Road Materials and Pavement Design ISSN: 1468-0629 (Print) 2164-7402 (Online) Journal homepage: http://www.tandfonline.com/loi/trmp20 Optimising water content in cold recycled foamed asphalt mixtures Wangyu Ma, Randy West, Nam Tran & Nathan Moore To cite this article: Wangyu Ma, Randy West, Nam Tran & Nathan Moore (2017): Optimising water content in cold recycled foamed asphalt mixtures, Road Materials and Pavement Design, DOI: 10.1080/14680629.2017.1389088 To link to this article: http://dx.doi.org/10.1080/14680629.2017.1389088 Published online: 24 Oct 2017. Submit your article to this journal View related articles View Crossmark data Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=trmp20 Download by: [UAE University] Date: 25 October 2017, At: 17:37 Road Materials and Pavement Design, 2017 https://doi.org/10.1080/14680629.2017.1389088 Optimising water content in cold recycled foamed asphalt mixtures Wangyu Ma ∗ , Randy West, Nam Tran and Nathan Moore National Center for Asphalt Technology, Auburn University, Auburn, AL, USA Downloaded by [UAE University] at 17:37 25 October 2017 (Received 15 August 2016; accepted 25 October 2016 ) During cold recycling, water is added to facilitate the dispersion of foamed asphalt in the mixture and to achieve uniform mixing and help compaction by providing suﬃcient lubrication. Too little water may cause diﬃculty in workability and compaction of the mixture, but too much water may extend the curing time and reduce density and strength. Therefore, the optimum water content (OWC) was considered as one of the most important factors in mix design procedures for cold recycling. Currently, mix design procedures for cold recycled foamed asphalt mixtures suggest adding water to the mixture at an optimum content to facilitate mixing and compaction. However, there is no standard method for determining the optimum total water content (OTWC) for cold recycling mixtures. Several empirical relationships were developed to determine the OTWC based on modiﬁed Proctor test results for Reclaimed Asphalt Pavement (RAP)/aggregate. However, the compaction eﬀort in the modiﬁed Proctor test for RAP/aggregate may not match that for mixtures, which is compacted using the Superpave Gyratory Compactor (SGC) or Marshall hammer. A study is underway to improve the design method for cold recycled foamed asphalt mixtures with 100% RAP. The purpose of this paper is to optimise the design procedure by developing a new method to determine OTWC. SGC was used to compact RAP instead of the modiﬁed Proctor test to match the compaction eﬀort recommended for foamed asphalt mixtures. A regression model was developed to calculate the OTWC for a mixture based on the determined OWC of the RAP, foamed asphalt content, and binder type as factors. The method for determining OTWC for a mixture was validated using six diﬀerent mixtures and was found to correlate well with the measured OTWC, even though two of six mixtures had underestimated OTWC due to diﬀerent binder source. Further comparisons with other two OTWC determining methods showed the mixtures at the proposed OTWC had improvement in indirect tensile strength. Keywords: cold recycling; foamed asphalt; water; modelling 1. Introduction Cold recycling is one of the most sustainable technologies in pavement construction and rehabilitation (Chan, Tighe, & Chan, 2010; Stroup-Gardiner, 2011; Thenoux, González, & Dowling, 2007). As it reclaims asphalt pavement up to 98%, cold recycling can conserve large quantities of natural resources (Diefenderfer, Bowers, Schwartz, Farzaneh, & Zhang, 2016). Also, since cold mixtures are produced at ambient temperatures, cold recycling can signiﬁcantly reduce fuel consumption and greenhouse gases generated during construction. Foamed asphalt was ﬁrst proposed for use in pavement recycling in 1957 (Csanyi, 1957). It is commonly used as a recycling agent for cold recycling. To make foamed asphalt, water and air are injected into hot asphalt binder. The process creates bubbles of water vapour within the *Corresponding author. Email: email@example.com © 2017 Informa UK Limited, trading as Taylor & Francis Group Downloaded by [UAE University] at 17:37 25 October 2017 2 W. Ma et al. binder that help expand the asphalt volume and reduce the binder viscosity. With the increased volume and lower viscosity, foamed asphalt can easily disperse among and mix with Reclaimed Asphalt Pavement (RAP) particles to provide non-continuous bonding (Diefenderfer et al., 2016; Muthen, 1998). During cold recycling, water is also added to facilitate the dispersion of foamed asphalt in the mixture and to achieve uniform mixing and help compaction by providing suﬃcient lubrication (Asphalt Recycling & Reclaiming Association [ARRA], 2015). Too little water may cause diﬃculty in workability and compaction of the mixture, but too much water may extend the curing time and reduce density and strength (Jenkins, 2000). Therefore, the optimum water content (OWC) was considered as one of the most important factors in the mix design procedure. Since there is currently no standard mix design procedure for cold recycled mixtures with foamed asphalt, diﬀerent methods have been used to determine the optimum total water content (OTWC) in mixture. Most of them are based on the OWC of RAP/aggregate determined by the modiﬁed Proctor test in accordance with American Association of State Highway and Transportation Oﬃcials (AASHTO) T180. Mohammad, Abu-Farsakh, Wu, and Abadie (2003) recommended using 100% of the OWC determined by the modiﬁed Proctor test. Wirtgen GmbH (2008) has recommended a reduction factor for OWC by the modiﬁed Proctor test. Muthen (1998) recommended using OWC by the modiﬁed Proctor test as the optimum total ﬂuid content, including water and foamed asphalt. Sakr and Manke (1985) proposed a regression model to calculate the OTWC using diﬀerent factors such as OWC determined by the modiﬁed Proctor test, proportion of ﬁnes, and proportion of foamed asphalt. While these methods for determining the OTWC have been used in the mix design, there are two key issues need to be addressed. First, the compaction eﬀort in the modiﬁed Proctor test for RAP/aggregate does not match the compaction eﬀort used to design the foamed asphalt mixture, which is typically done using Marshall Compactor or Superpave Gyratory Compactor (SGC). Second, the relationship used to determine the OTWC varied from one study to another because it was developed based on limited types of mixtures. These issues may aﬀect the accuracy of the determined OTWC, inﬂuencing the density and strength of cold recycled foamed asphalt mixtures. A new method for determining the OTWC for an improved mix design method for cold recycled foamed asphalt mixture was investigated in a recent study. This paper discusses the results of that study. 