Click for next page ( 8


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 7
7 In three separate rounds of testing, the Moustafa Test (now to directly measure development and splice strength in full- called the Large Block Pull-Out Test) was performed at differ- scale members. ent sites to determine its reproducibility across sites. In NASP The experimental work supporting the current require- Round I testing, the Moustafa Test was performed at Coreslab ments in the AASHTO LRFD Bridge Design Specifications Structures in Colorado and at Florida Wire and Cable Co. and ACI 318-05: Building Code Requirements for Structural (FWC). For the purpose of carrying out Moustafa Tests, FWC Concrete and Commentary (2005) for development of stan- built a completely automated testing machine so that the dard hooks in tension was conducted using a test setup rep- Moustafa procedures could be precisely followed. Round I test- resenting an exterior beam column joint. Because of the ing showed widely dissimilar results from the two testing sites. paucity of data on concrete strengths above 10 ksi, the evalu- In NASP Round II, the Moustafa Test and the PTI Bond Tests ation of uncoated and epoxy-coated bars terminated with were performed at three testing sites: Coreslab Structures, FWC, standard hooks in tension to normal-weight concrete with and the University of Oklahoma. Additionally, the NASP Bond compressive strength up to 15 ksi was performed using a sim- Test was introduced in an early form as a test very similar to the ilar exterior beam column joint test setup. PTI Bond Test except that a sand-cement mortar was used. The results of the initial work of NCHRP Project 12-60 Seven different strand samples were shipped to the different confirmed the basic premises stated in the original project testing sites. The trials were blind. Again, the Moustafa Test proposal. Thus, the efforts of the experimental program and failed to produce reproducible results across testing sites. Of the the order of priority of these efforts remained as originally three tests, the NASP Bond Test showed the highest statistical stated. The experimental program focused on the following correlation across testing sites. In the NASP Round III testing, major efforts listed in priority order: a more refined version of the NASP Bond Test again outper- formed the Moustafa Test in blind trials at the three testing sites. 1. Determining design equations for transfer and develop- In all three rounds of testing, when the Moustafa Test was used, ment length of strand in prestressed concrete bridge mem- it failed to produce results that were consistent across sites. The bers. Variables included concrete strength at release, con- NASP Bond Test proved more reliable at providing the same or crete strength at time of development length testing, use of similar results across testing sites in Rounds II, III, and IV. Be- air-entraining admixtures, "top bar effects," and strand size. cause of the NASP Bond Test's more consistent results, the 2. Development and splice length in tension of reinforcing NCHRP Project 12-60 testing program was built upon the bars. Variables included concrete strength, bar size, con- NASP Bond Test. crete cover/bar spacing, amount of transverse reinforce- The review conducted on testing for development/splice ment, epoxy coating, and casting position. length of deformed bars in tension showed that the generally 3. Development length in tension of bars terminated with recommended testing protocol for full-scale specimens because standard hooks. Variables included concrete strength, bar of both the relative ease of fabrication and the realistic state of size, concrete cover/bar spacing, amount of transverse stress achieved during testing is the beam-splice specimen. reinforcement, and epoxy coating. Thus, beam splice specimens were used in the development of experimental data related to development/splice length of mild A comprehensive article-by-article review of Section 5 of reinforcement during the course of the NCHRP Project 12-60 the 2nd edition of the AASHTO LRFD Bridge Design Specifi- study. It is well established that testing protocols to evaluate de- cations with the 1999, 2000, and 2001 interim revisions velopment and splice length requirements for deformed bars (AASHTO 1998) was conducted during the initial 6 months and wire in tension must be of an appropriate scale, containing of the NCHRP Project 12-60 study. In this review, the provi- more than one bar or wire, with due regard for a realistic trans- sions of Section 5 that directly or indirectly affect transfer and fer of force between concrete and steel reinforcement and development length of prestressing strand and develop- cover/bar spacing effects. Splice tests have in the past been ac- ment/splice length of mild reinforcement by the use of high- curate simulations of real conditions in structures; however, de- strength concrete were extracted and critically reviewed. velopment length tests have been largely conducted using pull- out tests, in which splitting failures are purposely avoided. As a result, the bond stresses developed along splices are low com- 2.2 Literature Review pared with the bond stresses along a bar in a pull-out test. This 2.2.1 Strand Transfer and Development difference in test methods is responsible for large differences in Length code-required anchorage lengths for splices and development of single bars. Pull-out failures occur in cases of high confine- A number of experimental investigations related to high- ment and short bonded lengths. In most structural applications, strength concrete have been conducted in North America and however, splitting failures tend to control. Beam-splice speci- overseas. Hence, a significant body of knowledge currently ex- mens are deemed to represent larger-scale specimens designed ists with respect to the performance of high-strength concrete

