Long-Term Performance of Epoxy Adhesive Anchor Systems (2013)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

45 This chapter summarizes the laboratory testing program used to investigate the effect of various parameters on the sustained load performance of adhesive anchors in concrete, the potential for using adhesive-alone testing to evaluate the sustained load performance of adhesive anchors, and the effect of early-age concrete on the short-term bond strength of adhesive anchors. The parameters considered, as well as the triage approach used to prioritize the parameters, are pre- sented. Finally, a detailed discussion on the testing program is presented in four sections: • Anchor pullout testing at the University of Florida, • Anchor pullout testing at the University of Stuttgart, • Adhesive-alone testing at the University of Florida, and • Early-age concrete evaluation at the University of Stuttgart. Research Plan The experimental program was implemented as follows. Baseline Tests Short-term (2-minute) pullout tests were performed on three adhesive anchor systems (A, B, and C) installed in con- crete to establish their baseline short-term strength. Addi- tional long-term tests were performed until failure on each adhesive anchor system at various percentages of the mean short-term strength. The resulting data were used to generate a stress versus time to failure for each adhesive anchor system under the baseline “control” conditions. Parameter Tests Short-term tests were conducted on each anchor system under a variety of installation and in-service conditions. An alpha-reduction factor for the short-term strength was deter- mined, which represents the effect that the parameter had on the bond strength at 2 minutes (duration of the short-term tests). Subsequent long-term tests were conducted on the adhesive anchor under the same variety of parameters and the result- ing stress versus time-to-failure relationship was evaluated. Influence Ratio If a given parameter has the same effect on the bond strength in the long term as it does in the short term, then the alpha- reduction factor at a given time to failure should be the same as the alpha-reduction factor evaluated at 2 minutes. Figure 34 shows the basic concept behind the use of the “stress versus time-to-failure” test method to evaluate the effect of a particular parameter on the sustained load per- formance of an adhesive anchor. The baseline “stress versus time-to-failure” relationship is shown as the solid line in Fig- ure 34. (Note that sample data points are not included in Fig- ure 34 for clarity.) For analysis purposes, an aST-baseline curve can be created on the stress versus time-to-failure plot (Figure 34) in which the alpha-reduction factor at any given time to failure is iden- tical to the alpha-reduction factor evaluated at 2 minutes. The stress versus time-to-failure relationship for a given parameter established from experimental data can be used to determine a long-term alpha-reduction factor (aLT) which is the stress to cause failure at a given time to failure divided by the baseline stress level at that particular time to failure. This aLT can there- fore be compared to the aST-baseline curve. If at any given point in time, the stress level to cause failure predicted by the aST-baseline is greater than the determination from the trend of experimental data, then the parameter has an adverse effect on the sustained load performance of an adhesive anchor. This can also be visualized by normalizing the short-term alpha-reduction factor (aST) by the alpha-reduction fac- tor determined from long-term testing (aLT) in what can be referred to as the influence ratio, as illustrated in Figure 35. C H A P T E R 2 Research Approach

46 = α α Influence Ratio ST LT If the influence ratio is greater than one, then the param- eter has an influence on sustained load. Conversely, influence ratios less than one indicate that the given parameter does not have an influence on the sustained load performance. Dogbone Tests A similar testing program (baseline tests, parameter tests, influence ratios) was conducted on dogbone specimens of the adhesive alone to determine if stress versus time-to-failure relationships determined from adhesive-alone dogbone ten- sile tests could be indicators of the long-term performance of adhesive anchors in concrete. Adhesive-Alone Tests Additional adhesive-alone (DMTA and creep) tests were con- ducted on a DSR machine to generate strain versus time (and compliance versus time) curves. These curves were compared to similar curves created from the dogbone tests and anchor pull- out tests to investigate if these simple and short-duration tests could be used to predict the long-term performance of adhesive anchor systems in concrete. Early-Age Investigation Finally, short-term tests were conducted on each of the three adhesive anchor systems at various days beyond concrete cast- ing to determine the effect of concrete age on adhesive anchor short-term bond strength. Parameters Identified for Testing The previous chapter identified many parameters that have the possibility of affecting the performance of adhesive anchor systems. Because of the project budget and timeline, not all parameters could be tested; therefore, a triage was conducted based on literature and the experience of the research team to Figure 34. “Stress versus time-to-failure” comparison of experimental, baseline, and aST-baseline trends. Figure 35. Influence ratio of a parameter versus time.

