A new fib Model Code proposal for a beam-end type bond test

A beam-end test is proposed in this paper and in the new fib Model Code 2020 to determine the effects of concrete type, confinement effects, and casting position on the anchorage strength of reinforcing bars. Two bars are cast in each specimen (in the two opposite corners), one in a “ good ” casting position and the other in “ poor ” casting position. The proposed test aims to be an economical bond test capable to represent actual conditions of anchored bars in real design practice. The test is also intended to verify whether existing provisions for bond and anchorage in the fib Model Code are suitable in nonconventional concrete. The validity of the proposed beam end type bond test is assessed by the results of more 60 tests on rebars having anchorage length of about 20 times the bar diameter in plain concrete, in fiber reinforced concrete, and in recycled aggregate concrete. Experimental results provide information both on the “ top cast effect ” in three types of concrete and on the effectiveness of fib Model Code provisions for anchorage when nonconventional concrete is used.


| INTRODUCTION
Bond behavior affects several aspects of the performance of RC members. At serviceability limit states (SLS), crack spacing, crack width, beam deflection, and tension stiffening are all influenced by bond. At Ultimate Limit State (ULS), bond governs the capacity of anchorages and lap joints but bond also influences rotational capacity in plastic hinge regions, depending on yield penetration away from the crack. 1,2 The capacity of anchorages and lap splices of rebars in concrete is a function of several factors as (1) the anchorage (or splice) length; (2) the resistance to splitting crack development provided by the concrete cover (including cover thickness and concrete tensile strength) 1,3-5 ; (3) the bar diameter, because the size effect decreases the bond strength (BS) with increasing bar diameter 6 ; (4) the rib geometry 7,8 which influences the interaction with the surrounding concrete; (5) the confinement provided by transverse reinforcement (if any) as links crossing the splitting surface [9][10][11] ; (6) the transverse compression due to the reaction at the beam supports (when present); and (7) the casting position since, after pouring concrete, the mix water rises and large aggregates tend to drop down. 12,13 Therefore, near to the top surface, concrete may have lower mechanical properties due to its higher porosity and to the plastic settlement of concrete around the bars.
There are two principal modes of failure of anchorages and splices: (1) a pull-out type failure where the concrete shears across the top of the ribs and leaves a quite relative smooth bore where the bar has been pulled out; this is a relatively ductile failure; (2) a splitting failure with longitudinal cracks along the bar axis (provoked by the radial stresses generated by the ribs in the concrete cover), which can cause spalling of the cover. The splitting failure is the weaker of the two failure modes and generally occurs when the cover is less the 2.5-3.0 times the bar diameter. 1,[3][4][5] This brittle failure can be controlled by transverse reinforcement crossing the potential splitting surface, 9,10 by the postcracking tensile strength of fiber-reinforced concrete (FRC), 14 or by transverse pressure (i.e., for anchorages at beam supports). 2 As documented in the fib Bulletin 72, 15 the capacity of an anchorage or a lap may be evaluated with the formulation proposed in fib Model Codel 2010 16 which has been reformulated for design in the new MC2020 (Chapter 20). 17 This semi-empirical model was calibrated on the basis of test results on anchorages and laps having a length longer than 10 d b in conventional concrete with a compressive strength ranging between 15 and 110 MPa. The arrival into the market of new types of concrete having mechanical and physical properties different from conventional concrete (such as FRC, recycled aggregate concrete [RAC], self-consolidating concrete, ultrahighperformance concrete, etc.) as well as new types of rebars (including nonmetallic rebars) may require the development of specific design formulation of anchorages. At the same time, by considering the presence of different types of deformed bars into the market, it may be necessary to control the rebar bond performance; therefore, a reliable and practical bond test, representative of practical applications, should be adopted.
