Professional Documents
Culture Documents
Synopsis le= embedment length for tie steel bars Kim S. Elliott
This is Part 2 of a paper dealing with the semi-rigid behaviour of lp= plastic hinge length BTech, PhD, CEng,
connections in precast concrete framed structures. In Part 11 the pw = yield stress of weld MICE
results of 28 full scale beam-to-column connection tests were pq = shear stress of dowel
School of Civil
presented in terms of flexural strength and stiffness. In Part 2, (1/r)cr = curvature of the beam (or beam + slab) based on the Engineering, University
equations are developed to predict moment-rotation (M–φ) flexurally cracked section of Nottingham,
behaviour, and a method to calculate the effective length α= column to beam flexural stiffness ratio with rigid
factors β for columns in sway frames containing semi-rigid connections Gwynne
connections is presented. α′ = modified α value with semi-rigid connection
The moment of resistance of the connection MRC is the sum of β= column effective length factor
Davies
the yield strength of all components present at the end of the δ= deformation BSc, PhD, CEng,
beam. The mean value of the ratio of the calculated-to-test γm = partial safety factor for material MIStructE, MICE
moment is 1.01. Connection rotation is the aggregate sum of λ= axial or shear stiffness of weld, cleat or dowel School of Civil
deformations at the interface, plus beam and column curvature φ= relative beam-to-column rotation Engineering, University
of Nottingham, UK
acting over a length related to the dimension of the beam or φc = calculated relative rotation at moment MRC
column. The correlation with experimental M–φ curves is in φR = relative rotation capacity of a simply supported beam
some cases better than 95% accurate. The analysis shows that η= lever arm factor in reinforced concrete flexural design Marcelo
whilst some connectors are suitable for semi-rigid frame design, Ferreira
non-compliant ones should be classified as pinned. The semi- Abbreviations MEng, MSc, PhD,
rigid behaviour of precast beam-column connections and the W= welded plate beam-column connector CEng
axial load capacity of columns can be obtained through rational B= billet beam-column connector University of Sao Paulo,
calculations. Brazil
Introduction Formerly Post-Doctoral
Notation The structural behaviour of framed structures is greatly influ- Researcher, Department
As = area of tie steel enced by the flexural strength and rotational stiffness of the beam- of Civil Engineering,
University of
Asd = area of dowel column connections. Indeed the main focus of attention in
Nottingham
Aw = area of weld structural steelwork and precast concrete frames has, over the
Ec = concrete Young’s modulus past 30 years, been aimed at characterising the behaviour of these
Es = steel Young’s modulus connections in terms of their own internal mechanisms as well as Halil Gorgun
F= force in weld, cleat or dowel for the global behaviour of the frame in which they exist.Although BSc, MSc, PhD
I= second moment of area design and computational convenience has favoured the use of Lecturer, Dicle
Ibeam = flexurally cracked second moment of area of beam (or pinned or rigid connections (the latter being fully moment resist- University, Turkey
beam + slab) ing), the majority of connections act in a semi-rigid manner as Formerly Research
Icol = flexurally cracked second moment of area of column illustrated by the experimental results on precast concrete connec- Assistant, Department of
Civil Engineering,
K= column axial load reduction factor tions given in Part 1 of this paper1. The experiments showed the University of
Ks = normalised joint stiffness = SL/4EI diversity of these connections, ranging from little more than Nottingham, UK
L= span of beam pinned (moment of resistance at the connector face MRC = 11kNm)
Madd = second order column bending moment to effectively rigid (MRC = 239kNm) with the ‘relative’ rotational
ME = experimental connector moment at limiting beam rota- stiffness S (between beam and column) varying from 200 to 27
A. A. Mahdi
tion 000kNm per radian, respectively. The work highlighted the need BSc, MSc, PhD,
MEc = predicted connector moment at limiting beam rotation to quantify MRC and S in terms of the geometrical and material MICE
MR = moment of resistance of beam properties of both the connection and the adjoining members. Hyder Consulting, UK
MRC = moment of resistance of connector In contrast with the large number of experimental investiga- Formerly Research
N= column axial load tions carried out to evaluate the semi-rigid behaviour of precast Assistant Department of
Civil Engineering,