2. Objective and scope The objective of this study was to develop a new method for determining the OTWC for an improved cold recycled foamed asphalt mixture design method. The new method uses a SGC at 30 gyrations to determine the OWC of RAP which matches the compaction eﬀort recommended for mixture (ARRA, 2015). This compaction eﬀort was considered equivalent to the ﬁeld compaction eﬀort (Cross, 2003). The new method also uses a regression model to determine the OTWC based on RAP type, binder type, and foamed asphalt content. In the following sections, material properties and the mixture production method are introduced followed by a discussion of the eﬀect of water on mixtures. Finally, a new method to determine OTWC in cold recycled foamed asphalt mixture was proposed and validated in the laboratory. 3. Materials Materials used in this study are discussed in this section. RAP is discussed ﬁrst followed by the asphalt binder used to produce foamed asphalt. Cold recycled foamed asphalt mixtures produced with these RAP and binder are discussed ﬁnally. Road Materials and Pavement Design 3 Table 1. Summary of residual binder for RAP. RAP type DF (A) DM (B) DC (C) VF (D) VC (E) Gradation categorya NMAS (mm) Residual binder content (%) Residual binder PG Fine Medium Coarse Fine Coarse 12.5 19.0 19.0 19.0 25.0 4.92 4.85 4.10 5.19 5.36 100-10 100-4 100-10 100-10 NAb a Classiﬁcation is based on relative location of the gradation curve compared to recommended gradation range by Wirtgen GmbH (2012). binder PG for Type E RAP was not tested due to availability of materials. Downloaded by [UAE University] at 17:37 25 October 2017 b Residual 3.1. RAP Table 1 shows the information of ﬁve diﬀerent RAP types (from A to E) used in this study. They were sampled from stockpiles in the NCAT Pavement Test Track or East Alabama Paving plant, which had been milled, crushed, and screened. Gradation was tested following procedures suggested in ARRA guidelines (American Standards for Testing and Materials (ASTM) C136 and C117). Residual binder content was tested using ignition method in accordance with AASHTO T308. Residual binder was extracted and recovered to determine performance grade following AASHTO M320. RAP types are listed in Table 1. The ﬁrst letter ‘D’ refers to the optimising design and ‘V’ refers to validation. The second letter (F, M, or C) represents the gradation category. For example, ‘DF’ RAP is design optimising RAP in the ﬁne gradation category. Types A–E indicate the sequence when they were used in this study. Residual binder content of these RAP was similar except for VC RAP with less binder content. Available residual binder PG grades of these RAP sources are also similar. Figure 1 plots ﬁve RAP gradations measured following ASTM C136 and C117. To categorise these RAP, the gradation range recommended by Wirtgen GmbH (2012) was also plotted as a reference. Most gradation curves were basically within the range or slightly oﬀset. DF gradation has more oﬀsets on the large sieves and DM gradation has more on the small sieves. These gradations could be altered by further processing RAP or blending with virgin aggregates, but this is out of scope of this study. 3.2. Asphalt binder Three types of virgin asphalt binder, including two PG 67-22 binders and one PG 58-34 binder, were used to produce foamed asphalt. The information for these binder types are shown in Table 2. Binder PG 67-22 is commonly used in state of Alabama for asphalt pavement construction. The two PG 67-22 binders were obtained from diﬀerent reﬁnery plants with diﬀerent sources. Even with the same performance grade, they were still considered diﬀerent binder types when producing foamed asphalt because foaming properties such as expansion ratio (ER) and half-life (HL) may be aﬀected by diﬀerent sources (Newcomb et al., 2015). The foaming properties of each binder were tested and summarised in Table 2. Binder 58-34 (B) produced a foamed asphalt with the highest ER and HL. Binder 67-22 (C) could not produce stable foamed asphalt at 160°C; so the foaming temperature was increased to 170°C. Binder C had marginal foaming properties in accordance with the criteria in the ARRA guidelines. 3.3. Cold recycled foamed asphalt mixes Thirteen cold recycled foamed asphalt mixtures were produced with the previously mentioned RAP and binder. The method to produce the mixtures is discussed later in experimental plan. 4 W. Ma et al. Table 2. Summary of asphalt binder PG and foaming properties. Foaming Properties Binder type Temp. (°C) Water Content (%) ER (times) HL (s.) Lake city, FL Alberta, CA Birmingham, AL 160 160 170 1.8 2.0 1.3 11 19 9 9 10 6 Portland cement was added to mix 1–10 at 1.0% to improve durability, asphalt–aggregate bonding, and moisture damage resistance (ARRA, 2015; Asphalt Academy, 2009; Wirtgen GmbH, 2012). It was assumed that the eﬀect of cement on the mixtures was small and consistent regardless of water content when equal amounts of cement was added because water evaporates quickly once curing starts. Mix 11 and 12 have no cement, allowing for the eﬀect of cement on OTWC to be evaluated. Mix 13 with 1.5% cement was only used for validating the proposed OTWC determination method (Table 3). 4. Experimental plan 4.1. Producing mixture Cold recycled foamed asphalt mixtures were produced in the laboratory for testing in this study following the ARRA guidelines with modiﬁed procedures for determining the OWC of RAP and testing indirect tensile strength (ITS). Figure 2 shows the basic steps for producing foamed asphalt mixtures. Each step is discussed brieﬂy in the following sections. FAC stands for foamed asphalt content and TWC stands for total water content. More detailed information about the procedures can be found in the ARRA cold recycling guideline (ARRA, 2015). 100 90 80 70 60 50 20 10 Figure 1. Gradation plot of diﬀerent RAP types. 