ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 110-S16

Lap Splices of Bars in Bundles

by John Cairns

European (EC2) and ACI 318-08 design code rules for lap splices parameters vary over a wide range.3,4 The data are, however,
of reinforcing bars within a bundle differ markedly—the former almost entirely restricted to splices between individual bars,
permits the same or shorter lap-splice lengths with respect to where all bars are spliced at the same cross section.
splices of individual bars, while the latter requires longer laps. This Although there is a small amount of test data on anchorage
paper reports an investigation into the performance of lap splices of bundled bars,5 there appears to be an almost complete
of individual bars within a bundle of two or three. The results show
absence of test data on lap splices of bars within a bundle.
that the bond strength of individual bars is not reduced on account
of their forming part of a bundle (contrary to the provisions of Stressing of a bundle differs markedly between these two
ACI 318-08). Staggered laps were found to be weaker but fail in situations: in an anchorage, all bars in the bundle are pulled
a less brittle manner when splices are staggered longitudinally as in the same direction, whereas in a splice, pairs of bars are
compared to equivalent laps, where all bars are spliced at the same pulled in opposite directions while the remainder are contin-
section whether or not bars were bundled. Therefore, the outcome uous throughout the splice length. Other, more general differ-
raises the question of the validity of reductions permitted by both ences may also exist between laps and anchorages: anchor-
the ACI 318-08 and European (EC2) codes, where only a portion of ages at end supports often benefit from transverse compres-
the bars is lap-spliced at a section. It is recommended that further sion generated by the support reaction, but lapped splices
investigation of the influence of the proportion of bars lapped at a will invariably be located where transverse stress is negli-
section on splice performance be undertaken. gible. The only study of splices of bundled bars known to the
Keywords: bond strength; brittleness; bundles; lap splices.
author is reported by Bashandy,6 but herein, entire bundles
of up to four bars were spliced at the same cross section
INTRODUCTION (giving a total of as many as eight bars in contact within
In situations where reinforcement is congested, it may be the splice length). Hence, detailing was not representative
advantageous to place bars in bundles of two, three, or four of normal practice nor compliant with ACI 318-08,7 which
states that bars are to be spliced individually and that splices
instead of fixing individual bars at equal spacing.1 Bundles
within a bundle are not to overlap.
permit flexibility in detailing where availability of larger
There are several factors that could be expected to enhance
bars is restricted and ease manual handling on site. Bundling
the performance of splices of single bars within a bundle
of bars allows for increased clear spacing, facilitating
when compared with similar splices of individual bars:
compaction of concrete between bundles. In comparison
1. Clear spacing between bars is increased (for a given
with reinforcement in layers, bundles are more efficient in
section breadth);
maintaining the effective depth of longitudinal reinforce-
2. For a given stirrup diameter and spacing, confinement
ment. Bundles may permit cross sections to be achieved
to each lap within a bundle is increased as splices are stag-
in reinforced concrete that would otherwise require
gered; and
prestressing. Bundling of bars does, however, require some 3. A proportion of the bars will be continuous where a
modifications in detailing of lap splices and anchorages with single bar within the bundle is spliced.
respect to provisions for individual bars. Other factors may have an adverse effect:
The bond action of ribbed reinforcing bars generates 4. The perimeter of the bar in direct contact with concrete
bursting forces that generate circumferential tensile stresses is reduced.
around the bar and tend to split the surrounding concrete Further differences may arise as a consequence of stag-
cover.2 Unless confinement is high (typically where concrete gering splices of individual bars in the bundle:
cover is greater than 5 times the diameter of the lapped bars 5. The distribution of bond stress throughout the splice
or where compressive stress is applied perpendicular to the length may alter; and
bar axis as at an end support, for example), bond failure 6. The share of tension force taken by individual bars may
of laps and anchorages usually occurs in a splitting mode alter as a result of differences in stiffness between spliced
with formation of longitudinal cover cracks throughout the and continuous bars.
bond length, leading to eventual spalling of the cover. Bond The aims of this investigation are therefore to assess
strength in this mode of failure is limited by the resistance of whether the rules in ACI 318-087 for lap joints of bars
the section to these bursting forces. Most design codes recog- within a bundle are soundly based and, if appropriate,
nize the importance of the splitting mode of failure through suggest revisions.
bond strength and detailing provisions linked to minimum
cover thickness and the area of secondary reinforcement
provided. Some codes also reflect the enhancement in bond ACI Structural Journal, V. 110, No. 2, March-April 2013.
obtained in the presence of transverse compressive stress. MS No. S-2011-090.R2 received August 15, 2011, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
Code rules for laps and anchorages of individual bars have reserved, including the making of copies unless permission is obtained from the
been validated against an extensive set of physical test data copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the January-February 2014 ACI Structural Journal if the discussion is
on splices and anchorages of individual bars in which these received by September 1, 2013.