Nmax = maximum value of N concrete beam-column connections2,3,4,5,6 there is still a lack of University of
Nbal = column design axial load capacity of a balanced section analytical equations to predict their semi-rigid behaviour, defined Nottingham, UK
Nuz = column design axial squash load by the moment-rotation (M-φ) curve in Fig 1. However, the
S= rotational stiffness = M/φ complexity of details found in precast connections means that the Received: 07/02
Sc = rotational stiffness = MRC/φc success of analytical modelling still depends on the knowledge of Accepted: 02/03
SE = experimental connector rotational stiffness at limiting the internal mechanisms of deformation in the connection zone, Keywords: Precast
beam rotation particularly where grouted-in steel billets, dowel bars, bolts and concrete,
Semi rigid, Frames,
SEc = calculated connector rotational stiffness at limiting stiffened cleats are present.Consequently experimental validation Beams, Columns,
beam rotation is required before an attempt to develop analytical equations for Connections,
X= depth to neutral axis use in connection design can be made. This is presented in Part 1 Equations, Effective
au = second order column sway deflection of this paper.The deformation patterns of cracking and curvature length, Design
recommendations
b= breadth of section found in the experimental tests provide vital evidence to justify
d= effective depth to rebar, weld, cleat or dowel the nature of the analytical equations for M-φ developed in this
fcu = compressive cube strength of concrete paper.
fy = yield stress of reinforcement A recent proposal made by Ferreira7 suggested that in calcu-
h= depth of section lating the important parameters of the M-φ curve, defined in Fig
hcol = depth of column 1 as MRC, S and φc, the requirements of flexural strength and rota-
le = column effective length = β lo tional stiffness would be met simultaneously.Additionally this also
lo = clear column length between restraints implies that rotational capacity φu should be determined, partic-
ularly if moment redistribution at the connection is being consid- MRC = fy As (d – 0.45X) = η fy As d ...(4)
ered. This is because the input data required for stability analy-
0.45f y A s
sis are determined from the M-φ curve at the intersection E of the Fig 2. where h = 1 -
0.67f cu 0.9bd
beam-line defined in Fig 1 – see Part 1 of this paper.(To distinguish Effective length
between point E obtained from the calculated and experimental factors used for For the welded plate connector alone (no slabs) the term fy As
M-φ curves, the calculated beam end moment is hereby defined as column buckling and in equations 3 and 4 is replaced with pw Aw for the tensile strength
MEc and the calculated secant stiffness as SEc). There are two second order of the weld.Similarly,for the billet alone pq Asd is used for the shear
possibilities for the output data MEc and SEc (see Fig 1a and 1b) analysis strength of the dowel, and so on.
depending on the relationship between the moment and the rota-
tion of the beam-end and the connector at the limit of propor-
tionality. The stiffness SEc may be more conveniently expressed
non-dimensionally in terms of the rotational stiffness of the beam
EcIbeam/L, as:
S Ec L
Ks = ...(1)
4E c I beam
where Ec is taken as the short term modulus of elasticity
32kN/mm2 (for grade C60 precast concrete), Ibeam is the second
moment of area of the flexurally cracked reinforced beam, and L
is the effective span of the beam (chosen typically as 6.0m).
Although the correlation between the experimental values of Ks
and ME was shown in Part 1 of this paper, Part 2 now presents the
derivation of the analytical equations necessary to calculate Ks
and MEc. These may then be used in the calculation of column
effective length factors leading to a proposed method for the design
of columns in semi-rigid frames.
The notion of using effective length factors β to assess the buck-
ling capability of a column has found favour with designers, Fig
2. Simple equations for β have been presented in terms of column
end boundary conditions and/or relative frame stiffness functions,
so that the designer may compute not only column buckling capac-
ities but also second order deflections au and ultimate second
order bending moments Madd. BS 81108 and EC29 adopt the
approach whereby column end conditions are equated to the total
relative stiffness ΣI/L of the column(s) to that of the beam(s)
framing into the ends of a column (defined as αc in BS 8110 and
as k in EC2).However,no reference is made to the strength or stiff-
ness of the connection thereby excluding semi-rigid connections.
