Effects of tbe Ignition Method on Aggregate Properties

by

Kevin D. Hall Associate Professor

Depanrntllt of Civil Engineering University of Arkansas

4190 Be!! Engineering Center Fayetteville, AR 72701

(501) 515-8695 (501) 575-7168 Fax kdh3@engr.uark.edu

Stacy G. Williams Senior Graduate Research Assistant

Depanment of Civil Engineering University of Arkansas sdg3@engr.uark.edu

Paper prepared for publication and presentation at the 1999 Annual Meeting of the

Association of Asphalt Paving Technologists

December I 1998

Errtrts ofth t Ignition Mtthod on Aggregatt Propwirs

i\IJSTRACf

Kevin D. Hall and Stacy G. Williams

Accurate determination of the asphalt binder content is a fundamental Quality COlllfoileSI for hot-mix asphalt concrete (IiMAC). Numerous tests for binder content have been used with varying degrees of success over the years. Two of tile more durable tests aTe solvent extraction and tile nuclear method Each of these tests carry potentially serious drawbacks Solvent extraction typically uses chlorinated solvents which involve strict slOrage, handling, and disposal regulat ions Nuclear methods require licensed radioactive sources, in addition, the nuclear test does not provide a clean aggregate for subsequent gradation test ing. Recently, the Ignition Method was developed fo r determining the asphal t binder content of HMAC [n the test, an ignition oven literally burns the asphalt birMler off the aggregates, providing not only a measure of asphalt content, but also clean aggregates for gradation testing. Some eoncem has been expressed regarding the effect of the extremely high temperature of the ignition oven on the properties of the aggregates. This paper reports on a study to determine the effects of the ignition method 011 aggregates used in Superpave mixes in Arkansas.

Specimens representing eight Superpave mixes, comprised of aggregates sampled directly from three hot-mix asphalt plants in Arkansas, were tested by the ignition method. The gradatioll of the aggregate blend{s) and the bulk specific gravity of tnc coarse fmllon of the blend(s) were tested both pre-igllition and post-ignition to determine changes ill properties due to the heat of the igllition oven. The results showed that for Arkansas mixes, the igtlition method can provide accurate, timely estimates ror asphalt binder content. In addition, aggregate blend gradation did IIO[ show significant change rrom pre- 10 post-ignition. Coarse aggregate bulk specific gravity decreased ollly slightly post-ignition; the change was n01 significant ror hot·mix asphalt applications. I! is recommended that the Arkansas State Highway alld Trallsportation Department adopt the igtli tion method as an alternative ror determining the binder COlllen{ of HMAC mixes.

EfTet'ts of the Igni tion Method on Aggrrgll tt Propenies

Kevin D Hall and Stacy G Williams

Introduction

One of the primary quality controlteSI$ performed on hot mix asphalt concrete (HMAC) mixes is the determination of asphalt binder content Many highway agencies, such as the Arkansas Slate Highway and Transponation Depanment (AHTD) pay paving contractors baSl'{\ on the aCC1J racy of binder content in the mix placed (I). Therefore, it is very imponantlo both the contraclor and highway agency that the binder content determination be accurate as well as timely The ignition method is a procedure for determining asphal t binder content that has many advantages. The ignition oven is a device thai heals the asphalt mixture to a temperature that literally igni tes the asphalt binder, leaving behind clean aggregate The loss of weight during this process is monitored, and an asphalt content is calculated. After the lest is performed, dean aggregate remains, which can then be tested for gradation acruracy. However, some concern has been expressed in the asphal t industry regarding the effett of the eKtremely high heat of the igni tion oven (i.e. 53 8 C) on the aggregates in an asphalt concrete mix, This paper repons on research regarding the accuracy of reported asphalt contents and the condition of the resulting aggregate after being burned, as compared to its original condition.

