- Email: [email protected]
Vibration-based nondestructive testing as a practical tool for rapid concrete quality control
Construction and Building Materials 104 (2016) 181–190
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Vibration-based nondestructive testing as a practical tool for rapid concrete quality control Rafael Aguilar a,⇑, Eduardo Ramírez a, Vladimir G. Haach b, Miguel A. Pando c a
Department of Engineering, Civil Engineering Division, Pontificia Universidad Católica del Perú PUCP, Av. Universitaria 1801, San Miguel, Lima 32, Peru Department of Structural Engineering, School of Engineering at São Carlos, University of São Paulo, Av. dos Trabalhadores São-carlense, 400, São Carlos – SP 13566-590, Brazil c Department of Civil and Environmental Engineering, University of North Carolina at Charlotte, 9201 University City Blvd, Charlotte, NC 28223-0001, USA b
h i g h l i g h t s Study related to vibration-based non-destructive testing for QC/QA of concrete. A correlation of dynamic E-modulus and compressive strength was identified. NDT were useful to track the evolution of static and dynamic E-modulus in time. A relationship of expected E-moduli from measurements at earlier ages is proposed. Proposed predictive correlations include compressive strength to dynamic E-modulus.
a r t i c l e
i n f o
Article history: Received 8 August 2015 Received in revised form 4 November 2015 Accepted 8 December 2015 Available online 15 December 2015 Keywords: Concrete quality control Concrete mechanical characterization Predictive formulation Non-destructive testing Impact resonance testing EMM-ARM method
a b s t r a c t This paper involves a practical approach to perform quality-control/quality-assessment of concrete using vibration-based NDT. The first component is the analysis of compressive strength and dynamic E-modulus of concrete samples from various construction projects. The second component involves continuous measurements of E-Moduli as a function of curing time in laboratory-controlled specimens. The experimental program allowed proposing a correlation to predict the expected static and dynamic E-modulus at 28 days from their measurement at any instant of the curing process. A similar relationship is proposed to predict compressive strength at 28 days based on the dynamic E-modulus measured at earlier ages. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction The quality control (QC) and quality assurance (QA) of concrete is very important in the construction of reinforced concrete structures. Traditionally, the axial compressive strength (fc) is the engineering property most commonly used for QC/QA assessment of concrete. The most common approach used to assess the compressive strength of concrete is to collect concrete cylinder samples from a project and measure this property through simple compression tests as outlined in the ASTM Standard C39 [1]. For structures under construction, the samples can be prepared using the fresh concrete being delivered to the project. For existing structures, this approach would require coring concrete specimens which may induce damage to the structure. Furthermore, for both cases, a ⇑ Corresponding author. E-mail address: [email protected] (R. Aguilar). http://dx.doi.org/10.1016/j.conbuildmat.2015.12.053 0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
large number of samples would be required to reliably estimate the concrete strength as concrete conditions are expected to vary greatly not only based on inherent variability between the different concrete batches delivered to a project, but also variability associated to external factors such as differences in concrete placement conditions, vibration, water/cement (w/c) ratio, curing conditions, and boundary conditions. Thus to obtain a reliable assessment of the compressive strength of the concrete placed in a structure using compressive tests usually requires a very large number of samples. Direct measurement of the in situ strength of concrete with destructive tests is usually expensive and time consuming. An alternative approach that has become more popular is the use of some direct compressive testing complemented with modern non-destructive tests (NDT). This is the main focus of this paper, specifically the use of vibration-based NDT methodologies to carry out QC/QA of reinforced concrete structures. Although
182
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
the NDT methodologies discussed in this paper cannot measure mechanical properties of concrete directly, they are useful for the indirect estimation of elastic mechanical properties of the concrete such as the strength and the elasticity modulus (E-modulus). The most remarkable advantage of the NDT techniques involved in this study is that they offer the capability to measure rapidly and costeffectively the concrete of most or the whole structure, thus allowing a QC/QA assessment of the concrete used in the structure under actual in situ conditions. The paper presents vibration-based NDT of concrete from two experimental programs. The first experimental program involved a comparison of impact vibration NDT results with results from direct compression tests. This comparison was done for 245 concrete cylinders obtained from various construction projects in the city of Lima, Peru. The second experimental program involved continuous ambient vibration monitoring of a concrete specimen from initial placement as fresh concrete until completely hardened at 28 days of curing. This second component is based on tracking the evolution of the E-modulus of the concrete as a function of curing age. The results obtained from both experimental components support the notion that vibration-based NDT tests can be a valuable approach to complement direct compressive testing of concrete samples commonly used for QC/QA of construction of concrete structures.
to 0.4 fc [2]. Finally the chord modulus (Echord) is defined as the slope of the line that connects the point of the stress–strain curve that has an axial strain level equal to 50 lƐ with the same point used to determine the secant modulus (i.e. point with stress equal to 0.4 fc) [3]. The use of NDT vibration-based experiments for purposes of QC/QA of concrete can be done by comparing mechanical properties (i.e. different E-moduli) from traditional compressive tests on concrete cylinder samples with corresponding values obtained from measurements of NDT vibration-based tests. As described later in this paper, the compressive tests were used to determine the compressive strength (fc) and the different types of Emodulus of the concrete in a direct form. The procedure for the QC/QA of the concrete based on vibration-based NDT tests is presented based on determining the dynamic E-modulus from impact resonance tests [4]. This paper also presents NDT evaluations of concrete using the E-Modulus measurement based on the Ambient Response Method (EMM-ARM) [5]. This second NDT technique has the advantage that it can be used to track the evolution of the Emodulus of the concrete as a function of curing time. The applicability of this second NDT test towards the QC/QA assessment of concrete is also assessed and discussed.
2. Mechanical properties commonly used in concrete QC/QA
The experimental program presented in this paper has two main components. The first component entailed a detailed evaluation of the feasibility of using impact resonance NDT for QC/QA of plain, unreinforced concrete. This was done by comparing the NDT based test results with direct measurements of the mechanical properties of unreinforced concrete samples from compression tests. The second experimental component involved the evaluation of the mechanical properties of a fresh concrete mix as a function of curing time. This section describes the main experimental methods used in this study, which are shown schematically in Fig. 2. The first experimental component entailed the direct measurement of the mechanical properties of plain, unreinforced concrete cylinders samples was done by mean of compressive tests, as shown schematically in Fig. 2(a). Compressive tests in 245 samples were carried out at the structures laboratory at the Pontificia Universidad Católica del Perú (PUCP). The concrete samples came from dozens of projects within the city of Lima, Peru therefore representing a wide variety of concrete mix designs and curing conditions and ages. The compression tests were carried out in general accordance with the standard test method described in ASTM C39 [1]. All concrete cylinder samples were plain, unreinforced concrete with a standard 0.15 m diameter and a 0.30 m height. The cylinders were tested using neoprene end caps as per specifi-
Before discussing the NDT methods it is useful to review the main mechanical properties commonly used when performing QC/QA of reinforced concrete structures. The main structural contribution made by the concrete in a reinforced concrete structural element is its strength and stiffness under compression. A schematic stress–strain curve for concrete is shown in Fig. 1(a). It can be seen in this curve that the behavior of concrete under axial compression has a small portion where the behavior can be considered linear elastic, but for the most part the behavior is considered nonlinear. The peak of the curve is often used to determine the compressive strength of the concrete (referred herein as fc). The level of axial strain when the peak compressive stress is reached is usually between 20,000 and 30,000 lƐ [2]. In terms of E-moduli or stiffness it is common to use one of the following: the initial tangential E-modulus (Et), the secant E-modulus (Es), and the chord E-modulus (Echord). The definition of these three E-moduli is presented graphically in Fig. 1(b). The initial tangential modulus (Et) is the slope of a line drawn tangent to the initial linear portion of the stress–strain curve [3]. The secant modulus (Es) is defined as the slope of the line that goes from the origin to the point in the stress–strain curve corresponding to a compressive stress equal
3. Experimental program
Fig. 1. Stress–strain curve of concrete under compression: (a) curve until maximum strength; and (b) detail of elastic modulus calculation.
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
183
Fig. 2. Main tests carried out in this study: (a) compressive test (direct test); (b) impact resonance NDT; and (c) EMM-ARM NDT.
cations provided in ASTM C1231 [6]. For each compressive test the stress–strain curve was obtained using the setup shown in Fig. 2 (a). The concrete mix used for the second NDT component was designed and prepared at PUCP considering a water to cement ratio of w/c = 0.54. The composition and main characteristics of the concrete mix used are summarized in Table 1. The gradation curves for the sand and coarse aggregates used in the EMM-ARM concrete mix are shown in Fig. 3. This figure also shows the gradation limits recommended by ASTM C33 [7]. It can be seen that the gradation curves of the sand and coarse aggregates do not comply very well with these ASTM C33 limits thus, the Fuller method [8] was used to determine the adequate mix proportions to meet the ASTM specifications [9,10]. For this experimental component besides de EMMARM, which included continuous ambient vibration monitoring, we also carried out impact resonance testing and direct compression testing of 20 concrete cylinder samples that were cast from the same concrete batch used to prepare the concrete-filled acrylic tube beam. The 20 concrete cylinder specimens prepared were divided into five test groups of four samples each for direct compression testing at curing ages of 2, 5, 11, and 28 days. All specimens were cured under constant environmental conditions with an average room temperature of 21 °C and a relative humidity of 95%. Table 1 Mix design information components and fresh concrete properties for concrete used in experimental Component 2.