The standard tests available in international codes and standards or adopted by the research community differ significantly. In the pull-out test proposed by RILEM/ CEB/FIP 18 and EN10080:2005, 19 the bar is in tension and the concrete is in compression. This (pull-out) test has been widely used because it is economical to manufacture and to test and because it represents the concept of bar anchoring. However, this configuration is not representative of any practical situation. Furthermore, the bar presents a short embedded length (5 d b ) which causes a quite uniform distribution of bond stresses along the bond length (while it varies in longer anchorages). Moreover, a large concrete cover (4.5 d b , which is not representative for practice) and the friction arising at bottom bearing plate tends to confine the specimen, thus making the test not suitable to investigate splitting failure mode. It should be also noted that the stress developed by the bar is not directly proportional to the bond length; therefore, it is not possible to evaluate the capacity of real anchorages (where bond length typically exceeds 20 d b ) on the basis of the results of pull-out test with a short embedded length. Finally, pull-out specimen addresses only what would be classified in design Codes as a good casting location since good compaction of the concrete without voids around the bars is expected. The top cast (TC) ratio (the ratio between the bond strength of a bar near the top surface of a concrete element and that close to the bottom surface) tends to increase quite markedly as the bond length ratio (bond length divided by bar diameter) increases. 12 In fact, with a bond length equal to 5-10 times the bar diameter, the TC ratio is close to 0.5 12,13 while, for practical values of bond length, the TC ratio increases to one since the top bars tend to have a bond strength similar to the bottom bars. 12 Therefore, the investigated bond length significantly influences the test results.
An alternative test is the beam test proposed by RILEM/CEB/FIP, 18 adopted by EN 10080:2005, 19 prEN 10080:2023, 20 and by Eurocode 2. 21 In this test, the anchorage is placed in good bond condition close to the beam support where concrete is in tension, thus making the test more representative of actual conditions in structural elements. The bond length is 10 bar diameters (10 d b , thus double than that of pull-out test but still not representative of practice) but the anchorage is characterized by a very high confinement of transverse reinforcement to avoid the shear failure of the beam specimen. In fact, the density of transverse reinforcement is about four times greater than that typical of practical design; therefore, the splitting failure of the anchorage, typical of real cases, is unlikely. The concrete cover is quasiconstant (minimum cover to the center of the bar is 50 mm), independent of the bar diameter; therefore, the cover to bar diameter ratio decreases with increasing bar size. Furthermore, in the beam test, the loaded end slip of the anchored bar (relevant for the crack widths and for tension stiffening) can be hardly measured. It should be also noted that it also is very difficult to handle the specimen during setting up the test because of its large dimensions (and weight) as well as the discontinuity at midspan (consequently manufacture of the specimens is not costeffective). For all these reasons, the beam test cannot be used with sufficient degree of confidence to investigate all the parameters governing anchorage capacity. Furthermore, a wide experimental campaign showed a very limited correlation between RILEM pull-out and RILEM beam end tests (with a significant scatter of test results) because of the markedly different bond conditions. 22 The potential purposes of a bond test were already identified and discussed in 2003 by Cairns and Plizzari. 23 These are mainly: (i) the routine quality control of reinforcement; (ii) the performance classification for novel materials; (iii) the development of constitutive bond-slip laws for numerical analysis; (iv) the benchmarking of research studies and (v) the provision of design data. Cairns and Plizzari stated that the "principal purpose of a Standard bond test is to verify that a bar can be used with existing design guidance, itself derived from tests on situations representative of practical use." 23 This underlines that "Performance classification" is the main aim for a Standard bond test among all the possible functions listed above. It should be noted that a Standard bond test may also provide useful information and data for benchmarking of research studies, which can benefit from a beam end test representative of the actual behavior of anchorages.
Aim of this paper is to outline the format of an enhanced test method for bond and anchorages and to begin the process of calibration of the required structural performance.

