Reinforcing Bar Pullout Bond in Tremie-Placed Concrete Cast in Drilling Slurry Environments by Kelly Costello, Sarah Mobley, Justin Bowen, and Gray Mullins

Home , Rebar

ACI STRUCTURAL JOURNAL TECHNICAL PAPER Title No. 116-S87 Reinforcing Bar Pullout Bond in Tremie-Placed Concrete Cast in Drilling Slurry Environments by Kelly Costello, Sarah Mobley, Justin Bowen, and Gray Mullins

Drilling slurry made from a mixture of water and mineral or tion testing for any excavation stabilizing product (slurry) polymer powder is often used to stabilize deep excavations, in used during this process. This is done to ensure that previ- which cast-in-place reinforced concrete foundation elements such ously unapproved products perform to the same level as as drilled shafts are constructed. At the time of concreting, drilling those currently in practice. This study shows the influence of slurry is displaced by heavier, highly fluid concrete tremie placed slurry type on reinforcing bar bond strength. at the bottom of the excavation from within the reinforcement cage. Despite the fluidity of the concrete, it must build up within the BACKGROUND reinforcing cage to a sufficient height before then radially pressing into the annular cover region. This flow pattern has been shown Tremie placement is the requisite method of concreting in to trap slurry near the steel reinforcement and affect reinforcing submerged conditions whereby a small diameter pipe (for bar bond. This paper presents the results of 227 reinforcing bar example, 6 to 10 in. [150 to 250 mm]) is used as a concrete pullout tests performed over a 6-year period from 2012 to 2018. delivery conduit to prevent segregation as the concrete flows Shaft specimens were tremie placed in various slurry types and through the water to the bottom of an excavation. Even within consistencies. This analysis concluded that present development the tremie pipe, means to prevent segregation are used which length values should be increased by 1.8 for bentonite and 1.9 for include: capping/sealing the bottom of the pipe to prevent the polymer slurry; water casting environments showed no significant intrusion of water as the tremie is lowered to the bottom of the change relative to dry conditions. excavation prior to concreting, or use of a separation mate- Keywords: development length; drilled shaft; mineral; polymer; rein- rial such as a foam plug in front of the concrete that is pushed forcing bar bond; slurry. through the tremie or pump line as concrete moves down to the bottom of the excavation. Applications for tremie placement INTRODUCTION include seal slabs, drilled shafts, cutoff walls, dams, or anytime Drilled shafts (also known as bored piles or cast-in- concrete is placed below water. drilled-hole piles) are large-diameter cast-in-place rein- During drilled shaft excavation and concreting, sidewall forced concrete foundation elements used to support bridges stability is often achieved through hydrostatic pressure wherein and tall buildings. Due to the axial and lateral capacity that a mixture of water and mineral or polymer powder is intro- can be developed, a single shaft can replace a group of duced into the excavation and maintained at an elevation at smaller-sized driven piles. Structurally, the design of shafts least 4 ft above the ground water table.4 The effectiveness of uses the same resistance factors as their above ground rein- this slurry in providing stability is highly reliant on the slurry forced concrete counterparts, which assumes the same level viscosity, which is tracked using the Marsh funnel test method. of quality assurance. This extends to those portions of the This method times how quickly a known volume of a fluid shaft below the water table where the cage is placed, and the discharges with falling head from a standardized funnel.5 The concrete is tremie-placed under submerged conditions. unit of measure is seconds per quart; the thicker the slurry the The two most common complications that arise during higher the Marsh funnel viscosity. Typically, the workable range drilled shaft construction stem from: 1) excavation stability; of these materials is 30 to 50 s/qt for mineral slurry6 and 50 to and 2) concrete-related flow properties. The latter is further 90 s/qt for polymer slurry.7 For reference, water has a Marsh complicated by reinforcing cage congestion/spacing and the funnel value of 26 s/qt, which is the lower limit (1 qt = 0.95 L). properties of the slurry being displaced during concreting.1 Research focused on tremie-placed seal slabs around This paper discusses the effects of submerged shaft steel H-piles and prestressed concrete piles showed that the construction on reinforcing steel bond where ground water fluid displaced by the concrete can affect the resulting bond or drilling slurry is displaced during concreting. between the pile and seal slab.8-11 Prior to those investiga- tions, the design of seal slabs assumed the bond to be negli- RESEARCH SIGNIFICANCE gible or was fully discounted. Results showed that in some Present specifications applicable to reinforcing bar bond cases, significant bond could be expected. However, when strength from the American Concrete Institute (ACI)2 and the American Association of State Highway and Transporta- ACI Structural Journal, V. 116, No. 4, July 2019. 3 MS No. S-2018-259, doi: 10.14359/51715570, was received July 12, 2018, and tion Officials (AASHTO) make no provisions for the effect reviewed under Institute publication policies. Copyright © 2019, American Concrete of tremie placement of concrete in wet conditions; however, Institute. All rights reserved, including the making of copies unless permission is 4 obtained from the copyright proprietors. Pertinent discussion including author’s some state guidelines require reinforcing bar bond verifica- closure, if any, will be published ten months from this journal’s date if the discussion is received within four months of the paper’s print publication.