ACI Structural Journal/March-April 2013 1

John Cairns is a Senior Lecturer in structural engineering in the School of the Built be lap-spliced comprises three bars. ACI 318-087 permits
Environment at Heriot-Watt University, Edinburgh, UK. His research interests include only one bar within a bundle to be spliced at any section.
bond between reinforcement and concrete and monitoring, residual strength assess- Compared to a lap splice of a pair of single bars, the lap
ment, and management of deteriorating concrete structures.
length for an individual bar within a bundle is increased by
20% and 33% for two and three bar bundles, respectively.
The increase reflects the reduction in the exposed perimeter
of the group relative to that of the same number of bars fixed
separately (Fig. 1). Lap-splice length for individual bars
within a bundle is, thus, effectively the same as the develop-
ment length for the entire bundle. In addition, as the equiva-
lent bar diameter is used in place of the actual diameter when
calculating the contribution of confinement, the influence of
confinement is effectively downgraded and, consequently,
the increase in splice length may be somewhat greater in
certain circumstances.
Fig. 1—Equivalent diameter and exposed perimeter of bundles. The provisions of EC28 are markedly different from those
of ACI 318-08.7 Where an entire bundle is stopped, anchorage
length is determined using an equivalent diameter in which
the bundle is replaced by a single notional bar with an area
equal to that of the entire bundle. If all bars in the bundle are
of the same diameter, the diameter of the equivalent bar dbn
is db nb , where db is the individual bar diameter and nb is
the number of bars in the bundle. If the equivalent diameter
for a bundle of two bars does not exceed 32 mm (No. 10),
both may be spliced at the same section. Splice length is
based on the equivalent diameter of the pair and splices need
not be staggered. Otherwise, only one bar in a bundle of two
or three bars (that is, three or four bars, respectively, within
the spliced zone) may be spliced at a section and splices
must be staggered longitudinally by a distance of 1.3 times
the lap length. No increase in splice length is required above
that for an individual bar, however.
Figure 2(a) and (b) shows the increase in development and
splice lengths required for bundled bars by the two codes
over that required for an individual bar. It is assumed that
all bars within the bundle are anchored at the same location,
but that only one of the bars within the bundle is spliced at a
section and splices are staggered longitudinally in accordance
with code provisions. The plot takes no account of possible
differences in minimum concrete cover or clear spacing
between bars as a result of bundling. The plots show that
EC28 requires greater increases than ACI 318-087 in devel-
opment (or anchorage) length with an increasing number
of bars in a bundle; for splices of bars within bundles,
ACI 318-087 requires an increase, whereas EC28 does not.
Thus, the approaches and the consequent bundled splice-
length factors in the two codes differ significantly.
The difference in detailing rules for bond of bundled bars
between the two codes reflects a difference in the physical
concept of bond action. The ACI 318-087 approach treats
bond strength as an interfacial shear stress on the bar surface
that is constant under any specific confinement condition. As
Fig. 2—Comparison of ACI 318-087 and EC28 bond lengths only the external surface of the bundle is considered active
for bundles. in bond (Fig. 1), the force that may be transferred is consid-
ered to be reduced if more than two bars are in contact.
RESEARCH SIGNIFICANCE The concept underpinning EC28 is that splice or anchorage
This paper evaluates the provisions of ACI 318-087 for lap capacity is determined by the available confinement from
splices of individual bars within a bundle and identifies defi- the surrounding concrete and transverse reinforcement—
ciencies in the rules. together with transverse pressure where present—to resist
the bursting force generated by bond action. The force to
Review of code provisions be transferred into a bar is the same whether the bar is part
Both ACI 318-087 and Eurocode 2 (EC2)8 impose a of a bundle or not; consequently, EC28 does not require an
maximum limit of four bars in contact (including within a increased bond length. The differences between the two
splice); hence, the maximum bundle within which a bar may documents are therefore not merely an issue of the numer-

