Investigation of Air Void Structure in Double Layer Porous asphalt based on X-ray Computed Tomography
Shuchen Gong
Master Thesis
KTH Royal Institute of Technology
School of Architecture and the Built Environment Department of Civil and Architectural Engineering Division of Building Materials
SE-100 44 Stockholm, Sweden TRITA-ABE-MBT-20561
© 2020 Shuchen Gong. All rights reserved. No part of this thesis may be reproduced without permission from the author.
ABSTRACT
The X-ray computed tomography is a technology to investigate air void structure of drilled asphalt cores, which provides a non-destructive alternative to traditional laboratory methods, usually destructive testing.
In this work, six in-situ specimens from a double layer porous asphalt pavement in Linköping, Sweden, were taken and analysed using both qualitative and quantitative methods of tomography. The qualitative study focused on identifying different features of the drilled cores, such as densification and air voids in the interface between the two porous layers. In the quantitative study, the air voids contents were quantified from processed tomography images. The tomography results of air voids content in all three directions (X, Y and Z), when increasing the calculated number of slices from 4 to 8, were compared to the measured air voids contents from a standardized laboratory method. Both t-test and F-test were applied to determine if a significant difference was found. Besides, the Evaluation Index (EI) was introduced to determine the most accurate combination slices and directions.
The results shown here indicate that a reduced number of tomography slices can give enough accuracy in the determination of air voids content for the porous layers. The results also showed that air voids content determined with tomography gave no significant difference compared to the laboratory results. The most accurate combination found was, in this case, the Y+Z direction. Future development will focus on automatizing the determination of air voids, as well as mastic and aggregate phases using the same methodology of comparing tomography results with laboratory results.
Key words: Tomography, non-destructive testing, destructive testing, air voids content, porous asphalt drilled cores
SAMMANFATTNING (IN SWEDISH)
Datortomografi (även kallad skiktröntgen) är en teknik för att undersöka hålrumsstrukturen av asfaltsborrkärnor, vilket ger ett icke-förstörande alternativ till traditionella, vanligtvis destruktiva, laboratorieprov.
I detta arbete togs sex borrkärnor från en dränerande beläggning lagd i två lager i Linköping vilka analyseras med hjälp av både kvalitativa och kvantitative metoder genom att använda datortomografi. Den kvalitativa studien fokuserade på att identifiera olika särdrag hos borrkärnorna såsom förtätning och hålrum i gränssnittet mellan de två dränerande lagren. I den kvantitativa studien kvantifierades hålrumshalten med hjälp av bearbetade datortomografibilder. Tomografiresultaten för hålrumshalten i alla tre riktningar (X, Y och Z), när antalet bilder ökade från fyra till åtta, jämfördes med den uppmätta hålrumshalten från en standardlaboratoriemetod. Både t-test och F-test användes för att undersöka om en signifikant skillnad fanns. Ett Evaluation Index, EI, introducerades för att avgöra vilken kombination av bilder och riktningar som hade lägst avvikelse.
De presenterade resultaten indikerar att ett mindre antal av datortomografibilder kan ge tillräcklig noggrannhet för att bestämma hålrumshalten för de dränerande lagren. Resultaten visar även att hålrumshalten som bestämts med datortomografi inte har en signifikant skillnad jämfört med laboratorieresultaten i denna studie. Fallet med högst noggrannhet var, i denna studie, kombinationen av Y + Z riktningarna. Vidare utveckling av metoden kommer att fokusera på att automatisera bestämmandet av hålrumshalten. Utvecklingen kommer även att fokusera på att bestämma mastix- och ballastfaserna med hjälp av samma metod som för bestämmandet av hålrumshalten samt att jämföra dessa resultat med traditionella laboratorietest.
Nyckelord: Datortomografi, icke-destruktiv provning, destruktiv provning, hålrumshalt, borrkärnor från dränerande beläggning
PREFACE
This master thesis has been carried out at KTH Department of Civil and Architectural Engineering, Division of Building Materials, in Stockholm. In addition, the Swedish National Transport Laboratory (VTI) is in collaboration.
I would like to express my sincere gratitude to three important groups of people: my KTH professors, VTI group and my family, without whom this thesis would not be performed smoothly. Due to the severe situation of COVID-19, all of you have shown greatest support and doubled the effort to help me finish the thesis.
First and foremost, I would like to thank my supervisor Prof. Alvaro Guarin for his excellent guidance. Thank you for giving me the opportunity to work as your student and investigate such an interesting topic. I would like to thank Prof. Denis Jelagin for examine my thesis. Dr.
Abiy Bekele’s assist in image processing software is acknowledged.
Secondly, I would also like to appreciate Prof. Sigurdur Erlingsson. Thank you for raising a lot of new ideas to make the thesis better and letting me think more comprehensively. Dr. Joacim Lundberg and Tiago Vieira are also appreciated. Thank you for giving many pertinent comments on the structure and details of my paper and giving suggestions when I was struggling. I also thank Dr. Olle Eriksson for his advice and review in statistics and significance tests.
Finally, I would like to thank my boyfriend Xuechang Shi and my family Shifeng Gong and Xingli Chen for their great understanding and support in life and in spirit during the hard time of COVID-19.
Wuhan, June 2020 Shuchen Gong
LIST OF ABREVIATIONS
CT Computed Tomography
CPX Close-Proximity tyre/road noise measurement DLPA Double Layer Porous Asphalt
EI Evaluation Index HMA Hot Mix Asphalt MSE Mean Square Error
NMAS Nominal Maximum Aggregate Size OGFC Open-Graded Friction Course SBS Styrene-Butadiene-Styrene SE Standard Error
SSE Sum of Squares of Error SSX Sum of Squares of X SLPA Single Layer Porous Asphalt SMA Stone Mastic Asphalt
SPB Statistical Pass-By roadside measurements
TRRL the United Kingdom’s Transport Research Laboratory VTM Voids in Total Mix
CONTENTS
ABSTRACT ... iii
SAMMANFATTNING (IN SWEDISH) ... iv
PREFACE... v
LIST OF ABREVIATIONS ... vi
CONTENTS ... ix
1. Introduction ... 1
1.1.Background ... 1
1.2.Aim ... 3
1.3.Methodology ... 4
1.4.Limitations ... 7
2. Literature Review ... 8
2.1.Traditional surface courses ... 8
2.2.Double-layer porous asphalt ... 10
2.2.1. The origin of Double layer porous asphalt ... 10
2.2.1. The material and structure of the double layer porous asphalt ... 11
2.2.2. Air void structure ... 12
2.2.3. The noise reduction effect ... 15
2.3.X-ray Computed Tomography ... 17
3. Site description ... 19
3.1.General description ... 19
3.2.Material properties ... 21
3.3.Sample laboratory analysis ... 23
4. Methods ... 24
4.1.Traditional laboratory determination of air voids content ... 24
4.2.Tomography ... 26
4.2.1. CT slices acquisition ... 26
5.2.Qualitative Analysis ... 43
5.2.1. Features of drilled cores ... 43
5.2.2. Characteristics ... 45
5.2.3. Discussions ... 45
5.3.Quantitative Analysis... 47
5.3.1. Layer thickness ... 47
5.3.2. Significance test ... 49
5.3.3. Accuracy analysis ... 51
5.3.4. Results of the air voids contents ... 52
6. Conclusions and Recommendations ... 53
6.1.Conclusions ... 53
6.2.Recommendations ... 55
Reference ... 56
Appendix A Gradation curves ... 61
Appendix B Calculation Table in the quantitative analysis ... 63
Appendix C Characteristics of the drilled cores ... 80
Appendix D Results of linear regression ... 84
Introduction
1.1. Background
With the rapid development of economy and society, people’s travel demands are increasing [1]. As the number of trips increases, people begin to pay attention to the environmental problems caused by traffic. For instance, the problems can be the sudden environmental pollution in leakage and seepage of fuel and different kinds of lubricants caused by traffic accidents or the impact on ecology, such as animal migration [2]. Another example is non- exhaust dust generation and suspension from the road surface that lead to air pollution [3].