| RESEARCH SIGNIFICANCE
A new beam end test is proposed in this paper and in the draft of Chapter 20 of the new Model Code 2020 17 that aims to improve the current Standard test procedures. The proposed standard bond test is mainly intended for "performance classification" of anchorages since the routine quality control of the ribbed bars can be adequately ensured by the measurement of the rib geometry after production as in EN ISO 15630-1:2019. 24 Moreover, it aims to be an economical bond test with limited dimensions but capable of representing actual conditions of anchored bars in real practice, and able to investigate the casting position effect. The proposed test method describes procedures to assess the anchorage capacity of steel reinforcing bars in concrete. This test may also be used to determine the limits of validity of the equations proposed in MC2020 17 since they were derived from a wide database collecting results from specimens made of conventional concrete. Thus, the proposed beam end test method is primarily intended to verify whether existing provisions for bond and anchorage in the fib Model Code are suitable also in non-conventional types of concrete; however, the proposed test may also be adopted to assess the performance of new types of reinforcement, especially when their geometries are not included in the existing standards.
The paper mainly focuses on the principles governing the design of a beam end test for the evaluation of anchorage capacity with an outline of the test setup specifications; however, the description does not aim to define all aspects for the purposes of a Standard. The effectiveness of the proposed beam end test is assessed through a wide experimental campaign for determining the anchorage capacity of steel rebars in plain concrete (PC), RCA, and FRC, which are nowadays often used in practical applications. The tests were carried out both at the University of Brescia (Italy) and at Heriot-Watt (Edinburgh, UK).