2 ACI Structural Journal/March-April 2013

ical value of coefficients but reflect significant differences in
the underlying concept of behavior.
Recent research by Bashandy6 demonstrated that the
equivalent bar approach is valid for simultaneous laps of a
pair of bars in a bundle (in fact, the investigation went well
beyond good detailing practice by simultaneously lapping
entire bundles of up to four bars—a practice not permitted in
either EC28 or ACI 318-087). However, the author has found
no evidence to validate either set of code rules for staggered
laps of individual bars within a bundle of two or three bars.
This investigation was therefore undertaken to assess the
validity of current rules for the dimensioning of lap splices Fig. 3—Influence of proportion of bars lapped at a section
of individual bars within a bundle. on distribution of bar strains through splice length.
Both codes may require that laps within bundles be
staggered. The effect of staggering of laps and its corol-
lary—maintaining continuity of a proportion of bars The overall stiffness of a splice also varies with the
through a splice zone—is not clear. Both EC28 and proportion spliced. The overall elongation dlap of a pair of
ACI 318-087 encourage staggering of laps through reduc- spliced bars over the splice length is the sum of two compo-
nents: 1) the elongation of a spliced bar over the splice
tion factors on lap length, linked with the area of reinforce-
length; and 2) the combined loaded end slips sb of the pair
ment provided in relation to that required by calculation in
of spliced bars (Eq. (2)).
the case of ACI 318-08.7 Although it might intuitively be felt
that bar continuity through a splice zone would be beneficial,
l
test data are scarce. Such test data are derived from tests in d lap = ∫0b e s dx + sb (2)
which a proportion of bars was continuous throughout the
entire span,9,10 whereas in practice, all bars are likely to be
lap-spliced but the splices staggered. In some cases, the lap where es is the axial strain in reinforcement at any point; lb is
lengths used were short10 and their consequent weakness the splice length; and sb is the bar slip at the ends of the splice.
may well have presented continuity of a proportion of bars The first component on the right-hand side of Eq. (2) is
in an unduly favorable light. represented by the area under the various curves in Fig. 3,
The distribution of bond stress over a tension lap length from which it is evident that the proportion of bars spliced
will be affected by the proportion of continuous bars at at a section will affect splice elongation (that is, will affect
the splice section. Consider a splice zone situated within a the stiffness of the splice). Differences in splice stiffness will
constant-moment zone, stressed within the elastic range, and affect the balance between the stress carried by continuous
of sufficient length to allow all bars in the section—whether and spliced bars.
spliced or not—to be under the same strain at midlength.
If all bars are spliced at the same section, the total cross- EXPERIMENTAL PROGRAM
sectional area of reinforcement within the splice length is Design
double that outside. As the force is divided equally between The test program comprised two groups of specimens,
them, then bar stress, and therefore strain, at midlength of each containing five specimens. One group was designed
the splice tends toward a value half that outside the splice around splices in a two-bar bundle and the other was
zone. If only a portion of bars in the section is spliced, the designed around splices in a three-bar bundle. To exclude
total cross-sectional area of reinforcement within the splice the influence of staggering of splice zones from the study,
length is less than double that outside; therefore, the strain splices were staggered in a consistent manner in both bundle
at the midlength must exceed half of that outside (Eq. (1) and individual bar splices. Some reference specimens with
and Fig. 3). all bars spliced at the same section were also included,
however, as a benchmark against the wider data population.
SAso The influence of staggering of laps is being investigated in
1
e ml = e so = e so greater detail in a related study, the initial results from which
SAml (1 + r1 ) (1) have already been reported.9
The reinforcement layouts for all specimens are shown
schematically in Fig. 4, with details of dimensions in Table 1.
where eml and eso are the bar strains at midlength and outside Splice length was set at 20 times the bar diameter throughout
the long splice, respectively; SAml and SAso are the areas of the investigation to ensure that bond failure would precede
reinforcement within and outside the long splice, respec- yield of reinforcement. Each group contained:
tively; and rl is the proportion of reinforcement spliced at 1. A pair of replicate bundle splice beams—designation
the section. “Type B”—in which each bar within a bundle was lap-
While Eq. (1) makes considerable simplifications about spliced with the splices staggered longitudinally. Longitu-
load sharing and bar/concrete slip, it does demonstrate that dinal reinforcement in these specimens comprised six 12 mm
the force to be transferred over the end half of a splice length (0.47 in.) diameter bars, arranged as either three pairs in one
will tend to be higher when only a portion of the bars is group (B2a and B2b) or as two triplets in the other (B3a and
spliced. Splitting bond failure in a long splice is initiated by B3b). The bundle size refers to the number of bars in contact
the peak bond stress near the ends of the splice; hence, the outside the splice zones; within the splice zones, the two- and
average bond strength over the entire splice length at failure three-bar bundle specimens had three and four bars, respec-
tends to decrease as the proportion of bars spliced increases. tively, in contact. The proportion of bars spliced at a section

