You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228614726

Performance of cement modified dense graded cold-mix asphalt and


establishing mathematical model

Article  in  Indian Journal of Engineering and Materials Sciences · January 2007

CITATIONS READS

16 744

3 authors, including:

Seref Oruc Atakan Aksoy


Karadeniz Technical University Karadeniz Technical University
4 PUBLICATIONS   173 CITATIONS    12 PUBLICATIONS   440 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Seref Oruc on 02 June 2014.

The user has requested enhancement of the downloaded file.


Indian Journal of Engineering & Materials Sciences
Vol. 13, December 2006, pp. 512-519

Performance of cement modified dense graded cold-mix asphalt and establishing


mathematical model
Seref Oruc*, Fazil Celik & Atakan Aksoy
Civil Engineering Department, Karadeniz Technical University, 61080 Trabzon, Turkey
Received 5 December 2005; accepted 25 September 2006
Cold-mix asphalts have several advantages regarding economical, ecological and logistical aspects over hot-mix
asphalts. But, cold-mix asphalts have poor performance and are deteriorated in case of additional external water entrance
before mixture curing is completed. This study has been focused on to solve these problems and present an alternative to hot
mixtures. Cold-mix asphalt specimens with dense graded aggregate were prepared according to Marshall method for
emulsified asphalt-aggregate cold mixture design. Portland cement has been substituted for mineral filler in an increased
percent, from 0% to 6%. Cold-mix asphalt specimens have been prepared with 4.2%, 5.2% and 6.2% of residual asphalt
contents. Indirect tensile test, static and repeated load asphalt creep tests, and water damage test have been performed on the
specimens. In addition to these tests, the moisture losses of specimens that took place in time are also measured. Based on
the experimental work, it has been concluded that dense graded cement modified emulsion mixtures are of great
performance compared to hot mixtures with the same aggregate grading and bitumen content. A mathematical model has
also been developed for determining the resilient modulus of cold-mix asphalt.
IPC Code: C08L95/00

The use of cement in asphalt mixtures is not a new cement immersed in water after stability tests
concept. Portland cement was used primarily as filler disintegrated after 24 h, whereas cement-treated
in hot-mix asphalt to prevent stripping of the binder specimens indicated no deterioration.
from previously dried aggregate. It was also used to Dardak6 also reported that addition of cement to
enhance the coating of wet aggregate with bitumen or sand emulsion mixtures increased stability of 200-
tar1. Schmidt and Graf2 showed also a dramatic water 300%. Because of this, thickness of bituminous
resistance and with some aggregates a large increase coating is normally 22 cm and it is decreased to 12 cm
in the dry modulus of resilience (Mr) of the hot mixes by using of cement added emulsion.
were imparted by adding the cement and lime as a
Uemura and Nakamori7 stated to be used to normal
slurry to the aggregate 24 h before the hot mix was
Portland cement in emulsion mixtures for years in
made.
Japan. They pointed out that performance of these
Schmidt et al.3 studied the effect of adding cement
mixtures was at acceptable level in environmental
in an attempt to improve the slow development of
conditions.
strength of emulsion-treated mixes. The cement was
A study8, which was carried out by OECD in
added to the aggregate at the time the asphalt
France, with regard to the spent energy by taking into
emulsion was incorporated. It was concluded that
consideration the production, transportation, mixture,
mixes treated in this way cured faster, developed a
and distribution of the area and spreading out of
high Mr more rapidly, and were more resistant to
emulsion mixtures compared with hot mixtures.
water damage. Terrel and Wang4 had shown that the
According to this comparison, it was reported that
rate of development of Mr in emulsion-treated mixes
cold mix named as “Grave Emulsion” supplied an
is greatly accelerated by the addition of cement.
energy saving of 50%.
Head5 reported that the addition of cement had a
In a study9 carried out on asphalt emulsion
very significant effect on mix stability; addition of 1%
mixtures, or cold mix was put forward that
produced an increase in stability of 250-300% over
mechanical properties were affected by a number of
that of untreated specimens. Specimens without
parameters, including binder grade, void content,
__________ curing time and additives such as cement. In addition
*For correspondence (E-mail: oruc@ktu.edu.tr) to this, field trials have been indicated that cold mix
ORUC et al.: MATHEMATICAL MODEL OF COLD-MIX ASPHALT 513