ACI Structural Journal/July 2019 183 of cage and cover region was required to develop enough pressure to push the concrete radially through the cage and into the cover. The differential height was affected by clear spacing, rate of concrete placement, and the maximum coarse aggregate size. When compared to dry tremie control shafts, the buoyant weight of concrete in slurry-cast shafts was shown to increase the differential height relative to the decrease in buoyant unit weight. These findings were a departure from the previously accepted notions that concrete flow was vertical and uniform across the shaft, easily displaced the slurry (like water under oil), and that upward flow in the cover would scour the soil side walls of slurry product buildup as well as the reinforcing bar. This also opened the eyes of researchers and owners and revealed that Fig. 1—Steel pile to seal slab bond.10 slurry could become trapped around the reinforcing bar and against the side walls. the displaced fluid was bentonite slurry, a notable reduction Numerous studies15-18 have shown that use of bentonite in bond resulted. Figure 1 shows the effect of fluid type on slurry to support excavations for drilled shaft results in tremie-placed seal slab to steel pile bond. In that study, full- reduced concrete to soil bond by as much as 50% when scale seal slabs were cast around W14 x 90 steel pile sections compared to polymer or water-cast shafts. Early explanations and 14 in. (0.36 m) prestressed concrete piles in cofferdams attributed the reduced side shear to a filter cake that forms as flooded with water or bentonite slurry; dry conditions were the bentonite suspension/slurry flows into the surrounding also prepared as controls. For the steel pile tests, the average soil and deposits the clay particles on the excavation walls. water-submerged conditions gave average bond values 4% While filter action is somewhat responsible for the presence less than dry, and the bentonite slurry environment resulted of the bond-reducing bentonite, trapping of bentonite caused in up to a 54% reduction. Similar results were noted for the by radial flow was shown to be a far more significant mech- concrete piles. No polymer slurries were tested. anism for the deposition of bentonite along the side walls.19 Based on a study conducted by Butler in 1973,12 the Bowen20 and later Mobley et al.1 showed that laitance Federal Highway Administration (FHWA) concludes: “The creases form as radially flowing concrete fills the cover current state of knowledge on this topic suggests that the use region. The creases radially project from the cage config- of mineral and polymer slurries for drilled shaft construc- uration to the surface of the shaft sides. The Deep Founda- tion does not reduce the bond resistance between concrete tion Institute (DFI) termed this phenomenon as mattressing, and reinforcing bars. There is currently no reason to account giving the appearance of a quilted mattress cover.21 The for the use of drilling fluids when considering develop- cover integrity, concrete strength, and corrosion protection ment length of rebar in drilled shafts.”6 However, that study were all shown to be substantively affected in bentonite cast included only two pullout tests on No. 8 (M25) reinforcing shafts.1 Coring the intersection of these creases revealed bar where tremie-placed concrete displaced bentonite slurry; the concrete was not contiguous, and for high-viscosity one failed in bond splitting failure with 2.25 in. (5.72 cm) bentonite slurry, the cores separated into four pieces defined of cover in a 30 in. (0.76 m) diameter shaft; the other from by the laitance crease locations.13,20 Figure 2 shows, from the same shaft failed through cover spalling from poor jack top left to bottom right: a) the buildup of concrete inside alignment. The bentonite slurry was not a pure bentonite the cage noted by Deese and Mullins14; b) laitance creases slurry but rather a bentonite-polymer blend denoted as high- showing the cage projection to the sides of the shaft; c) radial yield bentonite, a product that has half the suspended solids coring the intersection of creases; d) laitance crease inside of pure bentonite and which is often not accepted for bridge cored hole walls extending to cage depth; e) core extracted construction applications.4 Although not reported, the Marsh from water-cast shaft with no change in quality; f) core from funnel viscosity of the slurry would have been approx- 60 s/qt (1 qt = 0.95 L) polymer cast shaft with minor void on imately 30 s/qt (1 qt = 0.95 L) based on the documented cover side of main bar; g) core from 40 s/qt (1 qt = 0.95 L) mixture ratio of 0.21 lb/gal. (25 kg/m3)12 and mixture ratio bentonite-cast shaft with void behind main bar and laitance to viscosity correlations for this type of slurry.13 The study crease extending to the shaft side wall at top of picture; and also included 12 laboratory-prepared pullout specimens, six h) and i) four pieces from the four quadrants of the cores with deformed bars and six smooth. Deformed bar tests were extracted from 50 and 90 s/qt (1 qt = 0.95 L) bentonite-cast further broken out to have three cast-in-dry conditions and shafts, respectively. The four pieces of the cores (all 4 in. three “coated with mud slurry before casting in concrete,” [102 mm] diameter) were still coated along the crease inter- but not tremie-placed, slurry-displaced.12 In short, the faces with trapped bentonite despite the wet coring method FHWA guidelines are based on five tests, none of which are used. Figures 2(e) through (g) are aligned with the shaft-side representative of construction practices. walls at the top of the image. Deese and Mullins14 documented concrete build up within As use of bentonite slurry has been shown to affect all the reinforcing cage during tremie placement and where aspects of cast-in-place foundation performance including a critical differential concrete height between the center concrete-to-soil bond, corrosion resistance, seal slab-to-pile

184 ACI Structural Journal/July 2019 Fig. 2—Tremie-placed shaft specimens cored to show effects of radial concrete flow. bond, concrete cover strength, and integrity, it is reasonable DEVELOPMENT LENGTH AND BOND STRENGTH to assume that the reinforcing bar-to-concrete bond can be Whereas development length is a practical parameter similarly affected. Although the FHWA states no effects required to ensure proper reinforcing steel performance, result from slurry-placed concrete on reinforcing bar bond, its determination stems from studies23-27 focused on the the Florida Department of Transportation (FDOT) requires bond stress (u) between concrete and reinforcing bars. ACI all new slurry products to “demonstrate the bond between Committee 40828 discussed how resistance factors could the bar reinforcement and the concrete is not materially or should be computed and applied to development length affected by exposure to the slurry under typical construction equations. The report used a database of splice and pullout conditions, over the typical range of slurry viscosities to be test results to compute the mean bias (measured/predicted), used.”22 The motivation for this study stemmed from veri- standard deviation, and the coefficient of variation (COV) fication tests required to satisfy this specification. Thereby, values for that data.28 The committee report defined the the effects of various polymer and mineral slurry products reinforcing bar bond strength reduction or resistance factor were tested over a wide range of viscosities. (Øb) and the effective strength reduction factor (Ød = Øb/Ø) for use in calculating an appropriate development or splice