In addition, traffic noise on the environment, especially for the residents along the routes, is harmful which has become a serious impact on both physical and mental health of residents as a traffic environmental problem [4].
Road traffic noise is a kind of unsteady noise, its sound level is generally between 60-85 dB (A) [5]. Some 20 % and 30 % of the EU population are exposed to noise levels higher than 65 dB(A) in the daytime and 55 dB(A) in the night-time [6]. The harm of road traffic noise to a human can be divided into direct harm and indirect harm. The direct harm mainly: (1) affects the sleep and the concentration during study; (2) affects the hearing, so that people may suffer from hearing loss; (3) brings adverse effect to the mood, such as making the person agitated [7]. Indirect harm mainly: (1) increases the risk of physiological diseases, such as blood pressure increase and abnormal hormone secretion [8]; (2) may lead to cardiovascular diseases [9], respiratory diseases [10], type 2 diabetes [11], and adverse birth outcomes [12]
amongst others.
Apart from noise pollution, particle emissions are a challenging problem faced by countries all over the world. In Sweden, studded tyres are mounted to increase skid resistance (50-99
% of light vehicles use studded tyres in winter conditions, depending on the region in Sweden [13]). In contact with studded tyres, the wear rate of pavement increase, which leads to durability deterioration and generation of road dust and non-exhaust PM10 (Particulate Matter with an aerodynamic diameter smaller than 10 µm) [14]. Large-scale tests in a laboratory setting have shown the impact of both the increased use of studded tyre as well as the pavement type and inherent material types and properties [15].
In order to mitigate noise and particle emissions, the Linköping municipality in Sweden is required to develop an action plan to mitigate noise according to EU regulatory framework [16]. Since the main source of traffic noise for high speed roads is from tyre/road interaction, which accounts larger than engine noise [17], several measures to reduce noise at its’ source are put forward in the action plan. Porous asphalt pavement is one of the measures, which is expected to reduce noise levels by 4-8 dB [18]. During this work, part of the Industrigatan road was identified as suitable for a low-noise pavement (described in more detail in Section 3. Thus, Linköping municipality replaced an abrasion wear resistant stone mastic asphalt (SMA) with a double layered porous asphalt (DLPA) during August 2018.
Noise measurements were carried out at three different occasions: the first one was on 6th of June, 2018 before the porous pavement was constructed (pavement replacement started on the 6th of August 2018, week 32); the second measurement was after the porous pavement was laid on 19th of September 2018, and the third one was carried out in 2019. The complete follow up includes one measurement per year for a total of five years. The initial results indicate that the porous pavement results in a noise reduction of up to 5 dB for light vehicles, and up to 4 dB for heavy vehicles [15]. But a substantial acoustical inhomogeneity is also shown along the DLPA. Noise measurements and the drilled core analysis suggest that the bottom layer does not meet the designed interconnected air void content.
Based on previous work of X-ray CT, such as porous asphalt clogging performance and aggregates characteristics, a new procedure to investigate internal properties of porous asphalt can be put forward. For this purpose, six in-situ drilled cores of the double layered porous asphalt pavement from Industrigatan were taken and sent to KTH Royal Institute of Technology for testing. X-ray Computed Tomography (CT), a non-destructive technology (described in more detail in Section 4.2), was performed to investigate drilled asphalt cores properties. Compared to traditional laboratory methods, X-ray CT is able to investigate test objects without physically cutting or slicing. Since the composition of asphalt mixture can be divided into three major materials: mastics (i.e. bituminous binder and filler), aggregates, and air voids, all which have great differences in density, it can be accurately distinguished in CT scan images. Therefore, X-ray CT is an appropriate technique to visualize features in the interior of opaque solid objects to obtain digital information on their 3D geometry and properties, such as asphalt mixtures [19]. Using after image processing, different features of the drilled cores can be identified. In order to evaluate X-ray CT to investigate asphalt mixtures, air voids contents are quantified and the tomography results are compared with laboratory results for exactly the same drilled cores.
1.2. Aim
The overall aim of this master thesis is to develop a more holistic approach regarding pavement properties through a new technique: X-ray computed tomography. The research question of this thesis is: How can X-ray computed tomography be used to study the air void structure of a double layered porous asphalt?
This thesis will focus on the evaluation of X-ray CT scanned slices, air voids in particular, by calculating different kinds of air voids content and comparing those with results from traditional laboratory methods. To achieve the overall aim, the following objectives are used.
Objective 1: To explore the X-ray CT scanned slices to capture features, such as densification, and summarize characteristics and similarities of the drilled cores.
Objective 2: To quantify different kinds of air void contents, such as total and interconnected air void contents in the top and bottom layer respectively and the overall air void contents in every drilled core.
Objective 3: To compare tomography results of air voids content in top and bottom layer with laboratory results to investigate if CT is eligible to study internal properties of asphalt mixture core, regarding air voids content.
1.3. Methodology
This thesis is establishing a method to evaluate the air void structure of asphalt mixtures.
Based on the objectives, a methodology is proposed to give an overview for the choices of methods. Here, the study of the air void structure can be divided into two parts:
interconnectivity and air voids content.
Figure 1-1 illustrates a technology flowchart of the proposed research methodology used in this thesis.
Qualitative Analysis
Quantitative Analysis
Features Summary Problem Analysis
Research
Question Aims
Literature Review X-ray Scanning
Results
Interpretation and Conclusion
Slices Scrutiny
Volumetric Analysis
Accuracy
A literature study on how to investigate the internal properties of asphalt mixtures is performed in Section 2.3 to give some insight on the research technique. In order to analyse the internal structure of asphalt mixtures, both qualitative and quantitative research are carried out after all drilled cores are scanned through X-ray CT. The qualitative analysis is conducted first to help understand the overall internal structure and capture the characteristics of an asphalt mixture core. The following quantitative analysis aims to confirm the method to evaluate air voids content feasibility. The qualitative research deals with words and meanings to explore concepts in depth, such as densification, while the quantitative research deals with numbers and statistics to answer a research question by collecting and analysing data, such as interconnected air void content. In a word, qualitative methods are more flexible, while quantitative methods are more reproducible [20].