12.5 9.50 4.75 2.36 1.18 Sieve Size (mm) 0.075 0.15 0.30 0.60 0 31.0 30 25.0 DF DM DC VF VC Wirtgen coarse limit Wirtgen fine limit 40 19.0 Percent Passing Downloaded by [UAE University] at 17:37 25 October 2017 67-22 (A) 58-34 (B) 67-22 (C) Binder source Road Materials and Pavement Design Table 3. Downloaded by [UAE University] at 17:37 25 October 2017 Mix no. 1 2 3 4 5 6 7 8 9 10 11 12 13 5 Summary of mixtures with diﬀerent binder and RAP types. Asphalt binder RAP Additive (Type I/II cement) 67-22 (A) 67-22 (A) 67-22 (A) 58-34 (B) 58-34 (B) 58-34 (B) 58-34 (B) 58-34 (B) 67-22 (C) 67-22 (C) 58-34 (B) 58-34 (B) 67-22 (C) DF (A) DM (B) DC (C) DF (A) DM (B) DC (C) VF (D) VC (E) VF (D) VC (E) DF (A) DM (B) DF (A) 1.0% 1.0% 1.0% 1.0% 1.0% 1.0% 1.0% 1.0% 1.0% 1.0% No cement No cement 1.5% 4.1.1. Step 1: preparing and testing materials’ properties RAP materials were sampled from an East Alabama Paving’s plant stockpile. The RAP materials had been milled, crushed, and screened before stockpiling. After sampling, the RAP was stored in buckets and dried in front of a fan for a week before mixing. A mechanical splitter was used to reduce RAP to testing size following AASHTO R47. Gradation, binder content, and binder performance grade of the RAP were tested. Virgin asphalt was heated and split into 1-gallon cans for ease of foaming properties’ tests. ER and HL were tested using Wirtgen WLB10S laboratory plant. Also, the optimum foaming water content and temperature were determined to produce foamed asphalt. The machine in Figure 3 is the laboratory foaming plant. 4.1.2. Step 2: determining RAP OWC RAP can be compacted to its maximum dry density after mixing with the optimum amount of water. This water content is called the OWC and used as guidance for determining the OTWC in cold recycled asphalt mixture. The current ARRA guideline recommends using the modiﬁed Proctor method to compact RAP in accordance with AASHTO T180 (or ASTM D1557) and using 100% OWC of RAP as the OTWC for mixture. However, this compaction eﬀort does not match that used for compacting foamed asphalt mixtures, which is 30 gyrations in a SGC. Instead of using the modiﬁed Proctor method, the compaction eﬀort by SGC is adopted in this Step 1: Preparing and testing materials’ properties for RAP & binder Step 2: Determining RAP OWC based on proposed method using SGC Step 3: Mixing and compacting at desired FAC and TWC Step 4: Laboratory curing (40 °C, 3 days) Step 5: Testing compacted density and indirect tensile strength (d/w) Figure 2. Basic steps to produce cold recycled foam asphalt mixtures. Downloaded by [UAE University] at 17:37 25 October 2017 6 W. Ma et al. Figure 3. Laboratory pug mill (left) and foaming plant (right). Figure 4. Procedure to determine RAP OWC using SGC. study to compact RAP. Testing procedure for determining the OWC of RAP is similar to ASTM D1557 but has some changes. The basic steps of determining the OWC of RAP in this study are summarised in Figure 4. A description of each step follows. • Dry RAP with a fan for a week; • Remove particles above ¾-inch sieve for use with 100-mm diameter molds; • Add water to RAP at diﬀerent water contents; at each water content, prepare at least 5kg RAP (using mechanical splitter to reduce RAP to test size) in order to compact three replicates. Add water slowly and mix with a scoop by hand until uniform; • Load wet RAP into a 100-mm mold and compact for 30 gyrations at an internal angle of 1.16° in a SGC; • Record the compacted height and wet weight after compaction. Dry the compacted specimens overnight at 110°C to determine the molding water content, which is used to calculate dry density; • Plot dry density against water content and determine OWC at the maximum dry density. All the TWC levels selected for mixtures are based on the OWC of RAP determined from this procedure. The OWC of RAP was compared with that determined from the modiﬁed Proctor method. Four RAP types were selected for comparison. Tested water contents are listed in the experimental plan shown in Table 4. The OWC determined by the proposed method and the modiﬁed Proctor method are listed and compared in Table 5. Table 5 shows that the OWCs of RAP determined by the proposed method are lower than those determined by the modiﬁed Proctor method. However, the maximum dry density from the two methods is similar, while those determined by proposed method are slightly lower. The diﬀerence Road Materials and Pavement Design 7 Table 4. Experimental plan for determining RAP OWC. Selected water contents for testing (%) RAP types DF (A) DM (B) VF (D) VC (E) Proposed SGC method Modiﬁed Proctor method (ASTM D1557) 3, 4, 5, 6 1, 2, 3, 4, 5, 6, 7 2, 3, 4, 5, 6, 7 2, 3, 4, 5 5, 6, 7, 8 2, 3, 4, 5, 6 3, 4, 5, 6, 7, 8 4, 5, 6, 7 Table 5. Summary of RAP OWC and maximum dry density determined from proposed method and modiﬁed Proctor method. Maximum dry density (kg/m3 ) Downloaded by [UAE University] at 17:37 25 October 2017 RAP OWC (%) RAP types DF (A) DM (B) VF (D) VC (E) Proposed SGC Mod. Proctor Proposed SGC Mod. Proctor 4.0 3.0 4.0 3.0 6.5 3.8 6.5 5.3 2046.8 1940.8 1969.3 1899.6 2101.1 1944.6 2018.3 1955.5 between two OWC values of the same RAP sample may be due to diﬀerent compaction methods. However, further evaluation is needed in the future. 4.1.3. Step 3: mixing and compacting RAP was mixed with cement, water, and foamed asphalt in a laboratory pug mill, as shown in Figure 3. Materials were added following a recommended sequence mentioned in ARRA (2015) guideline, which is RAP, mineral additive, followed by water and foamed asphalt. Added water helps spreading foamed asphalt in RAP and aids compaction of mixture (ARRA CR101). The twin-shaft pug mill has 10–30 kg mixing capacity. At each time, 10–15 kg RAP was added for mixing. Mixing speed and mixing time are adjustable. This study applied a medium driving speed (about 72 rpm) and 60-s mixing time for each mixture. Wirtgen GmbH (2008) recommended 30s for aggregate/RAP blending with cement and water. Therefore, 15-s pre-mixing was applied for both cement and water in this study to prevent non-uniformity of these two non-asphalt materials in mixture. The compaction method recommended in ARRA (2015) guideline was applied. Cold recycled foamed asphalt mixture was compacted for 30 gyrations in SGC. Specimens are compacted to the geometry of 100-mm diameter and 63.5 ± 2.5 mm height. 4.1.4. Step 4: curing The curing procedure used in this study followed the ARRA (2015) guideline. Compacted specimens were cured in oven at 40°C right after compaction. Curing time was 3 days for all the specimens. 4.1.5. Step 5: testing density and ITS Dry density of each compacted specimen is an indicator of durability and compaction quality. It was determined based on dry weight of specimen and measured dimension after curing. 8 W. Ma et al. Table 6. Experimental plan for determining eﬀect of mixing water on density and dry ITS. Mix no. 2 2 2 Binder RAP FAC (%) MWC (% OWC) TWC (% OWC) 67-22 (A) 67-22 (A) 67-22 (A) DM (B) DM (B) DM (B) 3.0 3.0 3.0 50, 75 50, 75 50, 75 75 100 125 Downloaded by [UAE University] at 17:37 25 October 2017 ITS test was conducted for each mixture in both dry and wet conditions. Dry specimens were cured at 25°C 1 h before testing and wet specimens were cured in a 25°C water bath for 24 h before testing. The wet ITS testing method is simpliﬁed from ARRA (2015) guideline because vacuum saturation was not applied. This method was recommended in Wirtgen GmbH (2012) cold recycling manual. 