You are on page 1of 12

Magazine of Concrete Research, 2005, 57, No.

1, February, 9–20

Moment redistribution effects in beams
R. H. Scott* and R. T. Whittle†

University of Durham; Arup Research and Development

Moment redistribution in beams has traditionally been considered as an ultimate limit state (ULS) phenomenon
closely associated with considerations of reinforcement ductility. This paper demonstrates that a significant propor-
tion of this redistribution will almost always occur at the serviceability limit state (SLS) because of the mismatch
between the flexural stiffnesses assumed when calculating moments for the ULS and those actually occurring at the
SLS due to variations in the reinforcement layout along the member and the influence of cracking. This is
demonstrated analytically in the paper and comments are made concerning the current recommendations in
BS 8110 concerning member stiffness. Tests on 33 two-span beams are then presented, parameters investigated
being values of redistribution, beam depth, reinforcement arrangements, concrete strength and the effect of brittle
reinforcement. The results confirm that there is significant redistribution at the SLS and that there is scope for
increasing the permissible limits for redistribution beyond those currently prescribed in design codes. This is
supported by further modelling although it is shown that considerations of crack width may become a limiting
factor.

Introduction
Moment redistribution is useful for practical design
Normal practice is to use a linear elastic analysis for as it allows some flexibility in the arrangement of
calculating the bending moment and shear force distri- reinforcement. It can be used to transfer moment away
butions in a reinforced concrete structure. This has the from congested areas (e.g. beam–column connections)
virtue of simplicity as well as permitting results from a into less congested areas (e.g. mid-spans of beams) or
series of analyses to be combined using the principle of help to allow standard reinforcement layouts where
superposition. It is endorsed by major design codes small differences occur in the bending moment distri-
1 2 3
such as BS 8110, EC 2 and ACI 318. butions for a series of beams, thus avoiding the need to
The assumption of linear elastic behaviour is reason- detail each beam separately. In addition, useful econo-
able at low levels of loading but it becomes increas- mies can be achieved when moment redistribution is
ingly invalid at higher loads due to cracking and the applied to different load combinations, resulting in a
development of plastic deformations. Once an element smaller bending moment envelope which still satisfies
cracks the behaviour becomes non-linear but it is still equilibrium.
reasonable to assume that the tension reinforcement Implicit in the current use of moment redistribution
and the concrete in compression both behave elastically is the assumption that sections possess sufficient ducti-
up to yield of the reinforcement. lity for the requisite plastic deformations to occur. De-
Design codes permit elastic analysis to be used at the sign codes achieve this by specifying rules which
ultimate limit state (ULS) but acknowledge this non- ensure that the tension steel must have yielded, expli-
linear behaviour by allowing a limited amount of mo- citly in the case of ACI 318 (which specifies a mini-
ment redistribution from one part of the structure to mum reinforcement strain of 7500 microstrain) and
another. The permissible moment redistribution is implicitly in the case of BS 8110 and EC 2 (which link
4
linked to the ductility of the reinforcement at the ULS. percentage redistribution to neutral axis depth). This
leads to the question of whether an upper strain limit
should be specified in order to avoid rupturing the
* University of Durham, South Road, Durham, DH1 3LE, UK reinforcement since, with small neutral axis depths,
† Arup Research and Development, 13 Fitzroy Street, London, W1T very high reinforcement strains can be expected.
4BQ, UK 1
BS 8110 imposes a minimum neutral axis depth of
(MCR 1208) Paper received 1 December 2003; last revised 2 March 0.11d (where d is the effective depth to the tension
2004; accepted 5 May 2004 steel) although this was introduced for the practical
9

0024-9831 # 2005 Thomas Telford Ltd

5. tion approach). 1 . steel reinforcement. perhaps. Since the reinforcement is not known sequences of a finite limit to reinforcement ductility on until the end of the design process. that the EI for sagging and hogging moments is the This paper presents the results of an investigation same throughout the member (BS 8110’s concrete sec- which aimed to explore the nature of moment redistri. Influence of flexural stiffness on moment redistribution. by far. any assumptions or approximations made in its deter.2). leading to strains tion. assume ment strains developed at the ULS. These the reinforcement will have occurred by the time this are the concrete section (the entire concrete cross sec- neutral axis depth has been reached. cerning the value of flexural stiffness (EI) along the stricting reinforcement strains to a maximum of member. reinforcement strain at the ULS could be the control. as a consequence. and compare the effect limitations should be imposed on the level of reinforce.Scott and Whittle reason that the top surfaces of beams and slabs are mination will directly affect the level of redistribution often quite rough and. such as slabs. L/2 L/2 (a) Flexural stiffness considerations Moment redistribution involves adjustment of the bending moment distribution obtained from a linear 0·2 elastic analysis with the difference between the redis- EIsagging/EIhogging ⫽ 1 (0%) tributed and initial bending moment diagrams being an EIsagging/EIhogging ⫽ 2 (24%) 0·1 indicator of the amount of redistribution which has oc. Then analyse for a range of situations bution as load is increased on a structure and thus provide some design guidance on the issues outlined P above. ling parameter in the more lightly reinforced types of a rectangular continuous beam will have constant flex- members. the concrete. 2005. However. They demonstrated that used. com- axis depth and that failure of a section would always be bined with the reinforcement on the basis of modular initiated by crushing of the concrete in compression. the most frequently the failure mode of a section. It also raised the question of what may not be fully appreciated. BS 8110’s limit has the effect of re. BS 81101 which gives three options for calculating the this value is largely meaningless since gross yield of EI value at any particular section (clause 2. Consequently. ratio). ignoring the reinforcement). This prompted considerable ural stiffness along its entire length) but it also has discussion concerning the ductility requirements for implications for moment redistribution which. This has the practical virtue of simplicity (for example. of altering the EI value along its length. the gross section (the greatly in excess of this nominal value. to of linear strain distribution across the section. Consider the propped cantilever shown in Fig. 57. First. 28 000 microstrain. 0 0·5 1 % redistribution ¼ ⫺0·1   moment after redistribution ⫺0·2 moment before redistribution 3 100 (1) moment before redistribution ⫺0·3 Distance: L (b) The initial elastic bending moment diagram thus forms the baseline for the redistribution calculation and Fig 1. entire concrete cross section with the reinforcement designers have effectively worked on the assumption included on the basis of modular ratio) and the trans- that the reinforcement will be able to develop whatever formed section (the compression area of the concrete level of strain is actually required by a specified neutral cross section. (b) bending moment distributions 10 Magazine of Concrete Research. ignoring the concrete in tension. The first stage in this process is to examine some of the assumptions made when using the code method of elastic analysis to determine bending mo- ment distributions in a reinforced concrete structure. upper limit to moment redistribution should be per. in practical design is the concrete section. it was deemed calculated using the above expression. 1(a). (a) propped cantilever. Recommendations in design codes vary from. No. In reality. EIsagging/EIhogging ⫽ 5 (48%) Bending moment: PL EIsagging/EIhogging ⫽ 100 (96%) curred. An elastic analysis is controlled by assumptions con- tions. EC 22 which is non-specific. mitted in design codes and whether more stringent which has a central point load. sensible to restrict the lever arm used in design calcula. A further option would be to take the cracked The above view was challenged in 1987 by the work section properties including tension stiffening effects of 5 of Eligehausen and Langer who investigated the con. The percentage of moment redistribution at a 0 section along a beam is calculated as follows:. when making the usual assumption at one extreme.