ACI Structural Journal/July 2019 185 Fig. 3—Pullout testing methods: (a) pullout specimen; (b) beam end specimen; (c) beam anchorage specimen; (d) splice spec- imen28; and (e) tremie-placed specimen used in this study. length. The effective strength reduction factor simply slurry to evaluate the actual performance. ACI 318 and removes the effect of any resistance factor associated with a recommended 408R-03 equations for both pullout capacity given loading type (for example, Ø for bending or tension) (Eq. (2) and (4)) and development length (Eq. (1) and (3)) from the bond resistance factor. Hence, during the computa- are used as comparative baselines for reinforcing bar pullout tion of a given system resistance/strength (such as bending), capacities when cast in slurry. the reduction associated with Øb is fully in place without compounding the overlapping effects of two different resis- APPROACH tance factors. Equations (1) through (4) show how a resis- The ACI 408R-03 report28 defines four different pullout tance factor is introduced into development length and bond testing configurations, also shown in Fig. 3: Case a) the force equations from ACI 3182 and ACI 408R-03 (recom- pullout specimen; Case b) the beam end specimen; Case c) mended).28 While it is not explicitly stated whether or not the beam anchorage specimen; and Case d) the splice spec- Eq. (1), ACI 318 development length equation, presently imen.2 For the testing performed during this study, a combi- includes this resistance factor, it is assumed based on the nation of Case a) and b) was adopted: direct pullout of verti- ACI 408R-03 report28 that one has not been included and cally cast specimens with debonded regions (Fig. 3(e)). that perhaps such a resistance factor should be considered All shaft specimens were tremie-placed after the slurry and cage were already in place. The shaft specimens were   tailored to meet FDOT minimum diameter and preferred tightest cage spacing criteria.4 Specimens were 42 in.  3 f ΨΨΨ  l =  y tes  d (1) (1.07 m) diameter, 24 in. (0.61 m) tall, and were formed with d  40  b λ∅dcf ′  cKbt+ r  a cylindrical steel sheet metal form. Reinforcing bar spacing       db   provided 6 in. (152 mm) clear spacing for both the longitudinal and circumferential steel. Longitudinal reinforcing bar  40A  for pullout specimens were slated to be every other main bar Tf= 10.472∅ ′ c + tr l (2) bbcb sn  d and where pullout bars were not permitted to touch the steel stirrups. Outer steel stirrups were placed outside seven of the main bars that were not tested. One-half inch (13 mm)   plastic hoops (simulated stirrups) were used to separate the  fs  − ∅ω2400 αβ1λ inner and outer rings of main bars (Fig. 4). The inner seven  1 d   ' 4  vertical bars/test specimens (No. 8 or M25) were tied inside  ( fc )  l = d (3) the plastic hoops but between the outer seven bars. The d   b 76.3 cKω + tr tested bars were machined down to 7/8 in. (22 mm) diam- ∅d    db  eter on top, threaded, and were debonded in the upper 8 in. (0.2 m) of embedded length with thin walled PVC pipes. The ()14/ 59..905 2400 lower 10 in. (0.25 m) was similarly debonded, resulting in a Tlbb=+fc′ ∅∅dm()cdbb+ Ab  bonded length of 6 in. (152 mm) where nothing touched the (4)  c   NA  12/ 01..M 09 30.88 tr 3 bonded region and concrete was unobstructed from directly  +  ++ ttrd  fc′  cm   n  contacting the bonded test length (Fig. 3(e)). Debonding the upper portion of the bars was designed to reduce the effects Given the noted effect of slurry of various kinds on bond, of the jacking-induced compressive stress at the surface (as 28 this study compared the pullout capacity of 227 reinforcing noted by the ACI 408R-03 report ) and was verified through bar specimens tremie-cast in water, mineral, and polymer numerical modeling to drastically reduce the compression

