
Welded plate (b) Billet lar the beam – column connection. Shear and/or axial forces may
(c) Single cleat (d) have a small beneficial or deleterious effect, but these are gener-
Double cleat ally ignored. Deformations in precast connections are known to be
influenced by many effects, particularly those involving an assem-
bly of welded plates, grouted dowels or bolted brackets – the
(tension at the bottom) is acting. But when the floor slab, together majority of which contain precast-to-precast or precast-to-in situ
with stability tie bars, cast in situ topping and the cast in situ infill interfaces.When subjected to a hogging bending moment M,defor-
around the ends of the slabs (see Fig 5) are taken into account it mations at the interfaces compress the concrete locally at the
is obvious that the connection must possess appreciable flexural bottom whilst tensile strains open the interfaces at the top, result-
strength and rotational stiffness. It would seem reasonable to ing in a rigid body rotation φ of the end of the beam relative to the
assume that if a precast concrete connection behaved in a semi- column, as shown in Fig 9.
rigid manner, its finite strength and stiffness could be harnessed The connection is classified as ‘semi rigid’ and its behaviour is
to enhance the global stiffness of the structure.
The increased stiffness of the connection has the effect of reduc- 6a
ing αc (by apparently increasing the stiffness of the beam) and
hence reducing β. In allowing the continuity of in plane bending
moments the sway profile of the frame will be as shown in Fig 4d.
The stiffness of the connection not only has an effect on the sway
profile of the structure, and hence on the serviceability limit of
deflection, but also on the ultimate load capacity of the column. It
is therefore necessary that the connection possesses sufficient
strength, stiffness and ductility to ensure that a premature failure
will not occur in the connection. Columns may be then designed
for each successive storey using the appropriate β factor, provid-
ing that the beam end moment is less than the moment capacity 6b
MRC of the connection at the end of the beam.
The connections used at beam to column joints vary widely in
the precast industry.This was due,in part,to patented connections
in the 1960s and 70s, which stifled the widespread use of the
most efficient types of connections and led to the proliferation of
new, or novel, types of connections3. Fig 6 shows a range of
connections that would be classified as pinned during the
construction phase, but on completion would generate flexural
strength and rotational stiffness in responding to imposed gravity
and wind loads. Because large shear forces and bending moments
will inevitably be transmitted to the columns and floor units,
possibly leading to cracking (Fig 72) or crack widening (Fig 84), it
was necessary to undertake new research to increase the under-
standing of the rotational behaviour of precast concrete 6c
connections.
Collaborative projects were established at Nottingham 2,4, City5
and Southampton6 Universities, funded by the EPSRC and a
number of UK precast concrete frame manufacturers. The full
scale tests reported in this paper were carried out at Nottingham
and City by different research teams. The aim of the work was to
find out if a precast skeletal sway frame could be designed (and
constructed) using semi rigid flexural connections, and to provide
engineers with bona fide test data, empirical design equations and
a semi rigid frame analysis computer program.
This was done in four stages:
• to determine the rotational behaviour of full scale precast
concrete beam to column connections subjected to hogging
and/or sagging, cyclical and monotonic flexural loading. 6d
• to determine the ultimate load in unbraced frames containing
rigid, pinned and semi-rigid connections and thus propose
parametric equations for column effective length factors.
• to develop a finite element model of some of the connections
tested under item (i) to enable a wider range of joint geometry
and material properties to be studied.
• to propose a method for the design of columns in semi-rigid
frames, and validate the output using a 3-dimensional frame
analysis computer program.

This paper is Part 1 of two papers presenting the experimental


test results and the analysis included in items (i), (ii) and (iv).

16|The Structural Engineer – 19 August 2003


paper: elliott et al

Fig 7. defined by a moment-rotation (M-φ) diagram shown in Fig 10a.


Shear cracks in precast The rotation φ must not include rotations due to flexural curva-
column caused by ture of the beam or column, otherwise these will be counted twice
opposing moments in in frame analysis. (It is important to recognise that test results
sway frame tests2 cannot always separate these two effects.) The relevant properties
of semi rigid connections are defined in Fig 10a as:
• ultimate flexural strength MU
• rotational stiffness S
• ultimate rotational ductility capacity φU

In precast connections there is a zone of influence beyond which


beam end rotations are changing but are not influenced by frame
actions, namely, curvature of the beam or column. The length of
the zone is approximately equal to the depth of the adjoining
beam and/or column2,4. In this context it is very important that the
characteristics of the connection are tailored to suit the require-
Fig 8. ments of the adjoining members.
Transverse cracks in The beam-line method, illustrated in Fig 10b, is used to quan-
hollow cored slabs tify this relationship.The beam-line represents the characteristic
subjected to flexural M-φ behaviour of a beam under certain conditions of loading in a
tension flexurally cracked state.Consider a uniformly distributed imposed
load w acting on a beam of span L; the moment of resistance of
the beam (or composite beam + slab) MR at point A is equated to
the fixed end moment wL2/12.The beam is fully rigid theoretically
such that φ = 0. If the beam is pin ended, then at point B, M = 0
and φ = wL3/24EcI, where Ec is the short-term value for Young’s
modulus, and I is the flexurally cracked second moment of area.
It can be shown that the interaction between points A and B is
linear, having a gradient of –2EcI/L. A similar beam-line may be
constructed for any other type of loading, e.g. a point load at mid
span.
All points along the beam-line define the relationship between
Fig 9. the end moment and end rotation of the beam. If the
Definition of relative moment–rotation for the connection is plotted on the same
rotation of a beam to diagram the intersection point ‘E’ (E means beam End) simulta-
column. Note φ is neously defines the end moment and rotation of BOTH the beam
exclusive of nodal and connection. For a connection to be considered in frame design,
rotations and beam point E must exist, as shown in Fig 10b, and the rotational stiff-
curvature. ness may be taken as S. If the connection fails prior to reaching
point E (Fig 10c) the connection is classed as pinned (even though
there may be some strength and stiffness present).
The semi rigid behaviour of precast connections will be further
complicated by material non-linearity - firstly where concrete
strains exceed about 500µε, then when localised debonding of
reinforcement takes place, and finally when the reinforcement
yields4. The non-linear M-φ diagram, which is defined in Fig 10d,
has further important properties:
• initial (tangent) stiffness, Si

Table 1: Dimensions and material data for single sided tests


Test Beam and Distance to As top & Effective As Effective Beam Infill mortar Bars
ref. slab section1 point load2 bottom depth to bars depth to cube cube yield
b x h + hs bars beam bars in slab slab bars strength3 strength3 strength3
d ds fcu fcui fy
(mm) (mm) (mm2) (mm) (mm2) (mm) (N/mm2) (N/mm2) (N/mm2)
B1 300 x 300 2365 1608 250 57.2 42.5 520
B2 300 x 300 2365 1608 250 59.9 38.5 520
B3 300 x 500 1400 982 450 69.5 30.2 538
B4 300 x 500 1400 982 450 60.2 44.6 538
B5 300 x 600 2365 1608 550 56.9 55.2 520
B6 300 x 600 2365 1608 550 56.9 68.3 520
B7 300 x 300 2365 1256 250 982 450 34.2 41.8 504
+ 200 slab
W1 300 x 300 2365 1608 250 54.5 46.2 520
W2 300 x 300 2365 1608 250 54.5 49.8 520
W3 300 x 450 1400 982 400 59.9 49.8 523
W4 300 x 450 1400 982 400 65.2 55.3 523
W5 300 x 600 2365 1608 550 63.2 47.3 520
W6 300 x 600 2365 1608 550 54.5 48.2 520
W7 300 x 300 2365 1256 250 982 450 44.5 45.0 518
+ 200 slab
DC1 250 x 450 1400 982 400 59.0 50.2 533
DC2 250 x 450 1400 982 400 59.0 55.3 533
Notes: 1. b = beam breadth, h = beam overall depth, hs = hollow core slab depth (breadth taken as 1.0 m in I calculations); 2. Measured from
face of column; 3. As measured values; Even numbered tests testing in sagging bending mode. Odd numbers hogging.