The bending moment distribution post-yield. A further complication occurs once gross section and transformed section approaches are the section cracks as tension stiffening then causes the unsuitable for design calculation of bending moment EI to vary with the applied moment. To illustrate the point. This again changes the EI distribution obtained from analyses that use the concrete section prompting another analysis and another adjustment of approach. This additional. The percentages of redistribution ob. This redistribution will be very small prior to analysis and so on. even at the be adjusted to accommodate the new bending moment serviceability limit state (SLS). to the bution in this paper. then the rect’ EI values can be determined and thus ensure that bending moment distribution is that for EI constant all the calculated bending moment diagram is consistent along the beam and this is the situation for which the with the beam’s own moments of resistance. In the extreme plastic as is commonly assumed at present. be analysed assuming constant EI (the concrete section jority of beams. The reinforcement is then calculated but vary along the span. the reinforcement layout. in spite of the tribution can be quite considerable. mo. will differ from those distribution. moment redistribution will oc. not the absolute values at analysis a distribution of EI which is consistent with any particular section. reinforced than the spans and the effects of cracking. changes in EI and the reinforcement layout must now tions actually developed along the beam. Any change in the relative EI the actual stiffness distribution of the reinforcement values will cause moment redistribution to occur. distributions at the ULS. Consequently. of the gross section. Fig. span) relative to the hogging stiffness (applicable at the Total redistribution at the ULS will actually be the sum support) has the expected effect of shifting moments of the elastic and plastic components. 1(b).5. 1 11 . The based on either the gross or transformed section will beam is then re-analysed but the bending moment change along the span giving relative EI values which distribution will now have changed because of the are no longer unity. This is reinforcement is normally designed at the ULS. 1(b) empha. the beam could first ally be achieved at ultimate. The magnitude of this redis. It will not be all away from the support and into the span. 2005. (BS 8110’s transformed section option). ment behaves elastically. relative to the bending moment distribu. as will be recommendations of clause 2. plotted in Fig. This is likely to be inconsistent with the tion obtained using the concrete section approach. Plastic redistribution is the redis- situation of constant EI all along the beam and thus tribution with which design engineers have traditionally acts as a reference line for the other distributions. calculations. Consequently. been familiar and which they implicitly make use of Increasing the sagging stiffness (applicable in the during traditional moment redistribution calculations. Furthermore. 57. Moment redistribution effects in beams where the EI for sagging moments differs from that for under SLS conditions and that the reinforcement is hogging moments. The above Magazine of Concrete Research. redistribution will be termed plastic redistri- for EIsagging /EIhogging ¼ 1 corresponds. amount of redistribution for which the beam has actu- ment redistribution has occurred in all the other ally been designed and thus the plastic redistribution bending moment distributions even though all sections that actually occurs will be different from the design along the member are always behaving in a linearly value. the resulting reinforcement layout will approach). However. 2. are also shown in Fig. After the reinforcement yields EIsagging /EIhogging of 1. This means that actual EI values this will change the actual distribution of EI. known prior to undertaking the analysis so that ‘cor- the concrete section is used for the analysis. Similar comments can be made concerning use the level of redistribution. both the demonstrated later. 5 and 100 have been selected further redistribution will occur owing to further and the resulting bending moment distributions are changes in the relative values of EI. sises that it is the relative values of EI that control the An apparently attractive alternative is to use for bending moment distribution. of course. To be success- The above has consequences for understanding the ful this requires that the reinforcement layout be behaviour of reinforced concrete beams. It although it is important to recognise the simplifications should be noted that this redistribution is occurring and approximations which result from this. If. tribution caused by the supports being more heavily 1(b) in order to illustrate this point. iterative approach. Thus the bending moment distribu. For instance. when calculated relative to the plastic redistribution in order to offset the elastic redis- constant stiffness distribution. No. This causes yet another cur even though none was anticipated in the calcula. It is very difficult to achieve so an alternative is to try an also the bending moment distribution which will actu. The procedure can become un- cracking but after cracking the ratio of EIspan : EIsupport stable and is thus unsuitable for practical design pur- will normally increase with a corresponding increase in poses. change in the distribution of EI leading to yet another tions. as is usual. it is suggested The redistribution which occurs as a result of this that the only practical approach is to use EI values for mismatch between the actual and assumed EI values the concrete section when calculating these moments will be termed elastic redistribution in this paper. Elastic case when a stiffness ratio of 100 is used the behaviour redistribution can nearly always be expected even becomes very close to that for a simply supported though none may have been anticipated in the design beam. ratios of behaving elastically. a beam designed for no moment elastic fashion.2 in BS 8110. even though the reinforce. redistribution at the ULS will actually have to undergo tained at the support. Thus. in the vast ma. From the above it is suggested that.