ACI Structural Journal/March-April 2013 3

 in Specimens B2 and B3 was thus 50% and 33%, respec-
tively. Because of the scarcity of tests in which bars within
a bundle are spliced and in view of the potential for varia-
tions in workmanship causing differences in the compaction
of concrete around and between bundled bars, replicate tests
were conducted on these specimens. Specimens B2a and
B2b were replicate specimens, as were B3a and B3b.
2. One single-bar splice beam—designation “Type S”—
was also reinforced with six 12 mm (0.47 in.) diameter bars
but with bars arranged and spliced individually. Splices were
staggered longitudinally to match the arrangement in the
companion Type B bundled bar specimen.
3. One equivalent bar beam—designation “Type E”—
containing single bars with a cross-sectional area approxi-
mately equal to that of its companion bundled bar specimen.
A single 16 mm (0.63 in.) diameter bar was used in place
of each two-bar bundle (equivalent diameter of 17.0 mm
[0.67 in.]) and a single 20 mm (0.79 in.) bar replaced each
three-bar bundle (equivalent diameter of 20.8 mm [0.82 in.]).
Section breadth was adjusted to maintain a similar reinforce-
ment ratio in bundle and equivalent bar specimens. Indi-
vidual splices were staggered in a manner consistent with
Type B and E specimens. However, it was not possible to
match both the equivalent bar diameter and the proportion
of bars lapped at a section. Beam E2 was reinforced with
three 16 mm (0.63 in.) bars with three staggered splices
(33% spliced at a section), for example, whereas companion
Specimens B2a, B2b, and S2 were detailed with two splice
zones in which 50% of the longitudinal bars were spliced in
each zone.
4. One reference beam—designation “Type R”—using
bars of equivalent diameter to that of its companion bundled
bar specimen but with all bars lapped at the same section.
The purpose of these specimens was to allow splice strengths
measured in the tests herein to be benchmarked against the
wider database of tests in which splices were not staggered.
All beams contained approximately 1.2% longitudinal
reinforcement and minimum covers were 20 or 23 mm
(0.78 or 0.91 in.)—equal to or slightly greater than the
largest equivalent bar size tested. A modest quantity of
secondary reinforcement in the form of closed stirrups was
Fig. 4—Main reinforcement splice details (not to scale). provided in the lap zones of these specimens. The quantity

Table 1—Details of test specimens

Spacings
(Fig. 7) Stirrups
Concrete Bar Total Number Bars Splice Section Section
Beam cube strength diameter number of in % length breadth depth h, 2csi, 2csb, Number per Detail
reference fcu, MPa j, mm of bars groups group spliced lo, mm b, mm mm mm mm lap zone (refer to notes)
B2a 46.7 12 6 3 2 50 240 258 241 67 — 1 1
B2b 46.7 12 6 3 2 50 240 258 241 67 — 1 1
S2 46.7 12 6 6 1 50 240 258 260 51 — 1 1
E2 43.1 16 3 3 1 33 320 228 255 — 98 2 2
R2 43.1 16 3 3 1 100 320 226 254 39 — 3 3
B3a 41.2 12 6 2 3 33 240 258 241 158 — 1 1
B3b 41.2 12 6 2 3 33 240 258 241 158 — 1 1
S3 41.2 12 6 6 1 33 240 266 258 95 — 1 1
E3 42.0 20 2 2 1 50 400 224 303 — 161 2 2
R3 43.1 20 2 2 1 100 400 226 270 94 — 3 3
Notes on stirrup detail: 1 is center of each splice zone; 2 is 40 mm (1.57 in.) from both ends of each splice zone; 3 is 40 mm (1.57 in.) from both ends and center of each splice
zone; 1 MPa = 145 psi; 1 mm = 0.0394 in.