To date design guidance on semi-rigid connections,e.g.EC3,Annex
Fig 3.
Equilibrium of forces
at a connector
The total end relative rotation φc arises from three deforma- column beyond the top and bottom of the beam, i.e. twice hcol).
tions. Referring to Fig 4 these are:
1. Joint opening at the interface. This is associated with the
a) beam-column interface rotation due to the joint openings; elongation of the top reinforcing bars, or fixing cleats or plates as
b) beam end rotational deformation within the connection zone shown in Fig 4a. The rotation φ = δ/d, where the deformation δ is
(due to the beam end curvature along a plastic hinge length lp); equal to either:
c) column rotational deformation within the connection zone (due i) the yield strain in the steel bars fy/Es × embedment length le; or
to column curvature over a distance equal to the depth of the ii) the axial or shear force F/stiffness λ of a fixing cleat, dowel or
plate;
4(a) depending on the detail. The length le is taken as the lesser of
a length over which the stress distribution along the bar is
uniform, or the length available, as defined in Fig 5. If the bar is
stressed on both sides of the joint the maximum embedment
length is equal to half of the depth of the column. Then:
f y le F
zc =
Es d
or md
...(5)
4(c)
4(b)
Fig 4.
Components of
beam-to-column
relative rotation.
(a) Interface
deformation.
(b) Beam curvature.
(c) Column curvature
Table 1: Predicted moments of resistance, rotation and stiffness of connectors (billet and welded plate types only)
Test Connector Icracked for Icracked for Section Depth to Depth to Embed Plastic Connector Beam-end Beam-end
ref MRC beam column depth h weld cleat or length le hinge rotation moment stiffness
dowel length φc MEc SEc
lp
(kNm) (mm4) (mm4) (mm) (mm) (mm) (mm) (mm) (m.rad) (kNm) (kNm/mrad)
B1 –22.6 3.44E+08 3.44E+08 300 – 300 – 400 3.96 –22.6 0.53
B2 +16.0 3.04E+08 3.44E+08 300 – 160 – 400 17.03 +16.0 0.32
B3 –27.7 8.61E+08 3.44E+08 500 – 500 – 600 2.77 –27.7 1.27
B4 +27.6 7.47E+08 3.44E+08 500 – 310 – 600 7.28 +27.6 1.11
B5 –44.2 2.03E+09 3.44E+08 600 – 600 – 700 3.08 –44.2 2.48
B6 +32.0 1.77E+09 3.44E+08 600 – 310 – 700 9.64 +32.0 1.52
B7* –106.1 1.00E+09 3.44E+08 500 – 300 300 600 6.66 –106.1 6.23
W1 –73.4 3.44E+08 3.44E+08 300 200 – – 400 8.11 –73.4 2.59
W2 +32.8 3.04E+08 3.44E+08 300 100 – – 400 9.12 +32.8 0.73
W3 –74.0 6.65E+08 3.44E+08 450 200 – – 550 7.37 –74.0 4.36
W4 +95.5 5.78E+08 3.44E+08 450 250 – – 550 8.20 +95.5 5.90
W5 –73.6 2.03E+09 3.44E+08 600 200 – – 700 6.24 –73.6 4.44
W6 +156.3 1.77E+09 3.44E+08 600 400 – – 700 7.91 +156.3 10.88
W7* –153.1 1.00E+09 3.44E+08 500 200 – 300 600 8.77 –153.1 10.03
B8a +16.1 3.04E+08 3.44E+08 300 – 160 – 400 17.04 +16.1 0.32
B8b –22.6 3.44E+08 3.44E+08 300 – 300 – 400 3.67 –22.6 0.53
W8a +31.9 3.04E+08 3.44E+08 300 100 – – 400 9.06 +31.9 0.71
W8b –73.2 3.44E+08 3.44E+08 300 200 – – 400 8.10 –73.2 2.57
B9a* –230.4 1.00E+09 3.44E+08 500 – 300 150 600 5.20 –230.4 26.75
B9b* –230.4 1.00E+09 3.44E+08 500 – 300 150 600 5.20 –230.4 26.02
W9a* –264.9 1.00E+09 3.44E+08 500 200 – 150 600 5.82 –264.9 43.13
W9b* –264.9 1.00E+09 3.44E+08 500 200 – 150 600 5.82 –264.9 43.13
W10a* +114.9 1.26E+09 3.44E+08 500 300 – – 600 7.14 +114.9 6.24
W10b* –267.7 1.05E+09 3.44E+08 500 200 – 300 600 13.83 –226.3 19.35
* tests including 200mm deep slabs; Es = 200kN/mm2, Ec = 32kN/mm2; Fweld = 412.8kN, shear stiffness = 600kN/mm; Fcleat = 30.0kN, axial stiffness = 42.5kN/mm; Fdowel =
104.5kN (or 90.5 kN in tests B3 and B4), shear stiffness = 50kN/mm; All columns 300 × 300mm with 4 no. T32 bars at 50 centroidal cover
bring MRC into exact agreement with the test results for B9 and
W9.