pressive cube strength was 51. tension reinforcement provided over the centre support A programme of laboratory tests was undertaken to and D shows the specimen belongs to series D. D′ all in the same horizontal plane. The main tion 7. 250 mm of 10 mm. Other suffices (typically X) are used when specimens were repeated.7% (series bars versus small bars) C). the several specimens within each series (d) different concrete strengths (i. A typical example is B2T12D where B batched and delivered by truck.6. use of high. This enabled bar diameter effects to be examined since the combinations gave approximately Further details are given in the next section. values ranging (c) different arrangements of reinforcement (e. Fig 2. The remain.g. The high- for 30% redistribution from the centre support into the strength concrete used 10 mm aggregate and microsili- adjacent spans and will be described first. An H investigate the above points in greater detail.2 MPa). 2005. D and E) were used. B2T12DH. * indicates normal exit point for wires in gauged bars – see text and Tables 1 and 2. concrete and six with high-strength concrete. which investigated other aspects of Cement (CEM 1) and sand. D. However. Diagrammatic layout of test specimens 12 Magazine of Concrete Research. mal strength concrete had a maximum aggregate size cent spans. 1 . Depths of 400 mm (series A). 2. D′ cranked on plan at lap positions to achieve this. For bar diameters see Tables 1 and 2.25 with adequate workability being achieved A consistent coding system was used to reference the by the addition of admixtures. The average com- Each span was 2435 mm long and loaded approxi.2 m long and Seventeen specimens were cast using normal strength 300 mm wide. 30 mm cover to centre of longitudinal bars. C′. Depths vary – see Tables 1 and 2. C. large from about 0. overall water : cement ratio of 0. It was commercially specimens.8 MPa (standard devia- mately at its mid-point. ca (in slurry form) in addition to Ordinary Portland ing ten specimens. Stirrups: Series A and B: T10s at 200 centres.9 MPa) and the average indirect cylinder strength set of 23 specimens. are described later.e. The ef. suffix would indicate that the specimen was made with fects of the following parameters were investigated: high-strength concrete e. D. was designed was 3. A total of 33 two-span beams. 2T12 is the sections where the reinforcement layout is known. of course.0 MPa (standard deviation 0. All beams 5200 mm long overall ⫻ 300 mm wide. (a) depth of section Each series investigated a different percentage of (b) different values of design moment redistribution tension (top) steel at the centre support. No. 3T16 or 5T12. Some tests were repeated and nom- Specimen details and test procedure inally identical specimens are block shaded in Table 1. an aggregate : cement ratio of 5. glass fibre ferent combinations of bar diameters such as 2T20 or reinforced polymer (GFRP) bars]. The nor- tribution effects between the centre support and adja. as indicated in Fig. shown in Table 1. Bars at C.e.Scott and Whittle is. The average compres- Load Load 1250 1250 * * C A C′ D B D′ * * 2435 2435 All dimensions in mm. C′ and B. The overall water : cement moment redistribution behaviour. Bars at locations A and B all 1500 mm long.5 and an (series B) and 150 mm (series C. 5.g. ratio was 0. 57.3% (series E) up to around 1. were tested to investigate moment redis. the same total area of tension steel. all other specimens R6s at 100 mm centres. not true for the analysis of existing indicates high yield bar reinforcement. Laps all 300 mm. (with the exception of series A which comprised one strength concrete for some specimens) specimen only) achieved this percentage by using dif- (e) the effect of brittle reinforcement [i. All other reinforce- ment (bottom steel at the centre support and top and bottom steel in the spans) was the same for all beams within each series. Bars at A.