19 August 2003 – The Structural Engineer|17


paper: elliott et al

10a Fig 10. • secant stiffness at point E, SE


Moment-rotation M-φ • moment at point E, ME
diagrams. (a) Linear
elastic-perfectly-plastic. For very high values of stiffness the behaviour will resemble
(b) Beam-line closely that of a rigid joint, and may be assumed to be rigid in
approach to beam end frame analysis. In fact a reinforced concrete monolithic joint will
moment and stiffness have a finite stiffness even though we consider this to be infinitely
requirements. (c) M-φ large. Similarly, a connection of very low stiffness may be assumed
diagram where to be pinned. Anything in between will be semi-rigid, but the
connection is boundaries between the three classifications are somewhat arbi-
insufficient for beam trary. Designers of structural steelwork have attempted to define
end requirements. (d) the boundaries for connections in sway frames based on stiffness
Non-linear semi-rigid only according to EC312. Ferreira & Elliott13 have classified semi-
behaviour. (e) rigid precast connections as ‘partial strength’, ‘full strength’ and
10b Definition of calculated ‘effectively rigid’, setting conjugate limits for connection strength,
strength and stiffness stiffness and rotational capacity. Their research has therefore
of semi-rigid focused on the relationship between the M-φ plot and the beam-
connection line, leading to non-linear computer analysis to validate the use
of this method. In order to take this a step forward real M-φ data
are required for a wide range of precast concrete connections.

Precast beam – column connections used in the


experimental work
The preferred method of making beam-to-column connections in
precast skeletal frames is to attach the end of the beam to the
column face via a steel or concrete corbel or stiffened steel angle.
At this stage of construction the joint is most definitely pinned, but
when the connection is completed (with or without the floor slab)
it will be subjected to bending moments arising from imposed
gravity and wind loading. Although the moment capacity of the
10c
connection may be limited in the sagging mode (because the
bottom of the connection is in tension) it has considerable magni-
tude in the hogging mode especially where reinforcing bars are
added within the floor zone as shown in Fig 5. Sagging bending
moments due to wind load may, in many cases, be considered as
unloading of the hogging moment because of the predominant
magnitude of permanent gravity loads.Therefore the most impor-
tant bending mode for these connections is hogging.
The common types of beam-to-column connectors are shown in
Fig 6. They are all of the ‘hidden’ type, i.e. no downstand corbel is
visible upon completion.
• the welded plate, prefix ‘W’, Fig 6a
• the steel billet, prefix ‘B’, Fig 6b
• the single cleat, prefix ‘SC’, Fig 6c
10d • the double cleat type, prefix ‘DC’, Fig 6d

The welded plate connector is a modified ‘Cazaly hanger’14 where


the cantilever beam is replaced by a deep narrow mild steel plate,
typically 25mm thick × 500mm long and 100mm deep. The plate
is anchored in the beam by welding two bent bars (typically T25
or T32) one on each side of the plate. A solid section (called a
‘billet’) 100 × 100mm cast into the column projects a distance of
90mm from the column face. The level of the top of the billet is at
190mm from the bottom of the beam. After a double sided fillet
site weld is made to the narrow plate, 80mm long × 20mm thick-
ness using a grade 275 electrode, a structural grade (C40) cast in
situ infill concrete, containing an expansive cement, completes the
joint at the end of the beam. The centroid of the weld is therefore
at 200mm from the bottom of the beam. The connection contains
two precast-insitu interfaces as indicated in the figure.
10e The billet connector is based on the conventional steel haunch
using a grade 275 or 350 RHS section. The billet projects 100mm
from the face of the column. The section size for a 300mm deep
beam is 100 × 60 × 6.3mm. In 500 – 600mm deep beams it is 160
× 80 × 8mm.The level at the top of the RHS varies for each depth
of beam and is at 140 and 290mm, respectively from the bottom
of the beam. A steel plate cast into the end of the beam ensures a
positive bearing through which a 16mm diameter high tensile
threaded dowel is secured beneath the RHS and to a top fixing
angle 100 × 100 × 8mm.The top angle is fixed to the column using
a 16mm diameter grade 4:6 bolt.The annulus around the RHS is
filled by highly workable gravity fed grade C40 mortar contain-
ing expansive cement. The 10mm wide gap at the face of the
column is dry packed with mortar such that for the most part the

18|The Structural Engineer – 19 August 2003


paper: elliott et al

11a above) and part dry packed.The connection therefore contains two
precast-in situ interfaces.
The double cleat connector comprises a load bearing (seating)
cleat and a top locating cleat. The former is made using a rolled
structural angle section.The level of the top of the seating cleat is
at 140mm from the bottom of the beam. A steel plate in the end
of the beam ensures a positive bearing through which a 16mm
diameter grade 8:8 threaded dowel is secured to a stiffened locat-
ing cleat at 110mm from the top of the beam.The seating cleat and
the locating cleat are pre bolted to the column using 25mm diam-
eter grade 4:6 threaded (to M24) dowels that pass through the
column. The connection is completed in the same manner as the
billet connection, and therefore contains two precast-insitu inter-
faces.
A schedule of tests is given in Tables 1 and 2, respectively, for a
range of single sided (one beam and one column) and double sided
(two beams and one column) tests.The connections were designed
11b 11c and manufactured by casting a reinforcing cage (or steel sections
called a ‘shear box’) in the ends of the beams to provide 200kN
vertical shear capacity.The beams were reinforced symmetrically,
i.e. top steel = bottom steel. The columns were reinforced using
four T32 bars at a centroidal cover of 50mm. Full details of the
connections are available in the literature2,4,5. Match cast precast
concrete and infill mortar/concrete cubes and axial tension rebar
testing provided the mean values for the strength of the precast
concrete fcu, in situ infill concrete fcui and rebars fy, given in Tables
1 and 2.
Using the as-tested material data, the calculated moment of
11d 11e resistance of the column, ignoring axial forces, was about 175
kNm (fcu and fy varied in different columns).Thus the total design
moment of resistance in the connector was twice this figure, i.e.
350kNm. (When BS 8110 partial safety factors are included this
reduces to 320Nm.) The design moment of resistance of the beams
MR was calculated according to the BS 8110 stress block method
using the as-tested material data without partial safety factors,
e.g. for test beam B1, b = 300mm, d = 250mm, As = 1608mm2, fcu
= 57.2N/mm2 and fy = 520 N/mm2.Then Fs = 520 × 1608 = 836kN;
x = 81mm; z = 250 – 0.45 × 81 = 214mm; MR = 836 × 0.214 =
Fig 11. connection contains one precast-precast interface. 178.6kNm. Values for MR are given in Tables 3 and 4.
Cruciform sub- The single cleat connector comprises a seating cleat of rolled The moment capacity of the connector MRC is calculated at the
assemblies for beam- structural tee section, for example cut from a 457 × 152 × 72 face of the column.The internal lever arm z is the resultant of the
to-column connection Universal Beam, gusseted for strength, and pre-fixed to the various horizontal forces, which, depending on the type of connec-
tests. (a) Location of column face using four M24 HSFG bolts. Two M24 grade 4.6 site tor are due to the weld, threaded dowel, fixing bolts, locating cleats
sub-assemblies and bolts connect through a steel plate in the beam to slotted holes in and, in the case of the floor slabs, the reinforcing tie bars. There
loading regimes. (b) the cleat.The level of the seating is 140mm from the bottom of the may be two, or three, component forces acting. It is assumed that
Single sided beam beam. Infill grade C40 expansive mortar is gravity fed into the the forces contribute exclusively to the moment capacity, that
only. (c) Single sided rectangular space beneath the cleat, which is lightly reinforced they attain their yield strength, and no local bending effects, for
with floor slab. (d) with R8 loops. The 30mm gap above the cleat is part filled (as example along the length of the weld,are acting.It is also assumed
Double sided beam
only, sway loading. (e) Table 2: Dimensions and material data for double sided sway and gravity load tests
Double sided with Test Loading Beam and Distance to As top & Effective As Effective Beam Infill Bars
floor slab, sway or ref mode slab section1 point load2 bottom depth to bars depth to cube cube yield
gravity loading b x h + hs bars beam bars in slab slab bars strength3 strength3,4 strength3
d ds fcu fcui fy
(mm) (mm) (mm2) (mm) (mm2) (mm) (N/mm2) (N/mm2) (N/mm2)
B8a sway sagging 300 x 300 2365 1608 250 55.2 42.2 520
B8b sway hogging 300 x 300 2365 1608 250 55.2 42.2 520
W8a sway sagging 300 x 300 2365 1608 250 54.8 45.1 520
W8b sway hogging 300 x 300 2365 1608 250 54.8 45.1 520
SC1a sway sagging 300 x 300 2365 1608 250 58.2 44.3 520
SC1b sway hogging 300 x 300 2365 1608 250 58.2 44.3 520
B9a gravity 300 x 300 2365 1608 250 982 450 48.4 46.4/27.8 518
+200 slab
B9b gravity 300 x 300 2365 1256 250 982 450 50.3 46.4/27.8 518
+ 200 slab
W9a gravity 300 x 300 2365 1256 250 982 450 50.4 45.4/33.8 518
+ 200 slab
W9b gravity 300 x 300 2365 1256 250 982 450 54.9 45.4/33.8 518
+ 200 slab
W10a sway sagging 300 x 300 2365 1608 450 982 55.1 47.5/26.2 520
+200 slab
W10b sway hogging 300 x 300 2365 1608 250 982 450 55.1 47.5/26.2 520
+200 slab
Notes: 1. b = beam breadth, h = beam overall depth, hs = hollow core slab depth (breadth taken as 1.0m in I calculations); 2. Measured from
face of column; 3. As measured values; 4. Two values are for beam-column mortar and beam-slab infill concrete