OCR for page 7
8 members. Amongst the experimental data are various studies large number of research programs intent on measuring the dealing with transfer length and development length of pre- transfer and development of prestressing strands. Research stressing strand and splice length and development length of was performed at the University of Texas (Russell and Burns black and epoxy-coated reinforcement. In this study, a com- 1996, 1997), Florida DOT (Shahawy, Issa, and Batchelor prehensive and critical literature review was undertaken to 1992), McGill University (Mitchell et al. 1993), and Auburn gather and synthesize existing data and information related to University (Cousins et al. 1993). The arbitrary 1.6 multiplier the transfer length and development length of prestressing from the original FHWA moratorium is now incorporated strand with diameters up to 0.6 in., and development and into the AASHTO LFRD Bridge Design Specifications. splice length in tension and compression of individual bars, By the mid-1990s, it became apparent that the studies ex- bundled bars, and welded wire reinforcement and develop- amining the transfer and development of prestressing strands ment length of standard. had not resulted in a consensus on design standards. As a The literature review centered on collecting information whole, the research displayed a large scatter of the test results, on testing protocols for determining surface bond character- with measured transfer lengths for 0.5-in. strand ranging istics of strand, performance of members containing trans- from a low of less than 20 in. to a high of more than 60 in. verse reinforcement, bond and transfer length, and tests Thus, it became apparent that other variables were in play addressing deformation capacity. Information available from and that such variables were not properly accounted for in the field--including FHWA showcase projects and the either design equations or specifications. unpublished experiences of engineers, bridge owners, and Since the mid-1990s, research work has concentrated on producers--was reviewed and used to supplement other developing a standardized test to assess the bond characteris- work conducted in this study. tics of individual prestressing strands. It was suspected that The development of reliable code expressions for transfer different strand manufacturers produced strand with quite and development of prestressing strand is made more diffi- dissimilar bonding characteristics. Hence, it was important to cult by the large experimental scatter reported by researchers quantify the bonding characteristics of an individual strand over the past 40 years. The original code expressions for before the transfer length and development length data transfer and development length of pretensioned strands would be meaningful. To that end, three or four different were developed from testing performed in the late 1950s and testing programs were undertaken to assess the viability of early 1960s on Grade 250, stress-relieved strand (Hanson and various "standardized tests" and the suitability of such tests Kaar 1959; Kaar, LaFraugh, and Mass 1963; Tabatabai and for predicting the "bond-ability" of prestressing strand. Dickson 1993). Based on these early tests, the ACI Building The first such testing program was developed by Rose and Code (ACI 2005) and the AASHTO LRFD Bridge Design Russell (1997). The various testing programs found little Specifications adopted provisions governing the design for correlation between a "simple pull-out" test and measured strand transfer and development. Manufacturing innovation transfer lengths. From these research programs, the precast has brought about Grade 270 low-relaxation strand as the in- concrete industry adopted a set of standard test procedures dustry standard, while the code expressions for transfer and that were to be employed in performing "pull-out" tests. The development length have changed very little. set adopted is known as the "Moustafa Test." Early results Furthermore, contemporary strand production employs using the Moustafa Test indicated that the test could be used induction heating to stress relieve strand, whereas convection to compare the bonding characteristics of strand on a relative heating was used in the late 1950s and early 1960s. Convec- basis. Logan (1997) demonstrated that the Moustafa Test, at tion heating created hotter surface temperatures on strand the recommended threshold value, would provide strand that may have burned off much of the surface residues with bonding capability more than adequate to meet current remaining from the wire drawing process. Today's processes, design assumptions. using induction heating, may have created surface tempera- In the meantime, the Post-Tensioning Institute commis- tures lower than those created