In these tests, the ratio ultimate design force in the rebars to
the concrete (Asfy/fcubd) varied between 0.08 and 0.11, the average
being 0.087. Applying the aforementioned factor of 0.82 to this, it
is proposed that if Asfy > 0.07fcubd the joint is adequately reinforced
to give confidence in the semi-rigid data.
or measured M-φ curve occurs at a rotation much greater than φc, Fig 7. MEc = MRC ...(9b)
it has no bearing on the eventual outcome of the calculated beam- Experimental and
end moment MEc and stiffness SEc. In contrast to this, the single- predicted M-φ curves and the secant stiffness SEc is the lesser of:
sided welded plate connections (Fig 9 and 10) have a more defined for single sided billet
M RC
yield point which is predicted rather well by MRC and φc. These connectors in S Ec = ...(10a)
zc
plots do have a bearing on the values of MEc and SEc due to the hogging moment
greater moment capacity and stiffness of these connections. or, if φc is greater than the requirements of the beam-line:
The predictions for the stiffness of the double-sided hogging M RC _ 2E c I beam /L i
moment tests (Fig 11) are in very good agreement with experi- S Ec = ...(10b)
mental results – the mean of the four tests being 0.95 of the _ M R - M RC i
predicted value. The agreement may have been even better had Values for MEc and SEc are given in Table 1.
the calculation for φc included the contribution of the mechanical The stiffness factor Ks is thus determined from equations 1 and
parts of the connections (welded plate, dowel), although it can be 10.It is used to calculate the effective length of a column in a frame
shown that the stiffness of the reinforcing bars is the dominant containing the said connections. From this, second order bending
factor in the value for φc. The moment prediction is unconserva- moments Madd may be calculated and compared with the allow-
tive – up to 20% greater as strain measurements5 suggest that the able beam-end moment capacity MEc. If Madd < MEc the frame may
bars did not achieve their yield strength prior to bond failure in be designed using semi-rigid connections, considering the stabil-
the top of the section. Applying a factor of 0.82 to 0.84 to fy would ity of the column in each storey in turn. If two connection are
Parametric study
Precast concrete sway frames are analysed either as unbraced
frames, Fig 12a, or as partially braced frames, Fig 12b, where
shear walls provide lateral bracing up to a certain level.Three sub
frames, labelled F1, F2 and F3 in Fig 13, were identified for the
analysis. Sub frames F1 and F2 represent the upper floor and the
ground storey, respectively, in an unbraced frame. Sub frame F3
represents the upper storey in a partially braced frame immedi-
ately above the level of the bracing.The columns surrounding the
bracing walls in the sub frame F3 are fully encastre over the
storey height, and may therefore be considered fully rigid at their
upper end. In all cases the semi-rigid connections are located at
le = β lo ...(15)
The maximum axial load Nmax that the column may carry until
Madd = MEC is determined from equations 15–17. However, if N <
Nbal, then K = 1, and the solution for Nmax is given as:
2000M Ec
N max = ...(18)
hcol _ l e /b i
2
the present parametric equations, whilst that of α is of lesser Fig 10.