Portland cement type I Water Sand (dry weight) Gravel (dry weight)
Proportion of components (kg/m3)
Properties of fresh concrete (all in situ measured according ASTM standards)
370 200 902 922
Density (kg/m3) Air content (%) Temperature (°C) Water to cement ratio (w/c) Slump (m)
2435 1.60 19 0.54 0.08
The NDT evaluation of the 245 plain, unreinforced concrete cylinder samples for the experimental Component 1 was done by means of impact resonance tests. The setup for this test is shown schematically in Fig. 2(b). The impact resonance tests were carried out just before the traditional compressive tests to avoid any changes in mechanical properties due to aging. All samples were weighed and measured before NDT testing. Photos of the concrete cylinders and the impact resonance NDT are shown in Fig. 4 (a) and (b), respectively. The impact resonance testing was carried out in general accordance with ASTM Standard C215 [4]. The test involved striking the sample with a small hammer on one end and in the opposite end a piezoelectric accelerometer was attached to record the generated signal, as shown in Fig. 4(b). The accelerometer transducer was connected to a portable data acquisition system with 24 bits of resolution. The sensitivity of the accelerometer was 101.8 mV/g, the signal recording time was set to 3 s and the sampling frequency to 2048 Hz. The resonant frequency was obtained analyzing the recorded signal in the frequency domain using the Welch method [11]. The dynamic E-modulus, also known as the impact E-modulus for the impact resonance NDT, was determined as per ASTM C215 [4], using the identified dynamic modal behavior of the specimens. For this study, only the longitudinal resonant frequency was measured in the impact method as recommended by Batchelder and Lewis [12] taking into consideration that the determination of the dynamic E-modulus using the transverse frequency gives similar results. The dynamic E-modulus was computed using the following expression [4]:
Ed ¼ DMn2
ð1Þ
where n = fundamental longitudinal frequency in [Hz], D = 5.093 (L/ d2) in [m1] for a cylinder sample, L = length of cylinder specimen in [m], d = average diameter of the concrete cylinder specimen in [m], and M = mass of concrete cylinder specimen in [kg]. The second experimental component carried out as part of this research involved the EMM-ARM method [5]. The test setup is
184
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Fig. 3. Gradation curves of sand and gravel aggregates used in concrete mix design used for experimental Component 2.
Accelerometer Concrete specimen Hammer
Data acquisition device
(a)
(b)
Fig. 4. Impact resonance NDT used for experimental Component 1: (a) view of a subset of concrete cylinder samples before impact resonance NDT; and (b) photo of impact resonance test setup.
shown schematically in Fig. 2(c). As mentioned before, this method allowed tracking the E-modulus of a concrete sample as a function of curing age. The EMM-ARM method was used following the implementation suggested in Ref. [5]. In order to assess the evolution of the E-modulus of concrete as a function of curing time, a sample of fresh concrete was cast inside an acrylic tube of 1.7 m in length, and inner and outer diameters of 93 mm and 99 mm, respectively. The fresh concrete was placed inside the acrylic cylindrical tube initially arranged in a vertical position. A total of six thin metal rods were placed at equal spacing along the length of the acrylic tube. These thin metal rods were inserted across the acrylic tube as suggested in Refs. [5,13]. This was done to ensure the concrete infill and the tube had a good bond, strain compatibility, and no relative slippage to ensure a composite behavior of the composite concrete-filled acrylic tube. Voids and air pockets inside the concrete were minimized through vibration induced by means of lateral blows with a rubber mallet. Once the concrete pouring and vibration was completed, the upper end of the tube was closed and the tube was kept vertical for about 5 min. After this short waiting period, the cylinder was carefully rotated, placed on a vibrating bed aiming at consolidating the mix, and set to a position of a simply supported horizontal beam. Photos of the preparation of this concrete-filled acrylic tube EMM-ARM test setup are shown in Fig. 5. The monitoring system consisted of a lightweight piezoelectric accelerometer with a sensitivity of 963 mV/g placed on top of the mid span of the beam, as shown in Fig. 5(c), and an acquisition device with 24 bits of resolution. An automated algorithm developed in LabView [14] was used for monitoring the evolution of modal properties of the concrete beam. The system was programed to acquire events with 5 min of time signals of the ambient vibration response with a periodicity of 10 min during 672 h (28 days). With this configuration, 4122 events were finally recorded. Given that the maximum expected frequency was relatively low (around 40 Hz), the sampling rate was set to
200 Hz to obtain a proper resolution when analyzing the modal parameters. As shown in Fig. 6, The EMM-ARM NDT uses ambient vibration recordings to evaluate the first resonance frequency of the concrete-filled acrylic tube beam. The key assumption of the methodology is that ambient vibration, on average, represents a white noise excitation with equal energy in all frequencies. Fig. 6 shows acceleration time series recorded at different curing times (t0, . . . tn). This figure also shows schematically different response spectra in frequency domain. The largest peak of each response spectra is considered the predominant frequency of the first mode at the corresponding curing time when the recording was taken. The continuous measurement of changes in the predominant frequency is next correlated to the variation of the E-modulus of concrete. Then, an automatic modal identification tool is required to process the large amount of recorded data. In this case, the first flexural resonance frequency was continuously calculated using an automatic routine implemented following the procedure proposed in Ref. [13]. The flexural stiffness of the composite concrete-filled acrylic tube beam was determined using Eq. (2) [15]. Then, using the composite beam property that states that its flexural stiffness (EI) is equal to the summation of the flexural stiffness of its two components, the E-modulus of the concrete infill is obtained with Eq. (3) [13]. It must be noted that the influence of the small weight of the accelerometer placed at mid-span at the top of the beam was neglected in the determination of the concrete dynamic Emodulus primarily due to its very small value (0.025 kg).
EI ¼ Ec ¼
f
2L2
!2
p
EI Ea Ia Ic
m
ð2Þ ð3Þ
185
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Fig. 5. E-Modulus measurement based on the Ambient Response Method: (a) concrete batch manipulation; (b) composite beam construction process; and (c) test setup.
E-modulus Ac.
t0h
t0h
t672h
time
Ac.
t672h
Automatic System Identification
Equation 2
Curing time
frequency time
t0h
t672h
Fig. 6. Schematic showing the sequence of data recording and analyses used in the EMM-ARM method.
where E = E-modulus of the composite cross section in [Pa], I = inertia or second moment of area of the composite cross section in [m4], f = natural frequency of the first mode of the composite beam in [Hz], L = free span length of the composite beam in [m], m = distributed mass per length unit in [kg/m], Ec = E-modulus of the contained concrete in [Pa], Ic = second moment of area of the contained concrete in [m4], Ea = E-modulus of the acrylic tube in [Pa] (2.5 109 for the present case), and Ia = second moment of area of the acrylic formwork in [m4].
to the results obtained from direct compression tests carried out on the same concrete cylinders. The second part will present the results from the EMM-ARM test program which will be compared to results from several subsets of cylinder samples prepared from the same batch of concrete mix prepared directly for this study. The main focus on this second component is to assess the suitability of this NDT method to track the evolution of stiffness and strength of a plain concrete mix as a function of curing time. 4.1. Impact resonant NDT component (Component 1)
4. Experimental results This section presents the results of the two main NDT testing components described earlier. The first part presents the results from the impact resonance NDT tests carried out on concrete cylinder samples from a wide variety of construction projects in Lima, Peru. As mentioned, the results from these NDT tests are compared
For this component a total of 245 concrete cylinders were tested at the structures laboratory of PUCP. As described in the previous section, the impact resonance tests used for this component have the advantage that they constitute an easy NDT method for estimating the dynamic E-modulus of plain concrete cylinders. The dynamic E-moduli values obtained using this technique can
186
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Fig. 7. Relation between compressive strength test results and dynamic E-modulus from impact resonance NDT.