Each method has advantages and disadvantages. Comparisons between qualitative and quantitative analysis are conducted in the following aspects:
Accuracy and bias
Accuracy refers to the deviation of a measurement from its true value [21]. Such a deviation can have two components: a systematic and a random one. The systematic component is known as bias, which is discussed here. Since the scanned slices in qualitative analysis are numerous and are observed by a human, it is likely to add personal bias. Quantitative analysis can help make the results more accurate as it is difficult to add personal bias to numbers obtained when the correct data gathering procedures have been followed. Avoiding personal bias and confirmation bias in the case of analysis means that the data will be more accurate, as bias happens even when the results are managed to ensure that there is no bias impact [22].
Application range
In the quantitative analysis, only the air voids content is computed to evaluate the internal structure of the drilled cores. However, other internal properties can be observed during the qualitative analysis, including densification, shape of the air voids etc. In other words, the qualitative analysis gives a general impression of the air void structure while the quantitative analysis focuses on qualifying different kinds of air voids content.
Validity
In quantitative analysis, validity is the extent to which any measuring instrument measures what it is intended to measure, referring to whether a study can be replicated [23]. The calculation of average air voids content by increasing the number of slices is a replication method, which can minimize the error and ensure that the results do not happen by accident.
In qualitative analysis, validity is when a researcher uses certain procedures to check for the accuracy of the research findings [24]. However, it is difficult to check the findings since there
are numerous slices contained with too much information. Therefore, validity in quantitative analysis is relatively better than qualitative analysis.
Reliability
In quantitative research, one type of reliability identified by Kirk and Miller [25] is the similarity of measurements within a given time period. Here, if the measurements are repeated, the results of air void contents are the same. Even if different slices are used for calculation, the results are similar. In qualitative research, such definition of reliability is challenging and epistemologically counter-intuitive, so the essence of reliability for qualitative research lies with consistency [26]. However, there is so much information in a slice that the observer may notice different features when studying the same slice at different times. Therefore, reliability is better in quantitative research than qualitative research.
The detailed method for qualitative and quantitative analysis will be further discussed in Chapter 4.
1.4. Limitations
The thesis only focuses on one type of pavement: porous asphalt pavement. The other types of asphalt pavement, such as hot mix asphalt pavement, and cement concrete pavement are not investigated.
In order to investigate the newly constructed DLPA pavement, six cores were drilled, which are located in left wheel track, right wheel track and between the wheel track in the northbound lane, with one set (3 drill cores) taken more at the southern part and one set (3 drill cores) at the northern part. However, the samples may not be well representative for the entire length of the road section due to the difference in each place of the pavement.
Within the context of the thesis, the emphasis is on the air voids structure. The qualitative and quantitative analysis only focuses on the air voids content. Other parameters, such as size and number of air voids, are not taken into consideration. However, the air voids content is not the only factor affecting the performance of a porous asphalt. Even with the same air void content, the performance can be different. The air voids distribution is also known to affect the performance of asphalt mixtures [27]. Hence, the analysis is only the results of partial investigation of the inner structure.
In the quantitative analysis, the manual calculation procedure has limitations. For example, only a limited number of slices are used for calculation in order to balance accuracy and workload. The regions that are considered as air voids are decided manually based on grayscale level, which might lead to inaccuracy of the results since the procedure is not automized. Besides, the analysis is done in 2D which simplifies the procedures to identify interconnected air voids. But interconnected air voids are roughly estimated because the neighbouring slices are not considered.
Kommenterad [DJ1]: As far as I can see, you focus on one material only
What about HMA, PMB, etc.? Re-write.
Kommenterad [DJ2]: The major limitation here is that analysis is done in 2D only
I think you should indicate that.
Literature Review
1.5. Traditional surface courses
Hot mix asphalt surface course is the most commonly used type pf pavement [28]. Hot mix asphalt or HMA is the designation given to asphalt mixtures that are heated and poured at temperatures between 300 and 350 ℃. There are many different types of HMA pavements, which are mainly reflected in the maximum aggregate size, aggregate gradation and asphalt binder content/type. The most common types of HMA pavement are dense graded HMA pavement and open graded HMA pavement (see Figure 2-1) [28].
Figure 2-1 Hot mix asphalt mix type; Left: dense graded; Right: open graded [28].
Dense-graded HMA mix is versatile and the most common and most investigated mix type [28]. Dense-graded mixtures are produced from well-graded or continuously graded aggregates (gradation curves without any sharp slope changes, see the blue lines in Figure 2-2) for general use. Typically, larger aggregates "float" in a mastic matrix consisting of asphalt binder and fine particles. When properly designed and constructed, the dense graded mixture is relatively impermeable [29].
HMA mixtures generally aims at the maximum density. Although it has the advantages of small air voids content and good durability, its high temperature performance is poor. Under the traffic load of heavy vehicles, the traditional continuous dense graded asphalt pavement may be deformed by rutting, waves and shifting due to the lack of thermal stability. In addition, due to the small surface structure depth and poor surface drainage capacity, its skid resistance is low in rainy days, which is likely to cause traffic accidents and affect the driving safety [30].
Open graded HMA mixture has relatively uniform particle sizes, represented by the lack of intermediate-sized particles (the gradation curve almost drops vertically in intermediate size range, see the red line in Figure 2-2). The typical mixture of this structure is the permeable friction course, commonly referred to as the "open graded friction course" (OGFC) and asphalt-treated permeable bases. Open-graded mixtures are commonly used as wearing courses (e.g. OGFC) or as underlying drainage layers due to their particular advantage in porosity [29]. This type of asphalt mixtures is able to provide good skid resistance to sliding and drainage, which helps improve driving safety in rainy days [28].
Open graded friction courses are the basis of porous asphalt pavement, which will be further discussed in Section 2.2.
1.6. Double-layer porous asphalt
1.6.1. The origin of Double layer porous asphaltPorous asphalt is defined as an open graded wearing course with high air voids content (usually > 20%). It provides a good skid resistance and significantly reduces splash and spray in wet conditions [31]. Figure 2-3 illustrates a cross-section of porous asphalt.
Figure 2-3 Typical cross-section of porous asphalt under construction [32]
Porous asphalt was originally developed in Europe. The first porous asphalt pavement was developed by the United Kingdom’s Transport Research Laboratory (TRRL) in the 1940s, which was constructed by the British Royal Air Force to remove rain on airport runaways [33].
In the 1960s, research in Europe began to improve driving safety and comfort of OGFC on the basis of original pavement. Since the OGFC has high air voids content (up to 20 %), rain can seep into the pavement and be drained away to the edge through interconnected air voids.
Therefore, this kind of pavement was given a uniform name by the European community:
Porous Asphalt Mixture [34].
Some studies of porous asphalt pavement were performed in the early 1970s, but the results of TRRL showed that the type of pavement was not effective because the modified asphalt technology was not yet mature [33]. After the development of material technology in the 1990s, the TRRL in the UK carried out a number of tests on porous asphalt pavements, which proved the effectiveness in reducing noise and driving water film in rainy days. TRRL attributed the new results to the excellent performance of polymer modified asphalt, saying that without
aggregate size of 4-8 mm has better noise reduction performance than that of the asphalt mixture with maximum aggregate size of 6-16 mm. Compared with the traditional dense graded pavement, the lifespan of porous pavement is shorter. Based on the findings, the idea of the double layer porous asphalt pavement was developed, which consists of a fine porous asphalt layer and a coarse porous asphalt layer [35].