4.2. Mechanism and modelling The eﬀect of mixing and compaction water on mixture performance needs to be investigated to optimise water content. After this mechanism was fully understood and inﬂuencing factors were identiﬁed, modelling was conducted to ﬁnd a relationship between the OTWC and the inﬂuencing factors. 4.2.1. Mechanism In cold recycling, water is added to the cold mix to achieve uniform mixing and to facilitate compaction by providing lubrication in the mix (ARRA, 2015). This section investigates the role of water during mixing and compaction in a laboratory environment. Understanding this mechanism is essential to foamed asphalt mix design because it answers the question of why water is needed. First, the eﬀects of water during mixing and compaction on the compacted density and ITS were assessed separately. Only one mixture (no. 2) was used in this step. Water was added in two steps: mixing water was added before foamed asphalt and compaction water was added afterwards but before compaction. Then, the combined eﬀect was determined for six diﬀerent mixtures (nos. 1–6). 18.104.22.168. Eﬀect of mixing water During mixing, water helps distribute foamed asphalt. This concept was proposed in 1950s by Csanyi (1957). This eﬀect determines how uniform the foamed asphalt is distributed in the mixture. The diﬀerence in distribution may cause changes in ITS because strength is dependent on the uniformity of bonding between foamed asphalt and RAP particles. In this study, the eﬀect was evaluated quantitatively by examining dry ITS at diﬀerent mixing water contents (MWCs), while keeping total compaction water constant. Wet ITS was not tested because of a lack of availability of specimens. The experimental plan is summarised in Table 6. The same mixture was tested for density and dry ITS at two MWCs at each TWC level. Analysis of variance (ANOVA) was performed in Minitab 17 software to determine the eﬀect of MWC and compare it with TWC. 22.214.171.124. Eﬀect of compaction water During compaction, water provides lubrication between particles and facilitates the compaction process. The eﬀect of compaction water content may be observed from the change of compacted density or ITS at diﬀerent compaction water contents while keeping MWC constant. An experiment was designed based on the mix design procedure proposed by Wirtgen GmbH (2012), in which water is introduced in two steps. Wirtgen GmbH (2012) suggests adding 75% of the OWC before foamed asphalt to achieve ideal mixing and then Road Materials and Pavement Design 9 Table 7. Experimental plan for determining eﬀect of compaction water on density and dry ITS. Mix no. 2 2 2 2 Binder RAP TWC (%OWC) FAC (%) 67-22 (A) 67-22 (A) 67-22 (A) 67-22 (A) DM (B) DM (B) DM (B) DM (B) 50, 75, 100, 125 50, 75, 100, 125 50, 75, 100, 125 50, 75, 100, 125 1.5 2.0 2.5 3.0 Table 8. Experimental plan for determining combined eﬀect of total water on density, dry and wet ITSs. TWC (% OWC) Downloaded by [UAE University] at 17:37 25 October 2017 Mix no. 1 2 3 4 5 6 Binder RAP Density Dry ITS Wet ITS FAC (%) 67-22 (A) 67-22 (A) 67-22 (A) 58-34 (B) 58-34 (B) 58-34 (B) DF (A) DM (B) DC (C) DF (A) DM (B) DC (C) 50, 75 75, 100, 125, 150 100, 125, 150, 175 50, 75 75, 100, 125, 100, 125, 150 50, 75 75, 100, 125, 150 100, 125, 150, 175 50, 75 75, 100, 125 100, 125, 150 50, 75 75, 100, 125, 150 100, 125, 150, 175 50, 75 75, 100, 125 100, 125, 150 2, 3 2, 3 2, 3 2, 3 2, 3 2, 3 adding the remaining water after the foamed asphalt to facilitate compaction. In this study, 50% of the OWC was added ﬁrst and remaining water was added after foamed asphalt. Mixing was performed by hand to ensure that the water was distributed uniformly. The eﬀect of compaction water on dry density and ITS is examined at diﬀerent FACs. Table 7 shows the experimental plan for determining eﬀect of compaction water on density and dry ITS. 126.96.36.199. Eﬀect of TWC for each mixture The above experimental plan in Tables 6 and 7 investigated the eﬀect of water during mixing and compaction separately. However, water was added only once either in the laboratory mix design or during ﬁeld production. Same amount of water not only facilitates mixing but also compaction. Therefore, this water has a combined eﬀect on compacted mixture. This study added water only once to investigate the combined eﬀect on density and strengths. The combined eﬀects of water on density as well as dry and wet ITSs were evaluated for six mixtures. Wet ITS was included because more specimens were prepared for strength testing. Table 8 summarises experimental plan for determining the eﬀect of TWC on density and strength (dry and wet ITSs). Each mixture was tested at two FAC levels. This allowed an ANOVA to be performed to determine the eﬀect of TWC as well as FAC. Results of the analysis are summarised and discussed later (as shown in Table 13). Except for the TWC in mixture, some other factors may also aﬀect the performance of mixtures, and therefore, aﬀect the OTWC. These factors need to be investigated before modelling OTWC. 4.2.2. Modelling The material types with potential to aﬀect mix properties (density, dry and wet ITSs) selected for study were RAP type and binder type. Before investigating these two material types, the best mixture property indicator to determine OTWC should be selected. Therefore, correlations among density as well as dry ITS and wet ITSs were evaluated ﬁrst, followed by a factorial study, and the modelling of OTWC. 10 W. Ma et al. Table 9. Experimental plan for determining eﬀect of RAP type and binder type on dry ITS. No. of comparison 1 2 3 4 Downloaded by [UAE University] at 17:37 25 October 2017 1 2 3 4 5 6 Mixtures selected for comparing Determination of eﬀect of RAP type on dry ITS Mix 2 vs. 3 at 2%FAC Mix 2 vs. 3 at 3%FAC Mix 5 vs. 6 at 2%FAC Mix 5 vs. 6 at 3%FAC Determination of eﬀect of Binder type on dry ITS Mix 1 vs. 4 at 2%FAC Mix 1 vs. 4 at 3%FAC Mix 2 vs. 5 at 2%FAC Mix 2 vs. 5 at 3%FAC Mix 3 vs. 6 at 2%FAC Mix 3 vs. 6 at 3%FAC 188.8.131.52. Correlation between density and strengths Density, dry ITS, and wet ITS of six mixtures (nos. 1–6) were included in this analysis. Totally, 167 data points were used to correlate dry ITS with density and 184 data points were used to correlate dry ITS with wet ITS. Based on the level of correlation, dry ITS was found to correlate well with the two other performance results. Therefore, it was used to determine and model OTWC. The results of correlations are discussed later in this paper. 