be correlated with the compressive strength (fc) measured with the direct compression tests as shown in Fig. 7. This plot shows that in general the compressive strength and the dynamic E-modulus values obtained from impact resonance NDT have some correlation where in general increasing dynamic E-modulus results in higher compressive strength values. This is similar to what is reported by others [16,17]. The results show considerable scatter particularly given that the concrete cylinders tested came from a wide variety of construction projects in the Lima area. Therefore these cylinders are expected to involve a very wide range of concrete mix designs, different water/cement ratios, differences in aggregates, additives, curing times, etc. With respect to curing time, the 245 cylinders tested included curing ages that ranged from 3 and 72 days, but the curing age for many cylinders was unknown or not reported. Ideally, the correlation of compressive strength vs dynamic E-moduli should be developed for the commonly used age of 28 days. Nevertheless, to use this technique for QC/QA, it is often necessary to use generalized correlations from experimental databases that include a wide range of design conditions. 4.2. EMM-ARM NDT component (Component 2) As mentioned before, the EMM-ARM NDT allowed assessing the influence of curing time on the strength and stiffness of a particular concrete mix designed for this project. The EMM-ARM results were compared with the results obtained from destructive compression tests carried out on test cylinders cast from the same concrete mix batch. The results of stiffness and strength obtained from the traditional destructive compression tests on test cylinders are shown in Fig. 8(a) and (b), respectively. The stiffness or E-modulus values
presented in Fig. 8(a) correspond to tangential, secant, and chord values defined earlier in this paper. Fig. 8(a) shows that all three types of E-moduli increase with increasing curing time. Despite the slight scatter of the experimental data, it is possible to define an upper and lower bound of the data. As expected, the upper bound corresponds to the tangential E-moduli while the secant and chord E-moduli define the lower bound. In terms of concrete strength, the compression test results shown in Fig. 8(b), show a strength gain as a function of time which is normally expected to be most marked in the first 28 days (672 h) of curing [18]. As shown in this figure, about 70% of the 28 day capacity was reached after 7 days (168 h) of curing. Between 7 and approximately 14 days (336 h) of curing the strength gain with respect to the 28 day strength was about 20%. Beyond 14 days (336 h) of curing the strength gain with respect to the 28 day strength was only moderate in the order of 10%. Additional to destructive compression tests, impact resonance experiments were carried on the test cylinders which allowed assessing the dynamic E-modulus from this NDT methodology as a function of curing time. These results are shown in Fig. 9(a). The results of this test show that more than 65% of the 28 day dynamic E-modulus was developed in the first 24 h of curing. Furthermore, after 7 days (168 h) and 14 days (336 h) of curing time the dynamic E-modulus measurements were nearly 90% and 95%, respectively of the average 28 day dynamic E-modulus. Fig. 9(a) shows a shaded area that represents the upper and the lower bounds of the E-modulus values measured from the compressive strength testing that was presented in Fig. 8(a). It can be seen that the impact resonance E-modulus values fall very close to the upper bound limit of the Fig. 8(a) shaded area. This result is somewhat expected since the dynamic E-modulus corresponds to a stiffness
Fig. 8. Direct measurement of mechanical properties in the Component 2 concrete mix as a function of curing time: (a) E-modulus; and (b) compressive strength.
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
50
EMM-ARM E-modulus (GPa)
50
Dynamic E-modulus (GPa)
187
40 30 20 10
40 30 20 10 0
0 0
96
192 288 384 480 576 672
0
96
192 288 384 480 576 672
Curing time (hours)
Curing time (hours)
(a)
(b)
Fig. 9. Continuous monitoring of concrete stiffness as a function of curing time: (a) dynamic E-modulus from impact resonance; and (b) E-modulus from EMM-ARM.
at very low strain levels and thus is expected to have a high correlation with the measured initial tangent stiffness modulus. This good correlation was also reported by Mehta [2]. The concrete E-modulus values from the EMM-ARM NDT method are shown in Fig. 9(b). As mentioned earlier the EMMARM method allowed continuous tracking of the concrete hardening during the curing process from very early ages. Fig. 9(b) also shows the upper and lower bounds reported in Fig. 8(a) obtained from the compression testing. It can be seen that the EMM-ARM E-modulus fall closer to the lower bound line of the compressive strength E-modulus values. As mentioned before this lower bound correspond to the secant and chord E-moduli. This suggests the EMM-ARM E-modulus values may be considered representative of a ‘‘static” E-modulus that is equivalent to those obtained from a strain range that is beyond the initial tangent of the stress–strain compression curve (e.g., secant or chord E-moduli) [3]. This is consistent with results reported in recent studies [5,13]. The EMM-ARM E-modulus results presented in Fig. 9(b) show that the E-modulus values start to oscillate with slightly higher amplitudes at a curing age of about 400 h (17 days). This larger amplitude appears in the continuous monitoring until a curing age of about 570 h (24 days). However, the average values continue with the same general trend increasing gradually and slowly with curing age. The onset of this sudden instability at approximately 400 h is believed to be related to the retraction of the concrete while curing. In fact, after 28 days a small gap between the concrete core and acrylic tube was noticed which led to the decision to carefully unmold and monitor the concrete core to verify results. This unmolding was fast and did not create any damage to the core. Even though there is a consistency on the results, further studies should try to improve the test setup. Quantitatively, the results show that after 24 h of casted, the concrete E-modulus reaches around 19 GPa, which represents approximately 70% of the modulus value reached at 28 days (672 h). Moreover, after 7 days (168 h) there is no significant increment on the EMM-ARM modulus, as at this age the modulus has reached nearly 85% of the 28 day stiffness. Beyond 7 days of curing the EMM-ARM data shows that the stiffness increase is very slow and gradual until the final curing aged monitored. 5. Presentation and discussion of preliminary predictive formulations This section presents a discussion of the results obtained in the experimental program and also presents expressions that could be used for preliminary predictions of QC/QA efforts.
5.1. Estimates of compressive strength using impact resonance dynamic E-modulus measurements The results of direct measurements of strength of concrete (fc) and NDT measurements of dynamic E-modulus (Ed) from the impact resonance tests are shown in Fig. 10. The results of the destructive compression tests carried out on test cylinders in the laboratory for experimental Component 2 were also added to this figure. A statistical analysis was then performed to determine an expression that provides the best fit to correlate both parameters (Eq. (4)). Based on this expression and the experimental data, upper and lower reliability limits (Eqs. (5) and (6), respectively) were calculated considering a confidence level of 90%. The equations for the best fit, lower and upper 90% confidence limits are as follows:
f c ¼ 2:2e0:07Ed
ð4Þ
0:07Ed
f c lower ¼ 1:5e
ð5Þ
f c upper ¼ 3:2e0:07Ed
ð6Þ
The scatter in the data shown in Fig. 10 is quite large, but it is reasonable for this experimental database which involves concrete cylinders from a wide variety of projects in Lima, Peru involving different concrete types, w/c, aggregates, mix designs, additives and even curing ages. Nevertheless, the upper and lower 90% confidence limits can be used for rapid preliminary QC/QA purposes using the dynamic E-moduli measured using cheap and rapid NDT in the field. The proposed procedure is to enter along the yaxis with the minimum concrete strength specified for the design and use either the best fit or the lower confidence limit lines to obtain the minimum dynamic E-modulus required during field impact resonant NDT. These equations can be refined in future studies by trying to populate the database and also subdivide the results based on concretes of similar design mix characteristics, curing conditions and age. 5.2. Evolution of E-modulus with curing time With all the information of the static (assumed equivalent to the E-modulus from the EMM-ARM method due to its similarity to the chord and secant E-modulus from compression tests) and dynamic E-modulus collected from impact NDT at different times after casting, a mathematical model that predicts the evolution in time of these properties is proposed in Eq. (7). This equation considers the age of concrete from the casting instant (t) as the principal feature for the development of stiffness. The maximum value is limited by the expected modulus at 28 days (672 h). A
188
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Compressive strength (MPa)
50
Specimens from constructions fc upper = 3.1e0.07Ed Laboratory specimens Best fit 90% interval of confidence
40 30
fc = 2.2e0.07Ed R2 = 0.785
20 10
fc lower = 1.5e0.07Ed
0 0
10
20
30
40
50
Dynamic E-modulus (GPa) Fig. 10. Regression analysis of the relationship of concrete compressive strength and dynamic E-modulus for rapid QC/QA of concrete.
parameter b was calibrated for obtaining the best fit approximation of the proposed model against the available experimental data. For the present study, the proposed evolution of E-modulus over time is as follows:
EðtÞ
b ¼ E28 1 0:70 t
ð7Þ
where E = E-modulus in [GPa], E28 = expected E-modulus at 28 days in [GPa] (E28 = 26 GPa for static, and 40 GPa for dynamic), t = time in [h], and b curve fitting constant = 2.9 for static, and 3.2 for dynamic E-modulus. Fig. 11 shows a comparison of the experimental data with the proposed mathematical model. As shown, the proposed model compares very well as denoted by the robust and strong correlation with the experimentally measured properties. In the model, it must be noted that the growing ratio of the stiffness at the initial part of the curve is defined by the b factor. The selection of a proper expected value for the E28-modulus is essential for a correct estimation. The continuous information of the Es from the EMM-ARM method, and the correspondent values of Ed from the resonance impact method, allowed plotting the ratio of Ed/Es as a function of time as shown in Fig. 12. As shown in the figure, this ratio is not a constant and changes according to the age of the concrete with values typically falling within a range of approximately 1.3– 1.6. As shown, the variation of this ratio is relatively high in the first 8 days (192 h). After this initial period, the ratio remains almost constant varying within a smaller range of 1.5–1.6. These results are in agreement to what is shown in other studies. Sharma and Gupta [19], for instance, present a relationship for high strength concrete in which, if the values registered in the present study are substituted, the results of Ed/Es would be equal to 1.44.