19 August 2003 – The Structural Engineer|19


paper: elliott et al

Fig 12. connections in portal frames were tested at the Technical


General arrangement University of Tampere in Finland17, at the Centre d’Etudes et de
for single sided tests Recherches de l’Industrie du Beton (CERIB) in France18,19, and at
subjected to sway University of Sao Paulo in Brazil20.
loading The experimental testing of beam-to-column connections in
skeletal frames at the aforementioned British universities
attempted to co-ordinate test data with numerical studies and
computer software. 28 full scale connection tests, some of which
were coupled with precast concrete hollow cored floors, were
subjected to hogging and sagging bending moments resulting
from gravity or sway loads. As it was not possible to test the
connections in a multiple bay or multi-storey frame environment,
cruciform sub-assemblages were used as shown in Fig 11a. This
arrangement does not permit moment redistribution, but it does
that the strength of the concrete in the compression zone is based allow large rotations to be measured,uninhibited by the rotational
entirely on the infill strength fcui.The structural models and spec- limitation of the beam.
imen calculations are given in Appendix A. Values for MRC are The rationale behind the testing program was to collect M-φ
given in Tables 3 and 4. data from the following sets of tests:
• single sided connections,comprising the beam alone in separate
Experimental tests on beam – column connections hogging and sagging modes. Three different connectors were
The history of research into semi rigid precast connections dates studied.The beam depth varied from 300 to 600mm.Two tests
back about 25 years. The PCI document on connections15 alluded
to using semi rigid connections via the beam-line method, from Trent Concrete Ltd. of Nottingham manufactured the single cleat specimens.
Bison Ltd. of Slough manufactured the welded plate specimens. Composite
which the ultimate and service moment of resistance could be Structures of Eastleigh and Atcost Ltd. of Ashford in Kent designed and
found.Stanton16 tested several of the connections illustrated in the manufactured the double cleat specimens, respectively. Crendon Ltd., now
defunct, manufactured the billet specimens. Bison Ltd. of Litchfield supplied
PCI Manual14, many of which contain cast in situ r.c. cruciform hollow cored floor units. Specimens references B7, B9, W7 & W9 were
joints more suited to seismic situations. Beam-to-column head manufactured in the Civil Engineering Laboratory at Nottingham University.

Table 3: Results of single sided beam – column tests


Test Connector Connector Beam Beam-end Ratio of Ratio of initial Beam-end Beam Ratio3
ref test moment moment of moment test/calc MU/MR siffness secant stiffness2 Ks
ultimate capacity1 resistance1 at E MU/MRC Si stiffness at E 4EcIcr/L at E
MU MRC MR ME SE
(kNm) (kNm) (kNm) (kNm) (KNm/m.rad) (KNm/m.rad) (KNm/m.rad)
B1 -–20.1 –22.6 –179 –20.1 0.89 0.11 8.2 0.46 7.3 0.06
B2 +21.7 +16.0 +180 +21.7 1.35 0.12 7.7 0.44 6.5 0.07
B3 –37.4↑ –27.7 –228 –29.3 1.35 0.13 11.6 1.36 18.4 0.07
B4 +37.2↑ +27.6 +226 +33.5 1.35 0.15 14.0 1.38 15.9 0.09
B5 –49.9 –44.2 –429 –45.8 1.13 0.11 83.3 2.58 43.3 0.06
B6 +32.9 +32.0 +429 +31.7 1.03 0.07 85.0 1.51 37.8 0.04
B7* –58.0 –106.1 –288 –54.2 0.55 0.19 26.6 2.48 21.4 0.12
W1 –92.5 –73.4 –177 –91.6 1.26 0.52 23.6 3.93 7.3 0.54
W2 +28.6 +32.8 +177 +28.6 0.87 0.16 6.1 0.62 6.5 0.10
W3 –78.5 –74.0 –194 –73.1 1.06 0.38 36.6 4.27 14.2 0.30
W4 +75.0↑ +95.5 +195 +75.0 0.79 0.38 17.5 3.84 12.3 0.31
W5 –74.3 –73.6 –432 –74.3 1.01 0.17 23.2 4.49 43.3 43.3
W6 +125.1 +156.3 +428 +125.1 0.80 0.29 22.4 7.81 37.8 0.21
W7* –156.4 –153.1 –316 -127.9 1.02 0.40 60.8 7.26 21.4 0.34
DC1 –58.1↑ –36.2 –196 –31.8 1.60 0.16 1.8 1.34 13.8 0.10
DC2 +39.8↑ +31.8 +196 +29.2 1.25 0.15 11.1 1.05 12.0 0.09
Notes: 1. Beam and connector design calculation to BS 8110 with actual fcu and fy without partial safety factors according to Table A1; 2. L =
6000mm, Ec = 32kN/mm2, I = flexurally cracked second moment of area (including reinforcement); 3. At E, Ks = SEL/4EcI; ↑ signifies maximum
moment and rotation not attained during tests; * with 200mm deep slab

Table 4: Results of double sided beam – column and slabs tests


Test Connector Connector Beam Beam-end Ratio of Ratio of initial Beam-end Beam Ratio3
ref test moment moment of moment test/calc MU/MR siffness secant stiffness2 Ks
ultimate capacity1 resistance1 at E MU/MRC Si stiffness at E 4EcIcr/L at E
MU MRC MR ME SE
(kNm) (kNm) (kNm) (kNm) (KNm/m.rad) (KNm/m.rad) (KNm/m.rad)
B8a +17.8 +16.1 +178 +17.8 1.11 0.10 21.6 0.36 6.5 0.06
B8b –26.0 –22.6 –178 –26.0 1.15 0.15 11.9 0.63 7.3 0.09
W8a +40.0 +31.9 +177 +40.0 1.25 0.23 16.9 0.94 6.5 0.15
W8b –48.8↑ –73.2 –177 –48.8 0.67 0.28 13.5 1.39 7.3 0.19
SC1a +11.0 +17.4 +179 +11.0 0.63 0.06 2.3 0.21 6.5 0.03
SC1b –13.2 –15.1 –179 –13.2 0.87 0.07 2.6 0.29 7.3 0.04
B9a* –188.2 –230.4 –322 –188.2 0.82 0.58 179 14.99 21.4 0.70
B9b* –191.3 –230.4 –325 –188.2 0.83 0.58 172 14.71 21.4 0.69
W9a* –237.0 –264.9 –331 –232.0 0.89 0.70 482 25.16 21.4 1.18
W9b* –238.8 –264.9 –331 –237.0 0.90 0.72 280 27.07 21.4 1.27
W10a* +125.1 +114.9 +362 +120.5 1.09 0.33 210 6.70 26.8 0.25
W10b* –210.4 –267.7 –357 –170.5 0.79 0.48 37 10.21 22.3 0.46
Notes: 1. Beam and connector design calculation to BS 8110 with actual fcu and fy without partial safety factors according to Table A2; 2. L =
6000mm, Ec = 32kN/mm2, I = flexurally cracked second moment of area (including reinforcement); 3. At E, Ks = SEL/4EcI; ↑ signifies maximum
moment and rotation not attained during tests; * with 200mm deep slab; For tests W8b there was no intersection with the beam-line, therefore
no value for ME exists