1 13 . respectively. 51 strain gauges were tion. Specimen details: 30% moment redistribution Series Specimen code Overall depth: mm Support reinforcement Span reinforcement Top Bottom Top Bottom (A) (B) (C. one each at the locations A. which ensures that the bond characteristic gauges installed in a central longitudinal duct having a between the reinforcement and surrounding concrete is 6 cross-section of 4 mm 3 4 mm in bars of 12 mm dia.0 MPa) and the average indirect cylinder strength strain limit of 10–15% were used in areas where high was 4.63%) B5T12C** 5T12 (1. 2.63%) B3T10D** 3T10 (0. and are also shown in italics in Table 1. ** indicates four strain-gauged bars. High. wherever possible. Further details of the gauging rods 1 and 4.53%) 3T12 3T12 3T20 B B2T20B** 250 2T20 (0. the centre with normal strength concrete contained four gauged bar in a group of three was gauged to maintain symme- bars. bars where both ends were As indicated in Table 1. undisturbed.91%) B5T12B** 5T12 (0. Three specimens with bar reinforcement were rein- zones where rapid changes in strain were expected to forced for zero redistribution. or above but this yield wires instead of bar reinforcement but were other- number was reduced to 30 gauges in the 6 mm and wise a repeat of the strain-gauged specimens in series 8 mm dia.26%) 2T8 2T8 3T8 B2T8EX 2T8 (0. Gauged bars formed part of the main reinforce- contained strain-gauged reinforcing bars. Speci.5 mm 3 2. was not possible in locations where there were only two mens with high-strength concrete contained two gauged bars but the effects of the resulting asymmetry were bars. 2005. one for 55% redistribu- 50 mm in less critical areas.63%) E B2T8E** 150 2T8 (0. Shaded groups of specimens are nominally identical.58%) 3T10 3T10 3T12 B2T12DX 2T12 (0.68%) B2T12DH* 2T12 (0. and above but which was reduced to 3. Moment redistribution effects in beams Table 1.63%) B2T12DXX 2T12 (0. * indicates two strain gauged bars.70%) 3T12 3T12 3T20 B3T16C** 3T16 (1. Details in italics are for high-strength specimens.53%) D B2T12D** 150 2T12 (0.2 mm A further ten specimens. Three specimens (prefixed W) used ribbed high- installed in each bar of 10 mm dia. 15 of the 23 specimens used.83%) B5T12BHX 5T12 (0. one each at locations A and D and referred to as believed to be small. sive cube strength was 120. strain levels were anticipated. bars and to 2. B. Typically.28%) B4T6E** 4T6 (0. Specimens ment in each beam and.89%) B3T16BLX 3T16 (0.2 mm 3 3.30%) This table should be read in conjunction with Fig. D shown in try of stiffness across the beam width.58%) B2T12DHX 2T12 (0.89%) B3T16BL** 3T16 (0.1 MPa (standard deviation 0. as mentioned exited from one end of each bar (see Fig.95%) B5T12BH* 5T12 (0. shown in Table 2. 2) except for previously.93%) 3T12 3T12 3T20 B3T16B** 3T16 (0.93%) B2T20BHX 2T20 (0. Gauges generally had a tion and two specimens examined the effects of Magazine of Concrete Research.5 mm in D. 57.4 MPa (standard deviation strain limit of 3% but high elongation gauges having a 10. C9) (D. have been published elsewhere.83%) B2T20BH* 2T20 (0. The wiring normally strength specimens have the H suffix. Obviously this Fig.5 MPa). bars. 1 and referred to as rods 1–4. C. No. Gauged bars had electric resistance strain technique.63%) B5T8D** 5T8 (0. D9) A B3T16A** 400 3T16 (0. bars. the 8 mm and 10 mm dia. were in 10 mm dia.86%) C B2T20C** 150 2T20 (1. Gauge spacing varied from 12.5 mm in 6 mm tested to investigate other aspects of moment redistribu- and 8 mm dia.

38%) 3T12 3T12 2T16+1T12 F5F13B 30| 250 5 No.58%) 3T10 3T10 3T12 W3T10D** 30 3T10 (0. 25 mm diameter GFRP bars (1. redistributing moments from the spans back to the mens. The specimens in Table 3 will The test procedure was to load the beams incremen. This gave the closest possible match to specimen Table 3. respectively (shown as and a grillage of Demec points. Applied loads. All quoted Typical stress–strain relationships for all the reinfor. and Table 4 for the additional specimens. 2. values of redistribution are calculated relative to the cement types for strains up to 14 000 microstrain are uniform EI case (i. tally until failure occurred. 2005.43%) 3T12 3T12 2T20+1T16 B8T12B 0 250 8T12 (1. support reac- tions and reinforcement strain gauge readings (when Specimens designed for 30% moment redistribution appropriate) were recorded at every load stage. C9) (D. Reinforcement stress–strain relationships 14 Magazine of Concrete Research. be discussed first. B5T12B with the available materials.37%) 3T12 3T12 2T20+1T16 B3T12D 0 150 3T12 (0. Details of additional specimens Specimen code % Redistribution Overall depth: mm Support reinforcement Span reinforcement Top Bottom Top Bottom (A) (B) (C. an analysis based on the stiffness shown in Fig.44%) 3T10 3T10 2T12+2T10 B4T20B 30 250 4T20 (1. at selected load stages. bars over the centre sup- Moment redistribution behaviour port and two 25 mm dia. Surface The test results confirmed that a number of discrete strains were measured on one side face of all speci.90%) 3T12 3T12 2T16+2T12 B5T20B 60 250 5T20 (2.94%) 3T10 3T10 2T12+1T8 B2T10D 55 150 2T10 (0. ** indicates four strain-gauged bars. 57. No. 13 mm diameter 3T12 3T12 2 No.01%) GFRP bars This table should be read in conjunction with Fig. bars in each span (the only two diameters readily obtainable by the authors at the Results from the test programme are summarised in time). D9) W2T12D** 30 150 2T12 (0. negative values in Table 2). stages could be identified in moment redistribution 800 700 600 500 Stress: MPa 400 T8 Bar T10 & T12 Bars 300 T16 Bar T20 Bar 200 T8 & T10 Wires E Steel ⫽ 200 GPa T12 Wire 100 E GFRP ⫽ 50 GPa GFRP 0 0 2000 4000 6000 8000 10000 12000 14000 Strain: microstrain Fig.68%) B3T20B 0 250 3T20 (1.Scott and Whittle Table 2.63%) W5T8D** 30 5T8 (0. 1 . using a Demec gauge centre support by 30 and 60%. 3. for specimens designed for 30% redistribution. The final specimen (F5F13B) used GFRP bars for the main tension reinfor- cement with five 13 mm dia. of the concrete section).e. 3.