4 ACI Structural Journal/March-April 2013

was kept close to the permitted minimum to reduce uncer- Test results
tainties in the interpretation of the contribution of stirrups The load-deflection response of all beams was close to
to the strength of splices in which only a portion of the bars linear up to peak load. Minor departures were evident at
was spliced at a section. The number of 8 mm (0.31 in.) low loads prior to the initiation of flexural cracking, where
mild steel stirrups provided to each splice zone is detailed in stiffness was slightly greater, and close to failure, where the
Table 1. Stirrups were provided at 300 mm (12 in.) centers response softened slightly. Vertical flexural cracks formed
outside splice zones. first within the constant-moment zone, followed by slightly
inclined flexural cracks within the shear spans. Failure
Materials occurred suddenly on formation of a widening flexural
Longitudinal reinforcement was of Grade 500B in accor- crack near one end of a splice zone and longitudinal cracks
dance with BS 4449:200511 (characteristic of 0.2% of proof along longitudinal tension bars over the splice length. Load
strength of 500 MPa [72.5 ksi]). Bars had pairs of crescent- dropped immediately after the peak was reached in all tests.
shaped ribs on opposite sides of the bar that merge into the Table 2 lists the peak loads and bond strengths for all spec-
core. The relative rib area was not measured on these partic- imens. The average stress in reinforcement at peak load fsu
ular bars but has been found to typically lie in the range of is calculated using the rectangular stress block for concrete
0.055 to 0.065 from similar production. The concrete cover given in EC28 with safety factors taken as 1.0. This stress
to longitudinal bars was provided by proprietary spacers. The block is effectively identical to that in ACI 318-087 for the
concrete was of medium workability (Class S2 in accordance concrete strengths used in this investigation. Reinforcement
with BS EN 206-1:200012 with a slump of 60 mm [2.5 in.]) stress is obtained from the maximum applied moment. A
and a maximum aggregate size of 10 mm [0.375 in.]), quadratic expression is used to determine the tensile force
containing a water-reducing admixture and supplied by in longitudinal reinforcement, which is then divided by the
a local ready mix company. Concrete was compacted by total area of reinforcement at a section outside the splice
internal vibration and subsequently cured under damp burlap zone to obtain an average reinforcement stress at maximum
and polyethylene for at least 3 days before stripping and load. Ultimate bond strength fbu is then calculated by Eq. (3)
storage in the laboratory until testing. Standard cube control as an average value over the splice length.
specimens were taken from each batch and tested at the same
age as the splice specimens. fsu
fbu = (3)
 lb 
Test procedure 4
Beams were tested in four-point bending (Fig. 5) with the  db 
lap zones positioned within the constant-moment zone in all
specimens except E2, where the lap zones extended 28 mm where fsu is the peak load bar stress; lb is the splice length;
(1.1 in.) beyond the point loads. The tension force to be and db is the individual bar diameter. The results from repli-
developed in the reinforcement of Specimen E2 would none- cate bundled splice specimen pairs B2a/b and B3a/b differ
theless have been constant over the entire lap zone when
allowance is made for the effect of inclined cracking within
the shear span.
Load was monotonically increased to failure over a period
of approximately 30 minutes. Load was applied in incre-
ments of 10 kN (2.248 kips) with crack development marked
at each stage. Loading was continued until residual strength
decreased by at least 25% after the peak load was passed.
The rate of displacement was increased during this stage.
Load and midspan deflection were logged at 2-second inter-
vals throughout the loading sequence. Fig. 5—Load arrangement.

Table 2—Test results

Bond strength
Beam reference Maximum load, kN Test, Eq. (3) fb,test, MPa Calculated, Eq. (4) fb,calc, MPa Ductility Dres Bond strength ratio fb,test/fb,calc
B2a 92.5 5.30 3.45 0.76 1.54
B2b 86.7 5.09 3.46 0.70 1.47
S2 93.5 4.86 3.35 0.43 1.45
E2 76.9 4.66 3.85 0.28 1.21
R2 79.6 4.85 3.03 0.83 1.60
B3a 94.3 5.33 3.97 0.27 1.34
B3b 93.4 5.26 3.99 0.75 1.32
S3 101.3 5.26 3.65 0.20 1.43
E3 83.3 5.22 3.75 0.30 1.30
R3 98.9 5.38 3.17 0.23 1.69
Notes: 1 kN = 0.225 kips; 1 MPa = 145 psi.

ACI Structural Journal/March-April 2013 5

 splices where bars are spliced individually (“S” speci-
mens). The direct comparison of bond strengths presented
in Table 2 does not, however, consider differences attrib-
utable to bar size or confinement from concrete and stir-
rups. Minimum concrete cover was essentially constant
throughout the program, but clear spacing between splices
2cso or the greater side cover 2cso (Fig. 7) was greater where
fewer bars were spliced at a section. Various empirical
and semi-empirical expressions have been proposed to
account for the influence of minimum cover, clear spacing
between bars, and confining reinforcement.3,13 One such
expression, adapted by rearranging the design expression
Fig. 6—Typical plot of load versus deflection (Beam B2a) proposed by Darwin et al.14 and subsequently adopted into
showing calculation of deformability index Dres. (Note: 1 kN ACI 318-08,7 is given herein as Eq. (4) and is applicable to
= 0.225 kips; 1 mm = 0.0394 in.) standard ribbed bars in a bottom-cast situation.

  c ω + K tr′  d 
fb,calc = 0.375  b  + 12ω b  fc′ 0.25 (SI)
  db  lb 

(4)
  cb ω + K tr′  db  0.25
fb,calc = 15.5   + 500ω lb  fc′ (in.-lb)
  db 

where fb,calc is the average bond stress over the splice length
in MPa (psi); and fc′ is the measured concrete cylinder
compressive strength, taken as 0.8 times the cube compres-
Fig. 7—Definition of cover dimensions. sive strength in MPa (psi)

cb = cmin + 0.5db (5)

cmax
ω = 0.1 + 0.9 ≤ 1.25 (6)
cmin

where cmax and cmin in mm (in.) are defined in Fig. 7

K tr′ = 6t d SAtr fc′ ( lb n) (SI)

