2700 REASONS TO CONFIRM TESTS OF FRESH OR HARDENED CONCRETE

K.C. HOVER

School of Civil & Environmental Engineering, Cornell University, Ithaca, NY, USA.

SUMMARY Conventional wisdom has it that a truckload of concrete can be characterized by single, well-defined values of slump, air content, temperature, and compressive strength at a given age. What may be closer to reality is that a load of concrete has as many estimates of its fresh and hardened properties as there are tests performed. The question is not so much “What is the slump of this load,” but “What is the range of slump?” Results are presented from a US Federal Highway Administration testing program conducted annually for 12 years in which 6 independent teams sampled and tested concrete from one single truckload of concrete, evaluating fresh and hardened concrete, including NDT tests for estimating in-place strength. At least 225 tests were conducted on each truck for 12 years for over 2700 individual tests: Results warn that decisions about the acceptability of concrete, or the lack thereof, should not be made on the basis of a single, unconfirmed, test.

INTRODUCTION From 1990 to 2008 the Portland Cement Concrete (PCC) module of US Federal Highway Administration’s annual “NHI Highway Materials Short Course” included a field- and lab- testing phase to evaluate fresh and hardened properties for a single, 3.1 m3 load of ready-mixed concrete. Four to six testing teams independently sampled concrete from the truck and repeatedly tested the concrete for temperature, slump, and air content by pressure and by volume meter. Each team also cast a simulated pavement section, 1.2m x 1.2m x 200mm thick to demonstrate proper placing, consolidating, and finishing techniques, using an internal vibrator, screed, float, and broom finish. Finally, each team cast six 150mm x 300mm and 6 more 100mm x 200mm test cylinders in plastic molds, plus 3 flexural test specimens. Hardened concrete tests were conducted at an age of 3 days, at the same time that 3 100mm dia. cores were drilled from each slab and tested, along with performing NDT tests on the slab including Rebound Hammer (“Schmidt” or “Swiss” hammer), Penetration Resistance (“Windsor Probe”), Pullout (LOK test), and Ultrasonic Pulse Velocity. Each year specimens were cured under both field- and standard-laboratory conditions, and slabs were covered with wet burlap (“Hessian” or “Jute”) and plastic sheets, plastic sheets-only, and exposed to air. Instructional objectives included demonstration of the test procedures and showing by example the inherent variability in test results when the concrete is from a single load. References (1-3) provide a detailed description. On a typical afternoon 30 engineers performed at least 24 slump tests, 48 air tests, 24 temperature tests and cast 6 slabs, followed by loading 72 cylinders, 18 cores, and a host of NDT bangings, bouncings, soundings, shootings, and extractions. More than 2700 tests were conducted over 12 years. The entire program was logistically supported by host institutions and the crew and equipment of the FHWA Mobile Concrete Laboratory. Over 400 participants completed the module, which has been adapted for training elsewhere.

For all tests reported here the work was conducted at the University of Nevada at Reno. All concrete was provided by the same supplier, supervisory staff and test equipment remained the same. All test equipment was calibrated or checked for accuracy one-day prior to testing. Certified test technicians and examiners were present with each team to provide immediate hands-on instruction and correction. The concrete mixture and cement and aggregate sources varied somewhat (Table 1), and although average daily temperature was similar (Table 2), concrete temperature at delivery, and air temperature at delivery and over the first night varied as well. These factors, along with inherent variability in the sampling and testing methods themselves, combined to produce a broad range of reported values of fresh and hardened concrete properties across the years. Results varied widely within any given year’s experiment and within any given curing condition. The testing program revealed that when an atypically large number of tests were conducted on a single load of concrete, the results indicated far more within-batch variability than often assumed to be the case.