184.108.40.206. Factorial study Since dry ITS was selected to determine OTWC, factors aﬀecting dry ITS of cold recycled asphalt mixtures may also have eﬀect on OTWC. Therefore, the eﬀect of the two material types (RAP type and binder type) on dry ITS was investigated. For simplicity, additive (i.e. cement) was studied separately from these two material types. Cement was added for all six mixtures at constant rate (1.0% by weight of dry RAP). Mixtures without cement (nos. 12 and 13) were studied later to evaluate the eﬀect of cement on OTWC separately. The eﬀects were assessed by ANOVA for each pair of mixture combination. The TWC eﬀect was included in each ANOVA to compare with the other eﬀects. Experimental plan for assessing RAP type and binder type eﬀects is shown in Table 9. 220.127.116.11. Regression analysis After the factorial study, the eﬀects of factors aﬀecting dry ITS were known. Analyses results are discussed later in Tables 14 and 15. Both RAP type and binder type were found to have a signiﬁcant eﬀect on dry ITS. Therefore, these two material types were also included for modelling together with the TWC and FAC. To construct a model for determining the OTWC, the measured OTWC was obtained ﬁrst. The measured OTWCs for six cold recycled foamed asphalt mixtures (nos. 1–6) were determined using quadratic regression model (second-order polynomial), built by ﬁtting dry ITS results at diﬀerent TWCs at each FAC. The measured OTWC corresponded to the maximum dry ITS on the regression curve. Then, the measured OTWC and three determining factors were ﬁtted using a multiple linear regression. In this model, binder type was treated as a categorical factor instead of a continuous factor to estimate the OTWC because these two binders diﬀered in both grade and sources. RAP type in the model was represented by the OWC, which was used as a continuous factor since several RAP materials have diﬀerent OWC levels in this study. Finally, the regression Road Materials and Pavement Design Table 10. Mix no. 7 8 9 10 11a 12a Downloaded by [UAE University] at 17:37 25 October 2017 a No 11 Experimental plan for validating using diﬀerent mixtures. Binder RAP FAC 58-34 (B) 58-34 (B) 67-22 (C) 67-22 (C) 58-34 (B) 58-34 (B) VF (D) VC (E) VF (D) VC (E) DF (A) DM (B) 2.5 2.5 2.5 2.5 3.0 3.0 cement was added in mix 11 and 12. model was built by correlating one categorical factor (binder type) and two continuous factors (FAC and OWC) with the measured OTWC. Each binder type would have one regression model. 4.3. Validation The proposed model was validated using other six diﬀerent mixtures. Also, dry ITS at predicted OTWC based on the proposed model was compared to the dry ITS at other two OTWC levels determined by typical methods. 4.3.1. Validating using diﬀerent mixture types Six additional foamed asphalt mixtures (nos. 7–12) were used to validate the two models. Table 10 gives information of these mixtures, which are diﬀerent from the previous mixtures (nos. 1–6) in FAC, RAP type, and/or binder type. Also, mix 11 and 12 had no cement so that eﬀect of cement on OTWC can be evaluated. The mixtures were produced and tested for dry ITS at diﬀerent TWC levels. The measured OTWC of these mixtures corresponding to maximum dry ITS was determined by quadratic regression. 4.3.2. Compare with typical OTWC determining method The proposed OTWC determination method in this study needed to be compared with typical methods to evaluate if it improves dry ITS. Mohammad et al. (2003) suggested using 100% RAP OWC determined by the modiﬁed Proctor method (AASHTO 180) as OTWC. Wirtgen GmbH (2008) suggested using the same procedure but a reduced RAP OWC as OTWC. Mixtures tested at diﬀerent TWC in this study allowed for the comparison of dry ITS at diﬀerent water contents, including proposed OTWC and other two OTWCs (100% RAP OWC by the modiﬁed Proctor method and OTWC suggested by Wirtgen). Three mixtures were selected to compare the proposed method with 100% RAP OWC by the modiﬁed Proctor method and six mixtures were used for comparing with the Wirtgen method. The detailed mixtures’ information and OTWC from each method is summarised in Table 18, together with discussions. 5. Results and discussion The results from mechanism and modelling were analysed and discussed ﬁrst, followed by the validating results of proposed OTWC determining method. 12 W. Ma et al. Table 11. Eﬀect of water during mixing by ANOVA. Eﬀect on density Factors p-Value Signiﬁcance Eﬀect on dry ITS p-Value Signiﬁcance MWC .016 Y .144 N TWC Interaction .034 .880 Y N .007 .987 Y N Table 12. Eﬀect of water during compaction by ANOVA. Downloaded by [UAE University] at 17:37 25 October 2017 Eﬀect on density Eﬀect on dry ITS Factors p-Value Signiﬁcance p-Value Signiﬁcance TWC < .001 Y .479 N FAC .615 N < .001 Y Interaction .303 N .008 Y 5.1. Mechanism and modelling 5.1.1. Mechanism 18.104.22.168. Eﬀect of water during mixing Table 11 summarises the eﬀect of water on density, dry and wet ITSs. MWC has a signiﬁcant eﬀect on density but is not on dry ITS. The interaction of mixing water and total water has an insigniﬁcant eﬀect on either density or dry ITS, which means MWC and TWC have no combined eﬀect. Eﬀect on wet ITS was not analysed because only three specimens in dry condition were prepared at each water content. 22.214.171.124. Eﬀect of water during compaction TWC in Table 12 is the water content for compaction. As discussed in Section 4.2.1, this TWC composed of a ﬁxed MWC and a varying compaction water content. Densities at diﬀerent TWC levels are signiﬁcant diﬀerent. TWC has more inﬂuence than FAC and interaction on density, since p-value for FAC and interaction factor are much higher than 0.05. However, dry ITS is not statistically diﬀerent as compaction water increases (or decrease). Meanwhile, FAC and its interaction with TWC signiﬁcantly aﬀect dry ITS. To separate water for mixing and compaction, water was added in two steps for the above analysis. However, for in-plant or in-place recycling in the ﬁeld, water is added only once. Therefore, the eﬀect of water was further characterised using six cold recycled foamed asphalt mixtures. Since water was added only once here, it facilitates both mixing and compaction process. The eﬀects can be evaluated through change in compacted density, dry ITS, and wet ITS. 126.96.36.199. Eﬀect of TWC for each mixture Table 13 summarises eﬀect of TWC on density and strengths (dry and wet ITSs) for six mixtures. The eﬀect of FAC was also evaluated to compare with TWC factor. Interaction between two factors was included in analysis because their eﬀects may be dependent from each other. p-Value results from ANOVA are listed to determine the signiﬁcance each