47 determine which parameters were believed to have the poten- tial for the most significant impact on sustained load per- formance and to develop a test program to investigate those parameters. This triage approach established three categories: • High priority parameters. Parameters thought to have the potential for a significant impact on sustained load perfor- mance and definitely should be tested. • Medium priority parameters. Parameters thought to have some potential for impact on sustained load performance and should be tested if budget and time permit. • Low priority parameters. Parameters thought to have a minimal potential for impact on sustained load performance and are not recommended to be tested under this project. Table 12 lists the parameters identified earlier with their rated priority and test series identification as listed in the test matrices shown in Table 14 and Table 15. As noted in Table 12, all high- and medium-priority parameters were included in the planned test program. Explanation of Triage Results The following describes the rationale for the prioritiza- tion of the parameters listed in Table 12 and how the influ- ence of each parameter would be evaluated if chosen for testing. High Priority Parameters These parameters were identified as having a strong pos- sibility for affecting the sustained load performance of adhe- sive anchors. Elevated In-Service Temperature. In-service temperature has a significant effect on the sustained load performance of adhesive anchor systems especially for adhesives with dif- ferent glass transition temperatures. Test series 3 and 4 were tested at above 120°F (49°C) and at 70°F (21°C) respectively in order to investigate the effect of in-service temperature. Sustained load tests were performed with the bonded anchor system that showed the lowest glass transition temperature. These test series were intended to investigate if the relation- ship between long-term bond strength and short-term bond strength was influenced by the in-service temperatures of 70°F (21°C), 110°F (43°C) (baseline), and >120°F (>49°C). Moisture-in-Service. Test series 8 consisted of sustained load tests on an adhesive installed dry but maintained wet during the sustained load test. It was thought that the mecha- nisms that could potentially reduce the sustained load capac- ity due to in-service moisture were (1) plasticization of the Parameter High Priority Medium Priority Low Priority Test Series* In-service factors Elevated temperature X 3,4 Reduced temperature X Moisture-in- service X 8 Freeze–thaw X Factors related to the adhesive Type of adhesive X 1,2,21 Mixing effort X Adhesive curing time when first loaded X 22 Bond line thickness X Fiber content of adhesive X Chemical resistance X Installation factors Hole orientation X 5,6 Hole drilling X 13 Hole cleaning X 9 Moisture in installation X 7 Installation Temperature X 10,11 Depth of hole (embedment depth) X Anchor diameter X Type of concrete X 12,14,15 Concrete strength X Type of coarse aggregate X Cracked or uncracked concrete X Confined or unconfined test setup X 16 Early-age concrete X 17 * See Table 14 for description of test series 1–17 and Table 15 for description of test series 21–22. Table 12. Prioritization of identified parameters.

48 adhesive, (2) reduction in the adhesive bond strength due to moisture, and (3) degradation of the adhesive due to a high alkaline environment found in moist concrete. It was thought that exposure to a high alkaline environment was the con- trolling mechanism, and therefore sustained load tests were conducted on the adhesive that showed the highest sensitiv- ity to alkalinity as determined by the resistance to alkalinity test results provided in the ICC-ES AC308 evaluation service report (ESR) for each adhesive. Type of Adhesive. It is well known that adhesives perfor- mance differs significantly for given parameters and duration of loading. Three adhesives from different manufacturers and of different adhesive types were tested in this project. It was recommended to include at least an epoxy and a vinyl ester. Baseline short-term and sustained load tests were con- ducted on all adhesives (series 1, 2, and 21). Due to project budget and timeline, sustained load tests were not conducted on all adhesives for every identified parameter, but rather the adhesive that was the most sensitive to the given parameter in short-term tests. The three adhesive anchor systems chosen all met the assess- ment criteria of ICC-ES AC308, indicating that they are viable for structural applications. In addition, the test results of the extensive ICC-ES AC308 testing program could provide useful information in this research project. Adhesive Curing Time When First Loaded. If a sustained load is applied before the adhesive is completely cured, the ini- tial displacement and strain rate might be higher than that of a specimen with a completely cured adhesive. In order to inves- tigate the sensitivity to cure time, tensile creep adhesive-alone tests (test series 22) were conducted on adhesive dogbone cou- pons at varying degrees of cure time. Baseline short-term ten- sile creep tests were conducted at the manufacturer’s specified cure time and at 7 days. Sustained load tensile creep tests were conducted on all adhesives at 7 days cure time and at the man- ufacturer’s specified cure time on the adhesive that showed the most sensitivity to cure time from the short-term tests. Adhesive anchor pullout tests were not conducted on speci- mens at varying cure times due to the logistical difficulties and the time duration required to condition a specimen of concrete from the installation temperature to the testing temperature. Hole Orientation. The presence of voids has a significant effect on the bond stress of an adhesive anchor. It is well known that voids in the adhesive will occur more often with anchors improperly installed horizontally or overhead. Anchors were installed horizontally and vertically in test series 5 and 6 respec- tively. Sustained load tests were performed with two bonded anchor systems. Hole Drilling. Hole drilling has been shown to influence bond strength due to the resulting roughness of the sides of the holes from different hole drilling methods. Most manu- facturers recommend rotary hammer drills with carbide bits. While in general, holes were drilled by rotary hammer drills with carbide drill bits, test series 13 used a diamond core drill. The sustained load tests were performed with the product approved for core drilling that was shown to be the most sen- sitive to the type of drilling with respect to bond strength established by short-term tests. Hole Cleaning. It is well known that the degree of hole clean- ing can significantly influence the short-term bond strength. Test series 9 used a reduced cleaning effort (50% of the manu- facturer’s recommended cleaning procedure as specified by ACI 355.4-11 test series 7.5) on anchors installed in dry holes. The sustained load tests were performed with the product that was shown to be the most sensitive to a reduced cleaning effort with respect to bond strength established by short-term tests. Moisture in Installation. Due to the significant decrease in short-term bond strength for anchors installed in damp and submerged holes, it was highly recommended that the influ- ence of moisture during installation be tested in this project. Test series 7 installed anchors in a wet/damp hole and con- ducted sustained load tests in a dry condition. The sustained load tests were performed with the product that was shown to