- Email: [email protected]
Practical method for in-place density measurement of cold in-place recycling mixtures
Construction and Building Materials 227 (2019) 116731
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Practical method for in-place density measurement of cold in-place recycling mixtures Peter E. Sebaaly ⇑, Jorge A.C. Ortiz, Adam J. Hand, Elie Y. Hajj Pavement Engineering and Science Program, Department of Civil and Environmental Engineering, University of Nevada Reno, USA
h i g h l i g h t s In-place density is a critical property of CIR pavements for good long-term performance. There is no standard procedure for measuring the in-place density of the compacted CIR during construction. The sand cone method proved reliable for measuring the in-place density of the CIR layer during construction. The sand cone method must be coupled with the parafilm method for measuring the density of filed cores.
a r t i c l e
i n f o
Article history: Received 23 May 2019 Received in revised form 13 August 2019 Accepted 15 August 2019
Keywords: Cold in-place recycling In-place density Sand cone
a b s t r a c t Cold in-place recycling (CIR) has been used for the past two decades as an effective rehabilitation technique for deteriorated asphalt concrete (AC) pavements. The CIR process consists of milling the top portion of the deteriorated AC layer, mix the reclaimed asphalt pavement (RAP) with asphalt emulsion and additive such as lime or portland cement, and compact. The generated CIR layer serves as a good base under the new AC overlay. The most critical property of the CIR layer is its excellent resistance to reflective cracking which makes it a highly effective rehabilitation treatment for cracked AC pavements. Other rehabilitation alternatives such as straight AC overlay would suffer from reflective cracking at the rate of 25 mm per service year. There are five parts of the CIR process; 1) milling, 2) mixing, 3) laydown, 4) compaction, and 5) overlay. Parts 1, 2, 3, and 5 have been well established while part 4 is still requires major improvements. Up to this date, the measurement of the in-place density of the compacted CIR layer is un-achievable. In the case of the AC layer, the in-place density during compaction is measured through a nuclear density gauge that has been calibrated against cores cut from the compacted mat. The application of the same process for the CIR layer faces major limitations: a) in-ability of the nuclear gauge to measure density due to excess water and b) cores cannot be cut from the CIR mat until the 14 days curing has occurred. This paper describes a new method to measure the in-place density of the CIR layer during the compaction process. The new method uses the Sand Cone apparatus to measure the bulk density of the compacted CIR layer. The proposed method has been validated through laboratory testing and field evaluation. Published by Elsevier Ltd.
1. Introduction Cold in-place recycling (CIR) is a pavement treatment equivalent to the re-construction of the upper portion of the asphalt concrete (AC) layer without the application of heat. CIR is developed using reclaimed asphalt pavement (RAP) and an additive or a combination of additives, such as asphalt emulsion, foamed asphalt, lime or portland cement. Up to 200 mm of top portion of the ⇑ Corresponding author. E-mail address: [email protected] (P.E. Sebaaly). https://doi.org/10.1016/j.conbuildmat.2019.116731 0950-0618/Published by Elsevier Ltd.
deteriorated AC layer has been subjected to CIR. In most cases, it is critical that a minimum of 50 mm of the existing AC layer remains in place to provide stable support for the construction equipment and activities. CIR can be used to fix most types of pavement distresses and functions as an effective method for delaying reflective cracking [1]. The top portion of the asphalt concrete layer is pulverized and reused in-place. CIR reuses 100% of the RAP generated in the process without hauling the materials off the project site. This can lead to economic savings and a reduction in the environmental impact of a paving project.
2
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
2. Background The CIR construction process can be summarized by the following major steps: 1. 2. 3. 4. 5.
Milling and crushing of the existing AC layer Addition and mixing of a recycling agent and additive Lay down In-place compaction Placement of the overlay/wearing course
Step 4 has caused some serious problems with the construction and long-term performance of CIR projects throughout the U.S. Inplace compaction on AC pavement projects is controlled through the measurement of the density of the compacted layer and the calculation of the in-place air voids. In the case of AC mix, inplace density is measured by using nuclear density gauges calibrated with field cores cut from the compacted mat. This procedure is commonly used and well defined by agencies specifications. This same process has two serious limitations on CIR projects: a) absence of a calibrated nuclear density gauge due to the inability of cutting cores from the CIR layer before the complete curing of the CIR mix (i.e., 13–15 days) and b) the imprecise measurements of the nuclear density gauge due to the high moisture content of the CIR mixture. These limitations have led to the construction of CIR projects without effective control of the in-place compaction, affecting the long-term performance of the pavement. Industry experience and the literature both suggest that low density will lead to poor in-place material quality and thus poor performance [2,3]. This is supported by the recently completed NCHRP Project 09-51, which showed some correlation between density and stiffness properties of CIR existed [4]. Stiffness generally increases as density increases. However, stiffness can continue to increase as a function of time after final compaction, while density is not changing since no additional compaction occurs but curing continues [5,6]. In 2014, Sebaaly evaluated the impact of high in-place air voids on the resistance of CIR mixtures to moisture damage [7]. The study concluded that constructing a CIR layer without monitoring the in-place compaction could lead to in-place air voids in the range of 20–25%. This high range of in-place air voids would lead to serious early moisture damage, even with the presence of lime slurry. In other words, the addition of lime slurry could not overcome the negative impact of high in-place air voids. The guidelines of the Asphalt Recycling and Reclaiming Association (ARRA) on quality control and testing of CIR projects states the following [3]: ‘‘Target densities for recycled mix compaction are established using control strips as per ARRA CR101 or ARRA CR102. The compacted density is measured with a nuclear density/moisture gauge in accordance with AASHTO T 355 (ASTM D2950) or local agency approved method, since it is generally not possible to obtain cores during construction. The density obtained will be a ‘‘wet density” as conversion to a true ‘‘dry density” by the gauge is not possible with CR mixes. A dry density may be obtained by sampling the recycled mix at the nuclear gauge test location, determining the moisture content by drying and correcting the gauge wet density using the sample moisture content.” This advice confirms the inappropriateness of using the nuclear density gauge to monitor the in-place density of the CIR layer during the compaction process. In 2017, Teshale et al. evaluated the impact of air voids on the cracking resistance of CIR mixtures through the semi-circular bend (SCB) and the disk-shaped compact tension (DCT) tests [8]. The researchers concluded that the air voids content of the CIR sample has a significant impact on the cracking resistance of the CIR mix. As the air voids content increased from 12 to 16%, the fracture
properties of the CIR mix based on both the SCB and DCT tests were reduced by approximately 30%. The researchers concluded that SCB and DCT fracture parameters of the CIR mix showed strong inverse correlations to the air voids content. The primary objectives of on-going NCHRP Project 09-62 are to develop: (1) time-critical tests for asphalt-treated recycled materials and (2) a guide specification using these tests for process control and product acceptance that provides the agency with a basis for determining when the pavement can be opened to traffic and surfaced [9]. As part of this project, a stakeholder survey was conducted where sixty-three US/Canadian state/provincial and local agency specifications were reviewed [10]. By far the most common acceptance requirement was in-place density, which appeared in 98 percent of agencies CIR specifications reviewed. Density specification requirements ranged from 83 to 98 percent of a reference density (mix design or maximum test strip density). Based on the above cases, finding a practical method for inplace density determination during construction can improve the long-term pavement performance of CIR projects leading to longlasting pavement structures. 3. In-place density measurement techniques The selected test methods for estimating the in-place density of the CIR layer evaluated in this study were; the Sand Cone method (ASTM-D1556) and the Rubber Balloon method (ASTM-D2167). The two selected methods are commonly used to determine the unit weight and density of soils. 3.1. Sand Cone method The Sand Cone method is used to determine the in-place density and unit weight of soils as per ASTM D1556. The test method uses a sand cone density apparatus, which consists of a sand container, a sand cone (metal funnel with a valve), and a base plate. The sand used for the test must be clean, dry, with uniform density and grading. Fig. 1 illustrates the equipment used for the Sand Cone test method. The Sand Cone test method consists of the following steps: 1. Fill the excavated hole with free-flowing sand of a known density 2. The volume of the excavated hole is calculated from the mass of sand to fill the hole 3. The in-place wet density of the CIR mix is calculated as the wet mass of the removed material divided by the volume of the hole 4. The dry density of the in-place CIR mix is calculated using the wet mass of the material, the water content, and the volume of the hole 3.2. Rubber Balloon method The Rubber Balloon method is used to determine the in-place density and unit weight of compacted or firmly bonded soils as per ASTM D2167. The test method uses a rubber balloon density apparatus which consists of a calibrated vessel containing water within a thin, flexible, elastic membrane (rubber balloon), designed for measuring the volume of the hole, and a base plate. Fig. 1 illustrates the equipment used for the Rubber Balloon test method. The Rubber Balloon test method consists of the following steps: 1. Before excavating the hole, the operational pressure (determined during the calibration process) must be applied to the test location (i.e., should be reasonably plane and level) to measure the initial volume reading
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
3
Fig. 1. Sand Cone apparatus (left) Rubber Balloon apparatus (right).