eﬀect. p-Value less than .05 was considered signiﬁcant, which means density or strength is statistically diﬀerent as the factor changes. Exact value of p-value below .001 was not shown due to the decimal setting in Minitab software output. Road Materials and Pavement Design 13 Table 13. Summary of water eﬀect on density and strengths. Eﬀect on density Eﬀect on wet ITS Factors p-Value 2 TWC FAC Interaction TWC .270 .019 .156 .149 N Y N N .506 < .001 .153 .007 N Y N Y .134 .001 .576 < .001 N Y N Y 3 FAC Interaction TWCb .115 .152 .001 N N Y < .001 .509 .194 Y N N < .001 .184 .122 Y N N 4 FAC Interaction TWC < .001 .552 .547 Y N N < .001 .207 .048 Y N Y < .001 .695 .997 Y N N Mix no. 1 Downloaded by [UAE University] at 17:37 25 October 2017 Eﬀect on dry ITS Sig.a 5 p-Value Sig.a FAC .680 N .001 Y .007 Y Interaction .394 N < .001 Y .106 N TWC < .001 Y .405 N .001 Y FAC < .001 Y < .001 Y < .001 Y .018 Y .492 N .018 Y Interaction 6 p-Value Sig.a TWC < .001 Y .001 Y .041 Y FAC Interaction < .001 .260 Y N < .001 .185 Y N < .001 .550 Y N a ‘Sig.’ stands for signiﬁcance. b p-Value of TWC eﬀect on strengths became close to .05 as interval doubled. As shown in Table 13, TWC factor has signiﬁcant eﬀect on three mixtures’ densities (mix 3, 5, and 6), three mixtures’ dry ITS (mix 2, 4, and 6), and three mixtures’ wet ITS (mix 2, 5, and 6). Even though its eﬀects on strengths (dry and wet) for mix 3 are not signiﬁcant, as interval between TWC levels doubled (from 25% to 50% of OWC), the eﬀects became more signiﬁcant (p-value became .053 for dry ITS, and .060 for wet ITS). FAC has a signiﬁcant eﬀect on both strengths of all the mixtures, as well as densities of four mixtures (mix 1, 3, 5, and 6). This eﬀect is more signiﬁcant than the eﬀect of TWC. The results in Table 13 indicate water has a signiﬁcant inﬂuence on half of these mixtures tested in terms of density, dry and wet ITSs. Therefore, TWC needs to be selected carefully in mix design. 5.1.2. Modelling 188.8.131.52. Correlation between density and strengths Density and strengths are compared to determined their relationships. If they have good correlations, one factor can be used to select the OTWC without aﬀecting other two factors. Dry ITS was compared to density and wet ITS to determine correlations. Figures 5 and 6 show these two comparisons and their correlations were determined based on linear regression. These two ﬁgures demonstrate positive correlations, indicating density and wet ITS increase as dry ITS increases. Dry ITS correlates better with wet ITS than with density because of higher coeﬃcient of determination (R2 ) of ﬁtted model in Figure 6. The data points in Figure 5 are more scattered and include several outliers, which resulted in lower R2 value. The positive correlations indicate dry ITS can be used to select OTWC. 14 W. Ma et al. 184.108.40.206. Compare TWC with RAP type Two pairs of mixtures were used for comparison. With same binder type and FAC, ANOVA was performed to determine the eﬀects of TWC and RAP type. Each pair of mixtures contains two RAP types – DM (medium gradation) and DC (coarse gradation). Mixtures with DF RAP (ﬁne gradation) was not included because the tested TWCs were lower than mixtures with other two RAP types. Table 14 summarises the ANOVA results of four comparisons. RAP type has signiﬁcant eﬀect on dry ITS in the ﬁrst and third comparisons. These two comparisons used lower FAC (2%). Other two comparisons at higher FAC (3%) indicate RAP type is not signiﬁcant. During modelling, RAP type was represented by OWC of 2100.0 Density (kg/m3) 2050.0 2000.0 1950.0 y = 0.4702x + 1898.9 R² = 0.16 1900.0 1850.0 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 Dry ITS (kPa) Figure 5. Correlation between dry ITS and density. 350.0 300.0 250.0 Wet ITS (kPa) Downloaded by [UAE University] at 17:37 25 October 2017 220.127.116.11. Factorial study Since dry ITS was selected to determine OTWC of cold recycled foamed asphalt mixtures, material types related to dry ITS need to be evaluated. The OTWC is corresponding to the maximum dry ITS. Therefore, these material types may also aﬀect OTWC. Eﬀect of TWC on dry ITS has been evaluated previously. Besides TWC, the RAP type and binder type may also aﬀect dry ITS, which inﬂuences the OTWC. Therefore, their eﬀects on dry ITS were evaluated and compared with TWC. 200.0 150.0 100.0 y = 0.7549x + 17.348 R² = 0.6405 50.0 0.0 0.0 50.0 100.0 150.0 200.0 Dry ITS (kPa) Figure 6. Correlation between dry ITS and wet ITS. 250.0 300.0 350.0 Road Materials and Pavement Design 15 Table 14. Summary of ANOVA results for TWC and RAP type on dry ITS. Eﬀect on dry ITS Downloaded by [UAE University] at 17:37 25 October 2017 No. of comparison Pair of mixes for comparing Factors p-Value Signiﬁcance 1 Mix 2 vs. 3 at 2%FAC 2 Mix 2 vs. 3 at 3%FAC 3 Mix 5 vs. 6 at 2%FAC 4 Mix 5 vs. 6 at 3%FAC TWC RAP Interaction TWC RAP Interaction TWC RAP Interaction TWC RAP Interaction .256 < .001 .117 .319 .240 .437 .244 .027 .564 .789 .620 .835 N Y N N N N N Y N N N N RAP since it can distinguish RAP with diﬀerent gradations. Compared to RAP type, TWC eﬀect is not signiﬁcant to dry ITS since none of the comparisons shows signiﬁcance of TWC and the interaction. However, the analysis was based on limited types of mixtures with only two RAP types (DM and DC) and two TWC levels (100% and 125% OWC) available for comparisons. 18.104.22.168. Compare TWC with binder type Three pairs of mixtures were evaluated at two FAC levels. Each pair of mixtures has same RAP and FAC level. Binder types used for comparison are 67-22(A) and 58-34(B). Two TWC levels (50% and 75% OWC) were used in comparison 1 and 2. Three TWC levels (75%, 100%, and 125% OWC) were used in comparison 3 and 4. And three TWC levels (100%, 125, and 150% OWC) are used in comparison 5 and 6. ANOVA was performed in each comparison to evaluate the eﬀects of TWC and binder type and the results are summarised in Table 15. Binder type has a signiﬁcant eﬀect on dry ITS in four comparisons and TWC eﬀect is signiﬁcant in ﬁve comparisons. Binder type eﬀect is not signiﬁcant in comparison 4 and 6 where FAC is higher (3%). Therefore, TWC eﬀect on dry ITS is as signiﬁcant as binder type. Mixtures with higher FAC (3%) tend to have same dry ITS regardless of binder type. 22.214.171.124. Determining OTWC As discussed before, maximum dry ITS is corresponding to OTWC. Six foamed asphalt mixtures with combinations of three RAP types and two binder types were tested for dry ITS. For each mixture, OTWC determined based on quadratic regression is termed as measured OTWC. Figure 7 shows an example of the regression relationship between dry ITS and TWCs. The measured OTWC was determined at the peak ITS. Then, the measured OTWC and three factors (binder type, RAP type, and FAC) were used to build a mathematical model by multiple linear regression. The ﬁtted model was used to calculate the OTWC and compare with the measured OTWC. The results are summarised in Table 16. Data in columns 4, 5, and 6 of Table 16 were used to build regression models. Two models built are shown as Equations 1 and 2. Their adjusted R2 are 68.2% and 89.8%, indicating both models have good ﬁtting. Interaction term ‘OWC × FAC’ was included in model 2 because its eﬀect is signiﬁcant when 58-34 binder was used. For all six mixtures, the measured OTWC ranges from 2.07 to 4.51 and calculated OTWC is found between 2.07 and 3.96, which is a little lower than the measured range. 