Other researches, such as [17], found that this relationship varies from 1.32 to 1.39 for masonry mortar. 5.3. Final remarks The relationship of the dynamic E-modulus as a function of time given by Eq. (7) is further studied to understand its capability for providing predictive results of compressive strength. For this, a new expression to predict Ed at 28 days (Ed28) in function of a measured value in an instant ti (Ed(ti)) is presented in Eq. (8). The relationship of this new expression to compressive strength at 28 days (fc(28)) is given by the substitution of the predicted value of Ed28 into Eq. (5).
Ed28 ¼
Edðti Þ t 0:7 3:2 i
1h t i 672h
ð8Þ
Figs. 13 and 14 compare the accuracy of the formulations formerly presented. Fig. 13 plots Eq. (5) using the experimental data of Ed acquired in the resonance tests at different ages and compare them against the traditional compressive tests. Even though the proposed formulation has in the worst case a 36% of strength underestimation, the results are very promising considering that they are in the conservative side, only non-destructive testing were used, and an approximation up to 15% was achieved. This Figure also shows the correlation of other formulations that relates Es and the compressive strength. For this, the input data was considered as the continuous Es results of the EMM-ARM method. It is highlighted that the proposed formulation gives similar results to the correlation proposed by [20]. Fig. 14, on the other hand, plots the values of fc(28) provided by Eq. (5) and uses as input data the set of predicted values of Ed28 obtained from all the measured data at different instants ti. Similar to what was registered before, a
Fig. 11. Predicted evolution of static and dynamic E-modulus.
189
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Fig. 12. Evolution with time of the ratio between dynamic and static E-modulus.
Compressive Strength (MPa)
3d
7d
14d
28d
40
Curing time (days)
30
20 Experimental results Predicted values ACI [20] EN-1992-1-1 [21] A BNT NBR [22]
10
0 0
96
192
288
384
480
576
Curing time (hours) 672
Fig. 13. Comparison of the development of concrete strength in time according to different predictive formulations (See above-mentioned references for further information.).
Compressive strength (MPa)
40
30
20
Prediction
10
Minimun strength registered at 28 days
0 0
96
192
288
384
480
576
672
Age of prediction (hours) Fig. 14. Predicted compressive strength with dynamic E-modulus data registered at different ages.
maximum underestimation of 37% was obtained but errors of up to 9% in the predictive strength fc(28) were also registered. Once again, the NDT nature of the proposed philosophy is very attractive for practical QC/QA applications.
6. Conclusions This paper illustrates how impact resonance and EMM-ARM NDT testing techniques can be successfully applied for QC/QA efforts of concrete. The paper first presents results from an experimental component involving impact resonance NDT that was successfully used for QC/QA of concrete from a wide variety of construction projects from Lima, Peru. The results from this first component showed that the dynamic E-modulus (Ed) from this NDT technique can be used reasonably well to predict the compressive strength of concrete. The second component involved the EMM-
ARM continuous monitoring which was successfully used to monitor the evolution of stiffness of concrete as a function of curing time. The experimental results from this study were also used to investigate the relationship of the static E-modulus of concrete (Es), defined herein as a chord or secant modulus, and the Ed, which corresponds to the stiffness at very low strain levels or tangential. The experimental results show that the ratio Ed/Es is not constant and depends on the curing time elapsed from casting. The variation of this ratio is relatively large during the first 8 days of curing but then tends to stabilize and remains almost constant beyond this time. Furthermore, a new relationship for estimating the compressive strength of plain concrete as a function of Ed based on impact resonance tests was developed (Eq. (5)). The expression was found to fit in the lower limit of a confidence interval of 90% which is considered an acceptable agreement given the wide range of projects
190
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
and mix designs used for the concrete cylinders used in this study. It is important to point out that the correlation equation should be limited to concrete strengths of up to 50 MPa. This relationship, although shows an interesting practical application for field QC/ QA applications, should be used with caution and the user understand its limitations. Specific correlations can be developed for larger projects which can consider their particular conditions such as mix designs. The information of the evolution with curing time of the Es and Ed was obtained with the impact resonance and the EMM-ARM. A relationship was proposed (Eq. (7)) which can be used to estimate the expected E28-modulus with experimental data. This equation, combined with the relationship of compressive strength and Ed, can be used to estimate the 28 day concrete strength using Ed registered at any time. The application of this methodology evidences a very good agreement to the experimental data showing conservative predictions of compressive strengths at 28 days. Future studies should be aimed at improving the experimental programs in low and high strength concretes in order to get general relationships that will complement the models proposed in the present study. These studies should also consider the differentiation of the testing programs according to the type of cement and aggregates used. The relationship of the Es and compressive strength should also be further addressed following the same orientation but considering broader experimental programs with different types of cements, concrete mixes, and curing ages. Acknowledgments This research was carried out by the Engineering & Heritage research group at the PUCP with collaborations from the Department of Structural Engineering of the University of São Paulo at São Carlos and the Department of Civil and Environmental Engineering of the University of North Carolina at Charlotte. The authors would like to thank the fellowship funding for postgraduate studies for the second author by CONCYTEC, PERU. The authors also express their gratitude to Mauricio Gonzales who helped carrying out many of the experimental tests that are shown in this article. Finally, the authors acknowledge the support of the Structures Laboratory of PUCP (LEDI) during the performance of several material tests carried on the concrete samples retrieved from different construction sites.
References [1] ASTM C39/C39M-12a. Standard Test Method for Compressive Strength of Cylindrical Concrete Specimens. Philadelphia, 2012. [2] P. Mehta, Concrete: Microstructure, Properties, and Materials, third ed., McGraw-Hill, New York, 2006. [3] ASTM C469-02. Standard test method for static modulus of elasticity and Poisson ratio of concrete in compression. Philadelphia, 2002. [4] ASTM C215-08. Standard test method for fundamental transverse, longitudinal, and torsional resonant frequencies of concrete specimens. Pennsylvania, 2008. [5] M. Azenha, F. Magalhães, R. Faria, A. Cunha, Measurement of concrete Emodulus evolution since casting: a novel method based on ambient vibration, Cement Concr Res 40 (2010) 1096–1105. [6] ASTM C1231/C1231M-14. Standard Practice for Use of Unbonded Caps in Determination of Compressive Strength of Hardened Concrete Cylinders. Pennsylvania, 2014. [7] ASTM C33/C33M. Standard Specification for Concrete Aggregates, West Conshohocken, PA, 2013. [8] W. Fuller, S. Thompson, The laws of proportioning concrete, Trans. Am. Soc. Civil Eng. 59 (2) (1907) 67–143. [9] ASTM C138/C138M-14. Standard test method for Density (Unit Weight), Yield, and Air Content (Gravimetric) of Concrete. Philadelphia, 2014. [10] ASTM C1064/C1064M-12. Standard test method for Temperature of Freshly Mixed Hydraulic-Cement Concrete. Philadelphia, 2012. [11] P.D. Welch, The use of fast Fourier transforms for the estimation of power spectra: a method based on time averaging over short modified periodograms, IEEE Trans. Audio Electroacoust. 15 (1967) 70–73. [12] G. Batchelder, D. Lewis, Comparison of dynamic methods of testing concretes subjected to freezing and thawing, Proc. Am. Soc. Testing Mater. 53 (1953) 1053. [13] M. Azenha, L.F. Ramos, R. Aguilar, J.L. Granja, Continuous monitoring of concrete E-modulus since casting based on modal identification: a case study, Cement Concr. Compos. 10 (2012) 10–15. [14] Labview, LabVIEW User Manual, Release 8.0 , National Instruments Corporation, USA, 2006. [15] A. Chopra, Dynamics of Structures, vol. 3, Prentice Hall, New Jersey, 1995. [16] V. Malhotra, V. Sivasundaram, Resonant frequency methods, in: V.M. Malhotra, N.J. Carino (Eds.), Handbook on Non-Destructive Testing of Concrete, CRC Press LLC, Florida, 2004. [17] V. Haach, R. Carrazedo, L. Oliveira, M. Corrêa, Application of acoustic tests to mechanical characterization of masonry mortars, NDT & E Int (2013). [18] S.H. Kosmatka, Properties and performance of normal-strength and highstrength Concrete, in: E.G. Nawy (Ed.), Concrete Construction Engineering Handbook, CRC Press, Boca Raton, FL, 1997, pp. 5-1–5-44. [19] M. Sharma, B. Gupta, Sonic modulus as related to strength and static modulus of high strength concrete, Indian Concr. J. (Bombay) 34 (4) (1960) 139. [20] ACI, Building Code Requirements for Reinforced Concrete and Commentary. ACI 318-95/ACI 318R-95, American Concrete Institute, P.O. Box 9094 Farmington Hills, MI 48333, 1995. [21] Eurocode2. Design of Concrete Structures – Part 1: General Rules and Rules for Buildings, 2004, p. 225. [22] ABNT NBR 6118. Projeto de Estructuras de Concreto - Procedimiento. Associacao Brasileira de Normas Técnicas, 2004, p. 23.