2. The apparatus is placed on the excavated hole and the operational pressure is applied to completely fill the hole with the balloon 3. The total volume of the hole is the difference between the final and the initial volume readings 4. The in-place wet density is calculated as the wet mass of the removed material divided by the volume of the hole 5. The dry density of the in-place CIR mix is calculated using the wet mass of the material, the water content, and the volume of the hole 4. Experimental plan An experimental plan was developed to evaluate the applicability of the two test methods for measuring the in-place density of the CIR layer during construction. CIR slabs were produced in the laboratory to perform the two selected test methods. Core samples were drilled out of the slabs and used to measure the actual density of the compacted CIR layer. The measured cores densities were compared with densities estimated from the two selected in-place test methods. In total, seven slabs were fabricated and tested using
different CIR mix designs with the following variations; a) three asphalt emulsion types (A, B, and C), b) lime slurry percent (4.5% and 6.0%), and c) the in-place density testing time after compaction. The execution of the experimental plan is described in the following sections. 4.1. Preparation of the CIR mix The first step was to mix the RAP material with the asphalt emulsion and lime slurry to produce the CIR mix. This process was completed using a concrete mixer, as shown in Fig. 2. The times recommended by the mix design were followed; first mix the RAP with the lime slurry (4.5% or 6.0%) for 2 min, then add more water (1.5%) and mix for 1 min, and finally add the asphalt emulsion per the mix design and mix for 1 min. 4.2. Compaction of the CIR slab Once the CIR mix was prepared, the next step was to compact the CIR slab. The CIR mix was dumped inside the wooden frame and spread uniformly. The slabs wooden frame is a 107 cm square
Fig. 2. Laboratory CIR mix production and compaction using concrete mixer (left) and vibratory plate compactor (right).
4
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
with 64 or 90 mm thickness. A vibratory plate compactor was used to densify the CIR slabs as shown in Fig. 2. 4.3. In-place density measurements The next step after compaction was to perform the in-place density testing on the compacted CIR slab. The testing plan is shown in Fig. 3, where the small circles represent the locations of the Sand Cone and Rubber Balloon tests and the big circles represent locations for the cores. In-place density testing were performed 16 times per slab, eight Sand Cone tests and eight Balloon tests. Sand Cone and Balloon density tests were conducted at the same locations using the same hole to perform both tests. Table 1 presents a summary of the seven slabs tested for in-place
Fig. 3. Density testing plan for the laboratory CIR slabs.
density using the Sand Cone and Rubber Balloon methods and for obtaining cores. The in-place density testing for the first three slabs were performed on multiple days following compaction. One in-place density measurement of each test method was conducted on the slab on multiple days; starting on the compaction day and finishing on the 14th day after compaction. The days the tests were performed on the first three slabs were; 0, 1st, 2nd, 3rd, 6th, 9th, 12th and 14th after compaction. Table 2 presents an example of the in-place density test results for slab 1. The data showed that there is no significant difference between performing the in-place density tests on multiple days after compaction. Since the focus of this project was to determine the in-place density for a CIR field project immediately following compaction, it was decided to perform all the in-place density testing on the remaining four slabs one hour after compaction. It is important to note the practical limitations for each of the in-place density tests. For the Sand Cone test method, the standard (ASTM D1556) states that for a mixture with maximum particle size of 25 mm the minimum test hole size is 2000 cm3. With a base plate diameter of 150 mm and the depths of the CIR layers ranged between 65 and 90 mm, the volume of the holes varied approximately between 1150 and 1600 cm3 which is lower than the minimum requirement. The same problem was found for the Balloon test method, in which the standard (ASTM D2167) states that for a mixture with maximum particle size of 25 mm the minimum test hole size is 2000 cm3. With a base plate diameter of 100 mm and the depths of the CIR layers ranged between 65 and 90 mm, the volume of the holes varied approximately between 510 and 705 cm3, which is lower than the minimum requirement. To increase the volume of the excavated hole to be closer to the specification limits, it was decided to compact the first two slabs to a thickness of 90 mm. However, the thicker slabs and larger excavated hole did not improve the density results while still below the minimum requirement. Finally, since it was not possible to achieve
Table 1 Summary of the CIR Slabs Evaluated in the Laboratory Experiment. Slab
Emulsion
Slurry Lime (%)
Thickness (mm)
Multi-Day Testing
1 2 3 4 5 6 7
Type Type Type Type Type Type Type
6.0 4.5 4.5 6.0 4.5 4.5 4.5
90 90 64 64 64 75 75
Yes Yes Yes No No No No
B B B C C B C
Table 2 Balloon and Sand Cone Test Results for Laboratory Slab 1. Rubber Balloon
1
2
3
6
9
12
14
Wet Density, g/cm3 Dry Density, g/cm3 Air Voids, % Wc, %
1.94 1.83 22.0 6.0
1.99 1.91 18.3 4.3
1.95 1.87 20.3 4.3
2.02 1.93 17.4 4.2
1.92 1.87 20.0 2.5
2.00 1.95 16.7 2.5
1.99 1.95 16.8 1.9
1.91 1.87 20.0 2.0
Sand Cone
Day
Wet Density, g/cm3 Dry Density, g/cm3 Air Voids, % Wc, %* *
Day 0
0
1
2
3
6
9
12
14
2.03 1.92 17.9 5.81
1.90 1.82 22.3 4.59
1.93 1.85 21.0 4.39
2.04 1.96 16.3 4.23
1.98 1.93 17.7 2.68
1.98 1.93 17.4 2.50
1.97 1.93 17.4 2.10
1.95 1.91 18.3 2.01
Wc stands for Water Content.
Avg.
Std. Dev.
CV %
1.96 1.90 18.9
0.04 0.05 0.019
2.0 2.4 10.1
Avg.
Std. Dev.
CV %
1.97 1.91 18.5
0.05 0.05 0.020
2.4 2.5 11.0
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
5
the required minimum volume on a typical CIR mat thickness, it was decided to perform the testing with this limitation. 4.4. Obtaining cores The last step was to cut cores from the compacted CIR slabs and determine their densities in the laboratory. Once the CIR mixture was fully cured, 14 days after compaction, cores were cut from the slabs. The cores were cut right next to each excavated hole where the in-place density tests were performed, as shown in Fig. 4. It was assumed that, since the core location was close enough to the in-place density test location, the density at both locations would be the same. Since the cores were irregularly shaped with relatively high air voids, two test methods were performed to obtain their bulk specific gravity: Parafilm method (ASTM D1188) and Corelok method (ASTM D6752). All cores were tested following the two standards and the results were compared as shown in Fig. 5. The trends in the measured data are very similar to the results reported by Zhang et al. in 2014, where the air voids obtained from the parafilm method are about one percent higher than the air voids obtained from the Corelok method [11].