Background - Methods for Determining AC Content

Before detailing the study performed, it is useful to briefly review the major processes for delermining the asphalt content of hot-mix asphalt concrete (HMAC) mixes Such a review provides added incentive for investigations of the type described in th is paper The two most common methods used current ly for determining HMAC asphalt content are solvent extraction (ASTM 0 -2 172) and the nuclear asphalt content gauge (ASTM 4125)

Soh'fnl Extraction

The solvent extraction method has been around for almost one hundred years (1) This method uses a solvent to dissolve the asphah cement in an asphalt mixture. leaving only the mineral aggregate. During the extraction test, the solvent is added to the sample and as the asphalt cement dissolves, the liquid residual is removed through a filter by a centrifuge. vacuum, or renux, The major disadvantage of this method is that il is environmentally unfriendly Commonly used chlorinated solvents are non-biodegradable and carry strict regulations concerning safe storage, handling, and disposal. Newer, safer solvents being developt.'I.I still require additional work to gain widespread acceptance. Another disadvantage of this nlL'Ihod is that fine mineral aggregate panicles can be loS!, causing potential errors in the asphalt bioocr cantenl as well as the resulting gradation

An advantage of the solvent cxtractiun method is that by performing the teslto determine asph~h contcnt, the mineral aggregate is also prepared lor a gradation determination by sieve analysis Also, it is currently the only acceptable standard method for determining the asphalt content of Rcclaimed Asphalt Pa\'ement (RAP) (}j, althoogh some states are reponing successful tests {lr HAl) wit h ignition methods

Nlle/Cllr Aliphllft Content (jmlg/!

Nuclear gauges were first invented in 1956 By 1%9, the process had been refined such thaI it could determine asphalt conlent as accurately as the solvent extraction method (1). The basic principle behind the nuclear gauge involves a radioactive source thaI emits neutrons al an eXlremdy high velocity These lle\I\(ons are slOWed, or thermalised, by impacting hydrogen atoms contained in asphalt cement The nuclear gauge counts the number of slow namons and compares it to the number of neutrons that are still traveling at a high speed These counts are used in calculating tile asphalt content of an asphalt mixture sample.

There are advantages to the nuclear method of asphalt content determination, one of which is thaI it is a relatively quick method for obtaining accurate results. Once a calibrat ion CUf\'e has been established for a panicular HMAC mix, asphalt content tests take only sixteen minutes or less Some disadvantages are that the calibration process takes approximately four hours to complete. and no samples can be tested until this process has been done. Another drawback is that after asphalt content has been determined, hot-mix asphalt concrete remains: if the gradation of the mixture is to be tested, a solvent extraction or other process must be done in order to get the aggregate clean so thaI a sieve analysis can be performed. It is important to note (hal all personnel using a nuclear gauge must be certified to do so due to the use of a nuclear source Also, due to the nuclear source, owning and operating a gauge requires strict adherence to permitting regulations.

Upon reviewing the major methods currently used for determining asphalt content, it becomes obvious that a method that is relatively safe, provides timely and acrurate asphalt content results, and provides clean aggregate for subsequent gr.Idation testing is highly desirablc. The ignition method answers this need.

The Ignition Oven

The first studies of the Ignition oven were conducted in ! 969 under the sponsorship of the National Cooperative Highway Research Program (NCHRP) (3). The Ignition oven uses high heal to burn asphalt cement olf the aggregate in a mix This study showed that the asphalt content could be achieved wilh a temperature of about 843 C (1550 F); however, the· loss of mass of tile aggregate was a problem. In 1992, studies by U Yu were conducted using a muffle furnace operating at a temperature of a maximum 538 C (1000 F). This reduction in temperature greatly reduced the problem of aggregate loss and gradal ion differences that resulted from aggregate breakdown under such intense heat (~) Based on Yu's initial wor\(, a fun method for usinglhe ignition oven to determine asphalt conttllt was developed at the National Center for Asphalt

2

Technology (2,5) The Ignition Method has been standardized, and is given in AASHTO Provisional Standard TP53·95

AgKregate Com:ction Factor

When using the ignition oven to determine the asphalt content of a particular blend, a calibration factor, termed the "aggregate correction factor" (ACF), must be established This correction factor represents weight losses characteristic of a particular aggregate due to factors not associated with asphalt content. These losses are usually due to a release of moisture thaI is trapped in the aggregate and fine particles thaI escape from the catch pan and scale during the test. Although a given correction factor may be common to cenain types of aggregates, a specifiC value should be determined for each blend to be tested. The aggregate correction factor is typically determined in one of two ways. In the first, an HMAC specimen of 'known' asphah content is ignited and the oven-derived asphalt conlent is recorded; the difference in the measured asphalt content and the known asphalt content is taken to be the aggregate correction factor The second method igni tes a 'bJank' aggregate specimen (without asphalt binder), the recorded 'percent weight loss' is used directly as the correction factor. Both methods are listed in testing specifications.