by convection heating and sioned a study at Queen's University in Ontario (Hyett, thereby may have effectively changed the bonding character- Dube, and Bawden 1994). The study produced yet another istics of the surface of prestressing strands (Rose and Russell bond test, the "PTI Bond Test." The PTI Bond Test's primary 1997). purpose was to assess the bond characteristics of 0.6 in. di- In the mid-1980s, Cousins, Johnston, and Zia (1990) ameter strand and show the strand's suitability for use as a measured transfer lengths that exceeded the standard design rock anchor. In an appendix to ASTM A 416, the ASTM has predictions by a wide margin. Their findings led FHWA to adopted the PTI Bond Test on a provisional basis for 0.6 in. adopt a moratorium on the use of 0.6 in. diameter strands diameter strand that is to be used as rock anchors. and to increase the development length for other sizes of pre- Subsequent testing sponsored by the North American stressing strands. The FHWA action led to the creation of a Strand Producers Association (NASP) led to the development

OCR for page 7
9 of a third bond test, now called the "NASP Bond Test" portation agencies of the experimental results on transfer (Russell and Paulsgrove 1999b). In "blind trial" testing, the length and development length of strand in concrete. Moustafa Test, the PTI Bond Test, and the NASP Bond Test Round II of the NASP tests examined the proposed stan- were performed at multiple sites. The results of the blind trial dardized tests for repeatability and reproducibility. The re- testing indicated that the NASP Bond Test provided the best sults clearly indicated that the NASP Bond Test was the most repeatability. Based on these results and on as yet unpub- reliable test of the three tests examined. Results from the lished results from NASP Round III testing, the NASP Moustafa Test are shown in Figures 2.1 and 2.2. Note that in recommended the use of the NASP Bond Test as the stan- the Moustafa Test, results from a majority of strands tended dardized test to assess the bond characteristics of prestressing to cluster near the threshold level, and a more poorly strands. Overall, experimental results clearly show that performing strand was inconsistently rated. In a similar plot, inherent quality differences exist in the bond of prestressing Figure 2.3 compares results from two different test series per- strands from various manufacturers. Accordingly, it is im- formed at the University of Oklahoma (OU) featuring the perative in a testing program to evaluate the bonding charac- NASP Bond Test. Finally, Figure 2.4 compares the NASP teristics of the prestressing strands used. The standardization Bond Test results at two different test sites. The repro- process will make possible nationwide adoption by trans- ducibility of test results proved to be quite remarkable and 50 r2 = 0.92 STRESSCON=(0.597)OU+14.0 C STRESSCON DATA (kips) 40 M K B PW A 30 Z 20 J "PERFECT" TEST 10 0 0 10 20 30 40 50 OU DATA (kips) Figure 2.1. Comparison of Moustafa pull-out values from Stresscon and OU. 50 40 "PERFECT" TEST FWC DATA (kips) C 30 K B PW A M 20 Z J r2 = 0.88 10 FWC=(0.625)OU+5.25 0 0 10 20 30 40 50 OU DATA (kips) Figure 2.2. Comparison of Moustafa pull-out values from FWC and OU.

OCR for page 7
10 25 r2 = 0.97 OU SERIES TWO DATA (kips) 20 SERIES TWO = (0.83)SERIES ONE + 2.34 P C A K M 15 W B 10 Z 5 J "PERFECT" TEST 0 0 5 10 15 20 25 OU SERIES ONE DATA (kips) Figure 2.3. Comparison of NASP Bond Test results at OU in separate test series. can be seen in the figures. The test has received unanimous also cancel out the most compelling reasons to use high- endorsement by the NASP as its testing standard. strength concrete in pretensioned girder applications. Russell (1994) showed that 0.6-in. strands must be placed at a spac- ing of about 2.0 in. c/c to enable designs to take advantage of 2.2.1.1 Effects of Strand Spacing high-strength concrete. Historically, AASHTO limited the strand clear spacing to a The Auburn report (Cousins et al. 1993) was one of the minimum of three times the strand diameter (3 db). In bridge more recent works dedicated to investigating the effects of codes prior to the AASHTO LRFD Bridge Design Specifica- strand spacing on transfer and development lengths of pre- tions, this provision was made an explicit part of the design tensioned strands. In the Auburn study, 0.5-in. pretensioned code. It is likely that this code provision mirrored the stan- strands were fully stressed and placed at 1.75 in. c/c in some dard of placing 0.5-in. strands at 2.0 in. c/c. If this provision beams and 2.0 in. c/c in others. The research demonstrated were extended to the larger diameter 0.6-in. strands, then the that there was no substantive difference in transfer lengths 0.6-in. strands would have to be placed at 2.4 in. c/c. Never- measured on beams. For beams with strands spaced at 2.0 in. theless, using this strand spacing would cancel out the eco- c/c, the measured transfer lengths averaged 44 in. For beams nomic value inherent in the use of 0.6-in. strand and would with strands spaced at 1.75 in. c/c, the measured transfer 25 FWC SERIES TWO DATA (kips) 20 "PERFECT" TEST C P A 15 K M B W 10 Z r2 = 0.97 FWC TWO = (0.658) OU ONE + 3.21 J 5 0 0 5 10 15 20 25 OU SERIES ONE DATA (kips) Figure 2.4. Comparison of NASP Bond Test results at two different test sites.