5. Analysis of in-place density test results The percent air voids in the CIR cores were calculated per ASTM D3203 using the measured maximum theoretical specific gravity (Gmm) of the loose CIR mix per ASTM D2401 and the bulk specific gravity (Gmb) of the compacted CIR mix (i.e., cores) per ASTM D1188 and D6752. The air voids contents obtained from the cores using the two bulk specific gravity test methods were compared with the estimated air voids contents obtained from the two inplace test methods conducted on the CIR slabs. In-place air voids contents obtained from the Sand Cone and Rubber Balloon methods were compared with the air voids contents of the cores cut from the close-by location. As an example, Fig. 6 shows the comparison between the air voids content obtained from Sand Cone and Balloon methods and the air voids content obtained from the cores using the parafilm method for Slab 1. Figs. 7 and 8 show the overall summary of the data obtained from the seven slabs for the parafilm and Corelok methods, respectively. It is interesting to note that the majority of the data are above the equality line, which shows that in most of the cases the air voids content estimated from the in-place test methods is lower than the actual air voids content obtained from the cores.
Fig. 5. Relationship between air voids from parafilm and Corelok methods based on laboratory slabs.
5.1. Statistical analysis The Micro-Macro Multilevel Modeling package in the R statistical software was used to assist in identifying an appropriate method for measuring the in-place density of the compacted CIR layer [12]. An outlier detection method for the entire data set was used prior to performing the statistical analysis. The Box plot was selected as the outlier detection method for the entire data set. The inter quartile range (IQR) was used, which is the difference between the first and the third quartile. This method defines as an outlier any data point that is 1.5 times the IQR below the first quartile and 1.5 times the IQR above the third quartile. Outliers were detected using the difference between the air voids obtained from the cores (parafilm and Corelok) and the estimated in-place air voids by the two methods (Sand Cone and Balloon). The quartiles were calculated for each of the differences combinations (Balloon-parafilm, Balloon-Corelok, Sand Coneparafilm, Sand Cone-Corelok), and the IQR was estimated, as summarized in Table 3. Any data point above and below the range determined (1.5 times the IQR below the first quartile and above the third quartile) was considered an outlier and removed from the data set. There were no outliers detected for the data sets that estimated the air voids using the Balloon method, due to the high variability of the method, which led to a higher IQR value. 5.1.1. Modeling of the air voids data The general trend observed in the data set was that the air voids obtained from the cores, regardless of the method used for bulk specific gravity (parafilm or Corelok), were higher than the air
Fig. 4. Coring of the compacted and cured laboratory CIR slabs.
6
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
Fig. 6. Comparison of estimated air voids vs cores air voids using parafilm on laboratory slab 1.
Fig. 7. Comparison of estimated air voids vs cores air voids using parafilm for all 7 laboratory slabs.
Fig. 8. Comparison of estimated air voids vs cores air voids using Corelok for all 7 laboratory slabs.
voids estimated by both methods (Sand Cone or Rubber Balloon). This general trend can be observed in Fig. 9 comparing the estimated air voids from Sand Cone and Balloon methods with the core air voids obtained using the parafilm method. Based on these observations, it was decided to apply a correction factor to the air voids obtained from the estimation method improve the fit with the data obtained from the cores. The applied correction factor shifts the equality line of the data set for each combination. A linear regression analysis was selected
to determine the magnitude of the correction factor. The equation for the linear regression was as follows:
y ¼ b0 þ b1 x Since the equality line has a slope of one, then b0 = 0 and b1 = 1. The corrected equality line would have the same slope value as the equality line (b1 = 1) and would be described by:
y ¼ b0 þ x
7
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731 Table 3 Quartile and IQR Calculations for Outliers Identification for Laboratory Data.
Min. value 25th percentile 50th percentile 75th percentile Max. value IQR 1.5*IQR 1st-1.5*IQR 3rd + 1.5*IQR Number of Data Points Number of Outliers
Quartile
Balloon-Parafilm (%)
Balloon-Corelok (%)
Sand Cone-Parafilm (%)
Sand Cone-Corelok (%)
0 1 2 3 4
4.3 0.5 2.1 4.1 8.2 3.7 5.5 5.0 9.6 49 0
5.8 0.8 0.8 3.4 7.2 4.2 6.3 7.1 9.6 49 0
3.1 1.4 2.6 4.1 8.7 2.7 4.0 2.6 8.1 49 3
4.6 0.5 1.6 3.1 7.1 2.6 3.9 3.4 7.0 49 4
Fig. 9. Estimated air voids (Sand Cone and Balloon methods) vs air voids of the cores from laboratory slabs using parafilm.
b0 ¼ y x Each combination of test method and core method, for example Sand Cone with parafilm, has its own correction factor for the shifted equality line according to the average of air voids from the data set obtained in this study, as shown in Figs. 10–13. Following this procedure, the correction factors were determined as follows: The correction factor for the air voids obtained with the Rubber Balloon method when compared to the cores air voids using the parafilm method is: 2.4%.
The correction factor for the air voids obtained with the Rubber Balloon method when compared to the cores air voids using the Corelok method is: 1.2%. The correction factor for the air voids obtained with the Sand Cone method when compared to the cores air voids using the parafilm method is: 2.8%. The correction factor for the air voids obtained with the Sand Cone method when compared to the cores air voids using the parafilm method is: 1.9%. Following the correction stage, a statistical analysis was conducted on the data set of each combination to determine if the esti-
Fig. 10. Corrected equality line for estimated air voids (Balloon method) vs air voids of the cores from laboratory slabs using parafilm.
8
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
Fig. 11. Corrected equality line for estimated air voids (Balloon method) vs air voids of the cores from laboratory slabs using Corelok.
Fig. 12. Corrected equality line for estimated air voids (Sand Cone method) vs air voids of the cores from laboratory slabs using Parafilm.
Fig. 13. Corrected equality line for estimated air voids (Sand Cone method) vs air voids of the cores from laboratory slabs using Corelok.
mated air voids data obtained using the Sand Cone and Rubber Balloon methods are comparable to the air voids obtained from the cores. The selected statistical tool used to develop the analysis was the t-test, which is commonly used for testing the difference between two data sets with normal distributions and unknown
variances. The t-test is used to determine the probability of difference in the means of two populations. The t-tests were ran using the statistical software R [12]. Since the objective was to determine if the estimated air voids data (Sand Cone and Rubber Balloon) were comparable to the actual
9
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
air voids from the cores, the differences between the cores air voids and the estimated air voids were calculated for the following four combinations: Difference between the cores air voids using parafilm estimated air voids using the Sand Cone method Difference between the cores air voids using Corelok estimated air voids using the Sand Cone method Difference between the cores air voids using parafilm estimated air voids using the Rubber Balloon method Difference between the cores air voids using Corelok estimated air voids using the Rubber Balloon method
Use the 95% Confidence Interval of Difference for the combination of Sand Cone method and the cores with parafilm shown below:
and the
Difference ¼ AV%Parafilm ðAV%SandCone þ CF Þ
and the
Difference ¼ AV%Parafilm ðAV%SandCone þ 2:8%Þ
and the and the
The null hypothesis was that the difference between the cores air voids (parafilm or Corelok) and the corrected estimated air voids (Sand Cone or Balloon) is equal to zero (difference = 0). The t-test was run for each combination. If the parameter p-value from the analysis is lower than 0.05 (p < 0.05), the difference between the cores air voids and the estimated air voids is different than zero (i.e., difference – 0). If the parameter p-value from the analysis is higher than 0.05 (p > 0.05), then there is no significant evidence to prove that the difference between the cores air voids and estimated air voids is different than zero (i.e., difference = 0). The objective of the analysis was to compare the four combinations and to select the most suitable method for estimating the inplace density of the CIR layer during compaction. As shown in Table 4 for the four combinations, there is not significant evidence to prove that the difference between the cores air voids and the corrected estimated air voids is different than zero since all pvalues are above 0.05. The 95% Confidence Interval for each combination was determined from the following relationship:
r
CI95% ¼ x 1:96 pffiffiffi n where ‘‘r” is the standard deviation and ‘‘n” is the population. The lower the standard deviation indicates lower variability of the method and identifies the best alternative to determine the estimated in-place air voids. From the statistical analysis of the data generated in this experiment as summarized in Table 4, it was found that the best approach to estimate the in-place air voids is to use the Sand Cone method and compare the results with the cores air voids using the parafilm method for measuring bulk specific gravity. As can be seen, the selected alternative has the lowest standard deviation of 2.377. The 95% Confidence Interval of Difference provides the range over which there is 95% confidence that the mean of the differences will be contained.