The first double layered porous asphalt pavement was constructed in The Netherlands. This type of pavement is now serving in many European countries, such as Italy, Denmark, Belgium, Germany, Austria, Sweden, and Switzerland [36][37][38][39]. Several trial sections of double layer porous asphalt were also built in Japan and New Zealand [40].
1.6.1. The material and structure of the double layer porous asphalt
Since the surface of asphalt pavement has direct contact with vehicle tyres, it must provide good skid and wear resistances. The aggregate requirement of double layer porous asphalt pavement is similar to that of the traditional dense graded asphalt pavement, which requires its polishing stone value higher than 53 according to TRRL [41].
Due to relatively large air voids content, ravelling, rutting, and cracking are likely to occur.
Therefore, the quality of the asphalt binder is a critical factor. In Japan, high-viscosity modified asphalt is widely used in the porous asphalt pavement [42]. The styrene-butadiene-styrene (SBS) content of SBS modified asphalt in cold areas should reach 9-12 % to improve the durability [43]. Tsukamoto [40] pointed out that modified asphalt with different binder types should be selected for pavements in different regions. For instance, pavement stripping and deformation should be focused in high temperature and rainy areas, while cracking should be focused in cold areas.
Eijbersen [41] suggested that modified asphalt should also be used in the bottom layer, because water will be trapped in this structural layer for a long time, which aggravates the damage of the pavement structure. It is also possible to add other materials to asphalt binder to improve its service life. In Austria, wood fibre was used as an additive. In the Netherlands, synthetic fibre was used to enhance the skid resistance of the road surface [43].
In order to reduce noise from the road surface and tyre interaction, interconnected air voids on the surface should be small and dense. In the mid-1990s, the top layer of DLPA pavements in the Netherlands was 2-4 mm, and the bottom layer was 11-16 mm. However, it was found that although the structure had a significant noise reduction effect in the initial use, after a few months, the noise reduction performance declined rapidly because air voids of the pavement was relatively small and was easily clogged. In order to avoid clogging, the aggregate size of the top layer was increased to 4-8 mm in most areas [43].
Nielsen et al. [42] investigated the DLPA pavement structures in Japan. They are usually 2 cm thick in the top layer and 5 cm thick in the bottom. The aggregate size is usually 5-8 mm in the top layer and 13 mm in the bottom.
Hamzahmo et al. [44] used a rotating compactor to simulate the effect of excessive compaction of traffic load on single-layer and double-layer porous pavement. It is assumed that over-compaction can affect the air voids content, layer thickness and drainage time.
During compaction, the change of test thickness was observed, and the permeability parameters of the specimen were tested before and after compaction. The test results showed that after compaction, all the specimens became denser and the air void content decreased. The ability to resist over-compaction was the least for the single layer specimen with small aggregate particle sizes and ordinary asphalt binder while the porosity of the SBS modified asphalt mixture decreased less. After compaction, the thickness of the double-layer mixture decreases to the minimum. The anti-compaction capacity of the top layer with a nominal particle size of 10 mm is greater than that of the sample with a nominal particle size of 14 mm, and the anti-compaction capacity of the double-layer sample is lower than that of the single-layer sample.
1.6.2. Air void structure
Air voids are vital in porous asphalt mixtures. Researchers have evaluated the performance of porous asphalt mixtures, including drainability, strength and noise reduction and compared these with the air voids content, which is considered to have a major affect [50]. In general, air voids have a great influence on the drainage performance and noise reduction performance.
In DLPA pavements, air voids are mainly formed by the bottom layer, which contains coarse aggregates while the top layer of the pavement contains fine aggregates. The top layer is designed to avoid clogging and material lost. The fine aggregates in the top layer provide micro-pores absorbing rather than reflecting the noise. The coarse aggregates in the bottom layers create main air voids (see Figure 2-4). Noise is reflected in these main voids through extended transmission paths. The sound energy is dissipated by the friction between the air flow and void walls. The remaining sound energy is transmitted to the next downward layer.
Most of the noise can be absorbed by this pavement surfacing structure [48].
Arambula et al. [45] evaluated the relationship between the air voids distribution in asphalt mixes and their moisture susceptibility. It turned out that the specimen with the smallest range of air void content and air void radius was more susceptible to moisture damage while the specimen with the largest air void radii was the least moisture susceptible.
Liu and Cao [46] studied the material composition and performance of a porous asphalt pavement. They found that only the connected air voids, which connect to the air and therefore allowing water to flow through, play an important role in water drainage. They derived the relation between connected air voids content and permeability coefficient and proposed a relationship model between the two variables, see Figure 2-5, where the permeability coefficient increases with the air void in a nonlinear principle and the connective characteristic has an obvious effect on the drainage efficiency.
Figure 2-5 Relation between connective void and permeability coefficient. Source: [46]
Losa and Leandri [47] conducted experiments to test noise absorption coefficient on porous mixes with different air voids contents. Based on their findings, they proposed a comprehensive model to predict a noise absorption factor of porous mixes. This model takes into account aggregate shape and gradation as well as volumetric composition of the mix compacted in laboratory.
Liu et al. [48] carried out experiments to study the relationship between different air voids, different modified asphalt binder additives and sound absorption. Laboratory results show that the sound absorption coefficient of porous asphalt pavements has a close relationship with the air voids content. The sound absorption coefficients of the samples with low air voids and dense gradation are basically the same. For samples with higher air voids content, the absorption coefficient increases with the increase in air voids content. The porous asphalt
mixture with more than 20 % air voids has a high sound absorption coefficient in the frequency range between 400 - 1000 Hz in particular.
However, porous asphalt mixtures with the same air voids content can have significantly different performance. The microstructure of porous asphalt mixtures (i.e. distribution of air voids, size of air voids, and connectivity) also plays an important role on its performance.
Lefebvre [49] illustrates that only the interconnected air voids open from the surface are in favour of removing water and enhancing noise absorption. Figure 2-6 and Table 2-1 classify air voids. Voids a and b1 look similar from the surface, but b1 is semi-effective in water permeability. The influence of voids b2 and b3 on noise absorption remains uncertain This characterization of air voids is important to estimate interconnectivity and derive relationship between different kinds of air voids and noise absorption.
Figure 2-6 Air voids classification. a are continuously connected air voids, b are not continuous connected air voids, c are air voids not accessible to water and d are fully closed air voids. Source:
[49]
Table 2-1 Influence of Air Void Classification on Permeability and Noise [49]
Description of voids Type
Effectiveness with regard to
water
permeability noise absorption
“open”
(accessible)
continuously connected a + +
not continuous
accessible from the
surface b1 - +
entering the section b2 - ?
“blind canal” b3 - ?