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Vibration-based nondestructive testing as a practical tool for rapid concrete quality control Rafael Aguilar a,⇑, Eduardo Ramírez a, Vladimir G. Haach b, Miguel A. Pando c a
Department of Engineering, Civil Engineering Division, Pontificia Universidad Católica del Perú PUCP, Av. Universitaria 1801, San Miguel, Lima 32, Peru Department of Structural Engineering, School of Engineering at São Carlos, University of São Paulo, Av. dos Trabalhadores São-carlense, 400, São Carlos – SP 13566-590, Brazil c Department of Civil and Environmental Engineering, University of North Carolina at Charlotte, 9201 University City Blvd, Charlotte, NC 28223-0001, USA b
h i g h l i g h t s Study related to vibration-based non-destructive testing for QC/QA of concrete. A correlation of dynamic E-modulus and compressive strength was identified. NDT were useful to track the evolution of static and dynamic E-modulus in time. A relationship of expected E-moduli from measurements at earlier ages is proposed. Proposed predictive correlations include compressive strength to dynamic E-modulus.
a r t i c l e
i n f o
Article history: Received 8 August 2015 Received in revised form 4 November 2015 Accepted 8 December 2015 Available online 15 December 2015 Keywords: Concrete quality control Concrete mechanical characterization Predictive formulation Non-destructive testing Impact resonance testing EMM-ARM method
a b s t r a c t This paper involves a practical approach to perform quality-control/quality-assessment of concrete using vibration-based NDT. The first component is the analysis of compressive strength and dynamic E-modulus of concrete samples from various construction projects. The second component involves continuous measurements of E-Moduli as a function of curing time in laboratory-controlled specimens. The experimental program allowed proposing a correlation to predict the expected static and dynamic E-modulus at 28 days from their measurement at any instant of the curing process. A similar relationship is proposed to predict compressive strength at 28 days based on the dynamic E-modulus measured at earlier ages. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction The quality control (QC) and quality assurance (QA) of concrete is very important in the construction of reinforced concrete structures. Traditionally, the axial compressive strength (fc) is the engineering property most commonly used for QC/QA assessment of concrete. The most common approach used to assess the compressive strength of concrete is to collect concrete cylinder samples from a project and measure this property through simple compression tests as outlined in the ASTM Standard C39 [1]. For structures under construction, the samples can be prepared using the fresh concrete being delivered to the project. For existing structures, this approach would require coring concrete specimens which may induce damage to the structure. Furthermore, for both cases, a ⇑ Corresponding author. E-mail address: [email protected] (R. Aguilar). http://dx.doi.org/10.1016/j.conbuildmat.2015.12.053 0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
large number of samples would be required to reliably estimate the concrete strength as concrete conditions are expected to vary greatly not only based on inherent variability between the different concrete batches delivered to a project, but also variability associated to external factors such as differences in concrete placement conditions, vibration, water/cement (w/c) ratio, curing conditions, and boundary conditions. Thus to obtain a reliable assessment of the compressive strength of the concrete placed in a structure using compressive tests usually requires a very large number of samples. Direct measurement of the in situ strength of concrete with destructive tests is usually expensive and time consuming. An alternative approach that has become more popular is the use of some direct compressive testing complemented with modern non-destructive tests (NDT). This is the main focus of this paper, specifically the use of vibration-based NDT methodologies to carry out QC/QA of reinforced concrete structures. Although
182
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
the NDT methodologies discussed in this paper cannot measure mechanical properties of concrete directly, they are useful for the indirect estimation of elastic mechanical properties of the concrete such as the strength and the elasticity modulus (E-modulus). The most remarkable advantage of the NDT techniques involved in this study is that they offer the capability to measure rapidly and costeffectively the concrete of most or the whole structure, thus allowing a QC/QA assessment of the concrete used in the structure under actual in situ conditions. The paper presents vibration-based NDT of concrete from two experimental programs. The first experimental program involved a comparison of impact vibration NDT results with results from direct compression tests. This comparison was done for 245 concrete cylinders obtained from various construction projects in the city of Lima, Peru. The second experimental program involved continuous ambient vibration monitoring of a concrete specimen from initial placement as fresh concrete until completely hardened at 28 days of curing. This second component is based on tracking the evolution of the E-modulus of the concrete as a function of curing age. The results obtained from both experimental components support the notion that vibration-based NDT tests can be a valuable approach to complement direct compressive testing of concrete samples commonly used for QC/QA of construction of concrete structures.
to 0.4 fc [2]. Finally the chord modulus (Echord) is defined as the slope of the line that connects the point of the stress–strain curve that has an axial strain level equal to 50 lƐ with the same point used to determine the secant modulus (i.e. point with stress equal to 0.4 fc) [3]. The use of NDT vibration-based experiments for purposes of QC/QA of concrete can be done by comparing mechanical properties (i.e. different E-moduli) from traditional compressive tests on concrete cylinder samples with corresponding values obtained from measurements of NDT vibration-based tests. As described later in this paper, the compressive tests were used to determine the compressive strength (fc) and the different types of Emodulus of the concrete in a direct form. The procedure for the QC/QA of the concrete based on vibration-based NDT tests is presented based on determining the dynamic E-modulus from impact resonance tests [4]. This paper also presents NDT evaluations of concrete using the E-Modulus measurement based on the Ambient Response Method (EMM-ARM) [5]. This second NDT technique has the advantage that it can be used to track the evolution of the Emodulus of the concrete as a function of curing time. The applicability of this second NDT test towards the QC/QA assessment of concrete is also assessed and discussed.
2. Mechanical properties commonly used in concrete QC/QA
The experimental program presented in this paper has two main components. The first component entailed a detailed evaluation of the feasibility of using impact resonance NDT for QC/QA of plain, unreinforced concrete. This was done by comparing the NDT based test results with direct measurements of the mechanical properties of unreinforced concrete samples from compression tests. The second experimental component involved the evaluation of the mechanical properties of a fresh concrete mix as a function of curing time. This section describes the main experimental methods used in this study, which are shown schematically in Fig. 2. The first experimental component entailed the direct measurement of the mechanical properties of plain, unreinforced concrete cylinders samples was done by mean of compressive tests, as shown schematically in Fig. 2(a). Compressive tests in 245 samples were carried out at the structures laboratory at the Pontificia Universidad Católica del Perú (PUCP). The concrete samples came from dozens of projects within the city of Lima, Peru therefore representing a wide variety of concrete mix designs and curing conditions and ages. The compression tests were carried out in general accordance with the standard test method described in ASTM C39 [1]. All concrete cylinder samples were plain, unreinforced concrete with a standard 0.15 m diameter and a 0.30 m height. The cylinders were tested using neoprene end caps as per specifi-
Before discussing the NDT methods it is useful to review the main mechanical properties commonly used when performing QC/QA of reinforced concrete structures. The main structural contribution made by the concrete in a reinforced concrete structural element is its strength and stiffness under compression. A schematic stress–strain curve for concrete is shown in Fig. 1(a). It can be seen in this curve that the behavior of concrete under axial compression has a small portion where the behavior can be considered linear elastic, but for the most part the behavior is considered nonlinear. The peak of the curve is often used to determine the compressive strength of the concrete (referred herein as fc). The level of axial strain when the peak compressive stress is reached is usually between 20,000 and 30,000 lƐ [2]. In terms of E-moduli or stiffness it is common to use one of the following: the initial tangential E-modulus (Et), the secant E-modulus (Es), and the chord E-modulus (Echord). The definition of these three E-moduli is presented graphically in Fig. 1(b). The initial tangential modulus (Et) is the slope of a line drawn tangent to the initial linear portion of the stress–strain curve [3]. The secant modulus (Es) is defined as the slope of the line that goes from the origin to the point in the stress–strain curve corresponding to a compressive stress equal
3. Experimental program
Fig. 1. Stress–strain curve of concrete under compression: (a) curve until maximum strength; and (b) detail of elastic modulus calculation.
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
183
Fig. 2. Main tests carried out in this study: (a) compressive test (direct test); (b) impact resonance NDT; and (c) EMM-ARM NDT.
cations provided in ASTM C1231 [6]. For each compressive test the stress–strain curve was obtained using the setup shown in Fig. 2 (a). The concrete mix used for the second NDT component was designed and prepared at PUCP considering a water to cement ratio of w/c = 0.54. The composition and main characteristics of the concrete mix used are summarized in Table 1. The gradation curves for the sand and coarse aggregates used in the EMM-ARM concrete mix are shown in Fig. 3. This figure also shows the gradation limits recommended by ASTM C33 [7]. It can be seen that the gradation curves of the sand and coarse aggregates do not comply very well with these ASTM C33 limits thus, the Fuller method [8] was used to determine the adequate mix proportions to meet the ASTM specifications [9,10]. For this experimental component besides de EMMARM, which included continuous ambient vibration monitoring, we also carried out impact resonance testing and direct compression testing of 20 concrete cylinder samples that were cast from the same concrete batch used to prepare the concrete-filled acrylic tube beam. The 20 concrete cylinder specimens prepared were divided into five test groups of four samples each for direct compression testing at curing ages of 2, 5, 11, and 28 days. All specimens were cured under constant environmental conditions with an average room temperature of 21 °C and a relative humidity of 95%. Table 1 Mix design information components and fresh concrete properties for concrete used in experimental Component 2.