0:9% < Difference < 0:5% Apply the above relationship as follows: o Use the sand cone method to measure the in-place air voids of the compacted CIR layer as described in this paper o Add 2.8% to the in-place air voids determined from the Sand Cone measurement to calculate the adjusted in-place air voids o There is 95% confidence that the adjusted in-place air voids will be: 0.9 to +0.5 from the actual in-place air voids measured on cores using the parafilm method 7. Field verification In 2017, a field verification of the in-place density methods was conducted on a CIR construction project located on US-95 south of Fernley, Nevada. In-place density was measured using the Sand Cone and Rubber Balloon methods on the CIR construction project, as shown in Fig. 14. Testing was conducted at two different locations within the project. At each location, three in-place density tests, separated by 50 feet, were conducted using each method. The density testing (Sand Cone and Balloon methods) were conducted after the rolling pattern was completed and the full compaction was considered achieved. Cores were cut at the same locations of the in-place density measurements 14 days after compaction.
6. Recommended approach Based on the statistical analysis of the data generated from the laboratory testing on CIR slabs using the Sand Cone and Rubber Balloon methods and the measurements of the air voids on cores using the parafilm and Corelok methods, the following process is recommended to estimate in-place air voids of the compacted CIR layer.
Fig. 14. Balloon (left) and Sand Cone (right) test methods in-place at US-95 CIR project in Fernley, NV.
Table 4 T-test Developed on the Differences of the Air Voids for All the Combinations based on Laboratory Data. T-test
Balloon-Parafilm
Balloon-Corelok
Sand Cone-Parafilm
Sand Cone-Corelok
n t p-value 95% Confidence Interval Standard Deviation
49 0.658 0.514 [1.76 to 0.89] 4.63
49 0.003 0.998 [0.80 to 0.81] 2.80
49 0.667 0.508 [0.91 to 0.46] 2.38
49 1.037 0.305 [1.09 to 0.35] 2.51
10
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
Table 5 Verification of the 95% Confidence Interval of Difference for Air Voids Estimation using Sand Cone Method on US-95 CIR Project. Sand Cone - Parafilm Avg. AV: Sand Cone (%) Corrected Avg. AV: Sand Cone (%) Avg. AV: Cores (parafilm) (%) Difference (%) 95% Confidence Interval
Sand Cone - Corelok 19.4 22.3 22.0 0.3 [0.91% to 0.46%]
OK
Avg. AV: Sand Cone (%) Corrected Avg. AV: Sand Cone (%) Avg. AV: Cores (Corelok) (%) Difference (%) 95% Confidence Interval
19.4 21.3 20.0 1.3 [1.09% to 0.35%]
NO
Table 6 Verification of the 95% Confidence Interval of Difference for Air Voids Estimation using Rubber Balloon Method on US-95 CIR Project. Balloon - Parafilm Avg. AV: Balloon (%) Corrected Avg. AV: Balloon (%) Avg. AV: Cores (parafilm) (%) Difference (%) 95% Confidence Interval
Balloon - Corelok 22.7 25.1 22.0 3.1 [1.76% to 0.89%]
NO
The verification of the 95% Confidence Interval of Difference for each method was conducted by comparing the air voids obtained from the Sand Cone and Rubber Balloon methods and the air voids obtained from the cores using the parafilm and the Corelok methods, as shown in Tables 5 and 6. The data in Table 5 show that the 95% Confidence Interval of Difference was met for the Sand Cone method using the parafilm method on the cores, but it was not met for the Sand Cone method using the Corelok method on the cores. The difference between the average air voids from the cores using parafilm (22.0%) and the average air voids from Sand Cone plus the correction factor (19.4% + 2.8%) is inside the 95% Confidence Interval of Difference. The data in Table 6 show the air voids estimated by the Rubber Balloon method are higher than the air voids measured on the cores. This was not the regular trend found on the laboratory testing of the CIR slabs. Therefore, applying the correction factor obtained from the CIR slabs led to higher difference from the cores. This difference in the trend might be due to the higher variability of the Balloon test method. Finally, the results of the field validation trial confirm that the best approach to estimate the in-place air voids of the compacted CIR layer is to use the Sand Cone method and compare the estimated air voids with the air voids obtained from the cores using the parafilm technique. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgement The authors would like to acknowledge the Nevada Department of Transportation for supporting the research project that led to the development of the test method documented in this paper.
Avg. AV: Balloon (%) Corrected Avg. AV: Balloon (%) Avg. AV: Cores (parafilm) (%) Difference (%) 95% Confidence Interval
22.7 23.8 20.0 3.8 [0.80% to 0.81%]
NO
Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.conbuildmat.2019.116731. References [1] P. Murugaiyah, P.E. Sebaaly, E.Y. Hajj, S. Selvaratnam, Evaluation of Long-term Performance of Cold In-Place Recycling Pavements in Nevada Report No. WRSC-UNR-CIR-201402, Nevada Department of Transportation, Carson City, Nevada, 2014. [2] A. Nataatmadja, Some Characteristics of Foamed Bitumen Mixes Transportation Research Record No. 1767, Transportation Research Board, Washington, DC, 2001. [3] Asphalt Recycling and Reclaiming Association (2015). Basic Asphalt Recycling Manual, second ed. Annapolis, MD. [4] C.W. Schwartz, B.K. Diefenderfer, B.F. Bowers, Material Properties of Cold InPlace Recycled and Full-Depth Reclamation Asphalt Concrete Report 863, National Cooperative Highway Research Program, Washington, DC, 2017. [5] G. Thenoux, A. Gonzalez, R. Dowling, Energy consumption comparison for different asphalt pavements rehabilitation techniques used in Chile, Resour., Conserv. Recycling Vol (2007) 49. [6] Wirtgen, Cold Recycling Technology, third ed., Wirtgen GmbH, Windhagen, Germany, 2010. [7] P.E. Sebaaly, CIR Field Project Report, Regional Transportation Commission, Washoe County, Nevada, 2014. [8] E.Z. Teshale, D. Rettner, A. Hartleib, D. Kriesel. Application of Laboratory Asphalt Cracking Tests to Cold In-Place Recycled (CIR) Mixtures, Technical Paper, Journal of the Association of Asphalt Paving Technologists, Vol. 86, 2017. [9] National Academy of Sciences website: NCHRP Project 09-62, https://apps.trb. org/cmsfeed/TRBNetProjectDisplay.asp?ProjectID=4190, 2018. [10] B.K. Diefenderfer, D. Jones, A.J.T. Hand, B.F. Bowers, G. Flintsch, Rapid tests and specifications for construction of asphalt-treated cold recycled pavements available by request at, NCHRP Project 09-62 Phase I Interim Report, NCHRP, Washington, D.C, 2017. https://apps.trb.org/cmsfeed/TRBNetProjectDisplay. asp?ProjectID=4190. [11] Y. Zhang, L. Wang, B. Diefenderfer, W. Zhang, Determining volumetric properties and permeability of asphalt-treated permeable base mixtures, Int. J. Pavement Eng. 17 (2014). [12] R Core Team (2017). R: A language and environment for statistical computing – Version R3.4.1. R Foundation for Statistical Computing, Vienna, Austria.