OCR for page 7
11 lengths averaged 47 in. The researchers concluded that the cle) had transfer lengths that varied between 18 and 36 in. strand spacing had no effect on the measured transfer lengths. Other strand manufacturers provided strand that varied In the same study, beams were also tested for strand devel- between 18 in. and 21 in. In NASP Round II testing, nine opment. As with the transfer length measurements, the data different strand samples were tested. The NASP Bond Test demonstrated that beams performed similarly regardless of demonstrated significant and measurable differences whether strands were spaced at 1.75 in. or 2.0 in. c/c. The re- between strands. In the NASP Round III testing, 10 different searchers concluded that spacing 0.5-in. strand at 1.75 in. c/c strand samples were tested. In these tests, the differences in did not adversely affect transfer or development length of the pull-out test results were demonstrated to correlate directly strands. The researchers also concluded that the research results with strand transfer and development lengths. could be extended to the use of 0.6-in. strands at 2.0 in. c/c. From their research, Cousins et al. (1993) drew two con- 2.2.1.3 Influence of Concrete Strength clusions. First, "decreasing the strand spacing in preten- sioned, prestressed members from 2.0 inches to 1.75 inches Cousins et al. (1993) also tested for transfer and develop- has no significant effect on transfer length and does not result ment lengths in two different strength classes of concrete. The in splitting of members at transfer of prestressing force." normal-strength concrete mixture resulted in concrete Second, "decreasing the strand spacing in pretensioned, strengths between 6,000 and 8,000 psi. The high-strength prestressed members from 2.0 inches to 1.75 inches has no concrete mixture resulted in concrete strengths between significant effect on development length or nominal moment 10,000 and 12,000 psi. Transfer lengths measured in the high- capacity." With regard to 0.6-in. strand, Cousins et al. (1993) strength concrete were, on average, 37 in.; the transfer lengths make the following statement, ". . . for the results reported measured in the normal-strength concrete were, on average, herein for specimens prestressed with 0.5 inch diameter 51 in. The higher concrete strength resulted in transfer strand, the use of 0.6 inch diameter strand at a spacing of lengths that were about 25 percent shorter. The researchers 2.0 inches does appear reasonable." concluded that "increasing the concrete strength . . . reduces Deatherage, Burdette, and Chew (1994) also reported on the transfer length and development length." research performed to determine the effect that strand spac- Two other significant research programs examined the ef- ing had on transfer and development lengths. In their study, fects of concrete strength on transfer and development 0.5 in. diameter strand was placed in pretensioned beams length. The first, undertaken by Zia and Moustafa (1977), with 1.75-in. and 2.0-in. spacing. Also, strands of three dif- recommended code expressions for transfer and develop- ferent diameters (0.5 in., 0.525 in., and 9/16 in.) were placed ment length that included the concrete strength parameter. in beams with 2.0-in. spacing. In their studies, Deatherage, Nearly 20 years later, Abrishami and Mitchell (1993) also per- Burdette, and Chew (1994) concluded that a c/c spacing of formed transfer and development length tests. They also rec- 1.75 in. should be permitted for 0.5 in. diameter strands. Also, ommended that concrete strength be incorporated into the the researchers stated that their data indicated that the bond code provisions. However, as noted above, the lack of data strength of pretensioned strand was roughly proportional to that are consistent from one research program to another has its strand diameter, indicating that strand spacing did not prevented the development of a consensus for code expres- influence the bond characteristics of strand appreciably. sions related to transfer and development length of preten- Accordingly, the authors recommended that the spacing sioned strands. requirements for 0.5-in. strand be reduced from 4.0 strand diameters to 3.5 diameters. If this principle is applied to 2.2.1.4 Tests of Strands Pretensioned 0.6-in. strands, the authors would effectively recommend a in High-Performance Concrete 2.1-in. spacing for 0.6 in. diameter strands. In the 1990s, several research programs were undertaken by various states to design and build bridges using high- 2.2.1.2 Strand from Different Manufacturers performance concrete (HPC). Most, if not all, of these proj- Deatherage, Burdette, and Chew (1994) included 0.5 in. di- ects incorporated high-strength concrete as part of the HPC. ameter strands from various manufacturers. The researchers In several of the projects, strand transfer length was measured, provide strand transfer and development length test data, but and development length tests were conducted to ensure ade- provide little comment on differences between manufactur- quate bonding properties from the pretensioned strands and ers. The data indicate that differences in measured transfer to add to the body of knowledge regarding the transfer and de- lengths exist among strands made by different manufactur- velopment of pretensioned strands in high-strength concrete. ers. In the Deatherage, Burdette, and Chew study (1994), the Perhaps the first of these tests was performed in Texas by 0.5-in. strand provided by FWC (as designated in their arti- Gross and Burns (1995). In this research, two rectangular

OCR for page 7