Shaded groups of specimens are nominally identical. Results for additional specimens Specimen code % Designed redistribution Flexural failure Measured % redistribution Max. 4) specimen B2T12DH. rod 1 strain: microstrain Elastic Total A B3T16A 27–32 – Ungauged B B2T20B 25 – 2250 B3T16B 24–30 37 18 300 B3T16BL 10–20 21 4100 B3T16BLX 20 – Ungauged B5T12B 10–18 32 4000 B2T20BH Yes 25–33 38 27 000 B2T20BHX Yes 27 33 Ungauged B5T12BH 15–20 24 3700 B5T12BHX 12–28 34 Ungauged C B2T20C 12–18 23 10 800 B3T16C 20 32 16 400 B5T12C 5–10 27 12 700 D B2T12D Yes 15 27 31 000 B2T12DX Yes 34 38 Ungauged B2T12DXX Yes 25 34 Ungauged B3T10D Yes 16 26 37 500 B5T8D Yes 10–15 22 28 200 B2T12DH Yes 30–25 40 42 000 B2T12DHX Yes 28–30 45 Ungauged E B2T8E Yes 90–38 40 35 000 B2T8EX Yes 85–55 55 Ungauged B4T6E Yes 20–28 39 53 000 Details in italics are for high-strength specimens. No. The load increment after Magazine of Concrete Research. 4 for the strain-gauged From the start of the test until point A (see Fig. the small values of redis- between percentage redistribution at the centre support tribution being due to the influence of the different and experimental bending moment at the centre support percentages of tension steel along the beam on the over the full load history of this specimen. as illustrated in Fig. This figure shows the relationship the specimen was uncracked. Results for specimens designed for 30% moment redistribution Series Specimen code Flexural failure Measured % redistribution Max. 1 15 . rod 1 strain: microstrain Elastic Total W2T12D 30 Yes 22–15 25 32 600 W3T10D 30 Yes 10–16 32 33 000 W5T8D 30 Yes 14 27 20 000 B3T20B 0 10 13 Ungauged B8T12B 0 3 20 Ungauged B3T12D 0 Yes 18 0 Ungauged B2T10D 55 Yes 46–35 50 Ungauged B4T20B 30 5 0 Ungauged B5T20B 260 18 15 Ungauged F5F13B 30 5–25 25 Ungauged behaviour. values for the uncracked EI. Moment redistribution effects in beams Table 3. 57. Table 4. 2005.

it is important D was reached. this spread of developed. 6. iour) in the reinforcement at the centre support and at Ductility limitations were not an issue at any stage point F a similar situation occurred in the span. 4. Specimen B2T12DH: ultimate strain distribution along rod 1 Fig. with all reinforcement strains elastic. The final strain distribution in rod 1 from the absence of gross yield.e. Moment redistributions for beams of type B2T12D 16 Magazine of Concrete Research. 7. these load stages are shown in Fig. No. However. of the 40% total. indicate the flatter spread of strains resulting the end of the test. 2) for when the reinforcement was still behaving elastically. and values being caused by variations in the nature of the this zone corresponded to the SLS. all of which was elastic. These specimens step change in redistribution behaviour observable in all achieved the designed 30% redistribution at the Fig. ULS (albeit only just with B2T12D but comfortably so The span cracked at point D which reduced the span with the others) with the elastic contribution ranging EI of the beam. 25% was elastic redistribution D in Fig. development in both locations leading to plastic hinges High levels of redistribution were also achieved even being formed. B2T20B failed in shear with a peak strain in to the excessive deflections which occurred as a result rod 1 of only 2250 microstrain yet it achieved 25% of the mechanism action. The measured strain distributions in to note that some 25% redistribution was achieved rod 1 over the centre support (location A in Fig.Scott and Whittle 45 50000 G 40 40000 Strain: microstrain 35 D % Redistribution 30000 30 E C 20000 25 20 F 10000 15 B 0 10 A ⫺10000 5 0 400 800 1200 1600 0 Distance: mm 2 4 6 8 10 12 14 16 18 Experimental bending moment: kNm Fig. 4. Point E marks the crack patterns in the individual specimens. Strains had reached around redistribution at failure. The resulting flexural failure was char. 8. As Beeby has pointed 4 at the centre support is shown in Fig. which were all of UK origin. Development of moment redistribution in specimen B2T12DH the very localised nature of the gross yield associated point A caused a crack to form at the centre support with plastic hinge development. 1 . 4. Fig. F to G was a region of rapid strain very high strain levels reached in some instances. 2005. by during the test programme. strain distributions for rod 1 in this specimen. and only 15% was plastic redistribution. This indicates out. 7. Specimen B2T12DH: early strain distributions along rod 1 Fig. Between D and E a stable crack pattern from around 15% up to about 35%. 5. rotations associated with strain distributions such 1000 50 800 A Strain: microstrain 40 B % Redistribution 600 C D 30 400 200 20 B2T12D B2T12DX 0 10 B2T12DXX B2T12DH ⫺200 B2T12DHX 0 400 800 1200 1600 0 Distance: mm 2 4 6 8 10 12 14 16 18 Experimental bending moment: kNm Fig. These speci- onset of yield (i. 57. 6. with curves A to that is. and the strains in the top tension bars at this location Specimen B2T12DH achieved around 40% redistri- then increased with succeeding load stages until point bution by the end of the test. causing a significant reduction in five specimens of series D designed for 30% redistribu- the support EI relative to that in the span and hence the tion. mens all exhibited flexural failures. when the reinforcement was still behaving elastically at acterised by an inability to maintain applied loads due failure. The 42 000 microstrain and 18 000 microstrain at the sup. failed in spite of the microstrain. 5 corresponding to points A to D in Fig. shown in port and span locations respectively (rods 1 and 4) by Fig. None of the reinforcing which time the strain in rod 1 had reached around 5000 bars. 5. which are all plotted in Fig. 5 clearly shows how the crack developed between A similar pattern of behaviour was observed with the load stages A and D. initial deviance from linear behav.