20|The Structural Engineer – 19 August 2003


paper: elliott et al

was compacted by gravity feed and hand tamping. The infill


concrete was delivered by ready mix supplier as a grade C30
prescribed OPC mix without additives with a 50mm slump.
The precast floor slabs were 200mm deep × 1200mm wide Roth
type prestressed hollow core units (hcu), supported on the beams
over a 75mm bearing length. The length of the hcu, spanning
perpendicular to the plane of bending, was 1.0m. To complete the
connection two 25mm diameter high tensile tie bars were laced
inside of T10 projecting loops from the beams. The bars form an
integral part of the horizontal stability floor ties required by design
codes. In the double sided tests the tie bars passed through the
column as shown in Figs 5 and 13. In the single sided tests they
were cranked at 45° to the longitudinal axis and laced into project-
Fig 13. (above) ing loops from both the main beam and edge beams, see Fig 14.
Detail of double sided Floor ties were also placed into 500mm long milled slots in the hcu,
tests with floor slabs these were T12 bars at 300mm spacing.
and tie bars Loads P were applied incrementally using hand pumped
Fig 14. (left) hydraulic jacks and measured to an accuracy of 0.05kN by elec-
Detail of single sided trical resistance load cells. The distance a to the point load gives
tests with floor slabs the moment in the connector as M = Pa measured at the face of
and tie bars the column (even though it was obvious that the point of centre
of rotation was a short distance from the face of the column). M
includes the self weight of the beam and slab.
Beam to column relative rotations φ were measured using
either inclinometers or a set of four linear displacement trans-
ducers at known distances from the face of the column.The benefit
in using the latter method is that the individual relative rotations
between the column and in situ infill, and between the infill and
beam may be established.This enables a ‘zone of influence’ for the
connector to be deduced, which varied depending on the type of
connector. This distance, which varies from h to 1.2h, where h is
Fig 15. the total depth of the beam, is found to be the extent of the
Vertical deflection damaged zone due to connector behaviour4.
measurements Fig 15a shows the vertical displacements for the welded plate
beneath the beam for connector. The slope of the graph gives rotation φ measured from
(a) double sided the points at 300mm to 90mm from the column face.The rotation
welded plate at the interface at the end of the beam is approximately 1.5 times
connector (test W9) that at the column face. Fig 15b shows the same data for the billet
and (b) single sided connector. Here there is a reversal of vertical displacement caused
billet connector (test by a root point of rotation at the end of the billet (90mm from face
B7) of column). Although the measurement of φ is less certain it is

15a
involving a 300mm deep beam and 200mm deep precast floor
acting compositely were subjected to hogging moments. The
tests would represent the effects of sway loading at external
frame connections. See Fig 11a, b, c and Table 1.
• double sided connections, comprising two 300 mm deep beam
alone in simultaneous hogging and sagging modes. Two tests
involved 300mm deep beams and the same type of floor acting
compositely. These tests would represent the effects of sway
loading at internal frame connections. See Fig 11d, e and Table
2.
• double sided connections, involving two 300mm deep beams
and the same type of floor acting compositely in hogging mode.
These tests would represent the effects of imposed gravity
loading at internal frame connections. See Fig 1e and Table 2.

One of the main requirements in the testing arrangement was 15b


being able to apply sway loading to the column in such a way that
the reaction in each beam depended on the unknown stiffness and
strength of the connectors. 3m high columns were used for the
sway tests.They were mounted on rollers, enabling the bottom of
the column to sway relative to the top of the column, allowing the
true vertical reactions to form in the 2.5m long beams as shown
in Figs 11a and 12. In the gravity load tests, where both beams
are subjected to equal hogging moments, vertical point loads were
applied at the remote ends of the beams.
The assemblies were constructed according to specifications
drawn up by the precast companies; in fact professional site fixers
and welders supervised and executed some of the laboratory work.
Infill mortar was mixed in the laboratory using 3:1 sand:OPC with
a water/cement ratio of 0.45 and 75mm slump. An expanding
agent Conbex was added to the cement at 0.5% by weight. Mortar

19 August 2003 – The Structural Engineer|21


paper: elliott et al

because the effective depth to the connector dc is 110, 260 and


410mm, respectively. It is found that MU is approximately propor-
tional to dc–50mm, resulting in a horizontal tensile force of about
460kN in the connecting weld. The stress in the weld is therefore
460 × 103/20 × 80 = 288N/mm2, some 11% greater than the spec-
ified strength of 258N/mm2.In the billet tests the hogging moment
MU varies from –20 to –50kNm as the position of the connector
rises in the deeper beams, from 140 to 290mm from the bottom.
Consequently the lever arm changes,giving rise to small increases
in MU. For sagging moments of +22 to +37kNm there is no corre-
lation with depth.
The M-φ plots for the double sided sway tests are shown in Fig
18.The behaviour of corresponding single sided and double sided
tests (e.g. compare B2 and B8a) is surprisingly similar when the
behaviour of the column is considered. In the double sided tests
large field tension caused diagonal cracking in the column which
did not occur in the single sided test where diagonal tension does
not exist.
The M-φ plots for the single and double sided tests with floors
are shown in Fig 19. (Note: all beams are 300mm deep). The
obvious differences in these results are not down to the type of
measured from the gradient of displacements at 300mm and Fig 16. connector, although the welded plate once again out performs the
110mm from the column face. M-φ data for the single billet connector, but are due to the loading regime and sub-frame
The tests were terminated either when the ultimate capacity sided hogging geometry.The presence of opposing moments in the two beams in
MU of the specimen had been reached, or when the specimen moment tests the sway test W10 creates large bending moments and localised
failed in the mechanical connector itself, e.g. in tests ref. W5, B6 shear forces in the column, as shown in Fig 7. This has the effect
and W8. It is unfortunate that in the tests B3, B4,W4, DC1 & DC2 of reducing both the strength and stiffness of the connection. In
the full moment capacity of the connector was not always attained. this test the opposing moments MU = +125 and –210kNm
(This is because the main purpose of the tests (by Ragupathy5) was produced a total moment in the column of 335kNm, just less than
to validate a computer program, rather than determine the failure its moment capacity of 350kNm. The tangent stiffness in both
of the connector.) beams in W10 is just 7.8kNm/m.rad. By comparison, the tangent
stiffness in the gravity load test W9 (where beam moments are in
Results of beam-to-column connection tests balance) is 50kNm/m.rad. – more than six times larger. In spite
Moment – rotation (M-φ) behaviour of these differences MU are about the same, i.e. –210, –237 and
The M-φ plots shown in Figs 16 to 19 are divided into groups –238kNm.
according to the geometry of the assemblies, i.e. single or double The M-φ plots for the two single sided tests with floor slab (B7
sided, with or without floors.The important test results are given & W7) are also shown in Fig19. Their behaviour was very much
in Tables 3 and 4 for the single and double sided tests, respectively. dependant on the detailing of the 25mm diameter tie bars placed
The maximum moment achieved at the connector MU was gener- in the floor zone. The moment capacity achieved in the single
ally less than the calculated moment of resistance MRC. Fig 20 sided tests was 130kNm less than in the corresponding double
shows the relationship between MU and MRC. The mean relation- sided tests, e.g. –58kNm in B7 vs –188kNm in B9. The cracking
ship is: pattern in the top of the floor slab, shown in Fig 8, together with
strain data collected from the bars, suggests that the tie bars did
MU = 0.83 MRC ...(1) not all attain their uniaxial yield strain and are not fully effective.
This may be explained by the fact that the tie bars are angled at
The behaviour in the single sided tests is, in several instances, 45° to the direction of the tensile force. When the first cracks
characterised by a very large initial stiffness, typically Si = 30 to appeared in the in situ concrete infill (at approximately M =
80kNm/m.rad. (By comparison, Stanton16 found that Si in typical 40kNm) the bars were thereafter subjected to eccentric tie forces,
r.c. monolithic joints is about 90kNm/m.rad.) This was followed by thereby reducing their axial stiffness and limiting their axial
softer, ductile behaviour resulting in crack widening as shown in strength.
Fig 8. Relative rotations were generally in excess of 0.02 radians Fig 17.
at the ultimate moment. This corresponds to the rotational limit M-φ data for the single Relationship between moments and rotations in the beam and
of a simply supported pinned-jointed beam of 300 × 300mm cross- sided sagging moment connector
section and 6.0m length and shows that the rotational capacity of tests Fig 21 shows the beam-line from which the connector’s moment
the majority of connections is adequate for their respective beams.
The rotational requirements for sway will be even less than this
if the sway deflection /storey height limit is taken as 1/5001.
In general the welded plate (W) and double-cleat (DC) connec-
tors performed better than the billet (B) connector – the average
moment of resistance being 90kNm for the welded plate and
37kNm for the billet. This reflects the nature of the construction
of the connectors where a substantial horizontal shear force resist-
ance is present at the bearing of the welded plate connector, in
contrast to the shear resistance of the dowel in the billet connec-
tor. Further, the ratio MU/MR of the maximum test moment to the
design moment of resistance of the attached beam is at least 0.35
for the welded plate, 0.26 for the double cleat and 0.13 for the billet
connector.
The ultimate hogging (–ve) moment MU in the three welded
plate tests W1, W3 and W5 was almost the same: –74, –78 and
–92kNm, respectively. This is because the position of the connec-
tor, and hence the internal lever arm, is the same irrespective of
the depth of beam. However, in sagging MU in tests W2, W4 and
W6 increased from +28, > +75 to +125kNm, respectively. This is