TABLE 1 Concrete Mixtures

Portland Fine Coarse Crs Agg

Cement Fly Ash Agg Agg Nom.Size

Year kg/m3 kg/m3 kg/m3 kg/m3 mm w/c WRA AEA

1993 363 0 698 808 19 0.49 WRDA64 Micro-Air

1994 350 0 713 1009 19 0.45 WRDA64 CorAir

1995 356 0 771 998 19 0.45 MB300N MBVR

1996 356 0 785 1014 19 0.45 MB300N MBAE90

1997 363 0 754 1017 19 0.45 MB300N MBAE90

1998 356 0 833 990 19 0.45 MB300N MBAE90

2001 349 55 814 968 19 0.45 MB300N MBAE90

2002 Data not available

2003 363 0 831 974 25 0.45 Poly-1 DAREXII

2005 363 0 831 974 26 0.45 Poly-1 DAREXII

2006 363 0 831 974 27 0.45 Poly-1 DAREXII

2007 363 0 831 974 28 0.45 Poly-1 DAREXII

2008 363 0 831 974 29 0.45 Poly-1 DAREXII

Figure 1. Air temperature data from beginning of casting-day to ending of testing-day for each year of test.

EVALUATION OF FRESH CONCRETE Slightly more than 3 cubic meters of Concrete were delivered to the site in a ready-mixed concrete truck. It was sometimes necessary (particularly in warm weather) to bring the initial on-site slump to 100 to 125mm with water addition and thorough mixing prior to commencing testing. Teams repeatedly sampled concrete from the truck chute at approximate 20-30 minute intervals, and performed concrete temperature, slump, and air content testing with both pressure and volume meters. The truck drum was continuously revolving at slow agitation speed of less than 4 rpm for the duration of the exercise (in contrast to typical mixing speed of 12-15 rpm). Samples of fresh concrete properties with time are shown in Figures 2a,b,c. Overall variability for all years of the program is shown in Figures 3-5.

Figure 2a-c. Sample values from 1994: Fresh concrete properties vary with time and among

tests of the same batch at the same time.

Key observations from Figure 2 include a range of difference in fresh concrete temperature of more than 5C at any given time after batching, the undeniable time-dependency of slump, and a broad scatter in air content as influenced by time, testing team, and type of air meter. The time-dependency of slump is readily modeled as a linear function (85 mm slump loss per hour, for this mixture @ 20-25C in 1994). Interestingly, the same decay function used for radioactive half-life can be applied with equal accuracy to compute the “Half-Slump-Life.” (Hover 2013, ACI Transitions paper). For this concrete on this site the slump is cut in half every 40 minutes. Under specifications that require testing and discharge of concrete up to 90 minutes past batching, one could legally report this load of concrete as having a temperature of 17-23C, a slump of 60 to 170 mm, and an air content of 3 to 11%.

Figures 3a-d. Overall variability of test results by year.

One- and two-standard-deviation ranges of expected results are shown in Figs. 3a-d. For normally-distributed data, 95% and 99% of all test results would be expected to fall within the 1S and 2S bands, respectively. Figure 4 shows the values of COV, suggesting that for a 99% (2S) confidence level, the estimated range of concrete temperature would be ± 15% of the average in °C. Estimated slump would be ± 40% of average. Estimated range for pressure air meter would be ± 35%, and estimated range for volume air meter would be ± 40% of average value.

Figure 4. Coefficients of variation for each test by year.

EVALUATION OF HARDENED CONCRETE Evaluation of hardened concrete is shown schematically in Figure 5, detailing the numbers and sizes of specimens, and the curing of specimens and surrogate paving slabs. Additionally, all cylinders and cores of all sizes were randomly selected for testing by one of the two testing laboratories on-site. Both facilities used neoprene pad-caps in accordance with ASTM 511 and operated equipment per ASTM C 39. The University lab in Reno used a SATEC, MKIII-C 500QC machine with a maximum load capacity of 2,220 kN. The FHWA mobile laboratory used a Forney FX-600 (2670 kN capacity) for all years except 2008, at which time they installed a FX-700 (3110 kN capacity-no observed impact of the change in testing machine for the final year). Specimens from the laboratory moist room were stored in

20 liter buckets of water and removed from the bucket immediately prior to testing. Load was applied at the ASTM C 39 rate of 0.25 ± 0.05 MPa/sec, and loaded to destruction.