12 beams, 42 in. deep, were fabricated. Each beam employed pre- transfer length was 23.4 in. The concrete strength at release tensioned 0.6 in. diameter strands with 2.0 in. spacing. Trans- was 7,800 psi. fer lengths were measured and development length tested at The box beams were also tested for development length. each of the four ends. Concrete strengths were 7,040 psi at re- Concrete strength at the time of development length testing lease and 13,160 psi at the time of development length testing. was 11,000 psi. For embedment lengths in excess of 60 in., the From the four beam ends, an average transfer length of 14.3 strands demonstrated the ability to develop adequate tension in. was measured. This value is significantly less than the cur- force to support the flexural capacity of the beams. Subse- rent transfer length provision of 60 db found in the AASHTO quent failures were labeled as flexural failures. However, LRFD Bridge Design Specifications. Similarly, the development when the strand embedment length was set at 60 in. and 59 length for these 0.6-in. strands was found to be less than 78 in., in., web shear cracking formed in the webs of the box beams, which roughly corresponds to the development length given and strand anchorage failures ensued. The researchers re- by current AASHTO provisions. The history of these beams ported that the development length for the strand was 60 in. is also interesting. They were dubbed the "Hoblitzell-Buckner" Additionally, several research projects were undertaken in beams. Hoblitzell was employed by FHWA and was instru- the 1990s in part to investigate the transfer and development mental in developing the federal programs encouraging the length of 0.6-in. strands. Uniformly, these projects featured use of HPC. Buckner authored a report for FHWA titled, pretensioned 0.6 in. diameter strands and spaced at 2 in. c/c. An Analysis of Transfer and Development Lengths for Preten- The projects were sponsored by Texas (Barnes and Burns sioned Concrete Structures (Buckner 1994). Buckner reviewed 2000), Virginia (Roberts-Wollmann et al. 2000; Ozyildirim transfer length and development length data prior to 1992/ and Gomez 1999), and Georgia (Khan, Dill, and Reutlinger 1993 and developed some design recommendations based on 2002). Uniformly, these researchers concluded that 0.6 in. di- that earlier data. In his report, Buckner recommended that the ameter strands could be deployed safely using 2-in. spacing. design provision for transfer lengths be changed to reflect the The state of Virginia has also supported transfer length stress in the pretensioned strand prior to release (fpi) as testing of 0.6 in. diameter strains in HPC. Results reported by opposed to using the "effective prestress" after all losses, Ozyildirim and Gomez (1999) and Roberts-Wollmann et al. which is still found in the 318-02 Code (ACI 2002). Effectively, (2000) indicate that transfer lengths measured in HPC were Buckner's recommendation would have increased the re- substantially less than the transfer length predicted by the quirement for transfer length by about 25 percent. current code expressions. More interesting was Buckner's design equation for devel- Barnes and Burns (2000) reported on transfer lengths that opment length. In reviewing the data, Buckner concluded were measured on 36 AASHTO Type I beams pretensioned that the strain experienced by the prestressing steel at flexural with 0.6-in. strands. Strand spacing was 2 in. c/c. Concrete strength level was an important component in the develop- compressive strengths at release ranged from 3,950 to 11,000 ment of strand. His design equation required the design en- psi. Altogether, transfer lengths from 192 independent meas- gineer to increase development length requirements as the urements are discussed, and the report includes data on steel strain at flexural strength level increased. The Hoblitzell- strands that are fully bonded to the ends of the member and Buckner beams were designed, therefore, to develop ex- strands that are shielded, or debonded, at the ends of the tremely large strains in the prestressing steel at flexural member. The results of the Barnes and Burns study (2000) strength and test Buckner's proposal. In the subsequent de- indicate a definite trend in which transfer lengths tend to de- velopment length tests reported by Gross and Burns (1995), crease in inverse proportion to the square root of the concrete the strands were able to achieve their ultimate tensile capac- strength at release. A "best fit" line reported by the authors in- ity, undergo very large elongation strains, and adequately cludes the square root of the concrete strength at release in develop their tension capacities within the current AASHTO the denominator. This relationship is shown in Figure 2.5. design provision. The results of these tests suggested that However, the data demonstrate wide variation, and the sta- strand strain did not play an important role in strand devel- tistical correlation is relatively weak. Nonetheless, it appears opment, and therefore it would not be necessary to recom- that concrete strength is an important factor that may affect mend that the AASHTO LRFD Bridge Design Specifications the bond of pretensioned strand. should contain a development length provision based on Barnes and Burns (2000) also reported results on transfer predicted strand strain at flexural strength levels. lengths of strand from various strand manufacturers. Their The state of Colorado sponsored a research program results are illustrated in Figure 2.6. The data illustrated in Fig- specifically designed to assess the transfer length and devel- ure 2.6 demonstrate that wide variations in measured transfer opment length of 0.6-in. strands pretensioned in HPC box length may be the result of differences among strand manu- beams (Cooke, Shing, and Frangopol 1998). In these beams, facturers. This finding highlights the need to establish an in- 0.6-in. strands were spaced at 2 in. c/c. The average measured dustry standard for the "bond-ability" of prestressing strand.