tion varied from 10% up to around 30% with the 10– men F5F13B (Table 4). 3T16s or tribution at the ULS was to be achieved. 2005. 9. B3T12D) had slightly more strength concrete achieved a higher failure moment due reinforcement over the centre support than in the span to their larger moment of resistance). virtually all achieved the designed 30% total redistribu- Figure 9 shows redistribution plots for a range of B tion by the end of the test. which failed in flexure. This first cracked in the span. but the others all failed in shear. 57. these specimens cracked Bar diameter may have had an influence on behav. On first loading. 126 kN 1000 176 kN Specimens of series E. 8. 9) but then. F5F13B 40 100 A 30 80 % Redistribution B % Redistribution 20 B2T20B 60 B2T20BH 10 B2T20BHX 40 C B5T13B B5T12B 0 20 ⫺10 0 0 10 20 30 40 50 60 70 80 0 1 2 3 4 5 6 7 8 Experimental bending moment: kNm Experimental bending moment: kN m Fig. due to the much more localised effects of gross yield. due to variations in crack pattern resulting from the spans of up to 16%. Levels of elastic redistribu- series specimens including B2T20B and the FRP speci. 15% range being typical even in those specimens that mens B2T20BH and B2T20BHX had flexural failures failed in shear. achieved in this specimen. specimens which failed in flexure as shown in Fig. 11 for using several small diameter bars rather than a few bars specimen B3T12D. showed very high levels of redistribution very early in the tests. This is illustrated in Fig. men F5F13B. With B3T20B the reinforcement with larger diameter. 1 17 . bution (B3T20B. These tests. 11). The two high-strength speci. as shown by B5T12B in Fig. not a significant parameter as far as elastic behaviour Specimens designed for zero percentage of redistri- was concerned (although specimens made with high. leading to but not so with B3T20B and B8T12B which both failed high early levels of redistribution (see Fig. it plastic redistribution must occur if specimens designed closely followed the behaviour of B5T12B. Moment redistributions for beams of type B2T20B Fig.28%). supported by others in the Table 4 will now be discussed (except for F5F13B test programme. 8 can be comparable with those tic until point C was reached. 10. case with B3T12D (Fig. To summarise. Both support and span tension steel were elas- as those shown in Fig. Specimen B2T20B: strain distributions for rod 1 spans until cracks in the span (initiated at point B) caused moment to be transferred back to the centre support. B8T12B. Moment redistribution for specimenB2T8E Magazine of Concrete Research. Moment redistribution effects in beams 2500 40 kN had achieved 25% redistribution when it failed even 2000 61 kN though its main tension reinforcement had stress–strain Strain: microstrain 81 kN 1500 101 kN characteristics which were entirely linearly elastic.g. indicated that concrete strength was already dealt with above). achieving 25–33% The results for the additional specimens shown in redistribution. Specimens B2T20B. the resulting 90% redistribution meant that the beam effectively behaved as two simply supported Fig. Plastic redistribution was mens having similar reinforcement percentages but needed to offset the elastic redistribution if zero redis- different bar combinations (e. When the support cracked (point A in Distance: mm Fig. 6. 10). This supports the point made earlier that as further cracks developed along the specimen. this was not conclusive remained elastic until a shear failure occurred and there and there were no discernible differences in redistribu. This was the 5T12s). 10 for speci- ⫺500 0 400 800 1200 1600 men B2T8E. was no reduction in the 10% elastic redistribution tion percentages achieved at the ULS between speci. 2T20s. as illustrated in Fig. in shear. (see Table 2). B2T20BH and B2T20BHX (which all had two T20 bars as the main support tension steel) behaved very Additional specimens similarly over the elastic range. which had a low percentage 500 of tension steel over the centre support (typically 0 0. However. An interesting result is that for the FRP speci. ing moment) which caused elastic redistribution into 9. at the centre support (the location of the largest bend- iour in the elastic range. No.