be the most sensitive to a wet installation with respect to bond strength established by short-term tests. Type of Concrete. Anchor pullout tests at the University of Florida have shown that the short-term bond strength might be influenced by the composition of the concrete (e.g., amount of fly ash or blast furnace slag). The reasons for this might be due to the different porosity of the concrete com- pared to concrete without additives or perhaps due to the general surface condition of the drilled/cleaned hole. Test series 12 (standard DOT mix), 14 (20% fly ash), and 15 (50% blast furnace slag) were introduced in an attempt to address this question. The sustained load tests were performed with the product that was shown to be the most sensitive to the different concrete mixes. Medium Priority Parameters These parameters were identified as having a possibility for affecting the sustained load performance of adhesive anchors and/or they were recommended by the NCHRP panel for investigation during the proposal review. Installation Temperature. If anchors are installed at low temperatures, the final degree of curing is lower compared to installation at normal temperature. This might result in a reduction of the long-term bond strength. Therefore test series 10 and 11 were performed with anchors installed in concrete at the manufacturer’s lowest permissible installation temperature with the adhesive preheated to the manufactur- er’s lowest permissible adhesive temperature to ease adhesive injection. The adhesive with the lowest degree of cross-linking was chosen for testing. As any additional heating after instal-

49 lation causes additional curing of the adhesive, test series 10 was conducted at the manufacturer’s minimum temperature and test series 11 was conducted at 110°F (43°C). Unconfined Test Setup. Confined tests are used to ensure a bond failure. The bond failure may occur at the interface between the anchor and the adhesive and/or the adhesive and concrete and/or in the adhesive itself. In contrast, in uncon- fined tests, failure is often characterized with a concrete cone for shallow embedments and/or adhesives with high bond strengths. In order to ensure that both the short-term tests and sustained load tests in this program resulted in failures asso- ciated with bond strength, the confined testing method was used. Final design standards are based on unconfined bond strength established from short-term confined tests modified by a factor of 0.75 per ACI 355.4. In general, it is assumed that the ratio of long-term bond strength to short-term bond strength is independent of the type of support (confined, unconfined) provided bond fail- ure is the controlling factor of the unconfined condition and not concrete breakout. To check the validity of this assumption unconfined tests were performed (test series 16). Early-Age Concrete. It was suggested by the NCHRP proj- ect panel to investigate the effects of concrete age on the short- term bond strength. It is assumed that the synergistic effects of the low concrete strength and the high moisture content found in early-age concrete can affect the short-term bond strength of an anchor installed in early-age concrete. Test series 17 investi- gated the effects of concrete age by installing anchors in con- crete at various ages (3, 6, 13, 20, and 27 days) and conducting short-term anchor pullout tests after 24 hours of adhesive cure time. Sustained load performance due to installation in early-age concrete was not to be investigated in this project. Low Priority Parameters As the products used in this project had all met the assess- ment criteria of ICC-ES AC308, these parameters were iden- tified as possibly having a minimal effect (or none at all) on the sustained load performance of adhesive anchors used in this project. It was decided by the researchers and the NCHRP project panel that they not be tested during this project. Reduced In-Service Temperature. During approval tests of bonded anchors, according to ICC-ES AC308 or EOTA ETAG 001 Part 5, freeze/thaw tests are performed with anchors installed at normal ambient temperature in wet concrete. The anchors are loaded in tension with 55% the mean short-term pullout failure load. After 50 freeze/thaw cycles the residual bond strength is measured, which must be statistically equiva- lent with the short-term bond strength. It was recommended that the influence of long-term below-freezing temperatures on the long-term bond strength be considered low priority and not be investigated during the present research project. Freeze–Thaw. See discussion above regarding reduced in- service temperature. It was recommended that influence of freeze–thaw cycling be considered a low priority and not be investigated in this research project. Mixing Effort. The test program used bonded anchors with the adhesive delivered in cartridges. The anchors were installed according to the MPII. When using cartridges with their cor- responding mixing nozzle and the correct injection gun and following the manufacturer’s instructions (typically discard- ing the first inches of the mixed mortar) it may be assumed that the adhesive is thoroughly mixed. Incorrect mixing of the adhesive may only occur with these systems if the mix- ing nozzle is manipulated (e.g. shortened) or an inappropriate mixing gun is used. These are gross installation errors outside of the MPII and were not recommended to be evaluated in this research project. Incomplete mixing might occur with bonded anchors if the adhesive is delivered in bulk and mixed on site in an open con- tainer without controlled metering with a hand or machine mixer. These types of bonded injection anchors are not cur- rently addressed in this research project since they are out- side of the scope of ICC-ES AC308 (§1.2.4.2) and ACI 355.4 (§1.2.3). Bond Line Thickness. In general, all tests were performed with the gap thickness according to the MPII. A test series was proposed for consideration where the diameter of the hole was enlarged to check if the ratio of long-term bond strength to short-term bond strength was influenced by the hole diameter. The researchers and the NCHRP project panel chose not to test this parameter as it was deemed a gross installation error and to allow for testing of other higher priority parameters. Fiber Content of Adhesive. Since bulk mixing products were not considered by this project, the influence on sustained load performance of fiber content as modified by the installer was not addressed either. The influence of fiber content on sustained load performance could have coincidentally been examined if two of the three adhesives chosen were identical except for the amount of fiber content. However, this was not the case and the influence of fiber content was not a criterion when choosing the three adhesives to test in the project. Chemical Resistance. Chemical resistance is currently tested by ACI 355.4 §8.8 by two durability tests, a test for alkalinity and an optional sulfur dioxide test. As discussed earlier, both of these tests subject 13⁄16” slices to very harsh environments for long durations (2,000 hours). Test series 8, which tests for sensitivity to in-service moisture, will sub- ject the adhesive anchor to an alkaline environment since damp concrete is a naturally alkaline material. Since adhe- sive anchors installed in concrete are embedded much deeper than the 13⁄16” slices used in ACI 355.4 durability tests, the exposure to sulfur dioxide of a normal adhesive anchor will not be as extreme as the condition found in the tests.