22|The Structural Engineer – 19 August 2003


paper: elliott et al

a solution for ME/MR, as the value obtained from equation 4 is


always slightly less (therefore conservative) than that obtained
from equation 2. (In graphical terms the two functions are close
parallel lines, never meeting.) However, the intersection of the
beam-line at MRC gives a point from which the characteristic
secant stiffness SRC may be found, as shown in Fig 10e, as follows:

M R - M RC F c
S RC = <
M RC
2E I/L ...(6)

Note that SRC does not exist as a real stiffness: it is a theoreti-


cally convenient value that relates the strength of the connector
MRC to SE. The relationship between SRC and SE may be approxi-
mated as:

SE = 0.9SRC for the tests without floor slabs ...(7a)


and
SE = (0.08SRC + 1.6)2 for the tests with floor slabs ...(7b)

(Eq. 7 is dimensional: SE and SRC are in kNm/m.rad units)

and rotation requirements are satisfied relative to those at the end Fig 18. Solving equations 3 to 7 yields the following empirical equa-
of a beam. The beam-line is drawn for the design moment of M-φ data for the tions:
resistance MR of the beam (or composite beam in the test W9) and double sided sway R V
S 0.29M RC W
for the rotation of the same pin-ended beam given by φR = loads tests ME= MR S - 0.09 ...(8a)
MRL/2EcI, where Ec is taken as 32kN/mm2, and I is the second S _ M R - M RC i WW
T X
moment of area of the flexurally cracked beam. (The beam span for the tests without floor slabs, and:
of L = 6.0m is chosen as typical.) The intersection of the M-φ plot R V
M E = M R SS 0.064 E c I/L d n+
0.64 - 0.09 W ...(8b)
M RC
with the beam-line gives us the ‘beam-End’ requirement at point WW
S M R - M RC E c I/L
E where the moment ME and secant stiffness SE are measured and T X
presented in Tables 3 and 4. ME and SE are important parameters for the tests with floor slabs.
as they constitute input data for semi-rigid frame analysis and Equation 8 is dimensional: M in kNm, EcI in kNm2, and L in
column stability calculations presented in Part 2 of this paper. m.
The relationship between ME and SE,and the resistance MR and Table 5 shows a comparison of ME determined from equation
flexural stiffness EI/L of the beam is, according to Fig 10, given as: 8 and as obtained from the tests.

M R < S E + 2EI/L F
ME = SE ...(2) Calculation model for beam-end moment and stiffness
Equations 3 to 7 may be used to calculate the required beam-end
In the tests a wide range of results for ME and SE was to be moment ME and secant stiffness SE for any of the connectors and
expected given the large geometrical range over which the tests beams studied in this investigation. Partial safety factors must be
were carried out. The flexural stiffness of the beams themselves applied to material strengths (γm = 1.5 concrete, 1.05 rebars to BS
varied from 18 to 138kNm/m.rad. However, the relationship may 8110,) and to the resulting moment (γf = 1.6 for imposed gravity
be approximated as (Fig 22): loads, or γf = 1.2 for wind and imposed loads).
Two cases were studied,W1 and W9 as shown in Fig 21, chosen
M E = 50 S E - 15 ...(3) because of their similar connection type (welded plate) and beam
size (300 × 300mm), but with test W9 including two T25 tie bars
(Equation 3 is dimensional with ME in kNm and SE in in the floor slab. The design beam-end moment MEd is calculated
kNm/m.rad. It can only be used for the experimental results, not in Appendix B. Note that the flexural stiffness is based on the flex-
for the design values presented later in the paper.) urally cracked second moment of area.
The introduction of √SEis non scientific, except for the fact that Although the model given in Appendix B enables the design
S = M/φ = λz2,where λ is the axial stiffness of the connection,repre- Fig 19. moment and stiffness to be calculated and compared with specific
sented as a linear spring, and z is the internal lever arm of the M-φ data for the single test data, it is necessary to progress this work by developing
section. As the moment M is related to z, then M will be propor- and double sided analytical techniques to cater for other/generic precast connectors.
tional to some function of √S. Further, when the moment ME and hogging moment tests The development of the analytical theory given in Part 2 of this
stiffness SE are normalised with respect to the moment capacity with floor slabs paper is based on the compatibility of deformations observed in
and flexural stiffness of the beam there is remarkable correlation,
as shown in Fig 23, as follows:

M E = 0.8 K - 0.09 ...(4)


s, E
MR
where the ratio of the connector-to-beam stiffness Ks,E is given
as

K s, E = SE ...(5)
4E c I/L
Figs 22 and 23 both suggest that some initial stiffness is present
even at very low magnitudes of bending moment, a fact rein-
forced by the very high initial stiffness shown in some of the M-φ
plots. For example, in test B6 (600mm deep beam) the initial stiff-
ness is 85kNm/m.rad, but the eventual stiffness ratio Ks,E is only
0.04.
Although equations 2 to 5 appear to tie up the relationship
between ME and SE, they are as yet indeterminate until correla-
tion is found with the calculated moments of resistance MR or MRC.
Analysis of the results has shown that these equations do not yield

19 August 2003 – The Structural Engineer|23


paper: elliott et al

Table 5: Comparison of test results obtained from empirical equations ability of these connections in a semi-rigid frame analysis. The
Test arrangement Billet Welded plate Single / double cleats
connections were assembled using precast concrete beams and
Eq.8 Test Ratio Eq.8 Test Ratio Eq.8 Test Ratio columns, and in some cases hollow cored floors and continuity tie
steel, all of which are typical industrial components. Single sided
Single sided no slab 22.8 20.1 1.13 64.0 91.6 0.70 34.6 31.8 1.09
and double sided connections were subjected to gravity hogging
17.1 21.7 0.79 29.3 28.6 1.03 34.1 29.2 1.17
moments and opposing sway moments.The moment ME and stiff-
25.0 29.3 0.85 64.3 73.1 0.88
ness SE requirements of the connector at the end of the beam are
24.9 33.5 0.74 85.0 75.0 1.13
determined using the intercept of the M-φ plot with the beam-line,
45.1 45.8 0.98 66.2 74.3 0.89
33.6 31.7 1.06 135.7 125.1 1.08
drawn for the moment of resistance of the beam MR.