(7)
K tr′ = t d SAtr fc′ ( 2lb n) (in.-lb)
Fig. 8—Influence of bundling bars on bond strength ratio.

by less than 4%—well within the typical scatter of bond t d = 0.03db + 0.22 (SI)
strength measurements. (8)
t d = 0.78db + 0.22 (in.-lb)
An indication of the brittleness of failure is also given by
the quantity Dres, calculated as the ratio of residual load at a
deflection equal to 1.5 times the peak load deflection to the where SAtr is the total area of transverse reinforcement within
peak load itself (Fig. 6). splice length lb crossing the potential splitting plane in mm2
Reference Specimens R2 and R3 were included in the test (in.2); and n is the number of bars spliced at the section.
program to verify that splice strengths measured in the tests Further analysis of the results is based on the bond strength
herein were consistent with existing “best-fit” semi-empirical ratio—the ratio of measured bond strength to that estimated
expressions proposed by others and hence represent a valid by Eq. (4)—listed in the final column of Table 2.
benchmark against which other results reported herein may The influence of bundling of reinforcement is plotted in
be compared. These two beams were both reinforced with Fig. 8, in which the bond strength ratios for bundled bars
single bars, all lap-spliced at the same section—the same (beams designated “B”) are compared with those for beams
form as in the majority of specimens used to calibrate such with the same reinforcement but positioned and spliced
expressions. The strengths measured in Specimens R2 and individually (beams designated “S”). The value plotted for
R3 were 96% and 105%, respectively, of those estimated by bundled laps is the average of two individual tests, while that
the expression proposed by Zuo and Darwin,4 for example, for the individually spliced bars is a single value. Overall, the
and may therefore be considered representative of the larger difference between bundled and individual strength ratios is
body of test data. 2% and is not significant.
Bond strengths in a bundle of two or three (“B” speci- Figure 9 compares the bond strength ratio of bars lap-
mens) average 7% and 1% higher, respectively, than similar spliced within a bundle (beams designated “B”) with that of

6 ACI Structural Journal/March-April 2013

Fig. 9—Bundles and equivalent bars.
Fig. 10—Influence of proportion of bars lap spliced at
their equivalent single-bar (Type E) specimens. The equiva- section on bond strength ratio.
lent bar splice was 20% weaker in splices based on a two-bar
bundle, while in the three-bar bundle, the equivalent splice
was only 2% weaker. These results may appear inconsistent,
but this comparison between the bundled bar laps and their
companion “equivalent bar” specimens overlooks the differ-
ence in the proportion of bars lapped at a section. In beams
modeled on the two-bundle arrangement, 50% of the bars
were lapped at a section in the bundled bar specimens and
33% were lapped at a section in the equivalent bar speci-
mens. The proportions were reversed in specimens modeled
on a three-bar bundle.
Figure 10 plots the bond strength ratio for all specimens
tested herein and shows a reduction in the bond strength
ratio as the proportion of bars lapped at a section decreases.
The bond strength ratio decreased by an average of 20% Fig. 11—Influence of staggering of laps on bond strength ratio.
as the proportion of bars lapped at the section decreased (Note: 1 mm = 0.0394 in.)
from 100 to 33%. The results for lap splices within bundles
are consistent with those for individually lapped bars. A
direct comparison between the equivalent and reference
specimens (Fig. 11) provides further confirmation that
reducing the proportion of bars lapped at a section reduces
splice strength. The only difference between the equivalent
and reference splices was the proportion of bars lapped at a
section. Equivalent splices in which splices were staggered
were 25% weaker in both groups.
Splice strength decreased immediately after peak load in
all tests. No influence of bundle size on brittleness of failure,
as measured by the deformability index Dres, is apparent
(Fig. 12), but there is a marked correlation with the propor-
tion of bars spliced at a section (Fig. 13). The deformability
index Dres decreased from approximately 0.75 where 33% of
bars were spliced at a section to averages of 0.35 and 0.25, Fig. 12—Influence of bundle size on deformability.
where 50% and 100% of bars, respectively, were spliced
at a section. Even in the least brittle configurations, where
one-third of the bars were spliced at a section, residual bond
strength still dropped by 20 to 30% shortly after peak load.
The change is similar for individual lapped bars and for bars
within a bundle. Metelli et al.9 similarly observed less brittle
behavior, where a proportion of reinforcement is continuous
through a lapped splice.