50 Sustained load adhesive anchor chemical resistance tests were not conducted since the durability tests of ACI 355.4 are long-term tests and the reduction factor obtained from these tests, adur, was considered sufficient to account for chemical effects for both short-term and long-term loading conditions. Depth of Hole (Embedment Depth). Extensive short-term testing and analytical work has shown that the bond strength is not significantly influenced by the embedment depth in the ranges typically specified of about four to 20 anchor diam- eters. The authors feel that it can be safely assumed that the long-term bond strength is also not significantly influenced by the embedment depth. Therefore, all tests were performed with one embedment depth per anchor diameter. Anchor Diameter. For most bonded anchor systems the bond strength measured in short-term tests decreases some- what with increasing anchor diameter (Eligehausen et al., 2006b). The NCHRP project panel initially requested that anchor diameter be investigated, but later agreed to forgo this test parameter in order to evaluate a standard DOT concrete mix (test series 12). Concrete Strength. As discussed earlier, there is no direct correlation between concrete strength and bond strength. Since confined tests isolate the failure mode to the adhesive bond (eliminating the concrete cone failure mode) the effect of the concrete strength was not considered to be significant. As a result, influence of concrete strength on sustained load performance of adhesive anchors was not included in this test program. Type of Coarse Aggregate. This was not directly tested in this test program. Test series 12 (TS12) used granite aggregate but the concrete mix was different from the control in many ways. The effects of aggregates are accounted for in the ACI 355.4 test program via a series of round robin tests that evalu- ate the impact of regional differences on concrete mixtures. Cracked or Uncracked Concrete. Cracked concrete was not tested in this test program, but the ACI 355.4 test pro- gram contains test procedures for anchors to be qualified for use in both cracked and uncracked concrete. Testing Program Table 13 provides a summary of the testing program. Table 14 and Table 15 provide more detailed information on the anchor pullout testing program and the adhesive-alone testing pro- gram respectively. The equipment and tools used in the testing program are what was available at the laboratories and their use does not necessarily reflect an endorsement by the researchers. Anchor Pullout Testing Program Based on the triage approach discussed earlier, Table 14 presents the test matrix for anchor pullout testing program of threaded rods embedded in concrete for test series 1 through 17. Table 14 shows the test series, testing conditions with the tested parameter, explanations in notes at the bottom, number of tests per series, and location of testing. Tests were conducted at the University of Florida (UF) and the University of Stuttgart (US). Test series 1 through 16 began with short-term static load tests per the static tension test procedure per ASTM E-488. Five repetitions were conducted on each adhesive and their values averaged to determine the mean short-term load strength. Test series 1 through 16 concluded with a series of sustained load tests per the test procedure per AASHTO TP 84-10 with a few modifications. Three anchors were loaded until fail- ure at four stress levels for test series 1 and 2 and three stress levels for test series 3–16. The time to failure was evaluated at the time of rupture and as the time to tertiary creep per AASHTO TP 84-10. For laboratory logistics and in order to remove the effects of continued curing beyond the manufacturer’s stated cure time, all anchors were allowed to cure 7 days, and then were condi- tioned to the testing temperature for 24 hours prior to testing. Test series 17 only evaluated the effect of early-age concrete on the short-term bond strength. Its influence on the sus- tained load performance was not evaluated in this research project and therefore no sustained load testing was con- ducted for test series 17. Test Specification Test Series Description Data Short-term static load ASTM E488 1–16, 17 Anchor pullout from concrete Mean short-term load strength Sustained load AASHTO TP 84-10 (modified) 1–16 Anchor pullout from concrete at 3 or 4 stress levels Time to rupture (as measure of time to failure) Static load strength (dogbone specimen) ASTM D638 (modified) 21–22 Adhesive only Mean short-term load strength Sustained load (creep) (dogbone specimen) ASTM D2990 (modified) 21–22 Adhesive only Time to failure and time to tertiary creep Dynamic mechanical thermal analysis Adhesive only Stress, strain, and creep compliance Table 13. Summary test program.