“closed” not accessible to water under
atmospheric pressure c - -
full closed d - -
Jiang et al. [50] investigated the correlation among material composition, microscopic air voids features, and material permeability, acoustic absorption performance. They found that:
(1) the microscopic void features were significantly affected by gradation and nominal maximum aggregate size (NMAS); (2) in porous asphalt concrete mixtures with the same air voids content and NMAS, the coarser graded mixture had the larger equivalent diameter of air voids; (3) a linear relationship exists between equivalent diameter and material properties, including cohesion loss, dynamic stability, shear strength, anti-clogging performance and noise reduction.
1.6.3. The noise reduction effect
The main purpose for applying two-layer porous asphalt is noise reduction. The noise reduction mechanism of DLPA pavements is mainly through the air voids. When the sound wave reaches the pavement, part of it is reflected back from the surfaces and another part is transmitted through the interior structure of the pavement. During the transmission, the air in the voids vibrates. As a result of the friction on the solid wall, the sound energy is converted into heat energy. When the reflected sound wave reaches the surface a second or more times, the reflected sound wave may be transmitted to the surface again. Sound energy is transferred and dissipated, in other words, absorbed by the pavement [51].
Morgan [52] tested the noise reduction performance of the DLPA pavement in Sweden.
Statistical Pass-By (SPB) roadside measurements, the Close-Proximity (CPX) tyre/road noise measurements as well as tests for sound absorption on absorptive road surfaces was used.
For the DLPA pavement, the NMAS of top layer is 11 mm and the NMAS of bottom layer is 16 mm. Compared with SMA pavement, the noise generated by the vehicles is reduced by 9.1
dB(A) when passing at 110 km/h, and by 7.6 dB(A) after two years from construction of DLPA.
As a comparison, the initial noise reduction of single layer porous asphalt (SLPA) pavement (the NMAS is 11 mm) was 4.1 dB(A) and 3.4 dB(A) after two years. Then the noise reduction of SLPA pavement was increased to 5.2 dB(A) after four years. The increase of the SLPA pavement may be caused by the change of the absorption spectrum frequency caused by the pavement being clogged. When the pavement is new, the absorption spectrum frequency is different from the traffic noise frequency. When the pavement is clogged, the absorption spectrum frequency is similar to the traffic noise frequency, so the noise reduction performance is enhanced.
Nielsen et al. [53] used the SPB method to measure noise level of a DLPA pavement. Compared with the dense graded asphalt concrete pavement with NMAS of 13 mm, the noise generated on the DLPA pavement of light vehicles is reduced by 4 –7 dB(A), and that of heavy vehicles is reduced by 2 –5 dB(A).
Blokland and Peeters [54] studied the noise reduction performance of 1000 km DLPA pavement in the Netherlands. The results showed that the noise reducing capability can last for up to 10 years. DLPA is a standard solution for noise hotspots along highways in the Netherlands, and even after considering more frequent maintenance than standard dense asphalt concrete pavement, its life cycle cost is lower than noise barriers. The standard DLPA model has a 4/8 mm top layer. An extra 1 dB(A) reduction effect can be achieved when applying 2/6 mm top layer.
Chu et al. [55] investigated the sound improvement of different mix designs. If the porosity was changed from 25 % to 12 %, some changes in the sound absorption frequency characteristics was observed, but the contribution to the tyre-pavement noise reduction was negligible. On the other hand, the noise absorption coefficient and the tyre-pavement noise reductions were significantly decreased due to the clogging of pores. However, changing the thickness of asphalt mixture from 63 mm to 200 mm had no significant effect on sound absorption and the tyre/road noise reduction.
Wang et al. [56] applied finite element method to simulate acoustic absorption properties of a porous asphalt pavement structure. The noise reducing characteristics of the combination of porosity and the thickness of double layer porous asphalt pavement were analysed. The results indicated that the double layer porous asphalt pavement has better sound absorption performance when the surface porosity is 10 %, the bottom porosity is 25 %, the surface thickness is 2 cm, and the bottom thickness is 4 cm.
1.7. X-ray Computed Tomography
X-ray CT is a completely non-destructive technique used to display features inside opaque solid objects to obtain digital information about their three-dimensional geometry and properties [19]. The components of X-ray CT are shown in Figure 2-7. It consists of a X-ray source and a detector, with test samples placed in between [57].
Figure 2-7 Components of x-ray computed tomography system. Source: [57]
The X-ray source provides the energy beam that penetrates the specimen and provides the condition for the attenuation of X- ray in the specimen, which is basic for the construction of CT scan image [58].
The collimator can be divided into an anterior collimator and a posterior collimator. The anterior collimator diverges the conical ray into the fan ray, and the posterior collimator is used to screen the scattered signal [58]. The posterior collimator is generally narrower than the anterior collimator, so the effective size of X-ray beam is determined by the posterior collimator. The combination of anterior and posterior collimators can improve the imaging quality of CT system, which is decisive for the correct balance of various technical indexes of CT system [59].
The detector system is used to receive and measure the attenuation of X-rays. After the amplification and modulus of the system, it is transferred to the computer system for CT image reconstruction. The number of detectors directly affects the clarity of image reconstruction and the time required for image reconstruction [58].
During the CT scan of the specimen, the scanning system will translate or rotate it and adjust the spatial position of the X-ray source, specimen and detector. The image of each layer of the specimen can be obtained from multiple angles to provide imaging accuracy and reduce the noise disturbance [58].
The working mechanism of CT is as follows: when the X-ray source emits X-rays to the specimen on the detection platform, they will be partially absorbed after passing through the specimen. Due to the density difference in material, the X-ray absorption coefficient is
different. Therefore, the intensity of X-ray after attenuation varies with the material. When the detection platform rotates with a constant speed, the intensity of the remaining X-ray at different position will be transferred to the computer system. The computer system will calculate the absorption coefficient matrix, which can reflect the density distribution inside the specimen, so as to realize the imaging of CT scan [59].
CT scanning technology has been proven to be useful for asphalt mixtures [60]. It has been used in several studies to investigate the mesoscopic structure in asphalt concrete in terms of air-void space[61], spatial distribution of air content under axial loading [62], and clogging [63].
CT scanning technology has many advantages. It can obtain the internal structure of the specimen, such as aggregate distribution [64], distribution characteristics of air voids and cracks [65], and better reflect the contact form of the aggregate including contact distance distribution, contact length distribution and contact orientation [66]. This technique is non- destructive that requires no preliminary physical treatment of specimens, such as cutting and polishing [67]. CT scanning technology is almost immune to environmental changes, and can accurately reflect the characteristics of the aggregate, making the research results more accurate [68][69].
However, the cost of CT scanning technology is high, and the resolution of detailed features in the scanning process is also affected by the size of the specimen [70]. The major limitations in accurately describing the air voids content of asphalt mixtures are the large range of void sizes that exist in such materials. As mentioned above, nanoscale air voids content cannot be measured with this technique, only microscopic and macroscopic air voids content can be measured. The second limitation associated with this is that the sample size scales with achievable resolution: small sample sizes are required for better quality to characterize small air voids (10 mm or smaller), which may lead to the loss of macro information or sampling problems. Therefore, the scan strategy and resolution is crucial in best making use of the technique for air void characterization [71].