| Principle of the test
The principle of the test is to load a bar, that is cast into a concrete block representing the end section of a flexural element over a defined length, by a tensile force (F). The other end of the bar remains unstressed and a longitudinal equilibrating reaction is directly applied at the bottom portion of the specimen ( Figure 1). In order to balance the overturning moment, two opposite vertical reactions R have to be applied to the specimen ends which allow a strut-and-tie mechanism to be developed in the block (Figure 1b) with the anchorage length set in the tensile portion of the concrete block and with an inclination of the compressive strut of about 45 ( Figure 2).
The beam end specimen allows two bars to be tested in a single specimen. It is diagonally symmetric so that the second test can be done just by turning the specimen around. Two bars are arranged in the two opposite corners, in order to allow one bar to be tested in a "good" casting position and the other in a "poor" casting position at a distance of at least 300 mm from the bottom face, as defined in the draft MC2020 17 and in draft Eurocode 2. 25 The embedded length of the bar is set to a value dependent on the concrete compressive strength and steel grade and corresponds only to a part of the specimen length ( Figure 2a). Since the beam-end test aims at verifying anchorage performance, the anchorage length is to be designed to develop a bar stress f s,tgt (as determined from Equation (2), proposed in fib MC2010 16 ) of about 80% of the characteristic yield strength f yk . Plastic sleeves (bond breakers) are placed at both bar ends (>d b and > 5 d b at the loaded and unloaded bar ends, respectively) to control the position of the anchorage and to avoid a localized cone-type failure of the concrete at the loaded end of the specimen. The unbonded length at the free end aims to avoid the confinement provided by transverse pressure at the beam support ( Figure 2a).
The bar to be tested extends beyond the length of the specimen. Displacement gauges are placed to both ends of the test bar for measuring the bar-to-concrete slip (i.e., the relative displacement between steel and concrete).
The recommended concrete cover is equal to 1.5 d b to represent common practical situations where splitting failure is expected prior yielding of the anchored bars (Figure 2b,c). Closed steel links are placed to provide transverse reinforcement within the range used in current practice, corresponding to a stirrups index of confinement (K tr ) 16 of about 1%; the latter is defined as: where A st is the area of the cross section of one leg (mm 2 ); n st the total number of confining stirrups within the anchorage/lap length; n t the number of legs of a stirrup crossing the potential splitting failure surface (n t = 1 for corner bar); n b is the number of anchored bars or pairs of lapped bars; d b is the diameter of the bar (mm); and l b is the lap/anchorage length (mm).
It should be noted that the general form of the beam end test proposed is similar to that proposed by ASTM 944:2015 26 where there is only a single bar placed at the midface of the bottom side with larger concrete cover (equal to 38 mm). As laps in beams may often be located in a "poor" casting position near the top of the concrete element, the Standard test should cover this condition as well. In the proposed beam-end test, the recommended location of the bar is at a corner (Figure 2b), where concrete confinement may be weaker but links are more effective; in the corner location, the bar shall be oriented with one set of ribs pointing towards the adjacent corner in order to favor a side-face splitting failure in the concrete block corner (Figure 2c).
In case of splitting failure of the anchorage, the maximum stress in the bar (f stm ) at anchorage (splitting) failure may be calculated with the semiempirical where f cm is the mean concrete compressive strength, d b is the bar diameter, l b is the anchorage length. In Equation (2) Re-arranging Equation (2a) to determine the anchorage length gives Equation (3): where f s,tgt is the target anchorage strength, set equal to 0.8 f yk . The height (H) and the length (L) of the specimen are set by assuming an inclination of the concrete strut of 45 and that the vertical reactions are applied beyond the anchorage length through bearing plates 5 d b long and 10 mm thick, over the width of the specimen ( Figure 2a); a minimum distance of the top bar from the bottom surface of specimen of at least 300 mm shall be set in order to guarantee the "poor boor condition" of the top bar. Finally, the specimen width (B) may be set as the greater of 15 d b and 250 mm in order to avoid any interference between opposite bars in the failure mechanism as well as the concrete cone failure of the tensile portion of the block.
In summary, the specimen size should fit the following requirements: In Figure 3a,b, the anchorage length and the specimen depth are plotted against the bar size for B500 and B700 steel grade, respectively, by varying the concrete strength from C25/30 to C70/80 and assuming the confinement factors from cover and reinforcement α 2 = 1.11 and α 3 = 0.13, respectively (being α 3 = k m K tr and f cm = f ck + 8 MPa). It should be noted that the specimen depth tends to increase more than linearly since the l b /d b ratio is not constant with the bar size; furthermore, the specimen depth is governed by the requirement for guaranteeing "poor bond condition" of the top bar when small bar size is used (Figure 3). For a bar size commonly used in beams and slabs (≤16 mm), the sizes of the beam end specimen are rather limited, thus making the specimen easy to manufacture and handle.
As an example, for a corner bar diameter of 16 mm and a target anchorage strength f st,tgt = 0.8 f yk , an anchorage length of 308 mm (l b = 19.3 d b ) and a specimen depth of 400 mm (H = 25 d b ) should be provided when using a C25/30 concrete and a B500 steel grade (Figure 3a).
Finally, it should be observed that the beam end test might also provide local bond properties between reinforcement and concrete when a shorter embedded length is used (l b ≤ 5 d b ); in this case, details for and transverse reinforcement, test setup to promote a pull-out failure may be found in 27 where the bond-slip law of reinforcement bars with different rib geometries is investigated.

| Preparation of samples
The test bar shall be in "manufactured" condition without losing mill scale, preferably entirely free from rust. If the test bar is corroded, the conditions of the bar shall be described in the test report and possibly supported by photographs of the surface. The bar shall not be cleaned in any way that might change its roughness.
The fresh concrete is cast vertically in the molds with the bars laid horizontally (Figure 2). Concrete shall be compacted and cured as specified by the mix designer in a manner consistent with the method proposed for practical use of the material.

| Test setup
A sketch of a suitable test setup is shown in Figure 4. The top bar is pulled at one end while a bottom restrain prevents any longitudinal translation of the concrete block. The reacting bench may be a self-balanced frame consisting of two L-shaped rigid steel profiles simply supported by the floor. The steel bench should be stiff enough to ensure that the applied load remains parallel to the bar axis during testing. Two vertical ties connected by a transverse rigid steel beam are placed at the back of the specimen to avoid any rotation of the test block.  The loading system should be capable of measuring the applied tensile load (to an accuracy of ±2% of the applied force) by means of a load cell placed between the loading apparatus and the reacting steel bench (Figure 4). The compressive bottom portion of the specimen should be not greater than 5 d b (corresponding to about 0.20 H). Displacements of the loaded end of the longitudinal bar are measured by two linear variable differential transformers (LVDTs) having an accuracy of 0.001 mm (Figure 4b) both positioned on a plane through the bar axis. One LVDT placed at the free end of the bar measures the unloaded end slip. Further measuring devices may be placed on the specimen side to monitor the crack opening and its propagation.