importance over the range studied. Because the influence of Ks on Experimental and Also,if Nmax > Nuz then K = 0 and column moment cannot be crit-
β is different for values of Ks < 2.0 than for 2.0 < Ks < 10.0 sepa- predicted M-φ curves ical. In this case the maximum value of Madd is found by differen-
rate equations are presented as follows. for single sided tiating eq. 17 with respect to N such that
Referring to Figs 14a and 14b,the data for the upper storey sub- welded plate J 2 N
frame F1 may be approximated by using the following empirical connectors in K 1 d l n hcol O
K 2000 b O
relationship. The form of the equations is an extension of the sagging moment M add = c N N uz - N 2 m K ...(19a)
N uz - N bal O
effective length equations given in BS 8110, Part 2, clause 2.5. K O
K O
1 a L P
b=1+ + for 0.1 < K s # 2
0.2 + 10.0K s 2
0.3 + 1.8K - 0.45K s 2M add
...(12a) = N uz - 2N = 0 ...(19b)
2N
Fig 11.
b=1+ 1 + a for 2 < K s # 10 Experimental and Therefore Nmax = Nuz/2.
2
7.4 + 7.4K s - 0.4K s 1.6 + 0.3K s
...(12b) predicted M-φ curves
for double sided Values for Nmax are given in Tables 2 and 3. Several values of
Referring to Fig 14c, the data for the ground storey sub-frame billet and welded Nmax are very small, indicating that these connections are unlikely
F2 may be given as: plate connectors to be considered in a semi-rigid design.The most seriously affected
with floor slabs in have Ks less than 0.11. For the upper storey columns (Table 2)
b=1+ 1 + a for 0.1 # K s # 2 hogging moment those that are not limited by second order moments are:
2
2.0 + 2.0K s + 4.0K s 4.0 + 0.5K s
...(13a)
Referring to Fig 14d, the data for the upper storey sub-frame
F3 may be given as:
Table 2: Design of column in an UPPER storey precast frame. Design carried out to BS 8110 for an ultimate column axial
force Nmax that produces a maximum Madd
Test α Ks β Slenderness Column Column Deflection Maximum MEc Column design
ref ratio Nmax K factor5 au Madd limited by
(kN) (mm) (kNm) Madd?
B1 2.00 0.07 6.76 60.8 41 1.00 555 22.6 22.6 Yes
B2 2.00 0.05 7.62 68.6 23 1.00 706 16.0 16.0 Yes
B3 0.48 0.07 3.25 27.1 252 1.00 110 27.7 27.7 Yes
B4 0.48 0.07 3.25 27.1 251 1.00 110 27.6 27.6 Yes
B5 0.26 0.06 2.94 23.5 532 1.00 83 44.2 44.2 Yes
B6 0.26 0.04 3.36 26.9 295 1.00 109 32.0 32.0 Yes
B7 0.29 0.29 1.69 15.2 >1679 0.69 24 40.4 106.1
W1 2.00 0.35 3.54 31.9 482 1.00 152 73.4 73.4 Yes
W2 2.00 0.11 5.76 51.9 81 1.00 403 32.8 32.8 Yes
W3 0.65 0.31 2.11 17.9 1538 0.75 36 55.7 74.0
W4 0.65 0.48 1.82 15.4 >1679 0.69 25 41.6 95.5
W5 0.26 0.10 2.36 18.9 >1679 0.69 37 62.3 73.6
W6 0.26 0.29 1.66 13.3 >1679 0.69 18 30.8 156.3
W7 0.29 0.47 1.48 13.3 >1679 0.69 19 31.1 153.1
B8a 2.00 0.05 7.58 68.2 23 1.00 698 16.1 16.1 Yes
B8b 2.00 0.07 6.74 60.7 41 1.00 552 22.6 22.6 Yes
W8a 2.00 0.11 5.84 52.5 77 1.00 414 31.9 31.9 Yes
W8b 2.00 0.35 3.55 32.0 478 1.00 153 73 73.2 Yes
B9a 0.29 1.25 1.24 11.1 >1679 0.69 13 21.6 230.4
B9b 0.29 1.22 1.24 11.2 >1679 0.69 13 21.8 230.4
W9a 0.29 2.02 1.19 10.7 >1679 0.69 12 19.9 264.9
W9b 0.29 2.02 1.19 10.7 >1679 0.69 12 19.9 264.9
W10a 0.28 0.23 1.80 15.0 >1679 0.69 23 39.4 114.9
W10b 0.28 0.87 1.30 11.7 >1679 0.69 14 23.9 226.3
1. Beam and column I values based on flexurally uncracked section; 2. Beam length = 6000mm; 3. Column clear height = 3000mm – beam
depth; 4. Columns section 300 × 300mm with 4 no. T32 bars at 50mm centroidal cover. fcu = 50N/mm2, fy = 460N/mm2; 5. Column K factor =
(Nuz- N)/(Nuz – Nbal) where Nuz = 3358kN and Nbal = 938kN
a b
c d
bending moment Madd at the end of the column is not greater than column and no value of N would under any circumstances produce Fig 14.