Site description
1.8. General description
The investigated DLPA pavement is located on the Industrigatan road in northeast Linköping, Sweden (see Figures 3-1 and 3-2). It was partly constructed over the old SMA pavement after some treatment. The top layer of the DLPA pavement under study was designed to be 25 mm thick with maximum aggregate size of 11 mm, while the bottom layer was designed to be 55 mm with maximum aggregate size of 16 mm respectively, both with a designed air voids content of 25 %. The cross-section sketch is shown in Figure 3-3.
Figure 3-1 Photo of the site (Photo by Joacim Lundberg)
Figure 3-2: Location of the double layer porous asphalt section and the air quality measurement stations. Source: [15]
Figure 3-3 Cross section properties sketch
The road section is approximately 600 m long and has average annual daily traffic (AADT) of 14 700 vehicles, of which 7 % are heavy vehicles. In winter conditions, it is allowed to use studded tyres to provide adequate friction levels even with snow or ice on the road surfaces in Sweden. It is estimated that between 51 and 66 % of the light vehicles that drive over this road section with studded tyres. The use of studded tyres accelerates the process of the road surface worn out, thus affecting pavement abrasion wear particle and road dust generation, while also contributing to noise emissions [15].
The Swedish National Road and Transport Research Institute (VTI) has been commissioned to monitor and evaluate the noise reduction properties during its lifespan by the municipality of Linköping. At the same time, VTI decided to follow development of the road dust loading, mobile measurement of the suspension of road dust and stationary measurement of air quality of this pavement. As a reference for the road dust loading and the air quality measurements, the southern part was used, where the old pavement of type SMA was still intact, see Figure 3-2. For noise, the reference is the old pavement (SMA) prior to the DLPA.
Measurements suggest that this SMA seems to have a 16 mm nominal maximum aggregate size [15]. Six drilled cores were taken for investigation: located in left wheel track, right wheel track and between the wheel track in the northbound lane, with one set (3 drill cores) taken more at the southern part and one set (3 drill cores) at the northern part.
Top layer:
Thickness:25 mm Maximum aggregate size: 11 mm
Air void content: 25%
Bottom layer:
Thickness:55mm Maximum aggregate size: 16 mm
Air void content: 25%
Dense layer
1.9. Material properties
In mix design, different materials are used in the top and bottom layer respectively. It was constructed during august 2018 and the design values are shown in Table 3-1.
Table 3-1 Pavement recipe and design value
Top layer Bottom layer
Aggregate
Grain Density (g / cm3) 2.76
Tile Index (FI) <20(<=8 mm)
Cross Area (C) 100/0
Ball Mill Value (AN) <10 (8-11.2 mm) <10 (10-14 mm) Micro-Deval Value (Mde) <10 (10-14 mm) <25 (10-14 mm) Los Angeles Value (LA) <25 (10-14 mm) <10 (11.2-16 mm)
Asphalt binder
Binder Density (Mg / m3) 1.02
Binder Content (weight%) 6.6 6.4
Bulk Density (Mg / m3) 1.865
Compact Density (Mg / m3) 2.425
Marshall Cavity (Vol%) 23.1
Adhesion Number (ITSR%) >75
Additives
Wetfix BE Lev. Akzo Nobel 0.30%
Cellulosafiber, Onway 0,6 vikt%
Cement, Cementa 1%
Others
SS-EN 12697-13 Temperature (°C) 150-190
SS-EN 12697-1 Min binder content (%) 6.2 5.8
SS-EN 12697-8 Min air void content (%) 20
SS-EN 12697-8 Max air void content (%) 26
The design gradation for the top and bottom layer in DLPA pavement is shown in Table 3-2, Error! Reference source not found. and Figure 3-4.
Table 3-2 Design gradation
Sieve, mm 0.063 0.5 2 4 5.6 8 11.2 16 22.4 31.5
Top layer Passing% 4 8 12 16 40 95
Bottom Layer Passing% 3 6 9 11 13 91
Figure 3-4 Design gradation curve for the top layer
Figure 3-5 Design gradation curve for the bottom layer
1.10. Sample laboratory analysis
Sample laboratory quality analysis, including binder content and sieving, was performed after the drilled cores were taken from the field and scanned in the tomography.
Due to the mass required to do the analysis based on each layer, each sample is from the combination of left and right wheel tracks and between wheel track cores. This is valid for binder content and the sieving curve only. The results of binder content and sieving are shown in Tables 3-3 and 3-4. The gradation curves are available in Appendix A.
Table 3-3 Binder content of top and bottom layer in North and South location Location Top layer Bottom layer
North South North South Binder content, % 6.3 6 6.4 6.8
Table 3-4 Sieving results of top and bottom layer in North and South location
Sieve (mm) 0.063 0.125 0.25 0.5 1 2 4 5.6 8 11.2 16 22.4 31.5 45
Passing%
Top layer North 5 5 6 7 9 12 16 22 43 97 100
South 4.5 5 6 7 9 12 16 20 43 96 100
Bottom layer North 5.1 7 9 12 14 19 25 30 36 55 95 100
South 6.8 8 10 12 12 15 19 27 33 42 65 95 100
Methods
1.11. Traditional laboratory determination of air voids content
Asphalt mixture consists of three primary components—the aggregate, the asphalt binder, and air voids. Figure 4-1 shows a simplified schematic of a compacted asphalt mixture. Here, voids in total mix (VTM [%]) are measured in laboratory, which are the total volume of pockets of air in between the asphalt-coated aggregates in a compacted asphalt mix, expressed as a percentage of the total volume of the mix [72]:
𝑉𝑉𝑉𝑉𝑉𝑉, % = 𝑉𝑉𝑎𝑎 𝑉𝑉𝑚𝑚𝑚𝑚∙ 100
where 𝑉𝑉𝑎𝑎 is the volume of air voids (m3) and 𝑉𝑉𝑚𝑚𝑚𝑚 is the total volume of the compacted mixture (m3).
Figure 4-1 Conceptual block diagram. Source: [72].
The following laboratory method to calculate VTM is obtained by the Swedish standard (identical to European standard): Bituminous mixtures - Test methods - Part 8: Determination of void characteristics of bituminous specimens (SS-EN 12697-8) [73]:
The air voids content of a bituminous specimen is calculated using the maximum density of the mixture and the bulk density of the specimen. For mixtures with water in their composition (e.g. mixtures produced with bituminous emulsion or foamed bitumen), the bulk density of the specimen shall refer to its dry bulk density [73].
𝜌𝜌𝑚𝑚 is the bulk density of the specimen, in megagrams per cubic meter (Mg/ m3).
The bulk density is obtained by measuring the weight of the sample in air, in water (after immersing for 3–5 minutes), and then measuring the saturated surface dry weight (after patting the surface dry). The bulk-specific gravity, 𝜌𝜌𝑚𝑚(kg/m3) is determined from the following formula [72]:
𝜌𝜌𝑚𝑚= 𝐴𝐴 𝐵𝐵 − 𝐶𝐶 ∙ 𝜌𝜌𝑤𝑤 where
A is the weight in grams of the specimen in air (g)
B is the weight in grams of the saturated surface dry specimen in air (g) C is the weight in grams of the specimen in water (g)
𝜌𝜌𝑤𝑤 is the density of water (kg/m3)
The maximum density of the mixture is the ratio of weight of a unit volume of an uncompacted asphalt mixture sample to the weight of an equal volume of water at the same temperature.