Portland cement type I Water Sand (dry weight) Gravel (dry weight)
Proportion of components (kg/m3)
Properties of fresh concrete (all in situ measured according ASTM standards)
370 200 902 922
Density (kg/m3) Air content (%) Temperature (°C) Water to cement ratio (w/c) Slump (m)
2435 1.60 19 0.54 0.08
The NDT evaluation of the 245 plain, unreinforced concrete cylinder samples for the experimental Component 1 was done by means of impact resonance tests. The setup for this test is shown schematically in Fig. 2(b). The impact resonance tests were carried out just before the traditional compressive tests to avoid any changes in mechanical properties due to aging. All samples were weighed and measured before NDT testing. Photos of the concrete cylinders and the impact resonance NDT are shown in Fig. 4 (a) and (b), respectively. The impact resonance testing was carried out in general accordance with ASTM Standard C215 [4]. The test involved striking the sample with a small hammer on one end and in the opposite end a piezoelectric accelerometer was attached to record the generated signal, as shown in Fig. 4(b). The accelerometer transducer was connected to a portable data acquisition system with 24 bits of resolution. The sensitivity of the accelerometer was 101.8 mV/g, the signal recording time was set to 3 s and the sampling frequency to 2048 Hz. The resonant frequency was obtained analyzing the recorded signal in the frequency domain using the Welch method [11]. The dynamic E-modulus, also known as the impact E-modulus for the impact resonance NDT, was determined as per ASTM C215 [4], using the identified dynamic modal behavior of the specimens. For this study, only the longitudinal resonant frequency was measured in the impact method as recommended by Batchelder and Lewis [12] taking into consideration that the determination of the dynamic E-modulus using the transverse frequency gives similar results. The dynamic E-modulus was computed using the following expression [4]:
Ed ¼ DMn2
ð1Þ
where n = fundamental longitudinal frequency in [Hz], D = 5.093 (L/ d2) in [m1] for a cylinder sample, L = length of cylinder specimen in [m], d = average diameter of the concrete cylinder specimen in [m], and M = mass of concrete cylinder specimen in [kg]. The second experimental component carried out as part of this research involved the EMM-ARM method [5]. The test setup is
184
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Fig. 3. Gradation curves of sand and gravel aggregates used in concrete mix design used for experimental Component 2.
Accelerometer Concrete specimen Hammer
Data acquisition device
(a)
(b)
Fig. 4. Impact resonance NDT used for experimental Component 1: (a) view of a subset of concrete cylinder samples before impact resonance NDT; and (b) photo of impact resonance test setup.
shown schematically in Fig. 2(c). As mentioned before, this method allowed tracking the E-modulus of a concrete sample as a function of curing age. The EMM-ARM method was used following the implementation suggested in Ref. [5]. In order to assess the evolution of the E-modulus of concrete as a function of curing time, a sample of fresh concrete was cast inside an acrylic tube of 1.7 m in length, and inner and outer diameters of 93 mm and 99 mm, respectively. The fresh concrete was placed inside the acrylic cylindrical tube initially arranged in a vertical position. A total of six thin metal rods were placed at equal spacing along the length of the acrylic tube. These thin metal rods were inserted across the acrylic tube as suggested in Refs. [5,13]. This was done to ensure the concrete infill and the tube had a good bond, strain compatibility, and no relative slippage to ensure a composite behavior of the composite concrete-filled acrylic tube. Voids and air pockets inside the concrete were minimized through vibration induced by means of lateral blows with a rubber mallet. Once the concrete pouring and vibration was completed, the upper end of the tube was closed and the tube was kept vertical for about 5 min. After this short waiting period, the cylinder was carefully rotated, placed on a vibrating bed aiming at consolidating the mix, and set to a position of a simply supported horizontal beam. Photos of the preparation of this concrete-filled acrylic tube EMM-ARM test setup are shown in Fig. 5. The monitoring system consisted of a lightweight piezoelectric accelerometer with a sensitivity of 963 mV/g placed on top of the mid span of the beam, as shown in Fig. 5(c), and an acquisition device with 24 bits of resolution. An automated algorithm developed in LabView [14] was used for monitoring the evolution of modal properties of the concrete beam. The system was programed to acquire events with 5 min of time signals of the ambient vibration response with a periodicity of 10 min during 672 h (28 days). With this configuration, 4122 events were finally recorded. Given that the maximum expected frequency was relatively low (around 40 Hz), the sampling rate was set to
200 Hz to obtain a proper resolution when analyzing the modal parameters. As shown in Fig. 6, The EMM-ARM NDT uses ambient vibration recordings to evaluate the first resonance frequency of the concrete-filled acrylic tube beam. The key assumption of the methodology is that ambient vibration, on average, represents a white noise excitation with equal energy in all frequencies. Fig. 6 shows acceleration time series recorded at different curing times (t0, . . . tn). This figure also shows schematically different response spectra in frequency domain. The largest peak of each response spectra is considered the predominant frequency of the first mode at the corresponding curing time when the recording was taken. The continuous measurement of changes in the predominant frequency is next correlated to the variation of the E-modulus of concrete. Then, an automatic modal identification tool is required to process the large amount of recorded data. In this case, the first flexural resonance frequency was continuously calculated using an automatic routine implemented following the procedure proposed in Ref. [13]. The flexural stiffness of the composite concrete-filled acrylic tube beam was determined using Eq. (2) [15]. Then, using the composite beam property that states that its flexural stiffness (EI) is equal to the summation of the flexural stiffness of its two components, the E-modulus of the concrete infill is obtained with Eq. (3) [13]. It must be noted that the influence of the small weight of the accelerometer placed at mid-span at the top of the beam was neglected in the determination of the concrete dynamic Emodulus primarily due to its very small value (0.025 kg).
EI ¼ Ec ¼
f
2L2
!2
p
EI Ea Ia Ic
m
ð2Þ ð3Þ
185
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Fig. 5. E-Modulus measurement based on the Ambient Response Method: (a) concrete batch manipulation; (b) composite beam construction process; and (c) test setup.
E-modulus Ac.
t0h
t0h
t672h
time
Ac.
t672h
Automatic System Identification
Equation 2
Curing time
frequency time
t0h
t672h
Fig. 6. Schematic showing the sequence of data recording and analyses used in the EMM-ARM method.
where E = E-modulus of the composite cross section in [Pa], I = inertia or second moment of area of the composite cross section in [m4], f = natural frequency of the first mode of the composite beam in [Hz], L = free span length of the composite beam in [m], m = distributed mass per length unit in [kg/m], Ec = E-modulus of the contained concrete in [Pa], Ic = second moment of area of the contained concrete in [m4], Ea = E-modulus of the acrylic tube in [Pa] (2.5 109 for the present case), and Ia = second moment of area of the acrylic formwork in [m4].
to the results obtained from direct compression tests carried out on the same concrete cylinders. The second part will present the results from the EMM-ARM test program which will be compared to results from several subsets of cylinder samples prepared from the same batch of concrete mix prepared directly for this study. The main focus on this second component is to assess the suitability of this NDT method to track the evolution of stiffness and strength of a plain concrete mix as a function of curing time. 4.1. Impact resonant NDT component (Component 1)
4. Experimental results This section presents the results of the two main NDT testing components described earlier. The first part presents the results from the impact resonance NDT tests carried out on concrete cylinder samples from a wide variety of construction projects in Lima, Peru. As mentioned, the results from these NDT tests are compared
For this component a total of 245 concrete cylinders were tested at the structures laboratory of PUCP. As described in the previous section, the impact resonance tests used for this component have the advantage that they constitute an easy NDT method for estimating the dynamic E-modulus of plain concrete cylinders. The dynamic E-moduli values obtained using this technique can
186
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Fig. 7. Relation between compressive strength test results and dynamic E-modulus from impact resonance NDT.