COMPARISON WITH ACI 318-087 PROVISIONS

The original aim of this investigation was to determine
whether splice length of bars within a bundle should be
increased relative to that of individually spliced bars. The
results plotted in Fig. 8 show that bond strength is not
reduced in the bundled detail. Consequently, the increased
splice lengths in ACI 318-087 for laps of bars forming part Fig. 13—Influence of proportion of bars lapped at a section
of a bundle do not appear to be justified. The approach of on deformability.

ACI Structural Journal/March-April 2013 7

EC28 to the design of laps in bundles that does not require length currently permitted for continuity of bars through a
increases for bars in bundles and in which confinement splice zone.
factors are based on the individual bar size provides a more
consistent level of safety than the ACI 318-087 approach. SUMMARY AND CONCLUSIONS
These results indicate that the ability of a lap splice to This study set out to determine whether bond strength is
transfer force is determined more by the resistance to split- reduced where a single bar within a bundle of two or three
ting than by a notional shear stress over the exposed perim- is lap-spliced. Within the scope of the investigation, it is
eter of the bar. It is possible that in conditions of high cover, concluded that:
confining reinforcement, or transverse pressure—where 1. Bond strength is not reduced where an individual bar
a pullout failure occurs instead of a splitting failure—that within a pair or bundle of three bars is lap-spliced and appro-
this conclusion might be inappropriate. The splitting mode priate allowance is made for differences in confinement and
of failure is the weaker and more likely mode in structur- the proportion of bars spliced at a section.
ally significant members; consequently, design rules for lap 2. The results suggest that the strength of a lapped splice
splices are based on this mode. is controlled by resistance to the bursting forces generated
What does emerge from this study, however, is that lap by bond action rather than a notional shear resistance on the
splices of individual bars in bundles may be weaker than exposed portion of the bar surface.
individual lap splices because splice strength decreases with 3. Splicing only a portion of bars at a section reduced
the proportion of the bars spliced at a section. This trend is splice strength whether or not spliced bars formed part
not confined to bundled bars but is also evident in individual of a bundle. Consequently, the reductions in splice length
bar splices (Fig. 10 and 11). This trend is contrary to the currently permitted for continuity of bars through a splice
provisions of both ACI 318-087 and EC2,8 where staggering zone may be unsound, and further investigation as a matter
of lap splices and retaining continuity of a portion of bars of priority is merited.
through a splice zone is considered beneficial. EC28 permits 4. Less-brittle failures were observed where lap joints
splice length to be reduced where only a portion of the were staggered longitudinally over three zones instead of
bars is spliced at a section. If only one-third of the bars are one or two, regardless of whether the lap was between indi-
spliced at a section, a reduction in bond length of 23% is vidual bars or of an individual bar in a bundle. Failure could
permitted. ACI 318-087 effectively permits a similar reduc- not be classified as ductile, however.
tion in splice length for Class A splices. Splices are classi- 5. On the evidence of this study, the requirement of
fied as Class A where the area of reinforcement provided ACI 318-087 to increase the development length for bars
is: 1) at least double that required by analysis; and 2) no spliced within a bundle over that for an individual bar
more than 50% of bars are lapped at a section. Based on the appears to be unwarranted. However, as this increase
admittedly limited number of specimens in this investiga- compensates for an observed reduction in strength believed
tion, it could reasonably be concluded that an increase of to be attributable to staggering of splices, no changes in the
this amount would be justified. ACI 318-087 provisions for splices of bars in bundles should
Various factors that might influence the performance of be introduced until the influence of staggering on splice
splices of single bars within a bundle relative to similar resistance is better understood.
splices of individual bars were listed in the Introduction 6. It is desirable that the further work proposed previously
of this paper. Variations in confinement from cover and use larger-diameter bars more representative of circum-
confining reinforcement have been taken into account in stances in which lapping of bundled bars may be required
the analysis, and the reduction in exposed bar perimeter in practice.
does not appear to influence the results. It can therefore be
inferred that the reduction in splice strength is related to the ACKNOWLEDGMENTS
The assistance of G. Sorley with aspects of the testing program is grate-
distribution of bond stress along the splice length and differ- fully acknowledged.
ences in the share of load between spliced and continuous
bars (Fig. 3). This tentative conclusion must be treated with NOTATION
caution, however, as in these tests, the stagger between cmax = secondary cover dimension (refer to Fig. 7)
lap zones was less than specified in either ACI 318-087 or cmin = minimum concrete dimension (refer to Fig. 7)
EC28 due to physical constraints on the dimensions of test db = bar diameter
fb,calc = average bond stress over splice length calculated
specimens in this investigation. Further investigation on according to expression of Darwin et al.14
splices staggered in accordance with code requirements is fbu = bond stress at peak load from test
therefore required. In practice, bundling of reinforcement fc′ = concrete cylinder compressive strength
is employed only with large-diameter bars and further tests fsu = bar stress at peak load from test
K′tr = parameter representing amount of transverse reinforce-
should be designed accordingly. ment contributing to bond resistance
Strength is not the only aspect of structural performance lb = splice length
that needs to be considered, and structural robustness will n = number of bars spliced at section
be strongly influenced by the brittleness or deformability of nb = number of bars in bundle
splice behavior. The results show an increase in post-peak sb = bar slip at ends of splice
td = dimensionless parameter representing influence of
residual strength where only a portion of the bars are spliced bar diameter
(Fig. 13). The gain in structural robustness as a consequence es = axial strain in reinforcement; eml and eso are strains at
of reduced brittleness is difficult to evaluate and will depend midlength and outside splice
on the structural arrangement considered, particularly on the rl = proportion of reinforcement spliced at section
SAml, SAso = areas of reinforcement within and outside long splice,
degree of indeterminacy. Further detailed work is required respectively
to resolve this issue, but in the meantime, it would seem SAtr = total area of transverse reinforcement within splice
advisable at the very least to remove the reductions in splice length lb crossing potential splitting plane