OCR for page 7
13 1 MPa0.5 = 0.381 ksi0.5; 1 MPa-0.5 = 2.63 ksi0.5 70 fpt AASHTO LRFD lt = 0.22 MPa-0.5 db 60 f'ci ACI/AASHTO Standard Transfer Length (db) 50 40 fpt lt = 0.13 MPa-0.5 db f'ci 30 R = 0.37 20 10 0 0 50 100 150 200 250 300 fpt (MPa0.5) f'ci Figure 2.5. Data comparing transfer lengths to concrete strength at release (Barnes and Burns 2000). 1 MPa0.5 = 0.381 ksi0.5; 1 MPa-0.5 = 2.63 ksi-0.5 150 Manufacturer A fpt Manufacturer B lt = 0.5 MPa-0.5 db 120 f'ci Manufacturer C Transfer Length (db) Manufacturer D 90 Unknown AASHTO LRFD 60 ACI/AASHTO Standard 30 0 0 50 100 150 200 250 300 fpt 0.5 (MPa ) f'ci Figure 2.6. Data highlighting differences among strand manufacturers (Barnes and Burns 2000). In addition to the research projects explicitly discussed examine the effects of air entrainment on pretensioned bond. herein, there have been other projects across the United The research reported herein incorporates the use of air en- States that have incorporated the use of 0.6 in. diameter trainment; however, it should be noted that air entrainment strands and spaced at 2.0 in. c/c. Many of those projects have is not usually specified in combination with high-strength measured transfer lengths. One of the projects was performed concrete/HPC because air entrainment directly causes a by Kahn, Dill, and Reutlinger (2002). In some cases, the decline in concrete strengths. research reports are still in a preliminary format and use of the data is being reserved by the authors and the research 2.2.1.6 Water Reducers and High Range sponsors. However, it is safe to say that, uniformly, these Water Reducers projects are employing 0.6 in. diameter strands at 2.0-in. spacing without adverse effects. There is no evidence of a systematic testing program ex- amining the effects of water reducers (WRs) or high range water reducers (HRWRs) on the transfer and development of 2.2.1.5 Effects of Air Entrainment prestressing strands. Since WRs and HRWRs are used in There is no evidence that a systematic testing program more than 95 percent of the pretensioned prestressing plants examining the effects of air entrainment on the transfer and throughout North America, this is an important variable that development of prestressing strands exists. There is a need to warrants investigation.

OCR for page 7
14 2.2.2 Development and Splice Length is present in the development length. Tests have also shown for Mild Reinforcement that the development length of deformed welded wire rein- forcement is not affected by epoxy coating, and thus the To identify needed experimental research, the literature re- epoxy coating factor in the current ACI Code is 1.0 for epoxy- view focused on the analysis of test results from bond tests on coated deformed wire fabric. In recent years, welded wire development and splice length in tension of coated and un- fabric (WWF) has been used widely as shear reinforcement in coated bars and development length of coated and uncoated thin-webbed girders because of the ease of construction over bars terminated with standard hooks in tension. Based on the the use of conventional stirrups. Research conducted to date, reported bond performance of individual and bundled bars with concrete compressive strengths up to 12 ksi, indicates in compression, it was determined that no additional exper- that this reinforcement can be used effectively to resist shear imental work was required in this area. Compression devel- (Mansur, Lee, and Lee 1987; Xuan, Rizkalla, and Maruyama opment lengths are considerably shorter than tension 1988; Pincheira, Rizkalla, and Attiogbe 1989; and Zhongguo, development lengths because there are no transverse cracks Tadros, and Baishya 2000). It was shown that two cross wires in compression zones; the harmful effect of such cracks in ini- welded at a spacing of 2 in. at the open ends (top and bottom) tiating splitting is absent. However, the major difference be- of WWF cages provide satisfactory anchorage. Such anchor- tween tension and compression development and splice age was found to be more effective for deformed WWF than lengths is the ability of the bars in compression to transfer smooth WWF. The increase in concrete compressive strength load to the concrete directly by bearing. In tests conducted by has been shown to further improve the anchorage of this Pfister and Mattock (1963), bearing stresses equal to five reinforcement. times the cylinder strength of the concrete were attained at the square-cut ends of bars in compression splices. Addi- tional experimental work conducted at the Otto-Graf- 2.2.2.1 Databases Institute of the University of Stuttgart by Leonhardt and Teichen (1972) conclusively showed the following: There are two databases. One consists of 71 tension devel- opment and splice tests of specimens with top cast uncoated End bearing is responsible for the majority of splice failures reinforcing bars, 493 specimens with bottom cast uncoated in compression irrespective of the splice length tested. The reinforcing bars, 27 specimens with top cast epoxy-coated splice lengths varied between 9 and 38 bar diameters. bars, and 48 specimens with bottom cast epoxy-coated bars, The bearing capacity of the concrete at the bearing ends of for a total of 639 specimens. The other database consists of 33 the bars was increased by the presence of confining rein- specimens with uncoated bars terminated with standard forcement. Under such conditions, concrete bearing hooks and 13 specimens with epoxy-coated bars, for a total of stresses of 17 ksi were measured (for concrete with a uni- 46 specimens. axial compressive strength around 4 ksi). The provisions for development length of reinforcement in An increase in the thickness of the concrete cover over the Section 5 of the AASHTO LRFD Bridge Design Specifications compression splice resulted only in very minor improve- are based on the provisions of ACI 318-89 (ACI 1989). The ments in bond performance. 