186 ACI Structural Journal/July 2019 Fig. 4—Cages with debonded upper and lower portions; water (top left); polymer (bottom left); and bentonite with capped/ sealed tremie before introduction (right). stress in the bonded region.29 Debonding the lower portion In general, the mean values of the experimental results for a allowed the bonded length to be adjusted. given concrete strength agreed with the predicted capacities Pullout bars were cast in sodium montmorillonite, infor- from ACI 318 and ACI 408R-03 (Fig. 5). As most speci- mally known as bentonite (API section 10 clay), synthetic mens were cast with FDOT Class IV (4 ksi [28 MPa]) shaft partially hydrolyzed poly acrylamide polymer, or water concrete mixture, many of the predicted capacity values (natural) slurry. The polymer slurries used were from two were similar (vertical banding). different manufacturers, where the same amount of pullout The bias (measured/predicted) values for all pullout bars were tested per manufacturer. This study examines samples were computed using both prediction methods and pullout specimens cast over 6 years, from 2012 to 2018, and plotted versus the Marsh funnel slurry viscosity (Fig. 6). seven concrete placements. The mean bias values for each casting condition are also For pullout testing, an 8 in. (0.2 m) diameter load cell was shown as dashed horizontal lines. A general trend is shown placed on lead plates on the surface of the concrete followed of decreasing bias (and pullout capacity) with increasing by a 60 ton (533 kN) capacity hollow-core hydraulic ram. An viscosity, but like Fig. 5, significant variability is still upper steel plate and double nuts were used to complete the apparent but now for a given viscosity. Higher-viscosity loading assembly. While Fig. 3 Case (a) is not the preferred slurry is required for more porous/free-flowing soils, while testing configuration per ACI 408R-03, Case (e) provided lower viscosity is suitable for fine-grained, low permea- compromise creating a realistic casting condition while miti- bility soils. While lower-viscosity slurry (close to water) gating the effects of local compression stresses. performed better in bond, it is not reasonable to restrict the Pullout testing was completed using a manually oper- use of higher-viscosity slurry. Therefore, statistical eval- ated hydraulic pump to provide a slowly applied load rate uations of slurry effects were performed lumping the data of approximately 100 lb/s (0.4 kN/s). The load cell and a into the basic slurry types (water, bentonite, and polymer). displacement transducer were monitored with a computer- The raw data and details for each pullout test specimen are ized data acquisition system and sampled at 10 Hz to ensure included in the Appendix.* the peak force was captured. The testing was performed Shown in Table 1 are the mean bias, standard deviation, after the concrete reached a minimum compressive strength and COV values for the slurry types. Table 1 also includes of 4 ksi (28 MPa). All pullout testing was completed on the the previously published values for dry conditions, reported same day as compressive cylinder strength testing. Out of by Darwin et al.25 and the ACI 408 Committee.28 the 227 pullout tests, 131 were performed in mineral slurry, Using the mean bias and standard deviation, lognormal 56 in polymer, and 40 in water. probability density curves were prepared for the two prediction methods (Fig. 7). These curves do not include the effect TEST RESULTS Using Eq. (2) and (4), the predicted capacity was calcu- *The Appendix is available at www.concrete.org/publications in PDF format, lated for all test specimens where the resistance factor was appended to the online version of the published paper. It is also available in hard copy taken as 1.0. This was then compared to the measured value. from ACI headquarters for a fee equal to the cost of reproduction plus handling at the time of the request.

ACI Structural Journal/July 2019 187 Table 1—Mean bias and COV values for various conditions using ACI 318 and ACI 408R-03

Dry (ACI) Water Bentonite Polymer ACI Eq. 318 408R-03 318 408R-03 318 408R-03 318 408R-03 Mean bias (¯r ) 1.23 1.00 1.21 1.23 0.80 0.88 0.99 0.97 Standard deviation 0.30 0.12 0.31 0.32 0.22 0.21 0.36 0.36

COV (Vr) 0.24 0.12 0.25 0.26 0.28 0.24 0.36 0.37

Fig. 5—Measured strength versus predicted strength for ACI 318 (left); and ACI 408R-03 (right). (Note: 1 kip = 4.45 kN.)

Fig. 6—Bias versus slurry viscosity for ACI 318 (left); and ACI 408R-03 (right). (Note: 1 qt = 0.95 L.)

25 of a resistance factor (Eq. 2 and 4, Øb = 1). The vertical line Using the approach outlined by Darwin et al. to compute shown at 1.0 defines the threshold above which the measured resistance factors for reinforcing bar development length capacity is generally acceptable (≥ 1.0) or below which is (Eq. (5)), resistance factors were computed using the test unacceptable (<1.0). data (Table 1) bias ( ¯r ) and coefficient of variation (COV 28 Using the statistical values from ACI 408R-03 for a dead or Vr) values for both ACI 318 and ACI 408R-03 predic- load/live load ratio of 2, the mean measured/predicted load tion methods. While not fully expanded in Eq. (5), several bias, and the associated coefficients of variation for load variables are taken into consideration when determining (Table 2) along with the values from Table 1 for resistance, the mean value of the random loading variable (¯q ), and the Monte-Carlo simulations were run for each condition (dry, COV of the loading random variable (Vφq) which include: water, bentonite, and polymer).30 The simulations were actual-to-nominal dead and live load random variables, used to predict the probability of failure using one million dead and live load factors, nominal ratio of live load to dead randomly generated values for both the load and strength. load, and coefficients of variation pertaining to dead and live The failure ratios ranged from 1:2 for bentonite to 1:30 for loads (Table 2). These values were again obtained from ACI dry conditions (Fig. 7). It should be noted that the failure 408R-0328 aside from the COV for dead and live loading ratio is not assigned on the basis of the fraction of bias below which were obtained from Darwin et al.25 A reliability index 1.0, but rather where the random variations in load and β, value of 3.5, was used. strength result in a strength/load ratio below 1.0.

188 ACI Structural Journal/July 2019 Table 2—Parameters used in Eq. (5)

Load factor, DL 1.2 Load factor, LL 1.6 DL/LL ratio 2 Reliability index 3.5 Load bias, DL 1.03 Load bias, LL 0.975 Load COV, DL 0.093 Load COV, LL 0.25 Fig. 7—Lognormal probability density for ACI 318 (left); ¯q 0.76 and ACI 408R-03 (right). Vφq 0.1

Notes: DL is dead load; LL is live load.