mismatch between the assumption. which probably accounts both for the the test beams). quite considerable. 1 . central point load.l B4T20B (⫺30%) 10 B5T20B (⫺60%) 10 % Redistribution 5 0 0 0 0·5 1 1·5 2 2·5 3 ⫺5 ⫺10 ⫺10 ⫺20 ⫺15 ⫺20 ⫺30 EIspan /EIsupport ⫺25 0 10 20 30 40 50 60 70 Experimental bending moment: kN m Fig. 11. The results also suggested that total redistribu- 0 tion was not limited by considerations of (UK) reinfor- cement ductility. 2005. Beams designed for –30 and –60% redistribution load 18 Magazine of Concrete Research. However. Relationship between elastic percentage redistribu- tion and flexural stiffness ratio. is a matter not addressed by the design codes. 12 with negative values stiffness ratios. The following geometry/loading were initially being redistributed into the adjacent combinations were analysed to cover situations that spans. No. neither specimen achieved anywhere near its designed level of redistribution at the ULS. of a constant EI (flexural stiffness) value Results for the single specimen designed for 55% along the member and the actual EI values which are a redistribution are also shown in Fig. achieve its design value. (The actual value of this ratio for a indicating that support moments were being increased.d. central point load (simulating failed in shear. uniformly distributed load over concern. 13. The results for these two from spans to supports) for a range of EIspan : EIsupport specimens are shown in Fig. for purposes of tion in practice. W5T8D) behaved very similarly tive values confirm that the elastic contribution can be to the equivalent bar specimens of series D. at the SLS. Consequently. which is a potential cause for (d) fixed-ended beam.l 20 Fixed : Central point load 15 20 Fixed : u. respectively. particular beam will be dependent on the chosen rein- Early cracking over the support meant that moments forcement layout).Scott and Whittle 60 Design considerations 40 % Redistribution B3T12D The test results confirmed that total moment redis- B3T10D tribution has both elastic and plastic components with 20 the former making a significant contribution to the whole. spans. certainly not conclusive and further work in this area would be helpful. uniformly distributed load over and for the tendency to transfer moment back into the whole span. the results from these two tests are whole span. The specimens reinforced with ribbed wire The resulting relationships are shown in Fig. 13. as intended. W3T10D. Uniformly distributed Fig. there is a case for ⫺20 0 5 10 15 20 25 30 increasing the current 30% design limit on redistribu- Experimental bending moment: kNm tion prescribed in BS 81101 and EC 22 and the 20% Fig. 57. 12. but tions of a continuous beam. 11. u. at present. a programme of analysis was un- the quite remarkable value of 35% at the SLS. Negative values indicate that..d. moments were approximated to the end span and interior span condi- transferred back to the supports.d. dertaken to find how much elastic redistribution might Specimens B4T20B and B5T20B were designed to be expected (from supports to adjacent spans – no redistribute moments from the span to the support. Beams with zero and 55% redistribution limit in ACI 3183 by recognising the contribution made by elastic redistribution which.l. Elastic redistribution occurs at the SLS due to the for zero redistribution are actually to achieve this situa. but neither specimen ever seemed likely to (c) fixed-ended beam. rapid reduction in redistribution at the end of the tests (b) propped cantilever. This specimen consequence of the designed reinforcement layout. redistribution will occur from the span to the 40 Propped : Central point load 30 Elastic % redistribution Propped : u. Both specimens (a) Propped cantilever. Posi- (W2T12D. by attempt was made to model the effects of redistribution 30 and 60%. As span cracking developed. analysis. To actually achieved 50% redistribution at the ULS plus explore this further.

The first was signing for 60% total redistribution will give about that the structure must not have reached first yield at 35% elastic redistribution for the case of central point service load. Test results for beams with gauged should be treated as approximate and further work is reinforcement are also included and the agreement is needed were they to be used in design charts. loads on the beam were then increased from zero should this be considered desirable. Thus the curves in Fig. The beam geometry and loading arrangement mod- lier. 57. the reinforcement remains elastic under ser. Crack widths were also calculated at fixed) and 60% total redistribution respectively.3 mm in all cases but.3 mm was adopted. a further dual merits. No. Commonly. tribution. such as those in a continuous beam. the BS 8110 relationships between elastic redistribution and total crack width limit of 0. Fig. The moment– undesirable since the beam is being forced to behave in curvature relationships for the span and support loca- different ways at the SLS and the ULS. Fig. ‘Reverse redis. 2005. since the consequent deformations would load or about 20% elastic redistribution in the case of a be unacceptable. geometries and load combinations. exhibit elastic redistribution from the spans to the sup. Relationships between elastic and designed percent. considerations in greater detail before specific design It is important to ensure that. except for the elastic/total redistribution relationships for the four high-strength beams 150 mm deep for which a limit of conditions considered and reinforces the case that elas. This a given level of design redistribution.e.l 120 Fixed : Central point load 20 Fixed: u. the curves test programme. they rose sharply to unacceptable limits as cerning ‘reverse redistribution’. by the sive cube strengths of 35 and 100 MPa. Fig.. 45% would seem appropriate.d. Ratio of service load to yield load versus percentage age redistribution. 1 19 . 15. The curve load/yield load and % redistribution for beam depths for the propped cantilever/central point load case is and concrete strengths consistent with those used in the consistent with the test results but. for tribution to achieve the designed 0% at the ULS. an increase in would seem acceptable and this could be increased to the EIspan : EIsupport ratio and hence more elastic redis. for a propped cantilever. Second. Figure 15 shows the relationship between service ling exercise and these are shown in Fig. give a good and realistic overview of possible before yield at the SLS occurs. however. 15 indicates that over 50% redistribution is does. over whole.d. whatever upper limit is recommendations can be made covering a range of adopted. Thus the central support and at midspan.l. Midspan crack widths at tic redistribution is often a significant proportion of the the SLS were less than 0. through to the ULS and the bending moment distribu- The lower and upper ends of each curve in Fig. It also reinforces the points made above con. redistribution were also developed as part of the model. Uniformly distributed load design redistribution Magazine of Concrete Research. 45% if crack widths were not an issue. ing of the test beams a limit of 40% redistribution This will lead to greater span moments. Consequently. Moment redistribution effects in beams supports even though the beam was designed for redis.d. sort of ‘reverse redistribution’ effect. At this stage it is vice loads and that crack widths at service loads are still necessary to consider each situation on its indivi- kept within acceptable limits. even beams designed for zero redistribution will elled were the same as those used in the test beams. 14. or at least reduced. of course. for the midspan sections of the beam and then. u. de. support(s) fully each load stage. The procedure was to choose a reinforcement layout ports at the SLS and thus must undergo plastic redis. 14 good. series of analyses was undertaken to investigate these tribution at the ULS to be in the opposite direction—a two issues. overall. As discussed ear. Rather more comprehensive 0. A ratio of service load to yield load of uniformly distributed load. the yield load was approached for redistribution levels support moments. The work described in this paper indicates that there 40 Propped : Central point load Service load/yield load: % Elastic % redistribution 30 Propped : u. Two SLS measures were considered. calculate the ‘reverse redistribution’ at the SLS could be considered reinforcement at the central support. The applied level of designed redistribution which is selected. the figure indicates that. 14. 14 are likely to be More work is required to investigate crack width conservative in this situation. the supports. Overall.9 was assumed to be acceptable. 13 tion and hence percentage redistribution determined at correspond approximately to 0% (i.l 100 90% 10 80 0 60 Experimental 250 mm beam (C35 concrete) ⫺10 40 data points 150 mm beam (C35 concrete) 250 mm beam (C100 concrete) ⫺20 20 150 mm beam (C100 concrete) 0 ⫺30 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Designed % redistribution % Design redistribution in beam Fig. tions were then obtained for concretes having compres- tribution’ can be avoided. greater than 40%. for the geometry and load- are less than those pertaining to the fixed end case.