Anchor Size x hef (0) (0) (0) (1) (10) 1 Baseline tests UF 5/8x3 X X X 4 36 15 UF 2 Baseline tests US M12x80 X X X 4 36 15 US 3 X (2) >120°F 3 9 5 (11) US 4 X (2) 70°F 3 9 5 (11) US 5 horizontal X (3) X (3) 3 18 10 (11) US 6 overhead X (3) X (3) 3 18 10 (11) US 7 damp/dry 5/8x3 X (4) 110°F 3 9 5 (11) UF 8 dry/damp M12/80 X (4a) 70°F 3 9 5 (11) US 10 X (7) MFR min (6) 3 9 5 (11) US 11 X (7) 110°F 3 9 5 (11) US 12 DOT Concrete mix 75°F downward dry/dry full 5/8x3 DOT X 110°F confined 3 9 5 (11) UF 13 Type of drilling 75°F downward dry/dry full 5/8x3 Standard X 110°F confined 3 9 5 (11) UF 14 with FA X (8) 9 5 (11) UF 15 with BFS X (9) 9 5 (11) UF 16 Test setup (wide support) 75°F downward dry/dry full 5/8x3 Standard X 110°F un-confined 3 9 5 (11) UF Concrete Age (tested at 3 days) X X X 0 0 15 (11) US Concrete Age (tested at 7 days) X X X 0 0 15 (11) US Concrete Age (tested at 14 days) X X X 0 0 15 (11) US Concrete Age (tested at 21 days) X X X 0 0 15 (11) US Concrete Age (tested at 28 days) X X X 0 0 15 (11) US Sum 216 185 Notes: (0) Type A = vinyl ester system, type B = epoxy system, type C = epoxy system. (1) 4 sustained loads Np / Nu,m(reference) 0.75/0.65/0.55/0.45. Creep tests with Np = 0.55 Nu,m will be used to compare with current approach of AC308 3 sustained loads Np / Nu,m(reference) 0.70/0.55/0.40. (2) Only the product that is most sensitive to increased temperature (high ratio glass transition temperature to service temperature) will be tested. (3) Only the top two products that are most sensitive to installation direction (occurrence of voids) in static tests will be tested. (4) Only the product that is most sensitive to wet concrete in static tests will be tested. (4a) Product that is sensitive to high alkalinity will be tested. The tests are performed at normal ambient temperature because under increased temperature the concrete will dry out. (5) Only the product that is most sensitive to hole cleaning (no brushing) will be tested. (6) Concrete at manufacturer's lowest permissible concrete temeprature. (6a) Mortar at manufacturer's lowest permissible mortar preheating temperature. (7) Only the product that is most sensitive to low installation temperature (low degree of cross linking) will be tested. (8) Only the product that is most sensitive to fly ash concrete will be tested. (9) Only the product that is most sensitive to blast furnace slag concrete will be tested. (10) UF = University of Florida, US = University of Stuttgart. (11) It is assumed that the influence of the investigated parameter on the short-term bond strength is known from previous tests. If not, all products will be tested and the number of reference tests will increase. confined 17 75°F downward dry/dry full M12x80 Standard 75°F confined confined Concrete composition 75°F downward dry/dry full 5/8x3 110°F 5 (11) UF Installation temperature MFR min(6) & (6a) downward dry/dry full M12x80 Standard X (5) 110°F confined 3 9 confined 9 Hole cleaning 75°F downward dry/dry reduced 5/8x3 Standard Moisture during installation or service 75°F downward full Standard confined Installation direction 75°F dry/dry full M12/80 Standard 110°F confined confined Service temperature 75°F downward dry/dry full M12/80 Standard Number of sustained load steps Number of sustained load tests Number of reference tests Test Location 75°F downward dry/dry full Standard 110°F Concrete Composition Product Type A Product Type B Product Type C Test Temperature Type of support Test Series Test Description (Influencing parameter) Installation Temperature Orientation during installation Moisture of concrete during installation/ service Cleaning 3 3 Table 14. Proposed test matrix for anchor pullout testing.

52 Adhesive-Alone Testing Program Based on the triage approach discussed above, Table 15 pre- sents the test matrix for the tensile creep testing program of dogbone specimens of the adhesives for series 21 and 22. Test series 21 and 22 began with short-term load tests per the tensile testing procedure presented in ASTM D638. Five repetitions were conducted on each adhesive and their values averaged to determine the mean short-term load strength. Test series 21 and 22 concluded with a series of sustained load tests per the tensile creep test procedure from ASTM D2990 (2001). Three adhesive dogbone specimens were loaded until failure at four stress levels for test series 21 and 22. The time to failure was determined as time to rupture as discussed above. Adhesive-alone tests were conducted on a dynamic shear rheometer (DSR) to develop master curves using time– temperature and time–stress superposition to compare with creep compliance curves from dogbone and anchor pullout tests. Anchor Pullout Tests—University of Florida Overview The following test series (Table 16) were conducted at the University of Florida; see Table 14 for a detailed test matrix. The short-term and sustained load (creep) tests generally followed the test procedure found in AASHTO TP 84-10 with the following modifications: Concrete • AASHTO TP 84-10 specifies that the concrete mix should be plain concrete without any admixtures. For all tests except for test series 12, 14, and 15 the concrete mix did not have any admixtures or additives. Test series 12 had granite aggregate, water reducer, and fly ash. Test series 14 and 15 used the baseline concrete mix but replaced the cement with 20% fly ash and 50% blast furnace slag respectfully. • AASHTO TP 84-10 specifies that the concrete mix should have a compressive strength between 2,500 to 4,000 psi at time of testing. For this project, the NCHRP panel chose to specify concrete with a compressive strength between 4,000 and 6,000 psi at time of testing to better conform to typical DOT concrete mixes. Adhesive • Adhesives of different chemistries from three manufacturers were chosen to investigate their sensitivity to sustained load. • Only adhesive anchor systems that met the assessment crite- ria of ICC-ES AC308 were used. The adhesive chemistries are briefly described below: – Adhesive A: This product is a vinyl ester with acrylic monomers with a peroxide hardener and quartz filler. – Adhesive B: This product is an epoxy resin with amine hardeners and quartz filler. – Adhesive C: This product is an epoxy resin with an amine blend. Anchor • As allowed in AASHTO TP 84-10, a 5⁄8” diameter threaded rod was used to avoid a steel failure mode. • As allowed in AASHTO TP 84-10, to further reduce the possibility of steel failure, ASTM A354 grade BD steel with 130 ksi yield strength and 150 ksi ultimate strength was used, which is greater than the minimum specified strength of ASTM A193 grade B-7 steel. Short-Term Tests Sustained Load Tests (1) Test Series Test Description Cure Time Test Temperature Product Type A (0) Product Type B (0) Product Type C (0) Product Type A Product Type B Product Type C 21 Baseline 7 days 110°F (43°C) 5 5 5 12 12 12 22 Cure time mfr spec 110°F (43°C) 5 5 5 12 (2) Sum 10 10 10 24 12 12 Notes: (0) Type A = vinyl ester system, type B = epoxy system, type C = epoxy system (1) Four stress levels times three repetitions for baseline (2) Only the product that is most sensitive to load at reduced cure time will be tested mfr = manufacturer. Table 15. Proposed test matrix for tensile creep testing. Test Series Test Description 1 Baseline 7 Moisture during installation 9 Reduced hole cleaning 12 Standard DOT mix 13 Type of drilling 14 Concrete composition— fly ash 15 Concrete composition—blast furnace Slag 16 Test setup—unconfined Table 16. Test descriptions.