the moment capacity MEc at the end of the beam. In some cases a situation where Madd > MEc. In this case a semi-rigid connection Variation in column
Nmax is quite small, between 23 and 250kN, resulting from second is entirely satisfactory. effective length
order deflections of more than 100mm, showing that a semi-rigid The results presented in Tables 2 and 3 show that the connec- factor β with Ks and
connection would not be appropriate.However,in several cases the tions fall into two distinct groups: α. (a) Upper storey
maximum axial load was greater than the squash load of the I Those in which the capacity of the column is severely limited sub-frame F1 with 0
< Ks < 10. (b) Upper
Table 3. Design of column in an GROUND storey precast frame. Design carried out to BS 8110 for an ultimate column storey sub-frame F1
axial force Nmax that produces a maximum Madd with Ks < 2. (c)
Test α Ks β Slenderness Column Column Deflection Maximum MEc Column design Ground storey sub-
ref ratio Nmax K factor5 au Madd limited by frame F2 with Ks < 2.
(kN) (mm) (kNm) Madd? (d) Upper storey
B1 1.76 0.07 1.90 18.0 463 1.00 49 22.6 22.6 Yes sub-frame F3 with Ks
B2 1.76 0.05 1.91 18.2 324 1.00 49 16.0 16.0 Yes < 2.
B3 0.42 0.07 1.57 14.4 894 1.00 31 27.7 27.7 Yes
B4 0.42 0.07 1.57 14.4 891 1.00 31 27.6 27.6 Yes
B5 0.23 0.06 1.53 13.7 >1679 0.69 20 33.0 44.2
B6 0.23 0.04 1.54 13.8 >1679 0.69 20 33.4 32.0
B7 0.25 0.29 1.40 13.3 >1679 0.69 19 31.1 106.1
W1 1.76 0.35 1.73 16.5 >1679 0.69 28 47.3 73.4
W2 1.76 0.11 1.87 17.8 691 1.00 47 32.8 32.8 Yes
W3 0.57 0.31 1.47 13.6 >1679 0.69 19 32.4 74.0
W4 0.57 0.48 1.39 12.9 >1679 0.69 17 29.0 95.5
W5 0.23 0.10 1.50 13.5 >1679 0.69 19 31.9 73.6
W6 0.23 0.29 1.40 12.6 >1679 0.69 17 27.7 156.3
W7 0.25 0.47 1.32 12.6 >1679 0.69 16 27.6 153.1
B8a 1.76 0.05 1.91 18.1 325 1.00 49 16.1 16.1 Yes
B8b 1.76 0.07 1.90 18.0 463 1.00 49 22.6 22.6 Yes
W8a 1.76 0.11 1.87 17.8 670 1.00 48 31.9 31.9 Yes
W8b 1.76 0.35 1.73 16.5 >1679 0.69 28 47.4 73.2
B9a 0.25 1.25 1.15 10.9 >1679 0.69 12 20.8 230.4
B9b 0.25 1.22 1.15 10.9 >1679 0.69 12 20.9 230.4
W9a 0.25 2.02 1.10 10.4 >1679 0.69 11 18.9 264.9
W9b 0.25 2.02 1.10 10.4 >1679 0.69 11 18.9 264.9
W10a 0.25 0.23 1.43 13.1 >1679 0.69 18 30.1 114.9
W10b 0.25 0.87 1.20 11.4 >1679 0.69 14 22.9 226.3
1. Beam and column I values based on flexurally uncracked section; 2. Beam length = 6000mm; 3. Column clear height = 3000mm – 1/2 beam
depth; 4. Columns section 300 × 300mm with 4 no. T32 bars at 50 centroidal cover. fcu = 50N/mm2, fy = 460 N/mm2; 5. Column K factor = (Nuz-
N)/( Nuz – Nbal) where Nuz = 3358kN and Nbal = 938 kN
Table 4: Ratio of β factor obtained from empirical equations to computer program mechanical connectors;
Ks Frame F1 Frame F2 Frame F3
ii) beam curvature acting over a length which is related to the
α = 0.5 α = 1 α=2 α = 0.5 α = 1 α=2 α = 0.5 α = 1 α=2 depth of the beam;
0.1 1.058 1.051 1.158 0.968 0.958 1.038 0.960 0.935 1.019 iii) column curvature acting over a zone equal to twice the depth
0.2 1.069 1.075 1.185 1.034 1.000 1.055 1.034 0.984 1.042 of the column.