additional support from Arup Research and Develop- merical modelling. 1992. 93. ( f ) Concrete strength can influence total redistribution at the ULS since the moment of resistance of the section is increased. 75. ACI 318: Building Code Re- (d) A consequence of elastic redistribution is that quirements for Structural Concrete. the percentage of elastic redistribution search Report. Michigan. Ductility in reinforced concrete: why is it needed and how is it achieved? The Structural Engineer. Eligehausen R. avoidance of reinforcement yield at the SLS are likely when current limits are exceeded. with total BS 8110 relating to the neutral axis depth is only redistribution being the sum of the elastic and plastic related to plastic redistribution and 30% total redis- components. Scott R. W. 1997. 2. British Standards Institution. there with the current rules limiting neutral axis depth) now is now a strong case for increasing this value be applied just to the plastic component. Furthermore. at the SLS. No. 1987.g. 311–318. 1 . 1996. 2005. may be affected owing to variations in the distribu. British Standards Institution. tute. ENV 1992-1-1 Eurocode 2: Design of Concrete Structures. USA. undergo plastic redistribution before the ULS is 4. The limit given in contribution of elastic redistribution and. 6. Tarmac Topmix Ltd. Intrinsic mechanisms in reinforced concrete beam- column connection behaviour. and Langer P. sible redistribution limits given in BS 8110. large bars versus 5. (e) Reinforcement arrangement (e. British Standards Institution. ment. Discussion contributions on this paper should reach the editor by fluence seems minimal. However. 1. No. it is suggested that the 30% limit (together tribution should always be possible. there will the possibility of yield at the SLS should be frequently be scope for design engineers safely to ex. generations of project students. crack width considerations and loading arrangement. R.Scott and Whittle is no case for reducing the maximum permitted 30% (g) No failure of the reinforcing bars occurred in any limit to moment redistribution currently allowed in of the tests which indicates that there was always BS 8110 and EC 2. given in ACI 318 would appear to be over-conservative. H. London. References match between the uniform flexural stiffness as. 336–346. the limit of 20% sufficient ductility in the UK reinforcement. 57. Nevertheless. BS 8110: Structural Use of sumed and the stiffness values which actually Concrete Part 1: 1997: Code of Practice for Design and Con- occur due to variations in the reinforcement layout struction. (b) When carrying out elastic analysis at the ULS the Azizi and the crack width modelling was undertaken most sensible practical approach is to use the EI by Mr C. ACI Structural Journal. which occurs be. Beeby FREng. Farmington Hills. the assis- values for the concrete section. American Concrete Insti- beams designed for zero redistribution will. checked. 3. Rotation Capacity of Plastic small bars) has little effect on total redistribution. crack widths and to become limiting criteria. American Concrete Institute. 18. University of Stuttgart. EC 2 creased beyond 30% by taking into consideration the and ACI 318 should be reduced. Hinges and Allowable Degree of Moment Redistribution. No. tions of cracks. In addition. It should also be noted that. as defined in tance of the following is also gratefully acknowledged: BS 8110. Part 1-1: General Rules and The contribution of elastic redistribution is signifi- Rules for Buildings. The absolute although it is not now possible to specify a unique upper limit to total redistribution will depend on beam upper limit since the contribution of elastic redis- geometry and loading arrangement but could now be as tribution is dependent on support conditions and high as 55%. The strain-gauged beams were tested by Dr A. (h) There was no evidence that the maximum permis- It seems reasonable that the upper limit can be in. reached. 1 August 2005 20 Magazine of Concrete Research. London. Pope of Arup R&D. Elastic redistribution is caused by a mis. W. 2002. Beeby A. ceed the current 30% limit if they so wish. Re- However. However. 1997. technical staff of the University of Durham and several These are elastic redistribution. in fact. British Standards Institution. 3. Conclusions Acknowledgements (a) An investigation of moment redistribution effects Funding for this project was provided by the Engi- was undertaken comprising laboratory tests on 33 neering and Physical Sciences Research Council with two-span beams supported by programmes of nu. along the member and the influence of cracking. Professor A. the in. cause of the non-uniformity of EI and plastic redis- tribution which occurs after yield of the tension steel. the (c) Total moment redistribution has two components. However. cant even in members which fail in shear.