Fig. 8—Lognormal probability density for ACI 318 (left); and ACI 408R-03 (right) after applying resistance factors. Fig. 9—Effect of (cb + Ktr)/db term on water failure ratios; no limit (left); with 2.5 limit before (red) and after (black) 2212/ r −+VV∅ β ∅ = e ()rq (5) 1.65 load multiplier (right). b q tively. Monte-Carlo simulations were performed to assess the resulting failure ratios with the 2.5 limit and showed no Table 3 shows the computed resistance values Øb for the different slurry types along with the values reported for dry failures in 1 million trials, meaning that use of this limit (2.5) conditions from ACI 408R-03.28 is more conservative than the target reliability index of 3.5 Applying the calculated resistance factors to the predicted requires. Simulations were again performed where an addi- pullout resistances, the bias values were recomputed and the tional load multiplier was applied to the existing combined lognormal probability density curves were replotted (Fig. 8). live and dead load multiplier (1.33 for DL/LL ratio of 2) and With the application of the computed resistance factors, the progressively increased until the failure ratio provided a 3.5 majority of all curves lay to the right side of the 1.0 bias reliability index. Figure 9 shows the effect of the (cb + Ktr)/db value line. Monte-Carlo simulations were again performed limit on failure ratio for water specimens where the load and to confirm the resistance factor reliability. All failure ratios resistance are normalized by the unfactored load. The mean now conformed to the intended acceptable level associated load is shown to be approximately 1.0 for both cases (limit with a reliability index of 3.5, below 1:4149. and no limit), but only after increasing the load by a 1.65 multiplier does the failure ratio increase from 0:1,000,000 to EVALUATION OF FAILURE RATIOS 1:4545 (1:4149 max) when the 2.5 limit is applied. Hence, Implied by Fig. 7 is that the current development length the 1.65 value is a measure of the conservatism associated computations have failure ratios of 1:30 for both Eq. (1) with the 2.5 limit (additional safety factor). and (3) and through the application of resistance factors The same evaluation was applied to the bentonite and (Table 3) an acceptable level of reliability can be obtained polymer data sets where the (cb + Ktr)/db term was capped at (Fig. 8). However, all computations discussed to this point 2.5 and the load multiplier required to produce a reliability index of 3.5 was determined (Fig. 10). Where water pullout have put no limitations on the (cb + Ktr)/db expression, which is limited by ACI to values of 2.5 or less “to prevent pullout data required a load magnifier (1.65) to produce the target failures.”2 The effect of this limit is illustrated using the failure ratio, the bentonite and polymer data showed unac- pullout data from the water-cast specimens which in Fig. 7 ceptable failure ratios of 1:3704 and 1:2040. Applying the and Table 3 were shown to have similar statistical values same multiplier concept to the simulations, unconservative to that of the ACI reported values for dry conditions. For multiplier values of 0.99 and 0.93 would be required to bring all ACI 318-predicted capacity values for the water data, the failure ratios down to the target reliability for bentonite and polymer, respectively. Therefore, some consideration the (cb + Ktr)/db term was capped at 2.5 and the bias and COV values were recomputed to be 3.25 and 0.25, respec- for slurry casting conditions is needed and the multiplier

ACI Structural Journal/July 2019 189 Table 3—Resistance factors for all slurry conditions

Slurry type Water Polymer Bentonite Dry ACI 318 0.62 0.35 0.37 0.65 ACI 408R-03 0.61 0.33 0.47 0.74

Table 4—Development length factors (Ψslurry) for ACI 408R-03 values when in slurry environments

fc′, psi Water Bentonite Polymer

3000 1.3 1.8 2.6 4000 1.3 1.8 2.6 Fig. 10—Effect of (cb + Ktr)/db cap on bentonite (left) and polymer (right); before (red) and after load multipliers 5000 1.3 1.8 2.7 Case 1 (black); 1:3704 and 1:2040 failure ratios before.(Please 6000 1.3 1.8 2.7 refer to PDF version of this paper at www.concrete.org for 8000 1.3 1.8 2.8 full-color version. 10,000 1.3 1.9 2.8 β = 3.5 in the Monte-Carlo simulations, the multipliers were 3000 1.3 1.8 2.6 similar, where bentonite and polymer conditions require 4000 1.3 1.8 2.6 development lengths 1.7 and 1.8 times longer than water 5000 1.3 1.8 2.7 development lengths, again respectively, to provide the Case 2 same level of reliability. Water conditions were shown to be 6000 1.3 1.8 2.7 conservatively satisfactory when the (cb + Ktr)/db limit of 2.5 8000 1.3 1.8 2.8 was imposed. 31 3 10,000 1.3 1.9 2.8 Both ACI and AASHTO provide tables that further simplify the development equations dependent on concrete values are the same as resistance factors. To produce the strength, bar size, bar spacing or cover dimensions, and same level of conservatism presently in place for dry cast transverse reinforcement. Two categories are established specimens (based on the water cast specimen data), develop- that set the value of the term (cb + Ktr)/db in Eq. (1) to 1.5 ment length factors (Ψslurry) from Monte-Carlo simulations or 1. Case 1 refers to preferred main bar clear spacing, clear were determined by the ratio of load multipliers to be 1.7 cover, and stirrup spacing criteria (cb + Ktr)/db = 1.5; Case 2 31 (1.65/0.99) and 1.8 (1.65/0.93) for bentonite and polymer refers to all other conditions (cb + Ktr/db) = 1. This further casting environments, respectively. reduces the probability of failure, but using the same logic previously presented, slurry-cast reinforcing bar would DISCUSSION inherently be less reliable while still perhaps acceptable. Bar cleanliness and epoxy coatings are recognized to affect To date, the authors are unaware of any reinforcing bar reinforcing bar pullout capacity. Unclean conditions are to bond/development length-related failures in drilled shafts, be avoided and epoxy coatings are given a development which could make the argument there is no need to consider length factor (Ψe) of 1.2 or 1.5 depending on cage spacing the findings of this study. This could be due to the conser- and cover dimensions. The combination of slurry coating vatism built-in to development length equation via limits, and radial concrete flow around the reinforcing bar cage into or reinforcing bar splices in drilled shafts are not likely to the cover region (which is common to tremie placement) be in high moment regions (for example, directly under presents conditions that also affect reinforcing bar bond. the footing in overwater bridges). In such cases, develop- Accommodations for the associated reductions in resistance ment can be adequately provided via a hook in the dry cast and variability can be provided through resistance factors footing above or sufficiently long lengths within the shaft or ld equation cutoff limits for the (cb + Ktr)/db term. Using length below. Further, staggered splice locations are often either approach, the same level of reliability can be provided employed to reduce cage congestion and promote concrete to slurry casting conditions by increasing the development flow into the cover region; per AASHTO, “no more than length using ratio of load multipliers used for Fig. 9 and 10 50 percent of the reinforcement shall be terminated at any or the ratio of resistance factors from Table 3. The ratio of section.”3 Nevertheless, splices are designed to develop the resistance factors stems from using the Table 3 resistance bar without regard for the actual local moment demands factors in Eq. (1) where an alternate method of determining and if the same level of reliability is sought for slurry-cast a development length factor can be established for slurry, reinforcing bar, development lengths should be increased Ψslurry = ld slurry/ld dry, which simplifies to Ødry/Øslurry. For water, accordingly. No adjustment is necessary for water casting bentonite and polymer casting conditions, the resistance conditions. factor ratios from Table 3 values give development length The unexpectedly higher values for polymer (Ψslurry) were factors of 1.05, 1.76, and 1.86, respectively. Using the ratio attributed to the high COV for polymer data which may have of the simplistic load multipliers in Fig. 9 and 10 to satisfy stemmed from testing various manufacturer products. This suggests that further delineation of polymer slurry prod-