1989 provisions in the ACI Code were extensively modified Under long-term loading, the bearing pressure under the in the 1995 version of the ACI Code (ACI 1995) with a view ends of the compression bars diminishes because of creep; to formulating a more "user-friendly" format while main- hence, the splice performance improves. taining the same general agreement with professional judg- ment and research results. Tests conducted by Azizinamini et The available information on the anchorage in tension of al. (1993, 1999a) have indicated that in the case of high- welded wire reinforcement indicated that a significant exper- strength concrete, some minimum amount of transverse re- imental effort was not required as part of NCHRP Project inforcement is needed to ensure adequate ductility from the 12-60 (Furlong, Fenves, and Kasl 1991; Griezic, Cook, and splice at failure. A proposed modification to ACI 318-99 (ACI Mitchell 1994; and Guimaraes, Kreger, and Jirsa 1992). In the 1999), based on these tests, called for the determination of a case of plain wire fabric, the development in tension depends basic straight development length for bars in tension without on the mechanical anchorage from at least two cross wires. including the presence of transverse reinforcement, together Deformed welded wire reinforcement derives anchorage with a minimum area of transverse steel in the form of stir- from bond stresses along the deformed wires and from rups, Asp, crossing potential splitting planes. In these studies, mechanical anchorage from the cross wires. Current code ex- over 70 specimens with concrete compressive strengths rang- pressions for development length in tension of deformed ing between 5 ksi and 16 ksi were tested (Azizinamini et al. welded wire reinforcement assume that at least one cross wire 1993, 1999a).

OCR for page 7
15 Although the modification proposed in another paper by above 8 ksi. In order to assess whether the limit on f c can be Azizinamini and colleagues (1999b) was not adopted in the removed by examining the existing data for development and 2002 version of the 318 Code (ACI 2002), it was deemed an splice length of uncoated and epoxy-coated bars in tension, improvement over the current AASHTO LRFD provisions. the ratio of test to calculated bond strength is plotted versus Therefore, the 2005 318 Code (ACI 2005) provisions were the concrete compressive strength (f c) evaluated throughout used in NCHRP Project 12-60 as the basis for further exten- the range of concrete cylinder strengths. The bond strength sion of the AASHTO provisions to higher strength concrete. ratio is determined in terms of bar stresses at failure versus The experimental work conducted in the mild steel phase of calculated bar stress, using Equations 2.2 through 2.5: NCHRP Project 12-60 was focused on filling the gaps identi- ls 3 fs fied in order to extend the applicability of the present = (2.2) db 40 c + K tr AASHTO LRFD Bridge Design Specifications to normal- fc weight concrete with compressive strengths up to 15 ksi. db The 639-specimen database is shown in Figures 2.7 through c = ( c min + 0.5 * db ) (2.3) 2.10 by plotting the bond strength, utest, versus the concrete compressive strength, f c. The bond strength is defined as Atr * f yt K tr = (2.4) 1500 * s * n Ab f su utest = (2.1) To limit the probability of a pull-out failure, 318 Code (ACI db ls 2005) requires that In Equation 2.1, Ab is the area of bar being developed or c + K tr spliced, fsu is the stress in the bar estimated at failure using 2.5 (2.5) db moment-curvature type analysis and compatibility of defor- mations, db is the diameter of the bar, and ls is anchorage/ The additional parameters in the equations are the follow- splice length. As can be seen from Figures 2.7 and 2.8, there ing: fs is the stress in the reinforcing bar; cmin is the smaller of is a lack of data for development and splice lengths of un- minimum cover or one-half of the clear spacing between coated bars in tension above 16 ksi. Figures 2.9 and 2.10 show bars; Atr represents the area of each stirrup or tie crossing the that there are limited data for epoxy-coated bars in tension potential plane of splitting adjacent to the reinforcement 1.60 1.40 1.20 1.00 Utest (ksi) 0.80 0.60 0.40 0.20 0.00 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 ' (psi) fc Bottom Cast Uncoated Figure 2.7. Bond stress at failure (utest) versus the concrete compressive strength ( fc ) of bottom cast uncoated specimens.