16 W. Ma et al. Table 15. Summary of ANOVA for TWC and binder type. Eﬀect on dry ITS No. of comparison Mixes for comparing 1 Downloaded by [UAE University] at 17:37 25 October 2017 2 Factors p-Value Mix 1 vs. 4 at 2%FAC TWC .011 Y Mix 1 vs. 4 at 3%FAC Binder Interaction TWC .009 .145 .004 Y N Y Binder Interaction TWC Binder Interaction TWC Binder Interaction TWC < .001 .238 .168 .001 .970 .047 .856 .165 .018 Y N N Y N Y N N Y Binder Interaction TWC Binder Interaction .002 .447 .047 .055 .037 Y N Y N Y 3 Mix 2 vs. 5 at 2%FAC 4 Mix 2 vs. 5 at 3%FAC 5 Mix 3 vs. 6 at 2%FAC 6 Mix 3 vs. 6 at 3%FAC Signiﬁcance For mixtures with 67-22 (A): OTWC = 0.143 × FAC − 1.383 × OWC + 7.68. (1) For mixtures with 58-34 (B): OTWC = 3.520 × FAC + 2.750 × OWC − 1.260 × OWC × FAC − 4.37. Figure 7. Regression curve built to determine the measured OTWC (mix 6). (2) Road Materials and Pavement Design 17 Table 16. Summary of OTWC for each mixture. Mix no. 1 2 3 4 5 RAP OWC (%) FAC (%) Measured OTWC (%) Calculated OTWC (%) 67-22 (A) 67-22 (A) 67-22 (A) 67-22 (A) 67-22 (A) 67-22 (A) 58-34 (B) 58-34 (B) 58-34 (B) 58-34 (B) 58-34 (B) 58-34 (B) DF (A) DF (A) DM (B) DM (B) DC (C) DC (C) DF (A) DF (A) DM (B) DM (B) DC (C) DC (C) 4 4 3 3 3 3 4 4 3 3 3 3 2 3 2 3 2 3 2 3 2 3 2 3 2.76 2.26 3.69 4.51 3.63 3.74 3.59 2.07 3.21 3.19 3.51 3.01 2.43 2.58 3.82 3.96 3.82 3.96 3.59 2.07 3.36 3.10 3.36 3.10 Correlation between measured and calculated OTWC was illustrated in Figure 8. The data points are located near to the line of equality and a linear regression ﬁts these data with R2 as high as 86.7%. However, the capability of these two models built based on two binder types and three RAP types to calculate OTWC of foamed asphalt mixtures needs further validation using more types of mixture. 5.2. Validation 5.2.1. Validating with diﬀerent mixture types Table 17 lists information of the mixtures (nos. 7–12) used to validate the two models proposed in this study. Models (Equations (1) and (2)) were applied to predict OTWC of each mixture. Results of measured and predicted OTWC, as well as their diﬀerences are summarised in right three columns of Table 17. Measured OTWC ranged between 2.74% and 5.01% while predicted OTWC ranged between 2.07% and 3.89%. The two models under-predicted OTWC for these mixtures but the diﬀerences are below 0.9% except for the two mixtures with binder 67-22(B). This binder has the same PG as 67-22(A) but a diﬀerent source. The foaming properties such as 5 Calculated OTWC (%) Downloaded by [UAE University] at 17:37 25 October 2017 6 Binder 4 3 y = 0.8659x + 0.4355 R² = 0.8666 2 1 1 1.5 2 2.5 3 3.5 Measured OTWC (%) Figure 8. Comparison between the measured and calculated OTWC. 4 4.5 5 18 W. Ma et al. Table 17. Summary of model validation using diﬀerent mixtures. Mix no. 7 8 9 10 11a 12a RAP FAC Measured OTWC (%) Predicted OTWC (%) Diﬀerence (%) 58-34 (B) 58-34 (B) 67-22 (C) 67-22 (C) 58-34 (B) 58-34 (B) VF (D) VC (E) VF (D) VC (E) DFa (A) DMa (B) 2.5 2.5 2.5 2.5 3.0 3.0 3.33 3.43 4.28 5.01 2.74 3.98 2.83 3.23 2.51 3.89 2.07 3.10 0.50 0.20 1.77 1.12 0.67 0.88 cement was added in mix 11 and 12. ER and HL of this binder were also all less than 67-22(A). Therefore, this foamed asphalt may need more water to distribute uniformly in the mixture. The model should be able to determine OTWC with or without cement since removing cement does not signiﬁcantly aﬀect the prediction accuracy for mix 11 and 12. Additionally, the measured and predicted OTWC were plotted in Figure 9 for correlation. These data points are basically parallel with and close to line of equality, except for one data point representing OTWC of mix 9. The linear relationship between the two OTWC values has an R2 equals 53.3%, which could be improved to 85.3% excluding mix 9. This indicates that these two models may become more accurate if binder source is consistent. 5.2.2. Compare with typical OTWC determining methods Table 18 summarises the information of mixtures available for comparison. OTWC1 indicates 100% RAP OWC by the modiﬁed Proctor method or the Wirtgen’s recommended OTWC. OTWC2 represents the proposed OTWC determined from two models built. OTWC1* and OTWC2* are the actual water contents used to ﬁnd the peak dry ITS, which are the nearest to OTWC1 and OTWC2 (within ± 0.5%), respectively. For comparison with 100% RAP OWC by the modiﬁed Proctor method, eight mixtures had OTWC1 higher than the tested range of water content in this study. Also, other two mixtures without cement were not included because the models still need further calibration for no-cement mixtures. Therefore, only three mixtures are 6 5 Predicted OTWC (%) Downloaded by [UAE University] at 17:37 25 October 2017 a No Binder 4 3 y = 0.5718x + 0.7673 R² = 0.5332 2 Mix 9 1 1 2 3 4 Measured OTWC (%) Figure 9. Correlation between measured and model predicted OTWC. 5 6 Road Materials and Pavement Design 19 Table 18. Comparing proposed OTWC with 100% RAP OWC by the modiﬁed Proctor method and Wirtgen’s method. Mix no. 5 10 13 Downloaded by [UAE University] at 17:37 25 October 2017 2 2 8 9 10 13 Binder RAP FAC OTWC1 OTWC2 OTWC1* Compare proposed method with 100% RAP OWC by the modiﬁed Proctor method 58-34(B) DM 3.0 3.8 3.1 3.8 67-22(C) VC 2.5 5.3 3.9 4.9 67-22(C) DF 2.5 6.5 2.5 7.0 Compare proposed method with Wirtgen (2008) 67-22(A) DM 2.0 3.3 3.8 3.0 67-22(A) DM 3.0 3.3 4.0 3.0 58-34(B) VC 2.5 4.3 3.2 4.3 67-22(C) VF 2.5 5.2 2.5 5.7 67-22(C) VC 2.5 4.3 3.9 4.1 67-22(C) DF 2.5 5.2 2.5 5.0 OTWC2* 3.0 4.1 2.7 3.8 3.8 3.5 2.7 4.1 2.7 available for comparison. For comparison with Wirtgen’s OTWCs, six mixtures with OTWCs in the tested range were used. Dry ITS of mixtures at proposed OTWC and at other two recommended OTWCs were compared in Figures 10 and 11. The percent changes in dry ITS are summarised at the top of columns. Most mixtures show increases in dry ITS at the proposed OTWC except for mix 10. However, statistical analysis did not show