0.5 1.037 1.035 1.110 1.067 1.026 1.059 1.064 1.015 1.053 5. The correlation with experimental M-φ curves is in some cases
1 1.006 1.000 1.034 1.035 1.011 1.041 1.031 1.000 1.036 better than 95% accurate, particularly for the strongest con-
2 1.003 0.998 1.031 1.003 1.000 1.020 0.997 1.001 1.000 nections.
6. A second order stability analysis was used to determine effec-
by large second order deflections due to low values of connec- tive length factors β of columns in one-storey × one-bay sub-
tion strength and stiffness calculated from the equations. In frames in unbraced and partially braced sway frames.
general this group belongs to 300mm deep beams acting only 7. Column effective length factors β increase due to:
with mechanical connectors. i) an increasing number of semi-rigid connections per column;
II Those where second order deflections and moments have no ii) an increase in α, the relative stiffness of the columns to the
restriction on axial load capacity. These include all connections beam members;
incorporating tie steel in the floor slabs and some of the deeper iii) a decrease in Ks, the relative stiffness of the connection to a
beams acting with mechanical connectors. fully encastre beam member. The influence of Ks on β is
greater for values of Ks < 2 than when Ks > 2.
It should be noted that Nmax is an ultimate design load inclu- 8. Parametric equations have been presented for the variations
sive of partial safety factors for concrete (0.45fcu = 0.45 × 50) and in β with Ks and α. The results enable designers to determine
steel bars (0.95fy = 0.95 × 460), whereas MEc is calculated for β factors for situations currently not catered for in codes of
actual material strengths without partial safety factors. However, practice, in particular the upper storey in a partially braced
the division between groups I and II above is so large that this will frame.
not change the outcome of the final conclusion. 9. Second order bending moments, calculated according to BS
8110, are compared with the flexural strength of the connec-
Recommendations for designers tions MEc to determine whether a semi-rigid frame design is
The translation of the experimental results, via the develop- permitted.
ment of this analytical theory, to the application in design has 10. The main conclusion is that those connections which
involved the simultaneous confrontation of beam-column incorporate continuity tie bars such that the area of rein-
connections with the members they connect.The rotational stiff- forcement As > 0.07fcubd/fy, or where the span/beam depth
ness of the connector may be calculated for given geometry, is less than 13, are suitable for semi-rigid frame design,
connector type and material strengths.This yields an expression whilst the non-compliant ones should be classified as
for second order bending moments in a sway frame. The total pinned.