be correlated with the compressive strength (fc) measured with the direct compression tests as shown in Fig. 7. This plot shows that in general the compressive strength and the dynamic E-modulus values obtained from impact resonance NDT have some correlation where in general increasing dynamic E-modulus results in higher compressive strength values. This is similar to what is reported by others [16,17]. The results show considerable scatter particularly given that the concrete cylinders tested came from a wide variety of construction projects in the Lima area. Therefore these cylinders are expected to involve a very wide range of concrete mix designs, different water/cement ratios, differences in aggregates, additives, curing times, etc. With respect to curing time, the 245 cylinders tested included curing ages that ranged from 3 and 72 days, but the curing age for many cylinders was unknown or not reported. Ideally, the correlation of compressive strength vs dynamic E-moduli should be developed for the commonly used age of 28 days. Nevertheless, to use this technique for QC/QA, it is often necessary to use generalized correlations from experimental databases that include a wide range of design conditions. 4.2. EMM-ARM NDT component (Component 2) As mentioned before, the EMM-ARM NDT allowed assessing the influence of curing time on the strength and stiffness of a particular concrete mix designed for this project. The EMM-ARM results were compared with the results obtained from destructive compression tests carried out on test cylinders cast from the same concrete mix batch. The results of stiffness and strength obtained from the traditional destructive compression tests on test cylinders are shown in Fig. 8(a) and (b), respectively. The stiffness or E-modulus values
presented in Fig. 8(a) correspond to tangential, secant, and chord values defined earlier in this paper. Fig. 8(a) shows that all three types of E-moduli increase with increasing curing time. Despite the slight scatter of the experimental data, it is possible to define an upper and lower bound of the data. As expected, the upper bound corresponds to the tangential E-moduli while the secant and chord E-moduli define the lower bound. In terms of concrete strength, the compression test results shown in Fig. 8(b), show a strength gain as a function of time which is normally expected to be most marked in the first 28 days (672 h) of curing [18]. As shown in this figure, about 70% of the 28 day capacity was reached after 7 days (168 h) of curing. Between 7 and approximately 14 days (336 h) of curing the strength gain with respect to the 28 day strength was about 20%. Beyond 14 days (336 h) of curing the strength gain with respect to the 28 day strength was only moderate in the order of 10%. Additional to destructive compression tests, impact resonance experiments were carried on the test cylinders which allowed assessing the dynamic E-modulus from this NDT methodology as a function of curing time. These results are shown in Fig. 9(a). The results of this test show that more than 65% of the 28 day dynamic E-modulus was developed in the first 24 h of curing. Furthermore, after 7 days (168 h) and 14 days (336 h) of curing time the dynamic E-modulus measurements were nearly 90% and 95%, respectively of the average 28 day dynamic E-modulus. Fig. 9(a) shows a shaded area that represents the upper and the lower bounds of the E-modulus values measured from the compressive strength testing that was presented in Fig. 8(a). It can be seen that the impact resonance E-modulus values fall very close to the upper bound limit of the Fig. 8(a) shaded area. This result is somewhat expected since the dynamic E-modulus corresponds to a stiffness
Fig. 8. Direct measurement of mechanical properties in the Component 2 concrete mix as a function of curing time: (a) E-modulus; and (b) compressive strength.
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
50
EMM-ARM E-modulus (GPa)
50
Dynamic E-modulus (GPa)
187
40 30 20 10
40 30 20 10 0
0 0
96
192 288 384 480 576 672
0
96
192 288 384 480 576 672
Curing time (hours)
Curing time (hours)
(a)
(b)
Fig. 9. Continuous monitoring of concrete stiffness as a function of curing time: (a) dynamic E-modulus from impact resonance; and (b) E-modulus from EMM-ARM.
at very low strain levels and thus is expected to have a high correlation with the measured initial tangent stiffness modulus. This good correlation was also reported by Mehta [2]. The concrete E-modulus values from the EMM-ARM NDT method are shown in Fig. 9(b). As mentioned earlier the EMMARM method allowed continuous tracking of the concrete hardening during the curing process from very early ages. Fig. 9(b) also shows the upper and lower bounds reported in Fig. 8(a) obtained from the compression testing. It can be seen that the EMM-ARM E-modulus fall closer to the lower bound line of the compressive strength E-modulus values. As mentioned before this lower bound correspond to the secant and chord E-moduli. This suggests the EMM-ARM E-modulus values may be considered representative of a ‘‘static” E-modulus that is equivalent to those obtained from a strain range that is beyond the initial tangent of the stress–strain compression curve (e.g., secant or chord E-moduli) [3]. This is consistent with results reported in recent studies [5,13]. The EMM-ARM E-modulus results presented in Fig. 9(b) show that the E-modulus values start to oscillate with slightly higher amplitudes at a curing age of about 400 h (17 days). This larger amplitude appears in the continuous monitoring until a curing age of about 570 h (24 days). However, the average values continue with the same general trend increasing gradually and slowly with curing age. The onset of this sudden instability at approximately 400 h is believed to be related to the retraction of the concrete while curing. In fact, after 28 days a small gap between the concrete core and acrylic tube was noticed which led to the decision to carefully unmold and monitor the concrete core to verify results. This unmolding was fast and did not create any damage to the core. Even though there is a consistency on the results, further studies should try to improve the test setup. Quantitatively, the results show that after 24 h of casted, the concrete E-modulus reaches around 19 GPa, which represents approximately 70% of the modulus value reached at 28 days (672 h). Moreover, after 7 days (168 h) there is no significant increment on the EMM-ARM modulus, as at this age the modulus has reached nearly 85% of the 28 day stiffness. Beyond 7 days of curing the EMM-ARM data shows that the stiffness increase is very slow and gradual until the final curing aged monitored. 5. Presentation and discussion of preliminary predictive formulations This section presents a discussion of the results obtained in the experimental program and also presents expressions that could be used for preliminary predictions of QC/QA efforts.
5.1. Estimates of compressive strength using impact resonance dynamic E-modulus measurements The results of direct measurements of strength of concrete (fc) and NDT measurements of dynamic E-modulus (Ed) from the impact resonance tests are shown in Fig. 10. The results of the destructive compression tests carried out on test cylinders in the laboratory for experimental Component 2 were also added to this figure. A statistical analysis was then performed to determine an expression that provides the best fit to correlate both parameters (Eq. (4)). Based on this expression and the experimental data, upper and lower reliability limits (Eqs. (5) and (6), respectively) were calculated considering a confidence level of 90%. The equations for the best fit, lower and upper 90% confidence limits are as follows:
f c ¼ 2:2e0:07Ed
ð4Þ
0:07Ed
f c lower ¼ 1:5e
ð5Þ
f c upper ¼ 3:2e0:07Ed
ð6Þ
The scatter in the data shown in Fig. 10 is quite large, but it is reasonable for this experimental database which involves concrete cylinders from a wide variety of projects in Lima, Peru involving different concrete types, w/c, aggregates, mix designs, additives and even curing ages. Nevertheless, the upper and lower 90% confidence limits can be used for rapid preliminary QC/QA purposes using the dynamic E-moduli measured using cheap and rapid NDT in the field. The proposed procedure is to enter along the yaxis with the minimum concrete strength specified for the design and use either the best fit or the lower confidence limit lines to obtain the minimum dynamic E-modulus required during field impact resonant NDT. These equations can be refined in future studies by trying to populate the database and also subdivide the results based on concretes of similar design mix characteristics, curing conditions and age. 5.2. Evolution of E-modulus with curing time With all the information of the static (assumed equivalent to the E-modulus from the EMM-ARM method due to its similarity to the chord and secant E-modulus from compression tests) and dynamic E-modulus collected from impact NDT at different times after casting, a mathematical model that predicts the evolution in time of these properties is proposed in Eq. (7). This equation considers the age of concrete from the casting instant (t) as the principal feature for the development of stiffness. The maximum value is limited by the expected modulus at 28 days (672 h). A
188
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Compressive strength (MPa)
50
Specimens from constructions fc upper = 3.1e0.07Ed Laboratory specimens Best fit 90% interval of confidence
40 30
fc = 2.2e0.07Ed R2 = 0.785
20 10
fc lower = 1.5e0.07Ed
0 0
10
20
30
40
50
Dynamic E-modulus (GPa) Fig. 10. Regression analysis of the relationship of concrete compressive strength and dynamic E-modulus for rapid QC/QA of concrete.
parameter b was calibrated for obtaining the best fit approximation of the proposed model against the available experimental data. For the present study, the proposed evolution of E-modulus over time is as follows:
EðtÞ
b ¼ E28 1 0:70 t
ð7Þ
where E = E-modulus in [GPa], E28 = expected E-modulus at 28 days in [GPa] (E28 = 26 GPa for static, and 40 GPa for dynamic), t = time in [h], and b curve fitting constant = 2.9 for static, and 3.2 for dynamic E-modulus. Fig. 11 shows a comparison of the experimental data with the proposed mathematical model. As shown, the proposed model compares very well as denoted by the robust and strong correlation with the experimentally measured properties. In the model, it must be noted that the growing ratio of the stiffness at the initial part of the curve is defined by the b factor. The selection of a proper expected value for the E28-modulus is essential for a correct estimation. The continuous information of the Es from the EMM-ARM method, and the correspondent values of Ed from the resonance impact method, allowed plotting the ratio of Ed/Es as a function of time as shown in Fig. 12. As shown in the figure, this ratio is not a constant and changes according to the age of the concrete with values typically falling within a range of approximately 1.3– 1.6. As shown, the variation of this ratio is relatively high in the first 8 days (192 h). After this initial period, the ratio remains almost constant varying within a smaller range of 1.5–1.6. These results are in agreement to what is shown in other studies. Sharma and Gupta [19], for instance, present a relationship for high strength concrete in which, if the values registered in the present study are substituted, the results of Ed/Es would be equal to 1.44.