53 • A 31⁄8” embedment depth for the 5⁄8” diameter bars was chosen based on minimum recommendations from AASHTO TP 84-10 to ensure adhesive failure. Test Procedure • All tests were confined tests except for test series 16, which evaluated the effect of the test setup and were unconfined tests. • The stress levels set for the sustained load (creep) test were initially at 85%, 75%, and 65% mean static load for all test series and an additional stress level of 55% mean static load for the baseline tests. After testing began, it was decided to adjust the stress levels due to early failure times at 85% and 75% mean static load. • As allowed in AASHTO TP 84-10 the frequency of data readings for the long-term (creep) tests was reduced over time according to the following schedule: – Every 0.5 seconds during loading, – Every 5 seconds for 10 minutes (120 readings), – Every 30 seconds for 1 hour (120 readings), – Every 5 minutes for 10 hours (120 readings), and – Every hour thereafter until failure. Details on the anchor pullout testing program at the Uni- versity of Florida can be found in Appendix C. Anchor Pullout Tests—University of Stuttgart This section presents the test program conducted at the University of Stuttgart Institut für Werkstoffe im Bau wesen (IWB) to investigate the effect of various parameters on the sustained load performance of three adhesive anchor systems. Overview The following test series (Table 17) were conducted at the University of Stuttgart; see Table 14 for a detailed test matrix. The short-term and sustained load (creep) tests were per- formed in accordance with the test procedures described in AASHTO TP 84-10 with the following modifications: Concrete • The concrete mix design for all test series followed the requirements of Deutsches Institut für Normung (DIN) EN 206-1 (Part 1: Specification, performance, production and conformity). For this research project, the NCHRP panel chose to specify concrete with a compressive strength between 4,000 and 6,000 psi at time of testing to conform to typical DOT concrete mixes. This corresponds to a con- crete C25/30 according to DIN EN 206-1. Adhesive • Only adhesives that met the assessment criteria of ICC-ES AC308 were used. • Adhesives of different chemistries from three manufacturers were chosen to investigate their sensitivity to sustained load. • These were the same three adhesives used in the University of Florida tests. Anchor • Due to a limitation of the test rigs for the creep tests, the anchor was limited to M12 metric threaded rods with approximately ½” diameter (12 mm). • To avoid steel failure in short-term tests, steel grade 12.9 was used, corresponding to a 174-ksi ultimate strength and a 157-ksi yield strength. • The embedment depth was hef = 3.15” (80 mm). This depth was chosen in order to compare the results with the numerous creep tests that were previously performed at the IWB using the same embedment depth. • Generally the anchors were centered at the bottom of the borehole with the use of a centering guide except for tests that were specifically performed to examine the behavior under special installation conditions (horizontal and over- head installation direction). The special centering guide used was not part of any of the tested anchoring systems. The 0.6” (15-mm) high centering guide was placed in the bottom of a 3.75” (95-mm) deep hole providing a 3.15” (80-mm) embed- ment depth. The centering guide had a conical indention that guided the anchors during the installation procedure. Test Procedure • All tests were confined tests. • The stress levels set for the sustained load (creep) test were initially at 85%, 75%, and 65% mean static load for all test series and an additional stress level of 55% mean static load Test Series Test Description 2 Baseline 3 Service temperature: +120°F (+49°C) 4 Service temperature: +70°F (21°C) 5 Installation direction: horizontal 6 Installation direction: overhead 8 Moisture during installation 10 Installation temperature: mfr min Service temperature: mfr min 11 Installation temperature: mfr min Service temperature : 110°F (43°C) Table 17. Test descriptions.