signiﬁcantly diﬀerence between dry ITS at the proposed OTWC and dry ITS at other two OTWCs. If the measured OTWC is used instead of the proposed OTWC, dry ITS of mix 10 could be higher (about 258.3 kPa), which would be equal to that at 100% RAP OWC by the modiﬁed Proctor method and higher than that at Wirtgen’s OTWC. Therefore, dry ITS’ decrease of mix 10 shown in Figures 10 may be due to limitations of models’ predicting capability. Further calibration using more binder types and RAP types is needed to make prediction more accurate. Figure 10. Method. Comparing dry ITS at proposed OTWC and 100% RAP OWC by the modiﬁed Proctor Downloaded by [UAE University] at 17:37 25 October 2017 20 W. Ma et al. Figure 11. Comparing dry ITS at proposed OTWC and Wirtgen’s OTWC. 6. Conclusions and recommendations According to the results and analysis presented in this paper, the following conclusions and recommendations are oﬀered: • A new compaction method was proposed to determine OWC of RAP using 30 gyrations in a SGC. Compared to the modiﬁed proctor test method (AASHTO T180), the OWC determined by proposed method is lower. The maximum dry densities determined by two alternative methods are similar while those determined by proposed method are slightly lower. • The eﬀects of water during mixing and compaction were investigated, respectively. The result showed that mixing water has a signiﬁcant eﬀect on density but this eﬀect is not signiﬁcant on dry ITS. When the MWC is the same, the compaction water has a signiﬁcant eﬀect on compacted density and the eﬀect is more signiﬁcant than the eﬀect of FAC. However, the eﬀect of compaction water on dry ITS is not signiﬁcant. Although this evaluation only used one mixture, the method separating the two eﬀects of water would be helpful to future research. • Evaluation of the eﬀect of combination of water (including both mixing and compaction eﬀects) showed the TWC has a signiﬁcant eﬀect on density, dry and wet ITSs for three out of six mixtures. Although this eﬀect is not as strong as the eﬀect of FAC, the OTWC needs to be carefully determined for the mixture. • Besides TWC and FAC, the material types (RAP type and binder type) were also found to be signiﬁcant factors to dry ITS. However, their levels of signiﬁcance were found to be dependent on the FAC in the mixture. For mixtures with higher FAC, these RAP type and binder type were less signiﬁcant. Both factors were considered in modelling OTWC. Binder type was considered as a categorical factor and RAP type was represented by the OWC as a continuous factor. • Two models for determining OTWC using diﬀerent binder types were built by linear regression analysis. The relationship between OTWC and two determining factors (OWC and FAC) was established. Prediction capability of the models were validated by six different mixtures and good correlation between the predicted and the measured OTWC was found, although two of six mixtures had lower predicted OTWC due to diﬀerent binder source. Further comparisons were conducted to validate the improvement in mixtures’ Road Materials and Pavement Design 21 performance (dry ITS) using the predicted OTWC by the proposed method. Dry ITS at two diﬀerent OTWCs (100% RAP OWC and Wirtgen’s recommended OTWC) based on the modiﬁed proctor test were used for comparison. Dry ITS results were found to increase up to 22% at proposed OTWC. However, one mixture showed a decrease in dry ITS in both comparisons, which may be due to change of binder source. This indicates the developed model was not able to provide an accurate prediction for binders with same PG but diﬀerent sources. Therefore, further calibration using more binder sources is required. Acknowledgements Downloaded by [UAE University] at 17:37 25 October 2017 The authors thank Federal Highway Administration for supporting this research. The authors also acknowledge Buzz Powell, Jason Moore, Adam Taylor, and Jason Nelson from NCAT for their advice and assistance during this research. Disclosure statement No potential conﬂict of interest was reported by the authors. References Asphalt Academy. (2009). Technical guideline: Bitumen stabilised materials (2nd ed.). Pretoria: Asphalt Academy. Asphalt Recycling & Reclaiming Association (ARRA). (2015). Recommended mix design guidelines for cold recycling using bituminous recycling agents. Annapolis, MD: Author. Chan, P., Tighe, S., & Chan, S. (2010). Exploring sustainable pavement rehabilitation: Cold in-place recycling with expanded asphalt mix. 89th Annual Meeting Compendium of Papers, Transportation Research Board. Cross, S. (2003). Determination of Superpave® gyratory compactor design compactive eﬀort for cold inplace recycled mixtures. Transportation Research Record: Journal of the Transportation Research Board, 1819, 152–160. Csanyi, L. H. (1957). Foamed asphalt in bituminous paving mixtures (Highway Research Board Bulletin 160). Washington, DC: Highway Research Board. Diefenderfer, B. K., Bowers, B. F., Schwartz, C. W., Farzaneh, A., & Zhang, Z. (2016). Dynamic modulus of recycled pavement mixtures. Transportation Research Record: Journal of the Transportation Research Board, 2575, 19–26. Jenkins, K. J. (2000). Mix design considerations for cold and half-warm bituminous mixes with emphasis of foamed bitumen (PhD dissertation). University of Stellenbosch. Mohammad, L. N., Abu-Farsakh, M. Y., Wu, Z., & Abadie, C. (2003). Louisiana experience with foamed recycled asphalt pavement base materials. Transportation Research Record: Journal of the Transportation Research Board, 1832, 17–24. Muthen, K. M. (1998). Foamed asphalt mixtures: Mix design procedure (Contract report CR-98/077). Pretoria: CSIR TRANSPORTEK. Newcomb, D. E., Arambula, E., Yin, F., Zhang, J., Bhasin, A., Li, W., & Arega, Z. (2015). Properties of foamed asphalt for warm mix asphalt applications (NCHRP report 807). Washington, DC: Transportation Research Board. Sakr, H. A., & Manke, P. G. (1985). Innovations in Oklahoma foamix design procedures. Transportation Research Record: Journal of the Transportation Research Board, 1034, 26–34. Stroup-Gardiner, M. (2011). Recycling and reclamation of asphalt pavements using in-place methods (NCHRP synthesis 421). Washington, DC: Transportation Research Board. Thenoux, G., González, Á, & Dowling, R. (2007). Energy consumption comparison for diﬀerent asphalt pavements rehabilitation techniques used in Chile. Resources, Conservation and Recycling, 49(4), 325–339. doi:10.1016/j.resconrec.2006.02.005 Wirtgen GmbH. (2008). Suitability test procedures of foam bitumen using Wirtgen WLB 10S. Windhagen: Wirtgen GmbH. Wirtgen GmbH. (2012). Wirtgen cold recycling technology. Windhagen: Wirtgen GmbH.