190 ACI Structural Journal/July 2019 Table 5—AASHTO parameters and slurry factors published variations in load showed possible failure ratios of 1:2 and 1:3.5 for bentonite and polymer casting environ- Load factor, DL 1.25 ments, respectively. Specimens cast in a water environment Load factor, LL 1.75 showed similar failure ratios as dry conditions which were DL/LL ratio 2 on the order of 10 to 15 times better than slurry at 1:22 and Reliability index 3.5 1:30, respectively. Limiting the value of the (c + K )/d term to 2.5 in the Load bias, DL 1.05 b tr b development length equation (Eq. 1) increases reliability of Load bias, LL 1.15 reinforcing bar bond in slurry casting environments where Load COV, DL 0.1 the failure ratios fall to 1:3704 for bentonite and 1:2040 Load COV, LL 0.2 for polymer slurry conditions. Both fail to meet the target reliability index of 3.5 without a development length factor Ødry 0.65 for slurry (Ψslurry). For each casting environment, resistance Øwater 0.61 factors were computed using a reliability index of 3.5 and Øbentonite 0.37 Monte-Carlo simulations showed reinforcing bar pullout

Øpolymer 0.35 failure ratios to be acceptably low. These resistance factors when applied to ACI 318 or AASHTO development equa- Ψ 1.1 water tions suggest that present development length values should Ψbentonite 1.8 be increased by 1.9 for polymer or 1.8 for bentonite slurry Ψpolymer 1.9 environments (essentially doubled) to achieve the same reliability as dry conditions. When considering the coating ucts by exact chemical composition may be warranted, but factor for epoxy is as high as 1.5, these values are reason- at present, field construction practices make no distinction. able. For water casting environments, only subtle varia- Bentonite falls under strict API guidelines for uniformity.5 tions were noted from dry conditions, which were deemed When reviewing possible future use of the ACI 408R-03 inconsequential. development length computations, Eq. (3), the same development length ratio/slurry factor determination should AUTHOR BIOS be considered, but the simple resistance factor ratio is not ACI member Kelly Costello is a Postdoctoral Researcher in the Depart- ment of Civil and Environmental Engineering, University of South Florida, valid due to the formulation of that equation. The computed Tampa, FL, where she received her BS, MS, and PhD. Her research inter- development length multiplier (ld slurry/ld dry) for water condi- ests include use of self-consolidating concrete in drilled shafts, underwater tions showed a uniform multiplier (Ψslurry) value of 1.3, evaluation of drilled shaft concrete integrity, and the effective use of cement applicable for all concrete strengths. Bentonite and polymer and slag-cement binders in organic soil mixing. slurry conditions showed a subtle change through the range ACI member Sarah Mobley is a Doctoral Candidate in the Department of Civil of concrete strengths where the multiplier ranged from 1.8 to and Environmental Engineering, University of South Florida. She received her BS from the University of Alaska, Anchorage, AK, and her MS from the 1.9 and 2.6 to 2.8, respectively (Table 4). University of South Florida in civil engineering. Her research interests include Finally, when applying the same procedure for resistance assessing the corrosion potential of reinforcing steel and concrete strength vari- factor determination to AASHTO load factors, mean bias of ations that stem from tremie placement of drilled shafts. predicted loads, the corresponding coefficients of variation ACI member Justin Bowen is a Project Manager with Ceco Concrete 30 for AASHTO type loads, and the same dead load to live Construction, LLC, Tampa, FL, specializing in mid- and high-rise construc- load ratio (2.0) used in this study (Table 5), the computed tion. He received his BS and MS in civil engineering from the University of resistance factors using Eq. (5) were strikingly similar to South Florida’s Department of Civil and Environmental Engineering. those determined for the ACI 318 in Table 3. As a result, ACI member Gray Mullins is a Professor in the Department of Civil and the same development length multipliers computed for ACI Environmental Engineering at the University of South Florida, Tampa, FL, where he received his BS, MS, and PhD. His research interests include 318 prediction methods should be considered for AASHTO foundation design, construction, and quality assurance. applications as well. Details of this analysis can be found elsewhere.29 ACKNOWLEDGMENTS The authors would like to thank D. Darwin, University of Kansas, Lawrence, KS, for his insights during the analysis of the study data. The CONCLUSIONS authors also acknowledge the Florida Department of Transportation for Present ACI, FHWA, and AASHTO specifications make funding multiple studies that formed the basis for this work. However, the no adjustments to reinforcing bar development or lap splice opinions, findings, and conclusions expressed in this publication are those of the authors and not necessarily those of the Florida Department of Trans- lengths for tremie-placed, slurry-displaced concreting portation or the U.S. Department of Transportation. conditions. However, considerations for bar cleanliness (oil or mud) and epoxy coatings are well documented. This study NOTATION presented the reinforcing bar pullout test results from 227 Ab = area of bar being developed or spliced specimens tremie-cast in commonly used mineral, polymer, Atr = area of transverse reinforcement crossing potential plane of or natural (water) drilling slurry. While the average pullout splitting adjacent to reinforcement being developed cb = least of side cover, concrete cover to bar or wire, or one-half resistances were in line with present prediction methods, center-to-center spacing of bars or wires the statistical variations in resistance when coupled with cbot = bottom concrete cover for reinforcing bar being developed cM = maximum of cbot or cs, (cM/cm ≤ 3.5), in.