| Test procedure and interpretation of test results
Performance of the bar would be assessed by several measures, namely: peak bar stress (f s,ex ), bar stress at loaded end slip of 0.05 mm (δ L,0.05 , relevant to crack width, tension stiffening, and beam deflection), free end slip at peak load (δ U,max ), stress at unloaded end slip of 0.02 mm (δ uL,0.02 , related to anchorage stiffness); diagrams of force or stress (f s ) versus free end slip (δ U ) relations shall be provided. Measured displacements at the loaded end have to be adjusted to take into account the elastic deformation of the bar outside the embedded length. The development of side and face splitting crack width (w s and w f , respectively) might be also checked both on the vertical and on top surface of the specimen, respectively.

| EXPERIMENTAL PROGRAM
The format of the proposed beam end test is assessed within two broad experimental programs carried out at the University of Brescia (Italy) and at Heriot-Watt University (Edinburgh, UK). The former is addressed at comparing the anchorage behavior and the casting position effect of steel rebars in PC with those in RAC and FRC having similar compressive strength. This experimental campaign consists of four batches with three samples each (24 tests). One bar diameter was investigated (16 mm) placed at corner location of the beamend specimen with a 1.5 d b concrete cover and an anchorage length of 20 d b . The specimen was 450 mm long, 400 mm depth, and 300 mm wide ( Figure 5). Two stirrups having a diameter of 6 mm are arranged along anchorage length (K tr = 1.1%). In the fourth series with RAC, three more specimens were tested with double transverse reinforcement (K tr = 2.2%) to allow the anchorage to reach the yield strength of the bar (about 450 MPa).
The experimental study carried out at Heriot-Watt University mainly aims at demonstrating the conformity of the proposed beam-end test to practical conditions by comparing results from beam-end tests with the mean anchorage strength determined according to Equation (2) from fib MC2010. 16 Two series consisting of 10 specimens were tested with either the bar in corner or center location ( Figure 6). In the latter configuration, the anchorage strength may benefit from a higher concrete confinement with respect to the corner location, but the transverse reinforcing stirrups are less effective since the vertical links are not adjacent to the anchored bar.
In each series, two bar diameters are tested (10 and 16 mm) in normal strength concrete with the cover varying between 1.0 d b and 3.1 d b . The cross section of the beam-end specimen size was 200 Â 300 mm and the length of the specimen was 280 mm for the 10 mm reinforcing bar and 400 mm for the 16 mm reinforcing bar. The embedded lengths of main bars were set as 18.5 d b for a 10 mm and 19 d b for a 16 mm bar. Stirrups with a diameter of 6 mm were provided at 100 mm spacing over the anchorage length, corresponding to two and three stirrups for 10 and 16 mm bar, respectively (stirrup index of confinement of 3.06% and 1.74%). Table 1 exhibits all the geometrical and mechanical properties of the specimens, namely: concrete cover, anchorage length, diameter and number of the confining links and the stirrups index of confinement along with the maximum stress in the bar (f stm ) determined according to the MC2010 semiempirical equation (Equation 2) by assuming a concrete compressive-strength target f cm = 33 MPa.