Project Stope

Research into Superpave mixes recently completed at the University of Arkansas (6) used oorrently-stockpiled aggregates sampled from hot-mix asphalt plants; the project described in this paper used the aggregates arK! miKes developed during the Superpave research effon. Aggregates from four not·miK plants were sampled, representing most of the aggregate types currently used for HMAC in Arkansas. Superpave blends containing aggregates from three of these sources were tested in the ignition oven. Tables I and 2 list the aggregates and blends investigated

Delta Asphalt A_P.A.C L. J. [lirntsl Para~ould, Arkansas West Memphis, Arkansas Texarkana, Arkltnsas

1-1/8" Limestone 1- Jl.f' Limestone 1-1/8" Crushed Rock )/4" Crushed Gravel I" Limestone )/4" Crushed Rock

112" Limestone 3/4" Crushed Gravel 5/8" Limestone 1/4" Limestone 112" Crushed Gravel Dinv Screcnina.s Coarse Sand 1/2" Limeslone Clean ScreeninKS

Fine Sand LimestollC Screeninl!.S Natural Sand River Sand Donna Fill Field Sand Donna Fill

T ;tblt 1 Aggrtgalts Ttst ~d rrom AIJproved Arkansas Sources.

Delta ASllhall A.P.A.C. L. J. ~:anl es t

25.0 mm - Medium Blend 25.0 mm •• Coarse Blend

I" umestont -- ./6" /-/IS " CfllSW Rod·· 6m

•• Donna Fill·· 6.5% 'I 6%

refCI (0 thc gradation of the a"reptc blend in (he HMAC mix

Table 2. Aggregate Blends Tested in the Ignition Oven.

All mixes tested in the ignition oven contained the same asphalt binder, classified as a PO 64·22 The ignition method used in the investigation is that detailed in AASHTO TP 53·95. Specimens tested in the ignition oven contained approximately 2000 g of the aggregate blends listed in Table 2. For each blend. at least one sample contained no asphalt binder; additional specimens were carefully mixed at asphalt contents approximately equal to the optimum asphalt comenl for the blend. optimum. ± 0.5 percent. arK! optimum ± 1.0 percent. No aggregate correction factors were entered into the ignition oven cont rol panel al the lime of test. However. sullicient data was recorded for subsequent correction factor determination. Aggregate blend gradation, both pre·isnition and post·ignition. was determined using AASHTO T-II and T·27 Specific gravity of the aggregates used (bolh pre· and post·ignition) was determined using AASHTO T·84 and T·85.

Projetl Resulls

Agxrcgrlle Come/ion factor / Asphlll/ Binder Content

Table J summarizes some of the resuhs regarding aggregate correction factors and asphalt contenls measured during the project. The numbers shown in Table 3 represent the average of

live (5) separate specimens tested for each combination of asphalt conten! and aggregate blend Aggregate correction factor was calculated using two different methods. (I) the mass loss during ignition of a 'blank' aggregate specimen was taken directly as the correction factor, (2) The differences between the measured binder content and the 'known' binder content of multiple specimens were averaged, with the average differeoce reponed as the aggregate correct ion factor for a panicular blend of aggregate. Table J also shows the error in asphalt COnlent as estimated by the ignition method for multiple specimens, calculated using both values detennined for the aggregate correction factor.