ACI Structural Journal/July 2019 191 cm = minimum cover used in expressions for bond strength of bars 10. Mullins, G.; Sosa, R.; and Sen, R., “Seal Slab Prestressed Pile Inter- not confined by transverse reinforcement, minimum value ofc bot face Bond from Full-Scale Testing,” ACI Structural Journal, V. 98, No. 5, or cs, (cM/cm ≤ 3.5), in. Sept.-Oct. 2001, pp. 743-751. cs = minimum of (cso, csi + 0.25) or min (cso, csi), in. 11. Mullins, G.; Sosa, R.; Sen, R.; and Issa, M., “Seal Slab/Steel Pile csi = half bar clear spacing between bars, in. Interface Bond from Full-Scale Testing,” ACI Structural Journal, V. 99, cso = side concrete cover for reinforcing bars, in. No. 6, Nov.-Dec. 2002, pp. 757-763. db = nominal diameter of bar, in. 12. Butler, H. D., “A Study of Drilled Shafts Constructed by the Slurry fc' = concrete compressive strength (psi unless noted otherwise) Displacement Method,” Bridge Division, Texas Highway Department, fs = steel stress at failure, psi Austin, TX, Feb. 1973, 81 pp. fy = yield strength of bars being spliced or developed, psi 13. Mullins, A. G.; Winters, D.; Bowen, J.; Johnson, K.; DePianta, Ktr = factor representing contribution of confining reinforcement V.; Vomacka, J.; and Mullins, M., “Defining the Upper Viscosity Limit for across potential splitting planes; 40Atr/sn Mineral Slurries Used in Drilled Shaft Construction,” Final Report, FDOT ld = development length or splice length, bonded length, in. BDK84 977-24, Florida Department of Transportation, Tallahassee, FL, ld, dry = development length or splice length for casting dry conditions, Dec. 2013, 264 pp. in. 14. Deese, G., and Mullins, G., “ Factors Affecting Concrete Flow in ld, slurry = development length or splice length for slurry casting condi- Drilled Shaft Construction,” ADSC GEO3, GEO Construction Quality tions, in. Assurance/Quality Control Conference Proceedings, Bruce, D. A., and N = number of transverse stirrups or ties within development length Cadden, A. W. , eds, Nov. 2005, pp. 144-155. n = number of bars or wires being developed, or lap spliced along 15. Majano, R. E., “Effect of Mineral and Polymer Slurries on Perimeter plane of splitting Load Transfer in Drilled Shafts,” PhD dissertation, University of Houston, q¯ = mean value of random loading variable Houston, TX, 1992, 388 pp. Rr = relative rib area of reinforcement 16. Brown, D., “Effect of Construction on Axial Capacity of Drilled r¯ = mean test-prediction ratio Foundations in Piedmont Soils,” Journal of Geotechnical and Geoenvi- s = spacing of transverse reinforcement ronmental Engineering, ASCE, V. 128, No. 12, 2002, pp. 967-973. doi: Tb = total bond force of developed or spliced bar = Tc + Ts 10.1061/(ASCE)1090-0241(2002)128:12(967) Tc = concrete contribution to total bond force, bond force that would 17. Lam, C.; Jefferis, S.; and Martin, C. M., “Effects of Polymer and be developed without transverse reinforcement Bentonite Support Fluids on Concrete–Sand Interface Shear Strength,” Ts = steel contribution to total bond force, additional bond strength Geotechnique, V. 64, No. 1, 2014, pp. 28-39. doi: 10.1680/geot.13.P.012 provided by transverse steel 18. Lam, C., and Jefferis, S., “Performance of Bored Piles Constructed td = 0.78db + 0.22, term representing effect of bar size on Ts (Zuo and Using Polymer Fluids: Lessons from European Experience,” Journal of Darwin32) Performance of Constructed Facilities, ASCE, 2015, 9 pp. doi:10.1061/ tr = term representing effect of relative rib area on Ts (ASCE)CF.1943-5509.0000756, 04015024. 33 32 = 9.6Rr + 0.28 (Darwin et al. ; Zuo and Darwin ) 19. Caliari de Lima, L., “Construction Effects on the Side Shear of Vr = coefficient of variation of resistance random variabler (COVr) Drilled Shafts,” PhD dissertation, University of South Florida, Tampa, FL, Vφq = coefficient of variation of loading random variableq (COVq) Nov. 2017, 139 pp. β = reliability index 20. Bowen, J., “The Effects of Drilling Slurry on Reinforcement in λ = lightweight concrete factor; normalweight concrete is 1 Drilled Shaft Construction,” master’s thesis, University of South Florida, Ø = strength reduction factor for main loading Tampa, FL, Nov. 2013, 146 pp. Øb = overall strength reduction factor against bond failure 21. Beckhaus, K., “EFFC/DFI Best Practice Guide to Tremie Concrete for Ød = Øb/Ø, effective strength reduction factor for use in calculating Deep Foundations,” first edition, Deep Foundations