54 for the baseline tests. After testing began, it was decided to adjust the stress levels due to early failure times at 85% and 75% mean static load. • Due to a limitation of the measuring system, the frequency of data readings for the sustained load (creep) tests was not able to be varied and set to 10 minutes. Generally the first reading for a test occurred 120 seconds after the end of initial loading. Details on the anchor pullout testing program at the Uni- versity of Stuttgart can be found in Appendix D. Adhesive-Alone Tests—University of Florida This section presents the test program conducted at the University of Florida to investigate the isolated sustained load and short-term creep behavior of the adhesive alone. Overview The following test series (Table 18) were conducted at the University of Florida; see Table 14 for a detailed test matrix. The short-term tests generally followed the test procedure found in ASTM D638 with the following modifications: • Tested at 110°F (43°C) with an attached oven chamber and • Crosshead speeds were 0.1”, 0.4”, and 0.2” (2.5, 10, and 5 mm)/minute respectively for adhesive A, B, and C depend- ing on the brittleness of the sample. The sustained load (creep) tests generally followed the test procedure found in ASTM D2990 (2001) with the following modifications: • The weight for tensile creep was not directly applied to the specimen but through a lever arm system; • The strain was continuously measured by strain gauges; • Samples were conditioned as described in the following section; and • Stress levels were selected to be 35%, 45%, 55% and 75% of the adhesive’s maximum tensile stress obtained from short-term tests. Details on the adhesive-alone testing program at the Uni- versity of Florida can be found in Appendix E. Early-Age Concrete Evaluation— University of Stuttgart This section presents the test program conducted at the IWB laboratory of the University of Stuttgart to investigate the effect of early-age concrete on the short-term perfor- mance of three adhesive anchor systems. Overview The early-age concrete investigation is identified as test series 17. Refer to Table 14 for a complete description of the test program. The short-term confined tests generally followed the test procedure found in ASTM E488. Anchors were installed in concrete slabs of various ages (3, 6, 13, 20, and 27 days) and tested 24 hours later. Their short-term bond strength as well as other parameters (compressive strength, split tensile strength, initial surface absorption, hardness, and internal concrete temperature and relative humidity) were measured. Modulus of elasticity of the concrete was not considered. Details on the early-age concrete testing program at the University of Stuttgart can be found in Appendix F. Short-Term Anchor Pullout Data Reduction The following provides information related to data reduction. Displacement Adjustments As the anchors were initially loaded, the system took up slack (from the coupler, nuts, lading frame, etc.) producing large initial displacement readings. These large initial displacement readings at the beginning of the test were not due to interface slip between the adhesive and anchor or adhesive and concrete. Instead of adjusting the displacement readings for the initial slack in the system during testing, all data was recorded and adjustments were made after testing. The data acquisition sys- tem did however zero out the first position reading from the linear potentiometers (linear-pots) and all displacements read- ings were calculated from that initial position reading. The initial displacement readings were later adjusted to account for the slack in the system by extending a secant line through the load-displacement curve to the x-axis to deter- mine the x-intercept (Figure 36). The secant line intersected the load-displacement curve at approximately 10% and 30% of the peak load. The x-intercept was then used to adjust the load-displacement curve to intersect the origin. The displacement readings were also adjusted for the strain in the anchor between the concrete surface and the coupler. This was accomplished by adjusting the displacement reading by subtracting a strain correction factor (δcor) multiplied by the load reading. Test Series Test Description 21 Baseline 22 Manufacturer cure time Table 18. Test descriptions.

55 disp disp Nadj cor= − δi where dispadj = displacement adjusted for strain in anchor, disp = unadjusted displacement, and N = load. δcor eA E = l where l = distance between top of concrete and coupler, Ae = effective area of anchor, and E = modulus of elasticity of anchor steel. For the 5⁄8” diameter anchor pullout tests at the University of Florida: l = 2 in., Ae = 0.226 in.2, E = 29,000 ksi, and δcor = 0.000305 in./kip. For the 12 mm diameter anchor pullout tests at the Univer- sity of Stuttgart: l = 3.54” (90 mm), Ae = 0.131 in.2 (84.8 mm2), E = 29,000 ksi (200 GPa), and δcor = 0.000929 in./kip (0.0053 mm/kN). Determining Static Load Strength The static load strength is the strength of an adhesive determined from the short-term load test. Due to various possible failure modes, this might not be the maximum static load. The mean static load (MSL) is the average of the static adhesive strength values for an adhesive determined from a series of short-term load tests. This value is used to determine the percent load values in the sustained load (creep) test. There are several methods available to analyze the load- displacement behavior of a short-term load test in determining the static load strength which is referred to as Nadh by ACI 355.4. Section 10.4.4 of ACI 355.4 presents the following procedure: • Determine a tangent stiffness at 30% of the maximum static load (Nu), which is typically approximated as the secant stiff- ness from the origin to the point on the load-displacement curve at 0.30Nu. • If the displacement at 0.30Nu is less than 0.002 in., the ori- gin is shifted to the point on the load-displacement curve at 0.30Nu. 0 5 10 15 20 25 30 0.00 0.02 0.04 0.06 0.08 0.10 Displacement [in] Lo ad [k ip] secant line at 10% and 30% of peak load x-intercept Figure 36. Removing the effect of slack in the load-displacement graph.