| Material properties
Steel grade B500C (according to EN 10080 19 ) was used for all types of reinforcement. The 16 mm rebars tested in F I G U R E 5 Dimensions of beam end specimens tested at the University of Brescia (Italy).
Italy had a relative rib area f r = 0.079 and mean yield strength f ym = 526 MPa while those used in United Kingdom had a f r = 0.07 or 0.08 and a f ym = 570 and 560 MPa, for 10 and 16 mm rebars, respectively. The same mix design was used for the reference PC and FRC mixtures ( Table 2) tested in Italy, both with a target concrete class of C25/30; the concrete was supplied by local ready-mix company. Hooked-end steel fibers   The RAC was supplied by a ready-mix company with a C25/30 target class. In RAC, the water-cement ratio was 0.52 and 25% of natural coarse aggregate was replaced by recycled aggregate having a diameter ranging between 8 and 22 mm ( Table 2). The latter was obtained from blast furnace slags, whose density ranges between 3100 and 3250 kg/m 3 . At the time of the bond tests, the RAC had an average compressive cube strength (f cm,cube,ex ) of about 40 MPa (determined on 20 cube specimens). The slump of RAC was 170 mm (S4 workability class).
In Series 5 and 6 tested in the United Kingdom, PC had an average cube compressive strength of 40.6 or 44.9 MPa.

| Discussion of test results
The main experimental results of Series 1-4 are listed in Table 3 where maximum load (P u ), maximum bar stress (f s,ex ), maximum unloaded end slip (δ UL,max ), bar stress at a loaded end slip of 0.05 mm (f s,δL0.05 ) and at free end slip of 0.02 mm (f s,δUL0.02 ) are reported for each specimen along with the secant stiffness (K UL0.02 ) of the anchorage measured at free end slip of 0.02 mm. In Series 5 and 6, only the peak load at failure was recorded (Table 4). Figure 7 shows the comparison of the bar stress (f s,ex ) versus unloaded-end slip (δ UL ) curves between top and bottom anchorages for each series. The curves clearly show higher bar stresses and a stiffer behavior of bars in bottom location where good bond conditions are expected. None of the specimens showed any descending postpeak behavior, confirming the brittle behavior of anchorages with low confinement of transverse reinforcement (without transverse pressure). At the same time, it should be observed that top anchorages exploited larger unloaded end slip at peak load. This effect increases with the value of the stirrup index of confinement (K tr ), as expected in Series 4 ( Figure 7) where the amount of confining transverse reinforcement, which was doubled with respect to the other series, allowed a better control of the propagation of the splitting cracks.
As far as the failure mode is concerned, most of specimens failed because of the development of side-face splitting crack which quickly developed from the loaded end towards the free end (Figure 8a). However, the anchorage in bottom location of PC_2 specimen and all bottom anchorages in RAC of Series 4, presented a mixed failure mode with both a concrete cone at the loaded end and longitudinal splitting cracks (Figure 8b). Finally, in the specimen FRC_1 the bottom anchored bar yielded.

| Anchorage strength.
In order to better compare the performance of the anchorages, the maximum nominal bar stress measured    Tables 3 and 4.
In Figure 9a, the mean BS ratio is depicted for top and bottom bars of each series. For the bottom bars, the BS ratio ranges between 0.96 and 1.05 in PC (Series 1, 5,   Figure 9b, which evidences the strength reduction for top bars with respect to the bottom ones. The decrease of the anchorage resistance ranges between 29% and 21% in PC while it is 16% in FRC. A significant decrease of anchorage strength in poor casting position is measured for RAC (40%) which may be due to the density of recycled aggregates derived from blast furnace slags (BFS) that was 15% greater than that of the natural gravel. This difference may lead to a higher percentage of BFS aggregates in the lower part of the specimen. This kind of segregation may have caused different mechanical properties between the bottom and the top portion of the concrete element.
This different strength was confirmed by two pull-out tests on concrete cover (measured according to EN 1250-4-3 30 ) which provided a concrete strength of 33 and 16 MPa, of the bottom and top cover of specimen RAC_1, respectively ( Figure 10). This result suggests that the replacement of fine and coarse natural aggregate with BFS in RAC should be adequately proportioned to avoid any concrete segregation.
The experimental campaign provides also useful results on the brittleness of the anchorages in different concrete typologies, both in poor and good casting condition. In Figure 11, the maximum unloaded end slip (δ UL,max ) of the anchorages in top or in bottom locations is plotted for each series. A markedly less brittle behavior of bars in top location can be seen: the maximum end slip reaches a value close to 0.7 mm when transverse confinement is provided by steel fibers (Series 2) or by doubling the stirrups index of confinement (Series 4). The capability of FRC to control the anchorage failure can be observed also for anchorages in bottom location, since the maximum free end slip is five times greater than that of anchorages in PC.