Other researches, such as [17], found that this relationship varies from 1.32 to 1.39 for masonry mortar. 5.3. Final remarks The relationship of the dynamic E-modulus as a function of time given by Eq. (7) is further studied to understand its capability for providing predictive results of compressive strength. For this, a new expression to predict Ed at 28 days (Ed28) in function of a measured value in an instant ti (Ed(ti)) is presented in Eq. (8). The relationship of this new expression to compressive strength at 28 days (fc(28)) is given by the substitution of the predicted value of Ed28 into Eq. (5).
Ed28 ¼
Edðti Þ t 0:7 3:2 i
1h t i 672h
ð8Þ
Figs. 13 and 14 compare the accuracy of the formulations formerly presented. Fig. 13 plots Eq. (5) using the experimental data of Ed acquired in the resonance tests at different ages and compare them against the traditional compressive tests. Even though the proposed formulation has in the worst case a 36% of strength underestimation, the results are very promising considering that they are in the conservative side, only non-destructive testing were used, and an approximation up to 15% was achieved. This Figure also shows the correlation of other formulations that relates Es and the compressive strength. For this, the input data was considered as the continuous Es results of the EMM-ARM method. It is highlighted that the proposed formulation gives similar results to the correlation proposed by [20]. Fig. 14, on the other hand, plots the values of fc(28) provided by Eq. (5) and uses as input data the set of predicted values of Ed28 obtained from all the measured data at different instants ti. Similar to what was registered before, a
Fig. 11. Predicted evolution of static and dynamic E-modulus.
189
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
Fig. 12. Evolution with time of the ratio between dynamic and static E-modulus.
Compressive Strength (MPa)
3d
7d
14d
28d
40
Curing time (days)
30
20 Experimental results Predicted values ACI [20] EN-1992-1-1 [21] A BNT NBR [22]
10
0 0
96
192
288
384
480
576
Curing time (hours) 672
Fig. 13. Comparison of the development of concrete strength in time according to different predictive formulations (See above-mentioned references for further information.).
Compressive strength (MPa)
40
30
20
Prediction
10
Minimun strength registered at 28 days
0 0
96
192
288
384
480
576
672
Age of prediction (hours) Fig. 14. Predicted compressive strength with dynamic E-modulus data registered at different ages.
maximum underestimation of 37% was obtained but errors of up to 9% in the predictive strength fc(28) were also registered. Once again, the NDT nature of the proposed philosophy is very attractive for practical QC/QA applications.
6. Conclusions This paper illustrates how impact resonance and EMM-ARM NDT testing techniques can be successfully applied for QC/QA efforts of concrete. The paper first presents results from an experimental component involving impact resonance NDT that was successfully used for QC/QA of concrete from a wide variety of construction projects from Lima, Peru. The results from this first component showed that the dynamic E-modulus (Ed) from this NDT technique can be used reasonably well to predict the compressive strength of concrete. The second component involved the EMM-
ARM continuous monitoring which was successfully used to monitor the evolution of stiffness of concrete as a function of curing time. The experimental results from this study were also used to investigate the relationship of the static E-modulus of concrete (Es), defined herein as a chord or secant modulus, and the Ed, which corresponds to the stiffness at very low strain levels or tangential. The experimental results show that the ratio Ed/Es is not constant and depends on the curing time elapsed from casting. The variation of this ratio is relatively large during the first 8 days of curing but then tends to stabilize and remains almost constant beyond this time. Furthermore, a new relationship for estimating the compressive strength of plain concrete as a function of Ed based on impact resonance tests was developed (Eq. (5)). The expression was found to fit in the lower limit of a confidence interval of 90% which is considered an acceptable agreement given the wide range of projects
190
R. Aguilar et al. / Construction and Building Materials 104 (2016) 181–190
and mix designs used for the concrete cylinders used in this study. It is important to point out that the correlation equation should be limited to concrete strengths of up to 50 MPa. This relationship, although shows an interesting practical application for field QC/ QA applications, should be used with caution and the user understand its limitations. Specific correlations can be developed for larger projects which can consider their particular conditions such as mix designs. The information of the evolution with curing time of the Es and Ed was obtained with the impact resonance and the EMM-ARM. A relationship was proposed (Eq. (7)) which can be used to estimate the expected E28-modulus with experimental data. This equation, combined with the relationship of compressive strength and Ed, can be used to estimate the 28 day concrete strength using Ed registered at any time. The application of this methodology evidences a very good agreement to the experimental data showing conservative predictions of compressive strengths at 28 days. Future studies should be aimed at improving the experimental programs in low and high strength concretes in order to get general relationships that will complement the models proposed in the present study. These studies should also consider the differentiation of the testing programs according to the type of cement and aggregates used. The relationship of the Es and compressive strength should also be further addressed following the same orientation but considering broader experimental programs with different types of cements, concrete mixes, and curing ages. Acknowledgments This research was carried out by the Engineering & Heritage research group at the PUCP with collaborations from the Department of Structural Engineering of the University of São Paulo at São Carlos and the Department of Civil and Environmental Engineering of the University of North Carolina at Charlotte. The authors would like to thank the fellowship funding for postgraduate studies for the second author by CONCYTEC, PERU. The authors also express their gratitude to Mauricio Gonzales who helped carrying out many of the experimental tests that are shown in this article. Finally, the authors acknowledge the support of the Structures Laboratory of PUCP (LEDI) during the performance of several material tests carried on the concrete samples retrieved from different construction sites.
References [1] ASTM C39/C39M-12a. Standard Test Method for Compressive Strength of Cylindrical Concrete Specimens. Philadelphia, 2012. [2] P. Mehta, Concrete: Microstructure, Properties, and Materials, third ed., McGraw-Hill, New York, 2006. [3] ASTM C469-02. Standard test method for static modulus of elasticity and Poisson ratio of concrete in compression. Philadelphia, 2002. [4] ASTM C215-08. Standard test method for fundamental transverse, longitudinal, and torsional resonant frequencies of concrete specimens. Pennsylvania, 2008. [5] M. Azenha, F. Magalhães, R. Faria, A. Cunha, Measurement of concrete Emodulus evolution since casting: a novel method based on ambient vibration, Cement Concr Res 40 (2010) 1096–1105. [6] ASTM C1231/C1231M-14. Standard Practice for Use of Unbonded Caps in Determination of Compressive Strength of Hardened Concrete Cylinders. Pennsylvania, 2014. [7] ASTM C33/C33M. Standard Specification for Concrete Aggregates, West Conshohocken, PA, 2013. [8] W. Fuller, S. Thompson, The laws of proportioning concrete, Trans. Am. Soc. Civil Eng. 59 (2) (1907) 67–143. [9] ASTM C138/C138M-14. Standard test method for Density (Unit Weight), Yield, and Air Content (Gravimetric) of Concrete. Philadelphia, 2014. [10] ASTM C1064/C1064M-12. Standard test method for Temperature of Freshly Mixed Hydraulic-Cement Concrete. Philadelphia, 2012. [11] P.D. Welch, The use of fast Fourier transforms for the estimation of power spectra: a method based on time averaging over short modified periodograms, IEEE Trans. Audio Electroacoust. 15 (1967) 70–73. [12] G. Batchelder, D. Lewis, Comparison of dynamic methods of testing concretes subjected to freezing and thawing, Proc. Am. Soc. Testing Mater. 53 (1953) 1053. [13] M. Azenha, L.F. Ramos, R. Aguilar, J.L. Granja, Continuous monitoring of concrete E-modulus since casting based on modal identification: a case study, Cement Concr. Compos. 10 (2012) 10–15. [14] Labview, LabVIEW User Manual, Release 8.0 , National Instruments Corporation, USA, 2006. [15] A. Chopra, Dynamics of Structures, vol. 3, Prentice Hall, New Jersey, 1995. [16] V. Malhotra, V. Sivasundaram, Resonant frequency methods, in: V.M. Malhotra, N.J. Carino (Eds.), Handbook on Non-Destructive Testing of Concrete, CRC Press LLC, Florida, 2004. [17] V. Haach, R. Carrazedo, L. Oliveira, M. Corrêa, Application of acoustic tests to mechanical characterization of masonry mortars, NDT & E Int (2013). [18] S.H. Kosmatka, Properties and performance of normal-strength and highstrength Concrete, in: E.G. Nawy (Ed.), Concrete Construction Engineering Handbook, CRC Press, Boca Raton, FL, 1997, pp. 5-1–5-44. [19] M. Sharma, B. Gupta, Sonic modulus as related to strength and static modulus of high strength concrete, Indian Concr. J. (Bombay) 34 (4) (1960) 139. [20] ACI, Building Code Requirements for Reinforced Concrete and Commentary. ACI 318-95/ACI 318R-95, American Concrete Institute, P.O. Box 9094 Farmington Hills, MI 48333, 1995. [21] Eurocode2. Design of Concrete Structures – Part 1: General Rules and Rules for Buildings, 2004, p. 225. [22] ABNT NBR 6118. Projeto de Estructuras de Concreto - Procedimiento. Associacao Brasileira de Normas Técnicas, 2004, p. 23.