OCR for page 7
16 1.60 1.40 1.20 1.00 utest (ksi) 0.80 0.60 0.40 0.20 0.00 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 f' c (psi) Top Cast Uncoated Figure 2.8. Bond stress at failure (utest) versus the concrete compressive strength ( fc ) of top cast uncoated specimens. 1.00 0.90 0.80 0.70 0.60 utest (psi) 0.50 0.40 0.30 0.20 0.10 0.00 0 2000 4000 6000 8000 10000 12000 f' c (psi) Epoxy-Coated Bottom Bars Figure 2.9. Bond stress at failure (utest) versus the concrete compressive strength ( fc ) of bottom cast epoxy- coated specimens.

OCR for page 7
17 0.80 0.70 0.60 0.50 utest (ksi) 0.40 0.30 0.20 0.10 0.00 0 2000 4000 6000 8000 10000 12000 14000 f' c (psi) Epoxy-Coated Top Cast Figure 2.10. Bond stress at failure (utest) versus the concrete compressive strength ( fc ) of top cast epoxy-coated specimens. being developed, spliced, or anchored; fyt is the yield strength Tests have shown that the bar force is transferred rapidly of the stirrup reinforcement; s is the spacing of stirrups; and into the concrete, and the portion following a hook is gener- n is the number of bars being developed or spliced. The re- ally ineffective and can potentially be limited by the tensile sults of the evaluation indicated that the average of the ratio strength of the concrete. Marques and Jirsa (1975) reported for bars not confined by stirrups is 1.23 with a standard devi- on the results of 22 tests conducted using two #7 or two #11 ation of 0.28 for all f c values, and 1.23 with a standard devia- uncoated bars. Standard 90- or 180-deg hooks conforming to tion of 0.23 for concrete compressive strengths below 10 ksi. the 318 Code were used (ACI 2005). The concrete compres- In the case of bars confined by stirrups, the average is 1.23, sive strength was around 5 ksi. The specimens simulated and the standard deviation is 0.3 for all f c values. For f c values exterior beam column joints. Hamad, Jirsa, and D'Abreu de below 10 ksi, the average is 1.24 and the standard deviation is Paulo (1993) reported on the results of 24 tests to evaluate the 0.30. In members with confined bars, the stirrups are as- anchorage performance of epoxy-coated hooked bars. Based sumed to be uniformly spaced throughout the splice/devel- on these results, a 20-percent increase on the basic develop- opment length. The value for the members in the database, ment length was recommended for epoxy-coated hooked calculated by the ACI provisions, gives approximately the bars. It was shown that the relative anchorage strength of un- same scatter throughout the range of concrete compressive coated and epoxy-coated hooked bars was independent of bar strengths up to a maximum of 16 ksi for members with and size, concrete strength, side concrete cover, or hook geome- without stirrups. This conclusion supports the extension of try. The maximum concrete strength of the specimens was 7 these provisions to higher concrete compressive strengths ksi. These tests serve as the basis of the 318 Code anchorage with a few verification tests of uncoated bars at the upper provisions for bars anchored by means of standard hooks limit, mainly to establish the role of the minimum amount of (ACI 2005). The specimen and the test setup used in NCHRP transverse reinforcement on the mode of failure of splices in Project 12-60 was similar to the one used in the Marques and tension recommended in the Azizinamini et al. studies (1993, Jirsa (1975) and Hamad, Jirsa, and D'Abreu de Paulo (1993) 1999a). On the other hand, it is recognized that there is a studies. However, only 90-deg hooks were evaluated, since paucity of data on the performance of epoxy-coated bars in Hamad, Jirsa, and D'Abreu de Paulo found little difference in concretes with compressive strengths above 10 ksi. Therefore, the performance of 90- and 180-deg hooks. It should be a more intense verification testing effort was carried out in noted that sections of the AASHTO LRFD Bridge Design this study to close this gap. Specifications dealing with the anchorage of bars in tension