Single sided with slab 79.5 54.2 1.47 103.0 127.9 0.81 no tests
The main conclusions are:
Double sided sway no slab 17.2 17.8 0.97 28.6 40.0 0.71 25.3 11.0 2.30
• All connections tested are semi-rigid and non-linear.
22.8 26.0 0.88 63.8 48.8 1.31 22.0 13.2 1.67
• The tests ultimate moment MU varies from 11 to 238kNm, and
Double sided with slab 182.3 188.2 0.97 259.0 232.0 1.12 no tests the rotational stiffness S varies from 200 to 27 000kNm/radian
180.3 188.2 0.96 259.0 237.0 1.09 depending on beam size, type of connector and presence of tie
84.8 120.5 0.70 bars.
226.5 170.5 1.33 • The calculated moment of resistance of the connector MRC
predicts failure, on average for the 28 tests, to within 1% of the
test value MU, although individual values are much less accu-
rate.
• Knowing only MR and MRC, for which a calculation model is
given, the moment ME and stiffness SE requirements of the
connection at the end of the beam and may be determined
using empirical calculations. ME and SE are then the input
data for semi-rigid frame analyses.
• A calculation model is given for evaluating design values for
MEd and SEd for the welded plate connector with and without
floor slabs.

Appendix A
Calculation models for flexural strength MRC
The moment of resistance of the connector MRC (at the end of the
beam) is based on the equilibrium of all forces present in the
connector, i.e. the beam reinforcement is excluded, but the tie
steel above the beam is included.The following calculation model
the tests. Because of the way in which moment–rotation curves Fig 20. uses experimental observations to determine the various contri-
for precast connections resemble moment–curvature for mono- Relationship between butions to MRC together with the BS 8110 rectangular stress block
lithic r.c. joints, partial safety factors and theoretical relationships ultimate test moment using the as-tested material data.
adopted here are the same as for monolithic r.c. MU and calculated
The experimental results given in this paper show that propri- moment MRC in the (a) Billet connector in beam alone
etary precast beam-column connections can be used for the semi- connectors for all tests In the billet tests subjected to hogging moment, the ultimate
rigid analysis of precast frames. Part 2 of this paper will show mode of failure was by bending and slipping of the top locating
whether the existing code of practise (BS 8110) will accommodate Fig 21. cleat,leading to large cracks of up to 5mm wide.Although the hori-
this approach. Comparison of M-φ zontal capacity of the locating cleat was not measured in the test,
behaviour using its calculated capacity is based on the bending strength of a grade
Conclusions identical connectors 43 rolled angle, py = 275N/mm2, 100 × 100 × 12mm thick × 80mm
28 full scale bending tests have been carried out on precast for the double sided wide,subjected to a horizontal force acting at a lever arm of 45mm.
concrete beam-to-column connections using a wide range of test with floor slabs This gives Fs1 = 30.0kN. It is assumed that the full shear strength
proprietary mechanical connectors. The tests have generated and the single sided of the 16mm diameter dowel, taken (BS 5950, Part 1,Table 32) as
moment vs rotation (M-φ) data that are used to assess the suit- test without floor slabs pq = 0.69 × py = 520N/mm2, is mobilised given the displacement
observed at the seating level.The shear capacity is Fs2 = 201 × 520
= 104.5kN. The flexural strength of the infill concrete is taken as
0.67fcui even though the infill is not confined by links.
The depth to the neutral axis xc is given as:
Fs1 + Fs2
xc = ...(A1)
0.67f cui 0.9b
and
MRC = (Fs1 + Fs2) (dc – 0.45 xc) ...(A2)

where dc is the effective depth to the centroid of the summation


of all the internal forces.

As an example the hogging moment of resistance for the


connector in test B1 is calculated as follows: b = 300mm, effective
depth to the locating cleat = 300mm and to the seating billet =
140mm, fcui = 42.5N/mm2, dowel py = 754N/mm2, pq = 0.69py, dowel
area Asd = 201mm2. The tensile capacity of the locating cleat =
30.0kN. Then Fs1 = 30.0kN due to locating cleat, and Fs2 = 0.69 ×
754 × 201 = 104.5kN due to dowel in shear;‘net’ effective depth dc
= 30.0 × 300 + 104.5 × 140 /134.5 = 175.7mm; neutral axis depth
(eq. A1) xc = 17.5mm; zc = 175.7 – 0.45 × 17.5 = 168mm; MRC =
134.5 × 0.168 = 22.6kNm. Values for MRC are given in Tables A1
and A2.

24|The Structural Engineer – 19 August 2003


paper: elliott et al

of the beam in all cases, the calculated value of MRC is about


–74kNm.

In the sagging mode, where the top of the connection is in


compression:
Fw
xc = ...(A3)
0.67f cui 0.9b
and

MRC = Fw (dc – 0.45xc) ...(A4)

Values for MRC are given in Tables A1 and A2, The table over-
leaf is a summary.

(c) Billet and welded plate connectors in beam with floor slabs
Strains measured in the tie bars in the double sided tests attained
yield strains at more than 3000µε at ultimate failure.Thus, in the
bars Fs1 = fy As. It is assumed that the weld and/or dowel are fully
Fig 22. (b) Welded plate connector in beam alone mobilised as described above, i.e. Fs2, although this is unlikely to
Relationship between For the welded plate tests subjected to a hogging moment, the be the case as the neutral axis is close to the seating level.
beam-end moment ME weld appears to be fully mobilised at a tensile strength Fw = However, MRC is calculated according to equations A1 and A2.
and beam-end 412.8kN.As the centroid of the weld is at 200mm from the bottom In the single sided tests the tie bars are cranked at 45° to the
stiffness SE for all tests
Table A1: Calculation data for moment of resistance of connections in single sided tests.
NB: Design uses BS 8110 rectangular stress block method without partial safety factors.
Test Cross section Effective Effective Effective Shear Tensile Tensile Tensile Depth to Connector
ref at depth to level depth to depth to strength of strength strength strength NA of moment
1
connection of connector locating site bars threaded of weld3 of locating of site bars6 concrete capacity
bxh seating cleat dowel2 cleat4 MRC
(mm) (mm) (mm2) (mm) (kN) (kN) (kN) (kN) (mm) (kNm)
B1 300 x 300 140 300 104.5 30.0 17.5 –22.6
B2 300 x 300 160 104.5 15.0 +16.0
B3 300 x 500 190 390 90.5 30.0 22.0 –27.7
B4 300 x 500 310 90.5 11.2 +27.6
B5 300 x 600 290 490 104.5 30.0 13.5 –44.2
B6 300 x 600 310 104.5 8.5 +32.0
B7 300 x 300 140 300 450 104.5 30.0 198.0 44.0 –106.1
W1 300 x 300 200 412.8 49.4 –73.4
W2 300 x 300 100 412.8 45.8 +32.8
W3 300 x 450 200 412.8 45.8 –74.0
W4 300 x 450 250 412.8 41.3 +95.5
W5 300 x 600 200 412.8 48.2 –73.6
W6 300 x 600 400 412.8 47.3 +156.3
W7 300 x 300 200 450 412.8 203.5 75.7 –153.1
DC1 250 x 450 140 340 104.5 68.85 22.9 –36.2
DC2 250 x 450 310 104.5 12.5 +31.8
Notes: 1. Includes depth of floor slab; 2. Shear stress = as measured tensile yield strength × 0.69. ps = 520N/mm2, or 450N/mm2 nominal in
tests B3 and B4. Area of dowel = 201mm2; 3. Neglects any local bending in weld. py = 258N/mm2, length = 80mm nom., throat thickness =
20mm nom; 4. Least value of tensile capacity of 16mm dia. grade 4:6 set screw = 30.6kN, or bending capacity of 100 × 100 × 12 rolled angle ×
80 long = 30kN; 5. Tensile capacity of grade 4:6 M24 threaded rod. Stiffened cleat capacity assumed to be greater than this; 6. Effective stress
in the 45° cranked bars taken from strain gauge data as 0.4fy. fcui and fy given in Table 1