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Practical method for in-place density measurement of cold in-place recycling mixtures Peter E. Sebaaly ⇑, Jorge A.C. Ortiz, Adam J. Hand, Elie Y. Hajj Pavement Engineering and Science Program, Department of Civil and Environmental Engineering, University of Nevada Reno, USA
h i g h l i g h t s In-place density is a critical property of CIR pavements for good long-term performance. There is no standard procedure for measuring the in-place density of the compacted CIR during construction. The sand cone method proved reliable for measuring the in-place density of the CIR layer during construction. The sand cone method must be coupled with the parafilm method for measuring the density of filed cores.
a r t i c l e
i n f o
Article history: Received 23 May 2019 Received in revised form 13 August 2019 Accepted 15 August 2019
Keywords: Cold in-place recycling In-place density Sand cone
a b s t r a c t Cold in-place recycling (CIR) has been used for the past two decades as an effective rehabilitation technique for deteriorated asphalt concrete (AC) pavements. The CIR process consists of milling the top portion of the deteriorated AC layer, mix the reclaimed asphalt pavement (RAP) with asphalt emulsion and additive such as lime or portland cement, and compact. The generated CIR layer serves as a good base under the new AC overlay. The most critical property of the CIR layer is its excellent resistance to reflective cracking which makes it a highly effective rehabilitation treatment for cracked AC pavements. Other rehabilitation alternatives such as straight AC overlay would suffer from reflective cracking at the rate of 25 mm per service year. There are five parts of the CIR process; 1) milling, 2) mixing, 3) laydown, 4) compaction, and 5) overlay. Parts 1, 2, 3, and 5 have been well established while part 4 is still requires major improvements. Up to this date, the measurement of the in-place density of the compacted CIR layer is un-achievable. In the case of the AC layer, the in-place density during compaction is measured through a nuclear density gauge that has been calibrated against cores cut from the compacted mat. The application of the same process for the CIR layer faces major limitations: a) in-ability of the nuclear gauge to measure density due to excess water and b) cores cannot be cut from the CIR mat until the 14 days curing has occurred. This paper describes a new method to measure the in-place density of the CIR layer during the compaction process. The new method uses the Sand Cone apparatus to measure the bulk density of the compacted CIR layer. The proposed method has been validated through laboratory testing and field evaluation. Published by Elsevier Ltd.
1. Introduction Cold in-place recycling (CIR) is a pavement treatment equivalent to the re-construction of the upper portion of the asphalt concrete (AC) layer without the application of heat. CIR is developed using reclaimed asphalt pavement (RAP) and an additive or a combination of additives, such as asphalt emulsion, foamed asphalt, lime or portland cement. Up to 200 mm of top portion of the ⇑ Corresponding author. E-mail address: [email protected] (P.E. Sebaaly). https://doi.org/10.1016/j.conbuildmat.2019.116731 0950-0618/Published by Elsevier Ltd.
deteriorated AC layer has been subjected to CIR. In most cases, it is critical that a minimum of 50 mm of the existing AC layer remains in place to provide stable support for the construction equipment and activities. CIR can be used to fix most types of pavement distresses and functions as an effective method for delaying reflective cracking [1]. The top portion of the asphalt concrete layer is pulverized and reused in-place. CIR reuses 100% of the RAP generated in the process without hauling the materials off the project site. This can lead to economic savings and a reduction in the environmental impact of a paving project.
2
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
2. Background The CIR construction process can be summarized by the following major steps: 1. 2. 3. 4. 5.
Milling and crushing of the existing AC layer Addition and mixing of a recycling agent and additive Lay down In-place compaction Placement of the overlay/wearing course
Step 4 has caused some serious problems with the construction and long-term performance of CIR projects throughout the U.S. Inplace compaction on AC pavement projects is controlled through the measurement of the density of the compacted layer and the calculation of the in-place air voids. In the case of AC mix, inplace density is measured by using nuclear density gauges calibrated with field cores cut from the compacted mat. This procedure is commonly used and well defined by agencies specifications. This same process has two serious limitations on CIR projects: a) absence of a calibrated nuclear density gauge due to the inability of cutting cores from the CIR layer before the complete curing of the CIR mix (i.e., 13–15 days) and b) the imprecise measurements of the nuclear density gauge due to the high moisture content of the CIR mixture. These limitations have led to the construction of CIR projects without effective control of the in-place compaction, affecting the long-term performance of the pavement. Industry experience and the literature both suggest that low density will lead to poor in-place material quality and thus poor performance [2,3]. This is supported by the recently completed NCHRP Project 09-51, which showed some correlation between density and stiffness properties of CIR existed [4]. Stiffness generally increases as density increases. However, stiffness can continue to increase as a function of time after final compaction, while density is not changing since no additional compaction occurs but curing continues [5,6]. In 2014, Sebaaly evaluated the impact of high in-place air voids on the resistance of CIR mixtures to moisture damage [7]. The study concluded that constructing a CIR layer without monitoring the in-place compaction could lead to in-place air voids in the range of 20–25%. This high range of in-place air voids would lead to serious early moisture damage, even with the presence of lime slurry. In other words, the addition of lime slurry could not overcome the negative impact of high in-place air voids. The guidelines of the Asphalt Recycling and Reclaiming Association (ARRA) on quality control and testing of CIR projects states the following [3]: ‘‘Target densities for recycled mix compaction are established using control strips as per ARRA CR101 or ARRA CR102. The compacted density is measured with a nuclear density/moisture gauge in accordance with AASHTO T 355 (ASTM D2950) or local agency approved method, since it is generally not possible to obtain cores during construction. The density obtained will be a ‘‘wet density” as conversion to a true ‘‘dry density” by the gauge is not possible with CR mixes. A dry density may be obtained by sampling the recycled mix at the nuclear gauge test location, determining the moisture content by drying and correcting the gauge wet density using the sample moisture content.” This advice confirms the inappropriateness of using the nuclear density gauge to monitor the in-place density of the CIR layer during the compaction process. In 2017, Teshale et al. evaluated the impact of air voids on the cracking resistance of CIR mixtures through the semi-circular bend (SCB) and the disk-shaped compact tension (DCT) tests [8]. The researchers concluded that the air voids content of the CIR sample has a significant impact on the cracking resistance of the CIR mix. As the air voids content increased from 12 to 16%, the fracture
properties of the CIR mix based on both the SCB and DCT tests were reduced by approximately 30%. The researchers concluded that SCB and DCT fracture parameters of the CIR mix showed strong inverse correlations to the air voids content. The primary objectives of on-going NCHRP Project 09-62 are to develop: (1) time-critical tests for asphalt-treated recycled materials and (2) a guide specification using these tests for process control and product acceptance that provides the agency with a basis for determining when the pavement can be opened to traffic and surfaced [9]. As part of this project, a stakeholder survey was conducted where sixty-three US/Canadian state/provincial and local agency specifications were reviewed [10]. By far the most common acceptance requirement was in-place density, which appeared in 98 percent of agencies CIR specifications reviewed. Density specification requirements ranged from 83 to 98 percent of a reference density (mix design or maximum test strip density). Based on the above cases, finding a practical method for inplace density determination during construction can improve the long-term pavement performance of CIR projects leading to longlasting pavement structures. 3. In-place density measurement techniques The selected test methods for estimating the in-place density of the CIR layer evaluated in this study were; the Sand Cone method (ASTM-D1556) and the Rubber Balloon method (ASTM-D2167). The two selected methods are commonly used to determine the unit weight and density of soils. 3.1. Sand Cone method The Sand Cone method is used to determine the in-place density and unit weight of soils as per ASTM D1556. The test method uses a sand cone density apparatus, which consists of a sand container, a sand cone (metal funnel with a valve), and a base plate. The sand used for the test must be clean, dry, with uniform density and grading. Fig. 1 illustrates the equipment used for the Sand Cone test method. The Sand Cone test method consists of the following steps: 1. Fill the excavated hole with free-flowing sand of a known density 2. The volume of the excavated hole is calculated from the mass of sand to fill the hole 3. The in-place wet density of the CIR mix is calculated as the wet mass of the removed material divided by the volume of the hole 4. The dry density of the in-place CIR mix is calculated using the wet mass of the material, the water content, and the volume of the hole 3.2. Rubber Balloon method The Rubber Balloon method is used to determine the in-place density and unit weight of compacted or firmly bonded soils as per ASTM D2167. The test method uses a rubber balloon density apparatus which consists of a calibrated vessel containing water within a thin, flexible, elastic membrane (rubber balloon), designed for measuring the volume of the hole, and a base plate. Fig. 1 illustrates the equipment used for the Rubber Balloon test method. The Rubber Balloon test method consists of the following steps: 1. Before excavating the hole, the operational pressure (determined during the calibration process) must be applied to the test location (i.e., should be reasonably plane and level) to measure the initial volume reading
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
3
Fig. 1. Sand Cone apparatus (left) Rubber Balloon apparatus (right).