Average Asphalt Content Error ' Aggregate Corr~tiQn FactQr (ACE) 1!5ing AU ~ased~

Blank Average of Blank Average of Blend Aggregate "known" Mixes Aggregate "known" Mixes

Delt!! A~l!hah .el!:nd~

190 mm medium 013 0.18 012 0 11

250 mm coarse 0.40 017 0.23 015

APAC Inc. 8l!md ~

12.5 nun medium 0.23 035 0.13 0.12

19.0 mm medium 0.28 033 0.10 0.119

25 0 mm medium 0.23 0.44 021 0.10

L,J, Eam~l DJcmb

190 mm medium o 18 0.03 0.15 006

25 0 mm medium 0.12 O.oJ 0. 10 005

25.0 mm coarse 0.11 005 0.19 016

t Average asphatt conl ent error is lbe average oflbe dirre~ncc:s between lbe '. nown' asphalc t:ontcnl of a number of !iJlCCl l1lcns and Ihe asphall (;{mtenlmcasumi by Ille ignil ion oven. with the appropriate aggregate correwon r.1l:10r appl ied as SllOWlI

Table 3. Summary of Asphalt Content Results

Based on the work perfonned and the values shown in Table 3, some observations are offered relat ive to the use of the igni tion method for Arkansas Superpave mixes. • the 'accuracy' of the ignition oven, as represented by tile asphalt binder content errors sllown

in Table J, is comparable to the accuracy of asphalt binder content determination reponed for other metllods, and by other studies using the ignition method (5,7).

• in many cases, beller estimates of asphalt binder content are obtained L1sing the al:lSl~gillt' correction factor based un ignit ing an HMAC mix, rather than a 'blank' aggregate specimen

• the blends shown in Table J consist or mixtures or crushed stOlle (limestone) and crushed gravel (silica bast'll ). in varying proponions, no systematic di tlerentes are evident between the

1 ..... 0 aggregalc Iypes in terms of correction factor or accuracy of binder contellt estimate Values of aggregate correction factor appear to be more 5Ource-spetiflc, rather than aggregate type-spt:cilic.

The last observation listed could prove significant. in that aggregate correction factors ill Arkansas cannol be judged or est imated based on aggregate Iype alone. The results emphasize the tact that a correction factor must be determined for each aggregate bleoo al each source

The gradation of aggregate blends used in ignition oven specimens were determined by 'washed' sieve analysis prior to mixing with the PO 64-22 binder. Care was laken to collect all wash water (and subsequent fines) generated during the process for reintroduction into the aggregate blend After the igni tion test, cooled aggregates were again subjected 10 a washed sieve analysis Table 4 lists some of the resu lts of a comparison betwetn pre-ignition and post-ignition gradation. For ease of understanding. Figures I through 3 contain plots of the gradations listed in Table 4 The numbers shown in Table 4 represent the average of fou r (4) separate specimens tested for each aggregate blend Complete results from tht study are not included here; the numbers shown in Table 4 8nd the plots contained in Figures I through] are representative of all the blends tested The blends chosen for inclusion in this paper represent a predominately limestone based blend (A I'AC 190 mm medium), a predominately gravel (silica) based blend (Delta 19.0 mm medium) and a blend containing appreciable amounts of both aggregate types (U . Eamest 19.0 nun

medium).

Delta Asphalt APAC U . Earnest 19.0 mm medium 19.0 mm medium 19.0 mm medium

Sieve (mm) Before M" Before M" Before Me< 250 100.0 100.0 100.0 100.0 100.0 100.0

190 100.0 100.0 100.0 100.0 100.0 100.0

12.5 88.8 88.1 98.4 98.6 88.6 88.0

9 5 74.4 74.0 727 7\.9 73.1 72,

475 50.8 507 54.2 54.8 46.5 460

23, 31.0 30.7 n .7 33.9 30.5 31.2

1 18 22.3 21 .9 22.7 22.6 21.7 214

0.60 16.0 15.6 12.7 12.7 15.3 15.1

030 87 8.5 70 7.1 10.8 10.5

o 15 5.8 57 49 5.1 4.7 4.7

0075 4.15 4.08 3.92 4.05 2 73 2.65

Tllblt 4, Summary or Grada lion Resulls ror Seletled Bltllds

6

It is immediately obvious from the numbers presented in Table 4 and the gradation curves shown in Figures I through J lhal, for the aggregates tested, the ignition oven has little if any effect on the gradation of the aggregate blend . From a practical viewpoint, current AHTD gradation specifications allow a tolerance of t7 percent on any individual coarse aggregate (defined as plus 2.36 mm) sieve and a ±4 percent tolerance for fine (minus 2.36 mm) sieves (/J Given the type of results seen in this project, laboralOries may use the ignition oven without reservation - observed variability in gradation during field quality control should almost entirely renect "natural" material variability, and fIOt changes induced by the testing method itself