Table A2: Calculation data for moment of resistance of connections in double sided tests.
N:B Design uses BS 8110 rectangular stress block method without partial safety factors
Test Cross section Effective Effective Effective Strength of Tensile Tensile Tensile Depth to Connector
ref at depth to level depth to depth to bolts2 or strength strength strength NA of moment
connection1 of connector locating site bars threaded of weld4 of locating of site bars concrete capacity
bxh seating cleat dowel3 cleat5 MRC
(mm) (mm) (mm2) (mm) (kN) (kN) (kN) (kN) (mm) (kNm)
B8a 300 x 300 160 104.5 13.7 +16.1
B8b 300 x 300 140 300 104.5 30.0 17.6 –22.6
W8a 300 x 300 100 412.8 50.6 +31.9
W8b 300 x 300 200 412.8 50.6 –73.2
SC1a 300 x 300 160 113.0 14.1 +17.4
SC1b 300 x 300 140 113.0 14.1 –15.1
B9a 300 x 500 140 300 450 104.5 30.0 508.6 76.6 –230.4
B9b 300 x 500 140 300 450 104.5 30.0 508.6 76.6 –230.4
W9a 300 x 500 200 450 412.8 508.6 112.2 –264.9
W9b 300 x 550 200 450 412.8 508.6 112.2 –264.9
W10a 300 x 500 300 412.8 48.0 +114.9
W10b 300 x 500 200 450 412.8 510.6 107.5 –267.7
Notes: 1. Includes depth of floor slab; 2. Shear capacity of M24 grade 4:6 bolts = 56.5kN per bolt; 3. Shear stress = as measured tensile yield
strength × 0.69. ps = 520 N/mm2. Area of dowel = 201mm2; 4.Neglects any local bending in weld. py = 258N/mm2, length = 80 mm nom., throat
thickness = 20mm nom; 5. Least value of tensile capacity of 16mm dia. grade 4:6 set screw = 30.6kN, or bending capacity of 100 × 100 × 12
rolled angle × 80 long = 30 kN. fcui and fy given in Table 1

19 August 2003 – The Structural Engineer|25


paper: elliott et al

Comparison of test and calculated connector moment of


resistance
Billet connector Welded plate connector
Beam Hogging Sagging Hogging Sagging
depth
Test Calc Test Calc Test Calc Test Calc
300 –20.1 –22.6 +21.7 +16.0 –92.5 –73.4 +28.6 +32.8
450 –78.5 –74.0 +75.0↑ +95.5
500 –37.4 –27.7 +37.2↑+27.6
600 –49.9 –44.2 +32.9 +32.0 –74.3 –73.6 +125.1 +156.3
↑ signifies ultimate moment not achieved in test

direction of the principal plane and are therefore partially effec-


tive. Strain gauge data show that failure of the connection
occurred when the strains in the bars at failure were about
1050µε.4, i.e. the stress in these bars was about 0.4fy or 210N/mm2.
The force in the cranked region is therefore limited to Ft = 0.4 fy
As. Values for MRC are given in Tables A1 and A2. The following
table is a summary.
_1256x350 i + _ 982x550 i Fig 23.
Summary of MRC values from Tables 3 & 4 d= = 438mm Relationship between
2238
Double sided Single sided Double sided the moment ratio
gravity load sway load MRd = 978 × (438 – 0.45 × 161) × 10–3 = 357.3kNm ME/MR and the
Test Calc Test Calc Test Calc stiffness ratio KsE
Step 2. Connector capacity MRCd
Billet –188.2 –230.4 –58.0 –106.1 Fs = 0.95 × 460 × 982 × 10–3 = 429kN
–191.3 –230.4
Fweld = 215 × 80 × 20 × 10–3 = 344kN
_ 429 + 344 i x10 3
Welded –237.0 –264.9 –156.4 –153.1 –210.4 –267.7
plate –238.8 –264.9 +125.1 +114.9 xc = = 159mm (almost same as x = 161mm
0.45x40x0.9x300
Appendix B
Calculation of design values for beam end moment MEd and
stiffness SEd
Determine the bending strength and stiffness of the welded plate
beam-to-column connection shown in Fig B for a 6.0m long ×
400mm deep × 300mm wide precast concrete beam. In addition
200mm deep precast floor slabs are placed on top of the beam and
the gap between their ends is reinforced using two T25 bars, and
filled using grade C25 in situ concrete.
Use fcu = 50N/mm2 for the beam, fcui = 40N/mm2 for the infill, fy
= 460N/mm2 for the reinforcing bars, pw = 215N/mm2 for weld, Ec
= 32 000N/mm2 for concrete, Es = 200 000N/mm2 for rebars and
weld, centroidal cover to bars = 50mm from top of slab. Use BS
8110 rectangular stress block approach and partial safety factors.

Step 1. Beam capacity MRd


Fs = 0.95 × 460 × (1256 + 982) × 10–3 = 978kN

x= 978x10 3 = 161mm
0.45x50x0.9x300

Table B1: Calculation of design beam-end values for MEd and SEd for standard sizes of beams and standard welded Fig B: (above)
plate connectors Welded plate beam-
h Effective Effective xd MRd Effective xcd MRCd I cracked L SRCd SEd MEd to-column
depth to depth to depth to for composite connection
tie bars beam bars weld beam
(mm) (mm) (mm) (mm) (kNm) (mm) (mm) (kNm) (mm4) (mm) (kNm/m.rad) (kNm/m.rad)(kNm)
300 450 250 161 259.5 200 159 206.5 1.003E+09 6000 41.7 24.4 198.2
400 550 350 161 357.3 200 159 249.4 1.699E+09 6000 41.9 24.5 203.0
500 650 450 161 455.1 200 159 292.3 2.616E+09 6000 50.1 31.5 232.4
600 750 550 161 552.9 200 159 335.2 3.757E+09 6000 61.7 42.8 273.3
300 450 250 161 259.5 200 159 206.5 1.003E+09 8000 31.3 16.8 189.2
400 550 350 161 357.3 200 159 249.4 1.699E+09 8000 31.4 16.9 193.4
500 650 450 161 455.1 200 159 292.3 2.616E+09 8000 37.6 21.2 218.3
600 750 550 161 552.9 200 159 335.2 3.757E+09 8000 46.3 28.1 252.8
300 450 250 161 259.5 200 159 206.5 1.003E+09 10000 25.0 13.0 185.3
400 550 350 161 357.3 200 159 249.4 1.699E+09 10000 25.2 13.0 189.2
500 650 450 161 455.1 200 159 292.3 2.616E+09 10000 30.1 16.0 211.1
600 750 550 161 552.9 200 159 335.2 3.757E+09 10000 37.0 20.8 241.3
Variable parameters are: depth of beam = 300, 400, 500, 600mm; Beam span = 6.0, 8.0, 10.0m; Constant values are:
b = 300mm; hollow core floor slab depth = 200mm;
As in beam 4 T20 bars = 1256mm2 top & bottom; As in topping 2 T25 bars = 982 mm2; centroidal cover = 50mm
fcu = 50N/mm2; fcui = 40N/mm2; Ec = 32 kN/mm2; fy = 460N/mm2;
Weld size 80 × 20 root, weld py = 215N/mm2; weld tensile capacity = 344kN