2. The apparatus is placed on the excavated hole and the operational pressure is applied to completely fill the hole with the balloon 3. The total volume of the hole is the difference between the final and the initial volume readings 4. The in-place wet density is calculated as the wet mass of the removed material divided by the volume of the hole 5. The dry density of the in-place CIR mix is calculated using the wet mass of the material, the water content, and the volume of the hole 4. Experimental plan An experimental plan was developed to evaluate the applicability of the two test methods for measuring the in-place density of the CIR layer during construction. CIR slabs were produced in the laboratory to perform the two selected test methods. Core samples were drilled out of the slabs and used to measure the actual density of the compacted CIR layer. The measured cores densities were compared with densities estimated from the two selected in-place test methods. In total, seven slabs were fabricated and tested using
different CIR mix designs with the following variations; a) three asphalt emulsion types (A, B, and C), b) lime slurry percent (4.5% and 6.0%), and c) the in-place density testing time after compaction. The execution of the experimental plan is described in the following sections. 4.1. Preparation of the CIR mix The first step was to mix the RAP material with the asphalt emulsion and lime slurry to produce the CIR mix. This process was completed using a concrete mixer, as shown in Fig. 2. The times recommended by the mix design were followed; first mix the RAP with the lime slurry (4.5% or 6.0%) for 2 min, then add more water (1.5%) and mix for 1 min, and finally add the asphalt emulsion per the mix design and mix for 1 min. 4.2. Compaction of the CIR slab Once the CIR mix was prepared, the next step was to compact the CIR slab. The CIR mix was dumped inside the wooden frame and spread uniformly. The slabs wooden frame is a 107 cm square
Fig. 2. Laboratory CIR mix production and compaction using concrete mixer (left) and vibratory plate compactor (right).
4
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
with 64 or 90 mm thickness. A vibratory plate compactor was used to densify the CIR slabs as shown in Fig. 2. 4.3. In-place density measurements The next step after compaction was to perform the in-place density testing on the compacted CIR slab. The testing plan is shown in Fig. 3, where the small circles represent the locations of the Sand Cone and Rubber Balloon tests and the big circles represent locations for the cores. In-place density testing were performed 16 times per slab, eight Sand Cone tests and eight Balloon tests. Sand Cone and Balloon density tests were conducted at the same locations using the same hole to perform both tests. Table 1 presents a summary of the seven slabs tested for in-place
Fig. 3. Density testing plan for the laboratory CIR slabs.
density using the Sand Cone and Rubber Balloon methods and for obtaining cores. The in-place density testing for the first three slabs were performed on multiple days following compaction. One in-place density measurement of each test method was conducted on the slab on multiple days; starting on the compaction day and finishing on the 14th day after compaction. The days the tests were performed on the first three slabs were; 0, 1st, 2nd, 3rd, 6th, 9th, 12th and 14th after compaction. Table 2 presents an example of the in-place density test results for slab 1. The data showed that there is no significant difference between performing the in-place density tests on multiple days after compaction. Since the focus of this project was to determine the in-place density for a CIR field project immediately following compaction, it was decided to perform all the in-place density testing on the remaining four slabs one hour after compaction. It is important to note the practical limitations for each of the in-place density tests. For the Sand Cone test method, the standard (ASTM D1556) states that for a mixture with maximum particle size of 25 mm the minimum test hole size is 2000 cm3. With a base plate diameter of 150 mm and the depths of the CIR layers ranged between 65 and 90 mm, the volume of the holes varied approximately between 1150 and 1600 cm3 which is lower than the minimum requirement. The same problem was found for the Balloon test method, in which the standard (ASTM D2167) states that for a mixture with maximum particle size of 25 mm the minimum test hole size is 2000 cm3. With a base plate diameter of 100 mm and the depths of the CIR layers ranged between 65 and 90 mm, the volume of the holes varied approximately between 510 and 705 cm3, which is lower than the minimum requirement. To increase the volume of the excavated hole to be closer to the specification limits, it was decided to compact the first two slabs to a thickness of 90 mm. However, the thicker slabs and larger excavated hole did not improve the density results while still below the minimum requirement. Finally, since it was not possible to achieve
Table 1 Summary of the CIR Slabs Evaluated in the Laboratory Experiment. Slab
Emulsion
Slurry Lime (%)
Thickness (mm)
Multi-Day Testing
1 2 3 4 5 6 7
Type Type Type Type Type Type Type
6.0 4.5 4.5 6.0 4.5 4.5 4.5
90 90 64 64 64 75 75
Yes Yes Yes No No No No
B B B C C B C
Table 2 Balloon and Sand Cone Test Results for Laboratory Slab 1. Rubber Balloon
1
2
3
6
9
12
14
Wet Density, g/cm3 Dry Density, g/cm3 Air Voids, % Wc, %
1.94 1.83 22.0 6.0
1.99 1.91 18.3 4.3
1.95 1.87 20.3 4.3
2.02 1.93 17.4 4.2
1.92 1.87 20.0 2.5
2.00 1.95 16.7 2.5
1.99 1.95 16.8 1.9
1.91 1.87 20.0 2.0
Sand Cone
Day
Wet Density, g/cm3 Dry Density, g/cm3 Air Voids, % Wc, %* *
Day 0
0
1
2
3
6
9
12
14
2.03 1.92 17.9 5.81
1.90 1.82 22.3 4.59
1.93 1.85 21.0 4.39
2.04 1.96 16.3 4.23
1.98 1.93 17.7 2.68
1.98 1.93 17.4 2.50
1.97 1.93 17.4 2.10
1.95 1.91 18.3 2.01
Wc stands for Water Content.
Avg.
Std. Dev.
CV %
1.96 1.90 18.9
0.04 0.05 0.019
2.0 2.4 10.1
Avg.
Std. Dev.
CV %
1.97 1.91 18.5
0.05 0.05 0.020
2.4 2.5 11.0
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
5
the required minimum volume on a typical CIR mat thickness, it was decided to perform the testing with this limitation. 4.4. Obtaining cores The last step was to cut cores from the compacted CIR slabs and determine their densities in the laboratory. Once the CIR mixture was fully cured, 14 days after compaction, cores were cut from the slabs. The cores were cut right next to each excavated hole where the in-place density tests were performed, as shown in Fig. 4. It was assumed that, since the core location was close enough to the in-place density test location, the density at both locations would be the same. Since the cores were irregularly shaped with relatively high air voids, two test methods were performed to obtain their bulk specific gravity: Parafilm method (ASTM D1188) and Corelok method (ASTM D6752). All cores were tested following the two standards and the results were compared as shown in Fig. 5. The trends in the measured data are very similar to the results reported by Zhang et al. in 2014, where the air voids obtained from the parafilm method are about one percent higher than the air voids obtained from the Corelok method [11].