A visual inspection of the aggregate after ignition oven testing yielded some interesting observations Some aggregate particles had obviously ruptured during testing. exposing freshly fractu red faces Also, several panicles were notably cracked, but nO! broken Fractured aggregates should lead to a relatively finer gradation after ignition compared to pre-ignition However, in a few cases on the larger sieves (see Table 4) the percentage passing a given sieve IIClUaJly decreases, suggesting the aggregate had gOllen relatively coarser While interesting. these phenomenon did not account for enough of a change in overall gradation to affect quality control results.

Specific Gravity

For each blend tested, the bulk specific gravity of tile coarse aggregate (plus 4.75 mm) in the blend was determined after being subjected to the ignition method. An average of three test results was compared to an original calculated value for each blend This comparison is shown in Table 5. For all samples tested, the specific gravity of the post-ignition material was less than the pre-ignition value. Differences ranged from 0.20 percent to a 2.46 percent. In terms of the actual difference between specific gravity values, some of the differences shown in Table 5 fall wi thin the "acceptable range of two test resu!!s" (D2S) listed for the specific gravity test in AASHTO T85 This suggests that the difference in pre- and post-ignition values obtained may OOt be fully explained by the changes in the aggregate itself; some of the differences observed may result rrom the inherent variability of the testing method.

It is interesting to nole the aggregates that exhibited appreciable breakage after testing in the ignition oven were the 19.0 mm crushed gravel from Delta Asphalt, and the 12.5 nun and 190 mm crushed gravel from APAC. The blends that contain larger quant ities of these aggregates are the ones that show the larger changes in bulk specific gravity.

Summary I Conclusions

The ignition method is very effoc(ive for estimating tile asphalt binder coment of HMAC while also obtaining a "clean" sample of aggregate for determining the asgregale blend gradation without requiring the use of environmentally hazardous chemicals or radioactive sources Many states, including Arkansas, are looking 10 the ignition method as an alternative to solvent e~traction andlor nuclear methods. Some questions concerning the effeci of the extremely high

7

·

llulk Sp_ Gr. Bulk Sp Gr Olend Before I ~nition Mer I 'nition Difference Percent Chan 'e [)eh~

IQmm Medium 2.577 2,531 -0.046 -1.79% Deha

25mm Coarse 2688 H22 -0066 -246%

A.P A C. 12 )mm Medium 2.615 2.592 -0023 -0 88 %

A PAC. 19mm Medium 2.649 2.591 -0.052 -196%

A,P A C. 2~mm Medium 2.677 2,61 2 -0.065 -2 4J %

Texarkana 19mmMedium 2,591 2.575 -0022 -0 85 %

Texarkana 25mm Medium 2548 2.538 -0,010 -0 39 0;'

Texarkana 25mm Coarse 2547 2.542 -0005 -020%

Table 5 Coane Aggregate Bulk Specific Gravity Before and After Ignition.

temperature used by the ignition oven on aggregate blend properties have delayed implementation of the ignition method in many areas.

Based on the work petformed in this study using Arkansas aggregates. some potentially significant conclusions can be offered'

• the ignition method provides estimates of asphalt biooer content with similar accuracy as other binder content determination methods.

" the aggregate correction faCl or used in the ignition method should be determined using HMAC specimens with 'known' binder conlent, rather than 'blank' aggregate specimens.

for typical Arkansas limestone and crushed gravd aggregates. the aggregate correction faClor canoot be estimated based on aggregate type alone, a correction factor should be e,stablished for each aggregate blend individually.