Institute, Hawthorne, development/splice length NJ, 2016, 52 pp. Ødry = strength reduction factor for dry conditions 22. FDOT, “Standard Specifications for Road and Bridge Construction, Øslurry = strength reduction factor for slurry conditions Section 455 Structures Foundations,” Florida Department of Transporta- Ψe, β1 = coating on reinforcement factor; uncoated or zinc galvanized is 1 tion, Tallahassee, FL, 2018, pp. 545-609. Ψs = size of reinforcement factor; No. 7 or larger is 1 23. Orangun, C. O.; Jirsa, J. O.; and Breen, J. E., “Reevaluation of Test Ψslurry = development length factor for slurry casting conditions Data on Development Length and Splices,” ACI Journal Proceedings, Ψt, α = casting position factor; more than 12 in. of fresh concrete placed V. 74, No. 3, Mar. 1977, pp. 114-122. below horizontal reinforcement is 1.3, other is 1 24. Darwin, D.; McCabe, S. L.; Idun, E. K.; and Schoenekase, S. P., 28 ω = (cmax/cmin) + 0.9 ≤ 1.25 “Development Length Criteria: Bars Not Confined by Transverse Reinforce- ment,” ACI Structural Journal, V. 89, No. 6, Nov.-Dec. 1992, pp. 709-720. 25. Darwin, D.; Idun, E. K.; Zuo, J.; and Tholen, M. L., “Reliability-Based REFERENCES Strength Reduction Factor for Bond,” ACI Structural Journal, V. 95, No. 4, 1. Mobley, S.; Costello, K.; and Mullins, G., “The Effect of Slurry July-Aug. 1998, pp. 434-443. Type on Drilled Shaft Cover Quality,” DFI Journal, The Journal of 26. Esfahani, M. R., and Vijaya Rangan, B. V., “Local Bond Strength of the Deep Foundations Institute, London, UK, 2018, pp. 91-100. doi: Reinforcing Bars in Normal Strength and High-Strength Concrete (HSC),” 10.1080/19375247.2018.1468522 ACI Structural Journal, V. 95, No. 2, Mar.-Apr. 1998, pp. 96-106. 2. ACI Committee 318, “Building Code Requirements for Structural 27. Esfahani, M. R., and Vijaya Rangan, B. V., “Bond between Normal Concrete (ACI 318-14) and Commentary (ACI 318R-14),” American Strength and High-Strength Concrete (HSC) and Reinforcing Bars in Concrete Institute, Farmington Hills, MI, 2014, 519 pp. Splices in Beams,” ACI Structural Journal, V. 95, No. 3, May-June 1998, 3. AASHTO, “AASHTO LRFD Bridge Design Specifications,” eighth pp. 272-280. edition, American Association of State Highway and Transportation Offi- 28. ACI Committee 408, “Bond and Development of Straight Rein- cials, Washington, DC, 2017, pp. 5-172 to 5-178. forcing in Tension (ACI 408R-03),” American Concrete Institute, Farm- 4. FDOT, “Standard Specifications for Road and Bridge Construction,” ington Hills, MI, 2003, 49 pp. Florida Department of Transportation, Tallahassee, FL, 2018, 1213 pp. 29. Costello, K., “A Theoretical and Practical Analysis of the Effect of 5. API, Recommended Practice for Field Testing Water-Based Drilling Drilling Fluid on Rebar Bond Strength,” dissertation, University of South Fluids, Washington, DC, API Publishing Services, 2009, 121 pp. Florida, Tampa, FL, Dec. 2018, 114 pp. 6. Brown, D.; Turner, J.; and Castelli, R., “Drilled Shafts: Construc- 30. Nowak, A., and Collins, K., Reliability of Structures, McGraw Hill, tion Procedures and LRFD Design Methods,” NHI Course No. 132014, New York, NY, 2000, 338 pp. FHWA-NHI-10-016, FHWA GEC 010, U.S. Department of Transportation, 31. Kamara, M.; Novak, L.; and Rabbat, B., “Notes on ACI 318-08 Federal Highway Administration, Washington, DC, 2010, 7 pp. Building Code Requirements for Structural Concrete with Design Appli- 7. Mullins, G.; Caliari, L.; and Costello, K., “Effect of Polymer Slurry cations,” Portland Cement Association, Skokie, IL, 2008, pp. 4-1 to 4-5. Stabilization on Drilled Shaft Side Shear over Time,” Final Report, FDOT 32. Zuo, J., and Darwin, D., “Splice Strength of Conventional and High Contract No. BDV24 TWO977-19, Florida Department of Transportation, Relative Rib Area Bars in Normal and High-Strength Concrete,” ACI Struc- Tallahassee, FL, April 2018, 187 pp. tural Journal, V. 97, No. 4, July-Aug. 2000, pp. 630-641 8. Sosa, R., “Bond Capacity of Pile/Seal Slab Interfaces,” master’s 33. Darwin, D.; Zuo, J.; Tholen, M. L.; and Idun, E. K., “Development thesis, University of South Florida, Tampa, FL, Nov. 1999, 189 pp. Length Criteria for Conventional and High Relative Rib Area Reinforcing 9. Mullins, G.; Sosa, R.; and Sen, R., “Seal Slab/Pile Interface Bond,” Bars,” ACI Structural Journal, V. 93, No. 3, May-June 1996, pp. 347-359. Final Report, Florida Department of Transportation, Nov. 1999, 151 pp.

192 ACI Structural Journal/July 2019