The mixture is weighed in air, and then placed in a pycnometer with sufficient water to cover it. A vacuum pump is then connected to the pycnometer and turned on to remove entrapped air from the mixture. The mixture is agitated during this procedure to help remove the air. At the end of this step, the pycnometer is submerged in a water bath, and its weight (along with the mix inside it) is noted. The following expression is used for the calculation of 𝜌𝜌𝑚𝑚 (kg/m3) [72]:
𝜌𝜌𝑚𝑚= 𝐴𝐴 𝐴𝐴 − 𝐵𝐵 ∙ 𝜌𝜌𝑤𝑤 where
A is the weight of sample in air (g) B is the weight of sample in water (g) 𝜌𝜌𝑤𝑤 is the density of water (kg/m3)
1.12. Tomography
1.12.1. CT slices acquisitionIn this study, the KTH X-View TM X5000-CT X-ray CT scanner (see Figure 4-2) is used to obtain the detailed internal structure of the drilled porous asphalt cores from the Industrigatan road for further analysis. The X5000 CT scanner is a seven-axis X-ray imaging system designed to examine objects, which can accommodate to various shapes, sizes and weights of parts and produce up to 450kV of X-ray intensity.
Figure 4-2 X5000-CT scanner in KTH.
In order to obtain the CT slice, the following procures are needed: warming up the system, scanning, calibration and reconstruction.
A series of preparations and parameter setting are required before CT scanning of asphalt mixture specimens can be performed. The scanning process is presented below [59]:
(1) Start the X-ray scanner and the computer system. Select the warm-up mode according to the time since the last shutdown.
(2) Place the specimen on the rotating table (Figure 4-3). To allow the entire specimen to be exposed to X-rays, a lightweight foam block is placed on the rotating table and placed on one side of the specimen to make it tilt. In the subsequent processing, the imaging part of the foam
Figure 4-3 Left: X-ray sources and rotational table; Right: detector.
(3) Adjust the number of filter pieces. Filters can absorb low-energy rays, reduce their proportion in X-rays, and improve the quality of CT scans. Using too many filters can have the opposite effect, the optimal range for the number of filters needs to be determined by repeated debugging.
(4) After closing the protective door of the lead room, adjust the horizontal and vertical positions of the specimen, the focal point, rotating table and detector to obtain the best imaging effect. Rotate the rotating table 360 degrees and observe whether the specimen contour is beyond the scope of the imaging window. If there is any part beyond, the position of the specimen needs to be adjusted so that the specimen contour is within the scope of the imaging window at any angle.
(4) Adjust X-ray energy, beam current and focal point. Increasing the X-ray energy results in better penetration, but the detector might be overtuned. Using a small focal point improves the resolution of the image, but the maximum current decreases.
(5) Adjust the integration time. A longer integral time can improve the contrast of the image and reduce the noise, but it will make the scanning process longer and increase the workload of the equipment.
(6) Set the number of projections. Increasing the number of projections results in higher image resolution, but the time required for scanning and data reconstruction increases.
By setting each parameter, the CT scanning process can be completed.
After scanning, calibration is required to determine the characteristics of the X-ray signals read by the detector under scanning conditions and to reduce geometric uncertainty [74]. The two
main signal calibrations are offset and gain, which determine the readings when the X-ray is off and at different energy levels respectively [75].
CT images are obtained after reconstruction, the mathematical process of converting sinograms into two-dimensional slice images. During reconstruction, the original intensity data are converted into CT Numbers or CT values, which are determined by a computer system in a given range [75]. The CT scanner in the KTH laboratory uses a 16-bit scale, allowing values ranging from 0 to 65 535. CT number corresponds roughly with density where, for example, the CT number of air is 0 [76].
A CT image is referred to as a slice. The scanned slices in X, Y and Z direction are shown in Figure 4-4. Slices in X and Z direction are vertical while slices in Y direction are horizontal.
These directions are at a 90-degree angle from each other. Nearly 1500 slices were generated from one drilled sample in one direction.
(a) (b)
The intensity of the X-ray varies as it penetrates the drilled core due to difference in the density of the composition of the asphalt mixture. These different densities are represented by an image of 256 grayscale levels, in which lower density solids are represented by darker colours [77]. Hence, aggregate, mastic and air voids can be distinguished respectively (the black part is air voids, darker grey is likely mastics and brighter grey is aggregates, see for example Figure 4-5).
Three layers can be identified by different materials and air voids. In the top layer, aggregates are pure grey and loosely connected with large air voids. While in the bottom layer, aggregates are partly white and closely connected with a smaller air voids. In the dense layer, aggregates are darker grey with more mastics and almost no air voids.
Since the new DLPA pavement is partly constructed directly on the old SMA pavement, the bottom layer is above the dense layer. Traditionally, the top layer of the pavement contains fine aggregates and the bottom layer contains coarse ones with larger air void contents, which is in contrary to the drilled cores. Therefore, the research focus of this thesis is the top layer and bottom layer of the porous pavement.
Figure 4-5: A slice after X-ray CT scanning. The black parts are air voids and the gray parts are aggregates and mastics
1.12.2. Qualitative analysis
In the supporting CT image analysis software efX-CT 1.6 by company North Star Imaging, every drilled core is reconstructed as a digital 3D sample by combining thousands of slices together.
Top layer
Bottom layer
Dense layer
By adjusting the cutting plane, every slice can be observed and appraised in X, Y and Z directions. During this process, the representative slices for each drilled core can be found.
The criteria to choose the representative slices are shown below:
1) Interface between different layers (Figure 4-6);
Figure 4-6: Interface from top layer to bottom layer. From left to right: above the interface; the interface; under the interface)
2) Air voids content (relatively large or small) (Figure 4-7);
Figure 4-7: Left: relative large air void content in top layer; Right: relative small air void content in top layer
3) Composition of aggregate: material size in different layers.
4) Aggregate segregation; Segregation in an asphalt mixture can be defined as the separation of the coarse aggregate particles in the mix from the rest of the mass [78].
Figure 4-8: Left: aggregate segregation; Right: aggregate cracking
Although criteria (3), (4) and (5) are out of the scope of this thesis, they are used to characterize each core and describe the pavement internal properties.
After the specific slices are chosen, every drilled core can be investigated to describe air void structure to some extent by summarizing common and different features of the slices.
However, even if there are common features, the extents of these common features in each core are different. For instance, the segregation is more serious in Norr H than Syd M.
Therefore, comparisons are made to show the extent.
1.12.3. Quantitative analysis
The quantitative analysis focuses on the calculation of layer thickness and air voids contents.
1.12.3.1. Layer thickness
Slices in X-direction are used to calculate thickness for both top and bottom layer. In the computer software efX-CT, length can be measured directly, as shown in Figure 4-9. Since interface between different layers is not linear, one point is chosen randomly on the interface to calculate layer thickness in one slice.