8 ACI Structural Journal/March-April 2013

 REFERENCES 8. BS EN 1992-1-1:2004, “Eurocode 2: Design of Concrete Structures—
1. Standard Method of Detailing Structural Concrete: A Manual for Best Part 1-1: General Rules and Rules for Buildings,” British Standards Institu-
Practice, third edition, Institution of Structural Engineers, London, UK, tion, London, UK, 2004, 225 pp.
9. Metelli, G.; Cairns, J.; and Plizzari, G., “Influence of Bar Continuity
2006, 188 pp.
on Behaviour of Lapped Splices,” Proceedings of the 2010 fib Congress
2. fib Bulletin 10, “Bond of Reinforcement in Concrete: State-of-Art
and PCI Convention Bridge Conference, Washington, DC, June 2010,
Report,” fib, Lausanne, Switzerland, Aug. 2000, 434 pp. pp. 2904-2915.
3. Orangun, C. O.; Jirsa, J. O.; and Breen, J. E., “A Reevaluation of Test 10. Magnusson, J., “Bond and Anchorage of Ribbed Bars in High-
Data on Development Length and Splices,” ACI JOURNAL, Proceedings Strength Concrete,” PhD thesis, Chalmers University, Göteborg, Sweden,
V. 74, No. 3, Mar. 1977, pp. 114-122. 1997, 234 pp.
4. Zuo, J., and Darwin, D., “Splice Strength of Conventional and High 11. BS 4449:2005+A2:2009, “Steel for the Reinforcement of Concrete.
Relative Rib Area Bars in Normal and High-Strength Concrete,” ACI Struc- Weldable Reinforcing Steel. Bar, Coil and Decoiled Product. Specifica-
tural Journal, V. 97, No. 4, July-Aug. 2000, pp. 630-641. tion,” British Standards Institution, London, UK, 2009, 34 pp.
5. Jirsa, J. O.; Chen, W.; Grant, D. B.; and Elizondo, R., “Development 12. BS EN 206-1:2000, “Concrete. Specification, Performance, Produc-
of Bundled Reinforcing Steel,” Report 0-1363-2F, Center for Transporta- tion and Conformity,” British Standards Institution, London, UK, 2001,
74 pp.
tion Research, the University of Texas at Austin, Austin, TX, 1995, 103 pp.
13. Canbay, E., and Frosch, R. J., “Bond Strength of Lap-Spliced Bars,”
6. Bashandy, T. R., “Evaluation of Bundled Bar Lap Splices,” ACI Struc- ACI Structural Journal, V. 102, No. 4, July-Aug. 2005, pp. 605-614.
tural Journal, V. 106, No. 2, Mar.-Apr. 2009, pp. 215-221. 14. Darwin, D.; Lutz, L. A.; and Zuo, J., “Recommended Provisions
7. ACI Committee 318, “Building Code Requirements for Structural and Commentary on Development and Lap Splice Lengths for Deformed
Concrete (ACI 318-08) and Commentary,” American Concrete Institute, Reinforcing Bars in Tension,” ACI Structural Journal, V. 102, No. 6, Nov.-
Farmington Hills, MI, 2008, 473 pp. Dec. 2005, pp. 892-900.

ACI Structural Journal/March-April 2013 9

 Notes:

10 ACI Structural Journal/March-April 2013