5. Analysis of in-place density test results The percent air voids in the CIR cores were calculated per ASTM D3203 using the measured maximum theoretical specific gravity (Gmm) of the loose CIR mix per ASTM D2401 and the bulk specific gravity (Gmb) of the compacted CIR mix (i.e., cores) per ASTM D1188 and D6752. The air voids contents obtained from the cores using the two bulk specific gravity test methods were compared with the estimated air voids contents obtained from the two inplace test methods conducted on the CIR slabs. In-place air voids contents obtained from the Sand Cone and Rubber Balloon methods were compared with the air voids contents of the cores cut from the close-by location. As an example, Fig. 6 shows the comparison between the air voids content obtained from Sand Cone and Balloon methods and the air voids content obtained from the cores using the parafilm method for Slab 1. Figs. 7 and 8 show the overall summary of the data obtained from the seven slabs for the parafilm and Corelok methods, respectively. It is interesting to note that the majority of the data are above the equality line, which shows that in most of the cases the air voids content estimated from the in-place test methods is lower than the actual air voids content obtained from the cores.
Fig. 5. Relationship between air voids from parafilm and Corelok methods based on laboratory slabs.
5.1. Statistical analysis The Micro-Macro Multilevel Modeling package in the R statistical software was used to assist in identifying an appropriate method for measuring the in-place density of the compacted CIR layer [12]. An outlier detection method for the entire data set was used prior to performing the statistical analysis. The Box plot was selected as the outlier detection method for the entire data set. The inter quartile range (IQR) was used, which is the difference between the first and the third quartile. This method defines as an outlier any data point that is 1.5 times the IQR below the first quartile and 1.5 times the IQR above the third quartile. Outliers were detected using the difference between the air voids obtained from the cores (parafilm and Corelok) and the estimated in-place air voids by the two methods (Sand Cone and Balloon). The quartiles were calculated for each of the differences combinations (Balloon-parafilm, Balloon-Corelok, Sand Coneparafilm, Sand Cone-Corelok), and the IQR was estimated, as summarized in Table 3. Any data point above and below the range determined (1.5 times the IQR below the first quartile and above the third quartile) was considered an outlier and removed from the data set. There were no outliers detected for the data sets that estimated the air voids using the Balloon method, due to the high variability of the method, which led to a higher IQR value. 5.1.1. Modeling of the air voids data The general trend observed in the data set was that the air voids obtained from the cores, regardless of the method used for bulk specific gravity (parafilm or Corelok), were higher than the air
Fig. 4. Coring of the compacted and cured laboratory CIR slabs.
6
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
Fig. 6. Comparison of estimated air voids vs cores air voids using parafilm on laboratory slab 1.
Fig. 7. Comparison of estimated air voids vs cores air voids using parafilm for all 7 laboratory slabs.
Fig. 8. Comparison of estimated air voids vs cores air voids using Corelok for all 7 laboratory slabs.
voids estimated by both methods (Sand Cone or Rubber Balloon). This general trend can be observed in Fig. 9 comparing the estimated air voids from Sand Cone and Balloon methods with the core air voids obtained using the parafilm method. Based on these observations, it was decided to apply a correction factor to the air voids obtained from the estimation method improve the fit with the data obtained from the cores. The applied correction factor shifts the equality line of the data set for each combination. A linear regression analysis was selected
to determine the magnitude of the correction factor. The equation for the linear regression was as follows:
y ¼ b0 þ b1 x Since the equality line has a slope of one, then b0 = 0 and b1 = 1. The corrected equality line would have the same slope value as the equality line (b1 = 1) and would be described by:
y ¼ b0 þ x
7
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731 Table 3 Quartile and IQR Calculations for Outliers Identification for Laboratory Data.
Min. value 25th percentile 50th percentile 75th percentile Max. value IQR 1.5*IQR 1st-1.5*IQR 3rd + 1.5*IQR Number of Data Points Number of Outliers
Quartile
Balloon-Parafilm (%)
Balloon-Corelok (%)
Sand Cone-Parafilm (%)
Sand Cone-Corelok (%)
0 1 2 3 4
4.3 0.5 2.1 4.1 8.2 3.7 5.5 5.0 9.6 49 0
5.8 0.8 0.8 3.4 7.2 4.2 6.3 7.1 9.6 49 0
3.1 1.4 2.6 4.1 8.7 2.7 4.0 2.6 8.1 49 3
4.6 0.5 1.6 3.1 7.1 2.6 3.9 3.4 7.0 49 4
Fig. 9. Estimated air voids (Sand Cone and Balloon methods) vs air voids of the cores from laboratory slabs using parafilm.
b0 ¼ y x Each combination of test method and core method, for example Sand Cone with parafilm, has its own correction factor for the shifted equality line according to the average of air voids from the data set obtained in this study, as shown in Figs. 10–13. Following this procedure, the correction factors were determined as follows: The correction factor for the air voids obtained with the Rubber Balloon method when compared to the cores air voids using the parafilm method is: 2.4%.
The correction factor for the air voids obtained with the Rubber Balloon method when compared to the cores air voids using the Corelok method is: 1.2%. The correction factor for the air voids obtained with the Sand Cone method when compared to the cores air voids using the parafilm method is: 2.8%. The correction factor for the air voids obtained with the Sand Cone method when compared to the cores air voids using the parafilm method is: 1.9%. Following the correction stage, a statistical analysis was conducted on the data set of each combination to determine if the esti-
Fig. 10. Corrected equality line for estimated air voids (Balloon method) vs air voids of the cores from laboratory slabs using parafilm.
8
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
Fig. 11. Corrected equality line for estimated air voids (Balloon method) vs air voids of the cores from laboratory slabs using Corelok.
Fig. 12. Corrected equality line for estimated air voids (Sand Cone method) vs air voids of the cores from laboratory slabs using Parafilm.
Fig. 13. Corrected equality line for estimated air voids (Sand Cone method) vs air voids of the cores from laboratory slabs using Corelok.
mated air voids data obtained using the Sand Cone and Rubber Balloon methods are comparable to the air voids obtained from the cores. The selected statistical tool used to develop the analysis was the t-test, which is commonly used for testing the difference between two data sets with normal distributions and unknown
variances. The t-test is used to determine the probability of difference in the means of two populations. The t-tests were ran using the statistical software R [12]. Since the objective was to determine if the estimated air voids data (Sand Cone and Rubber Balloon) were comparable to the actual
9
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
air voids from the cores, the differences between the cores air voids and the estimated air voids were calculated for the following four combinations: Difference between the cores air voids using parafilm estimated air voids using the Sand Cone method Difference between the cores air voids using Corelok estimated air voids using the Sand Cone method Difference between the cores air voids using parafilm estimated air voids using the Rubber Balloon method Difference between the cores air voids using Corelok estimated air voids using the Rubber Balloon method
Use the 95% Confidence Interval of Difference for the combination of Sand Cone method and the cores with parafilm shown below:
and the
Difference ¼ AV%Parafilm ðAV%SandCone þ CF Þ
and the
Difference ¼ AV%Parafilm ðAV%SandCone þ 2:8%Þ
and the and the
The null hypothesis was that the difference between the cores air voids (parafilm or Corelok) and the corrected estimated air voids (Sand Cone or Balloon) is equal to zero (difference = 0). The t-test was run for each combination. If the parameter p-value from the analysis is lower than 0.05 (p < 0.05), the difference between the cores air voids and the estimated air voids is different than zero (i.e., difference – 0). If the parameter p-value from the analysis is higher than 0.05 (p > 0.05), then there is no significant evidence to prove that the difference between the cores air voids and estimated air voids is different than zero (i.e., difference = 0). The objective of the analysis was to compare the four combinations and to select the most suitable method for estimating the inplace density of the CIR layer during compaction. As shown in Table 4 for the four combinations, there is not significant evidence to prove that the difference between the cores air voids and the corrected estimated air voids is different than zero since all pvalues are above 0.05. The 95% Confidence Interval for each combination was determined from the following relationship:
r
CI95% ¼ x 1:96 pffiffiffi n where ‘‘r” is the standard deviation and ‘‘n” is the population. The lower the standard deviation indicates lower variability of the method and identifies the best alternative to determine the estimated in-place air voids. From the statistical analysis of the data generated in this experiment as summarized in Table 4, it was found that the best approach to estimate the in-place air voids is to use the Sand Cone method and compare the results with the cores air voids using the parafilm method for measuring bulk specific gravity. As can be seen, the selected alternative has the lowest standard deviation of 2.377. The 95% Confidence Interval of Difference provides the range over which there is 95% confidence that the mean of the differences will be contained.