Figure 5. Schematic diagram of compression testing program. Large and small circles indicate 150x300mm and 100x200mm cylinders. Circles inside slabs indicate 100mm (nominal) cores. Numbers on slabs designate the testing team.

Figure 6. Summary of test results for 12 years of the testing program. LC = standard laboratory curing, BP = wet burlap & plastic-covered field cure, PC= plastic covered field cure, AC= air cure, (no cover). Numbers 6 and 4 refer to cylinder diameter (inches), and C = drilled core.

Figure 6 shows a wide variation of 3-day compressive strength for the same basic mixture, but also shows a wide range of measurement of strength within any given batch, and for variations in curing and type of test specimen. in any given year the range of strength values obtained for a single batch of concrete varied by as little as 4 MPa (1994), or as much as 12 MPa (2006). These ranges, which amount to 1/3 to 1/2 of average laboratory-cured strength, are reduced by the fact that each data point represents the average of cores or cylinders of the same size representing the same curing conditions for the given experiment. The scatter would be greater if individual specimen test results were plotted.

Figure 7. Average relative compression strength for 100mmx200mm (“4-inch”) and 150x300mm (“6-inch”) cylinders and 100mm (“4-inch” nominal) cores, 1993-1998.

Figure 8. Adjusted average coefficient of variation for 100mmx200mm (“4-inch”) and 150x300mm (“6-inch”) cylinders and 100mm (“4-inch” nominal) cores, 1993-1998.

In the US, 100 mm dia. (“4-inch”) test cylinders have become common, but interest remains in comparing the results to those obtained with the “standard” 150 mm (“6-inch”) cylinders. Figure 7 uses laboratory-cured [23C, 95-100% humidity]. 150 mm cylinders as the basis of

comparison with 100 mm and 150 mm cylinders that have been field and lab-cured, along with 100 mm dia. drilled cores. Over the entire program and across multiple temperature regimes, highest 3-day compressive strength was obtained from cores drilled from slabs that had been cured with saturated Hessian and covered with plastic. Lowest 3-day strength was obtained from 150 mm cylinders, field cured with no protection. While Figure 7 suggests about a 5% increase in apparent strength via 100mm cylinders (see also French (1993) and Vandegrift & Schindler (2006), the variability of those results is also important, especially for statistical quality control. Figure 8 compares the coefficient of variation for all cylinder sizes and curing conditions, indicating that even when laboratory-cured, the 100mm cylinders are about 50% more variable than the “standard” 150 mm specimens, and that variability increases dramatically under for field-curing conditions. For all curing conditions at an age of 3 days, cores were more than twice as variable as 150 mm lab-cured cylinders. Values in Figure 8 have been adjusted to correspond to the ACI-318 Building Code definition of a “single test” as the average of 2 cylinder tests. Figures 9 and 10 augment these strength and variability analyses by comparing results from the two independent test facilities. In general the two labs show close agreement in average compressive strength, but sometimes large differences in variability, with no discernable pattern from year-to-year to reliably predict which of the two labs will report the highest or the most consistent results.

Figure 9a (top) and b(bottom) 3-day

compressive strength and coefficient of variation (COV) for 150 mm (“6-inch”)

cylinders.

Figure 10a (top) and b(bottom) 3-day compressive strength and coefficient of variation (COV) for 100 mm (“4-inch”)

cylinders.

NON-DESTRUCTIVE TESTING

Space does not permit a full discussion of the NDT phase of this study, but a schematic diagram of its scope is shown in Figure 11, and its overall outcome shown in Figure 12. In essence, at an age of 3 days field cured cylinders were the most accurate predictors of core strength, with estimates varying from ±10 to ±25% of average core strength provided by various NDT methods, as also influenced by curing conditions.