Figure 4-9 Measurements on layer thickness in Norr H
Five randomly chosen slices are used to calculate the average layer thickness. Table 4-1 gives an example in Norr H. Calculation tables in the other cores are presented in Appendix B.
Table 4-1 CT layer thickness in Norr H
Slices 1 2 3 4 5 Average Standard deviation
Top layer (mm) 25.77 26.25 21.10 21.10 23.98 23.64 2.47 Leg=25.77mm
Leg=44.703mm
1.12.3.2. Air voids contents
Different kinds of air void content are under study, which can be categorized into interconnected and total air voids content in the two layers respectively and the overall air voids content. Since different components are represented by different grayscale level, air void can be distinguished from aggregate and mastics. Image analysis software ImageJ (Version 1.0 published by National Institutes of Health) is used to threshold the proper grayscale level and separate air voids from other materials. Figure 4-10 shows the flowchart of operations carried out in ImageJ.
Figure 4-10 The flowchart of operations in the ImageJ software
After importing the slice into ImageJ, the image type should be changed from tiff to a 8-bit image to allow further operations. By cropping the top or bottom layer of a slice, total air void content in separate layers can be obtained. By cropping the entire slice, the overall air void content can be obtained. Figure 4-11 shows the different kinds of air voids contents after cropping and thresholding for a slice in Nor H in X direction.
Import the slice
Change image type
Crop the useful part
Interconnected air voids
Total/Overall air voids content
Select interconnected region
Measure interconnected area
Auto threshold
(a)
(b)
(c)
segment the slices, the “try all” option is used. This produces a montage with results from all the methods, allowing to explore how the different algorithms perform on a particular image.
Then the best suited method is chosen by comparing the results to the original slice based on the similarity of air voids area. Consequently, the Minimum auto threshold is the best method according to the montage results. The minimum method assumes a bimodal histogram, which is iteratively smoothed using a running average of size 3, until there are only two local maxima.
The threshold t is such that yt−1 > yt <= yt+1 [79].
An example of a slice of Norr H in Y direction is shown in Figure 4-12 and Figure 4-13.
Figure 4-12 Montage with results from all the auto threshold methods
Figure 4-13 The original slice and the best segmented slice using Minimum method
After thresholding, the original image is transformed into a binary image of black air voids and white solid phase after applying the threshold value, as shown in Figure 4-13. Consequently, air void content can be defined by dividing the area of air voids pixels with the total area of pixels.
Quantification of interconnected air voids is the next step. The interconnected air voids are identified by the Lefebvre air void classification (see Figure 2-6). They are connected to the external space which contributes to the permeability performance of porous asphalt [49].
Therefore, air voids in the water flow path are classified as interconnected. Based on the criteria, all interconnected air voids (the red area in Figure 4-14) are selected when looking from the surface using the wand tool in ImageJ and the total interconnected area is measured in each image. The interconnected air voids content is obtained by dividing the total interconnected area by the total area of the slice.
An example of the calculation is given for Norr H in Table 4-2, 4-3 and 4-4. The other drilled cores calculations are presented in the tables in Appendix B.
Note: Since only air void contents are of interest, it is not necessary to set scale in the image to present the real values. The uniform unit is pixels.
Table 4-2 Calculation table of Norr H in X-direction
Slice No. 1 2 3 4 Average
Overall air void1 11.85% 12.34% 13.17% 11.08% 12.11%
Interconnected area
(Pixel) 2.27 4.84 7.53 2.79 -
Total area (Pixel) 140.73 197.49 157.95 192.00 -
Interconnected content 1.61% 3.44% 5.35% 1.98% 3.10%
Top layer2 18.79% 19.70% 21.23% 19.40% 19.78%
Bottom layer 6.06% 7.64% 6.87% 5.66% 6.56%
1 The overall air void content
2 Total air void content in the top layer
Table 4-3 Calculation table of Norr H in Y-direction
Slice No. 1 2 3 4 Average
Total area (Pixel) 171.94
Top layer
Air void content1 22.23% 23.06% 19.16% 20.31% 21.19%
Interconnected area2 (Pixel) 23.17 20.48 19.65 20.59 - Interconnected content3 13.48% 11.91% 11.43% 11.98% 12.20%
Bottom layer
Air void content 6.71% 5.64% 6.43% 5.35% 6.03%
Interconnected area (Pixel) 3.90 2.91 2.33 2.79 - Interconnected content 2.27% 1.69% 1.35% 1.62% 1.73%
1 The overall air void content in the top layer
2 The area of interconnected air voids in the top layer
3 Interconnected air void content in the top layer
Table 4-4 Calculation table of Norr H in Z-direction
Slice No. 1 2 3 4 Average
Overall air void 10.49% 11.55% 12.77% 10.40% 11.30%
Interconnected area (Pixel) 2.78 8.01 8.90 1.41 - Total area (Pixel) 138.78 180.98 196.43 197.03 - Interconnected content 2.01% 4.42% 4.53% 0.72% 2.92%
Top layer 19.65% 20.54% 22.93% 20.10% 20.81%
Bottom layer 6.71% 7.46% 6.55% 5.53% 6.56%
1.12.3.3. Significance test
In the significance test, all air voids contents mentioned above are qualified in X, Y and Z directions. The results are analysed separately first and then together. For example, after both total and interconnected air voids content for top and bottom layers, as well as the combination between them for both X, Y and Z directions, are calculated, the average value is taken between X and Y to get the vertical contents to compare to Z. And the procedure proceeds in the sequence of X-Y, X-Z, Y-Z, and X-Y-Z. Apart from qualification between three directions, the calculating slices are also increased from 4 to 6 and then to 8 slices to test significant difference.
Later, linear regression is applied as an analysis tool, where the observations with the traditional method is the explanatory variable x and the measurements with the new X-ray CT method is the response y (not confuse with X, Y direction).
The relation between 𝑥𝑥𝑖𝑖 and 𝑦𝑦𝑖𝑖 is:
𝑦𝑦𝑖𝑖= 𝛽𝛽0+ 𝛽𝛽1∙ 𝑥𝑥𝑖𝑖+ 𝜀𝜀𝑖𝑖
where
𝑥𝑥𝑖𝑖 is the observed air void content using traditional laboratory method;
𝑦𝑦𝑖𝑖 is the observed air void content using X-ray CT method;
𝛽𝛽0 is the intercept of the linear regression equation;
𝛽𝛽1 is the slope of the linear regression equation;
𝑦𝑦� is the calculated linear regression air void content: 𝑦𝑦𝚤𝚤 � = 𝛽𝛽𝚤𝚤 0+ 𝛽𝛽1∙ 𝑥𝑥𝑖𝑖.
𝜀𝜀𝑖𝑖 is the residual between the observed air void content and the calculated linear regression air void content: 𝜀𝜀𝑖𝑖= 𝑦𝑦𝑖𝑖− 𝑦𝑦�. 𝚤𝚤
If the new method is good, then there is no significant difference between the results of traditional method and the CT method. The following findings should be observed:
(1) the observed intercept 𝛽𝛽0 is close to 0 (2) the observed slope 𝛽𝛽1 is close to 1