56 Source: Cook and Konz (2001) Figure 37. Typical strength-controlled failure. Source: Cook and Konz (2001) Figure 38. Typical stiffness-controlled failure. Source: Cook and Konz (2001) Figure 39. Typical displacement-controlled failure. • Multiply the tangent stiffness by 2⁄3 and project this line until it intersects with the load-displacement curve. • Nadh is taken at the intersection if the load at the intersec- tion is less than Nu. • Nadh is taken as Nu if the load at the intersection is greater than Nu. This method was analyzed and was not recommended, as it tended to drastically underestimate the static load strength in a few cases as can be seen in Figure 40. Another procedure was presented by Cook and Konz (2001), in which they classified three types of load- displacement response (strength controlled, stiffness con- trolled, and displacement controlled) and described meth- ods to determine the static load strength for each type of situation. The responses and methods of analysis are sum- marized below: • Strength controlled. This failure mode is defined by a very sharp peak in the load-displacement curve with a drastic reduction in the stiffness of the adhesive anchor beyond the peak. The static load strength is determined to be at the peak on the load-displacement graph. Figure 37 shows a typical curve of a strength-controlled failure. • Stiffness controlled. This failure mode is defined by a large initial stiffness and a drastic change in stiffness, which does not decrease but rather continues to increase at a lower slope. Due to the absence of a “peak” in the curve, the static load strength is determined by finding the point at a tangent stiffness of 30 kips/in (5 kN/mm). The tangent stiffness (slope) at a given data point can be approximated by calculating the slope between a point five data points after and five data points before a given point. Figure 38 shows a typical curve of a stiffness- controlled failure. • Displacement controlled. This failure mode has a load- displacement curve with a relatively constant stiffness above the stiffness-controlled threshold of 30 kips/in. (5 kN/mm). The maximum static load occurs at very high, and impracti- cal displacements. In this case, the static load strength is set at a point with a displacement of 0.1 in. (2.5 mm). While the 0.1 in displacement seems arbitrary, this failure mode usu- ally only occurs in inferior products. Since this research was limited to products that met the assessment criteria of ICC- ES AC308 (ACI 355.4), this failure mode was not expected and was not observed. Figure 39 shows a typical curve of a displacement-controlled failure. The method presented by Cook and Konz (2001) exhib- ited better results than the ACI 355.4 approach and was the approach chosen for the project. Figure 40 is a load-displacement

57 graph for a short-term load test conducted showing the static load strength calculated by three different methods. • The ACI 355.4 procedure estimated Nadh as 11,100 lbf. • The strength-controlled method estimated Nadh as 19,905 lbf. • The stiffness-controlled method estimated Nadh as 19,751 lbf. For each test, the static load strength was recorded and the mean static load for each adhesive was determined from the average of the tests. Static Bond Stress The static bond stress (tadh) was calculated as the static load strength (Nadh) divided by the adhesive bond area at the inter- face with the anchor Aadh, or: τ = N Aadh adh adh where Aadh = p d hef, d = diameter of anchor (0.625” at UF and 0.472” at US), and hef = embedment depth of hole (3.125” at UF and 3.150” at US). The static bond stress was calculated to compare the results between the laboratories at the University of Florida and the University of Stuttgart as different diameters and embedment depths were used. Sustained Load Anchor Pullout Data Reduction The following provides information related to data reduction. Determination of Time to Failure Time to failure was initially evaluated as both the onset of tertiary creep and as the time to rupture. Based on recommendations from NCHRP (2009) the change in slope method was used to determine the onset of tertiary creep. This method calculated the slope at a given point as the slope between it and the prior data point. The change in slopes between the given point and the following data point was plotted and examined over the region just prior to rupture. It was suggested that this examination be conducted on a normal graph (not log time). The rupture point was easily identified on the displacement vs. time graph by its near vertical slope. A suggested range for examining the change in slope was from 80% to 100% of time to rupture. Due to minor fluctuations in the displacement readings, the slope might change from positive to negative several times over this range. Tertiary creep was defined as the time the change in slope became positive for the last time prior to rup- ture. This method produced favorable results and a sample graph is shown as Figure 41. The time to rupture was identified as the point when the anchor pulled out of the hole, which is indicated by a verti- cal line on the displacement versus time graph. This proved to be a very easy and reproducible analysis and did not vary significantly from the initiation of tertiary creep. Both times were determined for each test and the values for the UF and US baseline series are listed in Appendix J. Apart from a few exceptions, there was an average 3% difference between the two values. In three cases there was a larger difference, but this was for three tests at US and was due to the very short failure time (20 minutes) in relation to the sampling reso- lution of 10 minutes. As the time to rupture and onset of Figure 40. Example of calculating static load strength from various methods.

58 8.15 -0.050 0.000 0.050 0.100 0.150 0.200 0.250 6.0 6.5 7.0 7.5 8.0 8.5 9.0 TIME (hr) D IS PL AC EM EN T (in .) SL O PE (in ./h r) -0.040 0.000 0.040 0.080 0.120 0.160 0.200 CH AN G E IN S LO PE (in ./h r2 ) Displacement Tertiary Point Slope Change in Slope FIRST POSITIVE CHANGE IN SLOPE JUST PRIOR TO RUPTURE INITIATION OF TERTIARY CREEP Figure 41. Example of the change in slope method. tertiary creep analysis produced essentially the same time to failures, it was decided to use the time to rupture as the determination of time to failure as it was a much simpler method. Assessment of a Parameter’s Impact on Sustained Load Performance Test series 1 through 16 and 21 through 22 evaluated a parameter’s influence on sustained load performance using the “stress versus time-to-failure” test method (either by anchor pullout tests or “dogbone” tensile creep tests) to evaluate the performance of adhesive anchors under sustained load. Unlike the “displacement projection” test method found in ASTM E1512-01, ICC-ES AC58, ICC-ES AC308 and ACI 355.4, the “stress versus time-to-failure” method does not rely upon pro- jections of measured displacements but simply records the time to failure of the anchor. The only disadvantage of this method is that it takes an unknown time to complete the tests since they are all conducted to failure. Suggested Improvements A few possible changes were identified to improve the sus- tained load anchor pullout test procedure performed at the University of Florida. • In case of an eccentricity with the loading rod, one of the two linear potentiometers could produce a negative dis- placement reading that would generate an error in the averaged displacement. It is suggested that either one linear potentiometer be placed concentric with the anchor axis or a coupler with three linear potentiometers be used. A few possible changes were identified to improve the dog- bone testing procedure. • Thinner samples would allow for lower loads during both the creep test and the static test. • Sustained load creep test frames could have better iso- lation from each other so that the falling weight of a failed sample will not disturb other adjacent running tests. Summary This chapter summarized the parameters that could pos- sibly affect sustained load performance of adhesive anchor systems. Due to project budget and timeline, a triage was conducted to prioritize the parameters in order to test those thought to have the most impact. A general overview of the test program and analysis procedures was presented. The fol- lowing chapter discusses the findings and applications.