moment, due to frame action and second order effects are
compared with the calculated moment of resistance of the Appendix A (see table below)
connector, and if satisfactory may therefore be classified in
design as a ‘semi-rigid’ connector. Raw output data for β factors from computer program
The design recommendations necessary to achieve this are: Ks Frame F1 Frame F2 Frame F3
1. The moment of resistance MRC of the connector is based on an α = 0.5 α=1 α=2 α = 0.5 α=1 α=2 α = 0.5 α=1 α=2
internal couple between the concrete at the end of the beam in 0 2.00 2.00 2.00 2.70 2.70 2.70
compression and tensile components of the connector. A BS 0.1 2.73 3.74 5.22 1.62 1.77 1.87 1.95 2.24 2.48
8110 rectangular stress block approach is used. 0.2 2.09 2.80 3.86 1.46 1.64 1.78 1.70 2.00 2.30
2. For typical precast concrete beam and column sizes, namely, 0.5 1.59 2.04 2.73 1.28 1.45 1.62 1.43 1.69 2.01
300-600mm deep, MRC should be at least 75kNm for use in 1 1.39 1.70 2.24 1.19 1.33 1.51 1.30 1.52 1.82
semi-rigid design. 2 1.28 1.52 1.94 1.14 1.25 1.42 1.22 1.37 1.68
3. The axial force capacity of the tie steel, or the tensile compo- 4 1.23 1.43 1.78 1.11 1.21 1.36 1.19 1.35 1.59
nents if no tie steel exists, should be at least 0.07fcubd. 10 1.19 1.37 1.67 1.10 1.18 1.32 1.16 1.30 1.53
4. The secant stiffness SRC of the connection is calculated and,
when normalised with respect to the flexural stiffness of the
beam (4EI/L), is used to determine the effective length of the
REFERENCES
column. The rationale for the new effective length factors is
the same as in BS 8110. 1. Elliott, K. S., Davies, G., Ferreira, M. A., Gorgun, H., Mahdi, A. A.: ‘Can precast concrete
5. The ratio of SRC to 4EI/L should be greater than about 0.3 to structures be designed as semi-rigid frames – Part 1 the experimental evidence’ The
have any significant influence on the column’s effective length Structural Engineer, 2003, 81/16, p 14-27.
factor. 2. Stanton, J. F., Anderson, R. G., Dolan, C. and McCleary, D. E.: ‘Moment resistant connec-
6. For a typical beam span of 6.0m, the depth of the beam, or tions and simple connections, PCI Special Research Project, No. 1/4, 1986
beam plus floor slab, should be at least 450mm. 3. Mahdi, A. A.: ‘Moment rotation effects on the stability of columns in precast concrete struc-
tures’, PhD Thesis, University of Nottingham, United Kingdom, 1992
Conclusions 4. Ferreira, M. A.: ‘Deformability of beam-column connections in precast concrete struc-
1. Theoretical equations are proposed to calculate the moment of tures’, PhD Thesis, School of Engineering of Sao Carlos, University of Sao Paulo, Brazil,
resistance MRC and rotation φc of a range of semi-rigid precast February 1999 (In Portugese)
concrete beam-to-column connections. This enables a 5. Gorgun, H.: ‘Semi-rigid behaviour of connections in precast concrete structures’, PhD
moment-relative rotation (M-φ) curve to be predicted from the Thesis, University of Nottingham, United Kingdom, 1997
geometric and material parameters of the connection. 6. Ragupathy, R.: ‘Semi-rigid connections in precast concrete frames’, PhD Thesis, City
2. Using the beam-line approach, the required beam-end University, London, 1993
moment MEc and secant stiffness SEc are obtained from the 7. Ferreira, M. A, and Elliott, K. S.: ‘Strength-stiffness requirement approach for semi-rigid
predicted M-φ data. precast connections’, University of Nottingham Report 2002
3. It is proposed that MRC is derived from the sum of the yield 8. BS 8110: 1997, The Structural Use of Concrete, British Standards Institute, London
strength of all components present at the end of the beam.The 9. ENV 1992-1-1, Eurocode 2, Design of Concrete Structures Part 1 - General Rules and Rules for
mean value of the ratio of the calculated moment to the test Buildings, British Standards Institute, London, 1991
moment is 1.01. 10. ENV 1994-1-1, Eurocode 3, Part 1.1 – General Rules for Buildings: Annex J on Steel Joints in
4. Connection rotation is the aggregate sum of three deforma- Building Frames, British Standards Institute, London, 1994
tions: 11. Aksogan, O. and Gorgun, H.: ‘The non-linear analysis of planar frames composed of flex-
i) beam-column interface due embedded reinforcing bars or urally connected members’, Cukurova University Journal, 8/2, December 1993, p 117-129
37|The Structural Engineer – 19 August 2003 19 August 2003 – The Structural Engineer|37