| Anchorage stiffness
Finally, the performance of anchorage behavior at SLS (related to crack width control) can be expressed by the bar stress (f s,δL0.05 ) at a loaded end slip of 0.05 mm. To this aim, in Figure 12a the mean value of the bar stress of each series is plotted both for top and bottom location. First, it should be noted that the bar stress is never >240 MPa, as required at SLS to control the crack width. 21 Furthermore, the top anchorages show a markedly softer behavior (with a decrease between 32% and 41% of the stiffness) with respect to the bottom bars. However, the test results present a large scatter with a coefficient of variation even >50% (Table 3).
A lower scatter of the anchorage stiffness can be obtained when measuring the bar stress (f s,δUL0.02 ) at an unloaded end slip of 0.02 mm (coefficient of variation ffi 11% in Series 1, 3, and 4, and ffi 28% in Series 4). The secant stiffness K δUL,0.02 of the bar stress (f s,ex ) versus slip (δ UL ) curve at an unloaded end slip of 0.02 mm is plotted in Figure 12b; it should be noted that the casting position also affects the anchorage stiffness of top bars with a decrease of 21% and 26% in PC and FRC, respectively, when compared with bottom ones. As already observed for the anchorage strength, the stiffness reduction is greater in RAC (equal to 46% and 65% in Series 3 and 4, respectively). These results may support the assumption of a more uniform distribution of bond stress in top location, which may mitigate the casting position effect in long anchorages/laps. 12

| CONCLUDING REMARKS
This paper outlines a beam end test that is particularly useful to verify the applicability of existing provisions for bond and anchorage in the fib Model Code when using nonconventional concrete; however, the proposed test may also adopted for assessing the performance of new types of reinforcement.
After presenting and discussion the geometry of the new beam-end specimen, experimental results of 64 tests of long anchorages of steel ribbed bars in different types of concrete (PC, FRC, and RAC) are presented. The tests were carried out in two different laboratories.
Based on the test results, the following remarks can be made: 1. The proposed bond test is cost-effective with limited dimensions and easy handling; this test is capable to represent actual conditions of anchored bars in real design practice and it allows the casting position effect to be better investigated; 2. The tests results confirm the reliability of the MC2010 formulation to predict the strength of long anchorages, and hence its viability to serve as the basis for the design of the anchorage length to be used in a standardized bond test; 3. Anchorages having a length of about 20 times the bar diameter show a strength reduction of about 25% in top location with respect to the bottom one in normal strength concrete; 4. The use of FRC leads to a significant increase in the anchorage strength and a slight reduction of the TC effect with respect to PC; 5. Anchorages in bottom location of RAC specimens show a resistance similar to that in conventional concrete; however, a greater TC effect is observed in RAC, probably due to the segregation of concrete with blast furnace slag. Therefore, the current provisions of fib-Model Code on casting position can hardly be applied to RAC. However, a proper mix design of the RAC concrete would guarantee a more homogeneous mixture to limit the cast effect of the top bars. anchorage length N number of tested specimen in each series n b number of anchored bars n st total number of confining stirrups along the anchorage length n t number of stirrup's legs crossing the splitting failure surface P, P u applied load, ultimate load TC ratio between the maximum bar stress developed by a top bar with respect with that measured in the corresponding bottom cast one w crack width α 2 confinement coefficient provided concrete cover (as expressed by MC2010) δ L slip of the bar loaded end δ U slip of the bar unloaded end