FIGURE 11. Schematic diagram of 1.2x1.2mx200mm thick slab used for NDT program. ① Three, 100mm cores extracted. ② Slab received a float finish with exception of one quadrant

that was broomed. ③ A minimum of 30, rebound-hammer impacts were recorded on a dry, broom-finished surface, and again after wetting the surface. ④ Thirty more rebound impacts

were recorded on dry, floated-finish; repeated wet. ⑤ Three penetration resistance probes were shot into concrete, using the metal template/baseplate for alignment and spacing. ⑥

Three Pull-out surface inserts, installed at casting, were extracted vertically. ⑦ Three Pull-out inserts, mounted on the edge form prior to casting, were extracted horizontally. ⑧ Pulse velocity was measured across the full width of the slab in 4 to 5 locations, selected to avoid core voids. ⑨ Pulse velocity was measured over length of cores prior to strength testing.

FIGURE 12. Summary of test results for 12 years of the testing program. LC = laboratory moist curing, FC = field cured in accordance with BP, PC, or AC. Numbers 4 and 6 refer to 100 and 150mm cylinder diameter, expressed in inches.

DISCUSSION AND CONCLUSIONS Among detailed applications identified in earlier reports (Hover, 2013,2014a& b), perhaps the most significant is recognition that when multiple tests are conducted on even a single load of concrete, the results will vary. This test program was not designed to identify whether that variation is due to inhomogeneity within a batch of concrete or inherent irreproducibility of the test methods themselves. Both sources are likely. Results of compression tests are the clearest indicators that when the same concrete is tested by two different, meticulous facilities, results can vary between the labs. Given the uncertainty in fresh and 3-day hardened concrete properties shown here, in which all participants were supervised closely and coached and corrected in test techniques on a one-on-one basis, one would expect even broader variability when testing is not performed in accordance with standard protocol. Perhaps most significant is the caveat that whenever critical decisions about load- carrying capacity or contractual acceptability of the concrete depend on test results, such results must be confirmed prior to making those decisions. For better or worse, one test result does not tell the whole story. ACKNOWLEDGEMENTS Work was conducted under the auspices of the National Highway Institute of the Federal Highway Administration, and host institutions included Purdue University, Arizona State University, University of Nevada Reno, and the Ministry of Communication, Riyadh, Saudi Arabia. Course co-developers were Robert Philleo and Ephraim Senbetta, who also actively participated in the lab work in the early, developmental years. Field work was only possible with the outstanding support of the FHWA Mobile Concrete Laboratory and expert crew. REFERENCES ACI 318 “Building Code for Structural Concrete,” American Concrete Institute, Farmington Hills, MI, 2011. French, C.W., Mokhtarzadeh, A., “High Strength Concrete: Effects of Materials, Curing and Test Procedures on Short-Term Compressive Strength,” PCI Journal, Vol. 38, No.3, May-June, 1993, pp 76-87. Hover, K.C., Highway Materials Engineering, Module VI: Portland Cement Concrete Lab Manual, NHI Course No. 131023, Publication No. NHI-04-127, U.S. Department of Transportation, Federal Highway Administration, Washington, October, 2003. Hover, K.C., “Observed Variability in Tests of Fresh Concrete Properties from the FHWA Highway Materials Engineering Course,” Transportation Research Record: Journal of Transportation Research Board, Issue 2342, September 2013, p 61–75. Hover, K.C., “Observed Variability in Cylinder and Core Strength from the FHWA Highway Materials Engineering Course,” Accepted for publication in the Transportation Research Record: Journal of Transportation Research Board, February 6, 2014. Hover, K.C., “Comparison of Non-Destructive Test Results with Core Strengths Observed in the FHWA Highway Materials Engineering,” Submitted for publication in the Transportation Research Record: Journal of Transportation Research Board, August 1, 2014. Vandegrift, D., Schindler, A., “The Effect of Test Cylinder Size on the Compressive Strength of Sulfur Capped Concrete Specimens,” IR-06-01, Highway Research Center and Department of Civil Engineering, Auburn University, Auburn, Alabama, August, 2006, 75 pp.