26|The Structural Engineer – 19 August 2003


paper: elliott et al

Table B2: Calculation of design beam-end values for MEd and SEd for standard sizes of beams and standard billet connectors.
h Effective Effective Effective xc MRCd I cracked L SRCd SEd MEd
depth to depth to depth to for composite
tie bars dowel top cleat beam
(mm) (mm) (mm) (mm) (mm) (kNm) (mm4) (mm) (kNm/m.rad) (kNm/m.rad) (kNm)
300 450 140 300 110 186.2 1.003E+09 6000 27.2 14.2 150.3
400 550 140 290 110 228.8 1.699E+09 6000 32.3 17.5 171.4
500 650 290 390 110 278.5 2.616E+09 6000 44.0 26.2 214.8
600 750 290 490 110 331.9 3.757E+09 6000 60.2 41.2 275.1
300 450 140 300 110 186.2 1.003E+09 8000 20.4 10.4 148.3
400 550 140 290 110 228.8 1.699E+09 8000 24.2 12.5 166.6
500 650 290 390 110 278.5 2.616E+09 8000 33.0 18.0 203.6
600 750 290 490 110 331.9 3.757E+09 8000 45.1 27.2 255.0
300 450 140 300 110 186.2 1.003E+09 10000 16.3 8.4 149.1
400 550 140 290 110 228.8 1.699E+09 10000 19.4 9.9 165.7
500 650 290 390 110 278.5 2.616E+09 10000 26.4 13.8 198.4
600 750 290 490 110 331.9 3.757E+09 10000 36.1 20.2 243.7
Variable parameters are: depth of beam = 300, 400, 500, 600mm; Beam span = 6.0, 8.0, 10.0m; Constant values are:
b = 300mm; hollow core floor slab depth = 200mm;
As in beam 4 T20 bars = 1256mm2 top & bottom; As in topping 2 T25 bars = 982mm2;
fcu = 50N/mm2; fcui = 40N/mm2; fy = 460 N/mm2;
Grade 8:8 dowel; area = 201mm2; shear strength py = 375N/mm2; shear capacity = 75.4kN
Top locating cleat tensile capacity = 30.0kN

in beam) Step 5. Beam-end secant stiffness SEd


_ 344x200 i + _ 429x550 i Eq. [7] SEd = (0.08 × 41.9 + 1.6)2 = 24.5kNm/m.rad.
dc = = 394.3mm
773
Step 6. Beam-end stiffness ratio Ks,Ed
MRCd = 773 × (394.3 – 0.45 × 159) × 10–3 = 249.4kNm Eq. [5]

Step 3. Second moment of area for the composite beam, I K s, Ed = 24.5x6000 x10 9 = 0.68
cracked 4x32000x1.7x10 9
Modular ratio = 200/32 = 6.25. Net value (less concrete area Step 7. Beam-end moment of resistance MEd
replaced) = 6.25 – 1 = 5.25. (i) from experimental results
Height above bottom of beam to neutral axis in cracked section: Eq. [4] MEd = (0.8 ×√0.68−0.09) × 357.3 = 203.5kNm

300x /2 + _ 6.25x982x550i + _ 6.25x1256x350i + _5.25x1256x50i


2

xc = c or (ii) directly from the beam-line


300xc + 9 6.25 _ 982 + 1256i C + _5.25x1256i Eq. [2]

∴ I cracked = 1.7 × 109mm4. M Ed = 24.5 x357.3 = 205.3kNm


24.5 + < x32000x1.7 F
2
6000
Step 4. Calculate secant stiffness SRCd
Eq. [6] Calculations for the design moment and stiffness for a range of
9
beams sizes, beam spans and connector type are given in Tables
S RCd = 249.4 x 2x32000x1.7x10 # 10 - 9 = 41.9kNm/m.rad B1 and B2.
_ 357.3 - 249.4 i 6000

REFERENCES

1. BS 8110: 1985, The Structural Use of Concrete, British Standards Institute, London of Civil Engineering Structural Connections, COST C1 International Conference,
2. Mahdi, A. A., ‘Moment rotation effects on the stability of columns in precast Liege, September 1998, p 65-74
concrete structures’, PhD Thesis, University of Nottingham, United Kingdom, 12. ENV 1994-1-1, Eurocode 3, Part 1.1 – General Rules for Buildings: Annex J on Steel
1992 Joints in Building Frames, British Standards Institute, London, 1994
3. Elliott, K. S., ‘Multi-storey Precast Concrete Framed Structures, Blackwell Science, 13. Ferreira, M.A. and Elliott, K. S., ‘Strength-stiffness requirement approach for
Oxford, United Kingdom, 1996, 624 pp semi-rigid precast connections’, University of Nottingham Report 2002
4. Gorgun, H., ‘Semi-rigid behaviour of connections in precast concrete struc- 14. Ife, J. S., Uzumeri, S. M. and Higgins, M. W., ‘Behaviour of the ‘Cazalay Hanger’
tures’, PhD Thesis, University of Nottingham, United Kingdom, 1997 subjected to vertical loading, PCI Journal, 13/6, Dec. 1968, p 48-66
5. Ragupathy, R., ‘Semi-rigid connections in precast concrete frames’, PhD Thesis, 15. Precast / Prestressed Concrete Institute, Design and Typical Details of Connections
City University, London, 1993 for Precast and Prestressed Concrete, 2nd. Ed., Chicago, IL, USA, 1988
6. Guo, M., ‘Finite element analysis of confined concrete in building frame compo- 16. Stanton, J. F., Anderson, R. G., Dolan, C. and McCleary, D. E., ‘Moment resist-
nents and joints’, PhD Thesis, University of Southampton, UK, 1998 ant connections and simple connections’, PCI Special Research Project, No. 1/4,
7. Elliott, K. S., Davies, G., Gorgun, H. and Adlparvar R., ‘Stability of semi-rigid 1986
precast concrete structures’, PCI Journal, 43/2, March-April 1998, p 42-60 17. Keronen, A., ‘Effect of semi-rigid connections in reinforced precast concrete
8. Elliott, K. S., ‘Semi-rigid connections in precast concrete structures and bridges’, portal frame – load tests’, Report 69, Technical University of Tampere, Finland,
Control of the Semi-rigid Behaviour of Civil Engineering Structural Connections, COST 1996
C1 International Conference, Liege, September 1998, p 1-10 19. Comair, F. and Dardare, J., ‘Model testing of precast semi-rigid beam-to-column
9. Elliott, K. S., Davies, G., Mahdi, A. A., Gorgun, H., Virdi. K. and Ragupathy, P., connection’, COST C1 Proc. 1st Workshop, Semi-rigid Behaviour of Civil Engineering
‘Precast concrete semi-rigid beam-to column connections in skeletal frames’, Structural Connections, E.N.S.A.I.S., Strasbourg, France, 1992, p99-119
Control of the Semi-rigid Behaviour of Civil Engineering Structural Connections, COST 20. DeChefdebien, A. and Dardare, J., ‘Experimental investigations on current
C1 International Conference, Liege, September 1998, p41-50 connections between precast concrete components’, COST C1 Proc. 2nd
10. Guo, M. and Jolly, C. K., ‘Finite element models for the analysis of steel and Workshop, Semi-rigid Behaviour of Civil Engineering Structural Connections, CTU
concrete in construction’, Advances in Computational techniques for Structural Prague, Czech Republic, 1994, p 21-30
Engineering, Ed. B. Topping, Civil-Comp Press, 1996, p 283-294 21. Ferreira, M.A., ‘Deformability of beam-column connections in precast concrete
11. Jolly, C. K., Guo, M., Virdi, K. and Ragupathy, P., ‘Application of numerical analy- structures’, PhD Thesis, School of Engineering of Sao Carlos, University of Sao
sis to connections in precast concrete frames’, Control of the Semi-rigid Behaviour Paulo, Brazil, February 1999 [In Portugese]

19 August 2003 – The Structural Engineer|27

You might also like