• for the Arkansas aggregates tested, the gradation of the HMAC aggregate blend does not change signi fi cantly after being subjected to the ignition method

• for the Arkansas aggregates tested, only slight changes in the bulk speci fic gravity of the coarse aggregate in an HMAC aggregate blend were measured after the aggregate has been subjected to the ignition method; in many cases these changes may be explained by variability in the testing procedure as well as changes due 10 ignition testing .

•

Uased on Ihe work performed in this study, it is recommended that Ihe ignition methud lor detennining the binder content of 'IMAC be adopted in Arkansas as an alternative to the currently sptCified nuclear method.

Acknowledgment

This paper reports on research performed at the University of Arkansas under project TRC-9604, sponsored by the Arkansas Stale Highway and Transportation Department and the US Department of Transportation. Federal Highway Administration. The authors wish to acknowledge the efforts of Graduate Research A&sislant Conie Walton·Macaulay during lhe execution of the research; and Alan Meadors and Terry Hardison of the Arkansas State Highway and Transportation DepartmtJ1t for project coordination and guidance.

The contents of this paper rened the views of the authors, woo are responsible for the accuracy of tile faclS and data herein. The contents do not necessarily reneet the ollicial views of the Arkansas State Highway and Transportation Department or lhe Federal Highway Administration This paper does not constitute a standard, specification, or regulation

References

I. Standard Speci/icaliQIIS for llighway COlis/merion, Edition of 1996, Arkansas State Highway and Transportation Department, Little Rock, Arkansas, 1996.

2 Brown, E.R., and Murphy, N., ~ Asphah Content Determination by the Igni tion Method", Final Report - Project #930.285, HPR Report IIOB, Alabama Department ofTransportalion, Montgomery, Alabama, July 1994.

3 Antrim, J.D , and Busching, H.W , "Asphalt Content Determination by lite Ignition Method", Highway Research Record No. 213, Highway Research Board, Washington, DC , t96()

4 Vu, Li, "Determination of Asphalt Content and Aggregate Gradation of HMA by Ignition Heating", Master's Thesis, Auburn University, Auburn, Alabama, 1992

S Brown. E.R, and Mager, S , "Asphalt Content by Ignilion - Round Rooin Study", paper prepared for 75· Annual Meeting of tlle Transportation Research Board, National Center ror Asphalt Technology, Auburn, Alabama, 1996.

6 Hall, Kevin D, and Goad, S.D., "Superpave Implementation Issues in Arkansas", paper submined 10 1998 Annual Meeting of the Association of Asphalt Paving Technologists, July 1997.

7 West, R., and Murphy, N , "Comparison of Ignition Testers for Determining Asphalt Contcnt ofHMA Mixtures", Technical Report 96·2, APAC Material Services, Smyrna, Georgia, July 1996.

"

Delta Asphalt 19.0 mm Medium Blend

100

90

80

'" 70

0

•• 60 • • "- 50 -0 • 40 u -• Pre-lgnition

• "- 30 --- Post-lgnition

20

10

0 0.075 2.36 9.5 25.0

Sieve Opening (mm~ raised 10 0.45 power

Figu rr I. Gradalion or Drlla Asphall1 9.0 mm Aggrtgalr Blend Prt- and Posl·lgnil ion

10

100

90

80

en 70 0

• 60 • • .. 50 ~

0 •• 40 0 • • .. 30

20

10

0 0.075

APAC • West Memphis 19.0 mm Medium Blend

2.36 9.5

I Pre·lgnition

--- Post·lgnition

25.0 Sieve Opening (mm), raised to 0.45 power

I'igurt 2. Gradarioll of APAC 19.0 111111 Aggrrgalt Dlentl Pre- and Posl-Ignilioll

"

100

90

80

en 70 c • 60 • • "- 50 l' • 40 0 , • "- 30

20

10

0 0,075

LJ. Earnest 19.0 mm Medium Blend

2,36 9,5

• Pre-lgnition

--- Post-Ignition

25.0

Sieve Opening (mm), raised to 0.45 power

Figurr J. Gradation of L.J. Earnest 19.0 mm Aggregatr Blend Pre- and Post-Ignition

II