0:9% < Difference < 0:5% Apply the above relationship as follows: o Use the sand cone method to measure the in-place air voids of the compacted CIR layer as described in this paper o Add 2.8% to the in-place air voids determined from the Sand Cone measurement to calculate the adjusted in-place air voids o There is 95% confidence that the adjusted in-place air voids will be: 0.9 to +0.5 from the actual in-place air voids measured on cores using the parafilm method 7. Field verification In 2017, a field verification of the in-place density methods was conducted on a CIR construction project located on US-95 south of Fernley, Nevada. In-place density was measured using the Sand Cone and Rubber Balloon methods on the CIR construction project, as shown in Fig. 14. Testing was conducted at two different locations within the project. At each location, three in-place density tests, separated by 50 feet, were conducted using each method. The density testing (Sand Cone and Balloon methods) were conducted after the rolling pattern was completed and the full compaction was considered achieved. Cores were cut at the same locations of the in-place density measurements 14 days after compaction.
6. Recommended approach Based on the statistical analysis of the data generated from the laboratory testing on CIR slabs using the Sand Cone and Rubber Balloon methods and the measurements of the air voids on cores using the parafilm and Corelok methods, the following process is recommended to estimate in-place air voids of the compacted CIR layer.
Fig. 14. Balloon (left) and Sand Cone (right) test methods in-place at US-95 CIR project in Fernley, NV.
Table 4 T-test Developed on the Differences of the Air Voids for All the Combinations based on Laboratory Data. T-test
Balloon-Parafilm
Balloon-Corelok
Sand Cone-Parafilm
Sand Cone-Corelok
n t p-value 95% Confidence Interval Standard Deviation
49 0.658 0.514 [1.76 to 0.89] 4.63
49 0.003 0.998 [0.80 to 0.81] 2.80
49 0.667 0.508 [0.91 to 0.46] 2.38
49 1.037 0.305 [1.09 to 0.35] 2.51
10
P.E. Sebaaly et al. / Construction and Building Materials 227 (2019) 116731
Table 5 Verification of the 95% Confidence Interval of Difference for Air Voids Estimation using Sand Cone Method on US-95 CIR Project. Sand Cone - Parafilm Avg. AV: Sand Cone (%) Corrected Avg. AV: Sand Cone (%) Avg. AV: Cores (parafilm) (%) Difference (%) 95% Confidence Interval
Sand Cone - Corelok 19.4 22.3 22.0 0.3 [0.91% to 0.46%]
OK
Avg. AV: Sand Cone (%) Corrected Avg. AV: Sand Cone (%) Avg. AV: Cores (Corelok) (%) Difference (%) 95% Confidence Interval
19.4 21.3 20.0 1.3 [1.09% to 0.35%]
NO
Table 6 Verification of the 95% Confidence Interval of Difference for Air Voids Estimation using Rubber Balloon Method on US-95 CIR Project. Balloon - Parafilm Avg. AV: Balloon (%) Corrected Avg. AV: Balloon (%) Avg. AV: Cores (parafilm) (%) Difference (%) 95% Confidence Interval
Balloon - Corelok 22.7 25.1 22.0 3.1 [1.76% to 0.89%]
NO
The verification of the 95% Confidence Interval of Difference for each method was conducted by comparing the air voids obtained from the Sand Cone and Rubber Balloon methods and the air voids obtained from the cores using the parafilm and the Corelok methods, as shown in Tables 5 and 6. The data in Table 5 show that the 95% Confidence Interval of Difference was met for the Sand Cone method using the parafilm method on the cores, but it was not met for the Sand Cone method using the Corelok method on the cores. The difference between the average air voids from the cores using parafilm (22.0%) and the average air voids from Sand Cone plus the correction factor (19.4% + 2.8%) is inside the 95% Confidence Interval of Difference. The data in Table 6 show the air voids estimated by the Rubber Balloon method are higher than the air voids measured on the cores. This was not the regular trend found on the laboratory testing of the CIR slabs. Therefore, applying the correction factor obtained from the CIR slabs led to higher difference from the cores. This difference in the trend might be due to the higher variability of the Balloon test method. Finally, the results of the field validation trial confirm that the best approach to estimate the in-place air voids of the compacted CIR layer is to use the Sand Cone method and compare the estimated air voids with the air voids obtained from the cores using the parafilm technique. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgement The authors would like to acknowledge the Nevada Department of Transportation for supporting the research project that led to the development of the test method documented in this paper.
Avg. AV: Balloon (%) Corrected Avg. AV: Balloon (%) Avg. AV: Cores (parafilm) (%) Difference (%) 95% Confidence Interval
22.7 23.8 20.0 3.8 [0.80% to 0.81%]
NO
Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.conbuildmat.2019.116731. References [1] P. Murugaiyah, P.E. Sebaaly, E.Y. Hajj, S. Selvaratnam, Evaluation of Long-term Performance of Cold In-Place Recycling Pavements in Nevada Report No. WRSC-UNR-CIR-201402, Nevada Department of Transportation, Carson City, Nevada, 2014. [2] A. Nataatmadja, Some Characteristics of Foamed Bitumen Mixes Transportation Research Record No. 1767, Transportation Research Board, Washington, DC, 2001. [3] Asphalt Recycling and Reclaiming Association (2015). Basic Asphalt Recycling Manual, second ed. Annapolis, MD. [4] C.W. Schwartz, B.K. Diefenderfer, B.F. Bowers, Material Properties of Cold InPlace Recycled and Full-Depth Reclamation Asphalt Concrete Report 863, National Cooperative Highway Research Program, Washington, DC, 2017. [5] G. Thenoux, A. Gonzalez, R. Dowling, Energy consumption comparison for different asphalt pavements rehabilitation techniques used in Chile, Resour., Conserv. Recycling Vol (2007) 49. [6] Wirtgen, Cold Recycling Technology, third ed., Wirtgen GmbH, Windhagen, Germany, 2010. [7] P.E. Sebaaly, CIR Field Project Report, Regional Transportation Commission, Washoe County, Nevada, 2014. [8] E.Z. Teshale, D. Rettner, A. Hartleib, D. Kriesel. Application of Laboratory Asphalt Cracking Tests to Cold In-Place Recycled (CIR) Mixtures, Technical Paper, Journal of the Association of Asphalt Paving Technologists, Vol. 86, 2017. [9] National Academy of Sciences website: NCHRP Project 09-62, https://apps.trb. org/cmsfeed/TRBNetProjectDisplay.asp?ProjectID=4190, 2018. [10] B.K. Diefenderfer, D. Jones, A.J.T. Hand, B.F. Bowers, G. Flintsch, Rapid tests and specifications for construction of asphalt-treated cold recycled pavements available by request at, NCHRP Project 09-62 Phase I Interim Report, NCHRP, Washington, D.C, 2017. https://apps.trb.org/cmsfeed/TRBNetProjectDisplay. asp?ProjectID=4190. [11] Y. Zhang, L. Wang, B. Diefenderfer, W. Zhang, Determining volumetric properties and permeability of asphalt-treated permeable base mixtures, Int. J. Pavement Eng. 17 (2014). [12] R Core Team (2017). R: A language and environment for statistical computing – Version R3.4.1. R Foundation for Statistical Computing, Vienna, Austria.