can be produced using conventional hot mix plant and Table 1— Aggregate gradation
laid using similar techniques. Besides, it has been
found that emulsion droplet coalescence was affected Sieve size Gradation limits Used gradation
by pressure, bitumen type, emulsifier level, cement % Passing % Passing
Inch mm
and temperature and that cement causes emulsion
3/4" 19.1 100 100
charges to become more positive (or less negative)
but other parameters had no effect on charge. 1/2" 12.7 77-100 86
An experimental investigation was conducted on a 3/8" 9.52 66-84 74
three-phase cement-asphalt emulsion composite No. 4 4.76 46-66 56
(CAEC) by Li at al.10. Laboratory tests on fatigue,
strength, rigidity, temperature susceptibility and No. 10 2.0 30-50 38
stress-strain relationship were performed to evaluate No. 40 0.42 12-28 18
the mechanical properties of the CAEC. The No. 80 0.177 7-18 9
preliminary test results showed that CAEC possessed
No. 200 0.074 4-10 6
most of the characteristics of both cement and asphalt,
namely the longer fatigue life and lower temperature Table 2 — Aggregate properties
susceptibility of cement concrete, and higher
Test name Test result
toughness and flexibility of asphalt concrete.
Potti et al.11 declared in their studies that the Coarse aggregate (ASTM C127)
purpose of their research program was to create a Bulk Specific Gravity (g/cm3) 2.698
methodology for the development of slow-setting Apparent Specific Gravity (g/cm ) 3
2.714
cationic emulsion and their use in the manufacturing
of an extend range of road paving materials. Absorption (%) 0.33
The use of asphalt emulsion in Turkey is largely Fine aggregate (ASTM C128)
restricted to various types of surface treatment (such Bulk Specific Gravity (g/cm3) 2.683
as slurry surfacing and surface dressing) and 3
bond/tack coat. Recently, efforts have been to the use Apparent Specific Gravity (g/cm ) 2.735
of emulsions in mixtures used for trench Absorption (%) 0.62
reinstatements and patching. It is used as a binder in Mineral filler (ASTM D854)
cold mix for structural layers. This has attracted
Apparent Specific Gravity (g/cm3) 2.743
relatively little attention, largely because of the
problems associated with the time taken for full L. A. Abrasion (%, ASTM C131) 23.57
strength to be achieved after paving and the Polishing Value (BS 813) 0.47
susceptibility to early life damage by rainfall. The
Sodium Sulphate Soundness (%, ASTM C88) 2.44
research was done in order to solve the problems and
provide an improved insight into how the mechanical
properties of cold mix might be improved. Besides, a
Cationic slow setting emulsion (CSS-1) was used
mathematical model has been developed for resilient
throughout the experimental program. The results of
modulus using results obtained from experimental
the characterization tests of the emulsion and
studies. It was shown that the resilient modulus can be
emulsion residue determined by ASTM are presented
estimated using such mathematical model.
in Table 3, along with their specification limits.
Experimental Procedure All Marshall Specimens produced for this research
Materials were prepared according to Marshall method for
The aggregate used in this study is crushed emulsified asphalt-aggregate cold mixture design12.
limestone and the gradation is shown in Table 1. Emulsified asphalt’s ability to coat an aggregate is
Physical properties of the aggregate, which contains usually sensitive to the pre-mix water content of the
coarse aggregate, fine aggregate and mineral filler aggregate. This is especially true for aggregates
determined by ASTM and BS standards are shown in containing a high percentage of material passing a 75
Table 2. Cement used in the study was PC 42.5 µm (No. 200) sieve, where insufficient pre-mixing
cement with a specific gravity of 3.108 g/cm3. water results in balling of the asphalt with the fines
514 INDIAN J. ENG. MATER. SCI., DECEMBER 2006

Table 3 — Test results for as delivered emulsion and emulsion residue


Test Test results Specification limits
Min. Max.
Viscosity, Saybolt-Furol, 25oC, s. (ASTM D88) 22 20 100
Settlement, 5 day, % (ASTM D6930) 0.1 --- 5
1 Day Storage Stability, % (ASTM D6930) 0.03 --- 1
Charge (ASTM WK11296) Positive Positive
Sieve Test, % (ASTM WK11294) 0.01 --- 0.10
pH (ASTM D1293) 5.37 --- 6.5
Residue by Distillation, % (ASTM D6997) 63.0 57 ---
Residue tests
Penetration, 25oC, 100 g, 5 s. (ASTM D5) 125 100 200
Ductility, 25oC, 5 cm/min, cm (ASTM D113) 84 40 ---
Solubility, % (ASTM D2042) 99 97.5 ---
Ash, % (ASTM D5040) 0.5 --- 2.0

and insufficient coating. For this reason, the coating were demounted after about 20 h and then cured in an
tests were performed at various aggregate water environmental conditional room.
contents. According to aggregate gradation, The diametral resilient modulus test method
percentage on aggregate mix of the emulsion was developed by Schmidt13 and detailed in the ASTM
made of at 6.69%. It was decided to have a pre-mix test method for indirect tension test for resilient
water is 3.5% along with natural water content of modulus of bituminous mixtures (D-4123) was used
aggregate from coating tests. Total water content of in this study. Specimens of cold mix were first tested
the mixture was 5.97%. Mixture properties are closely after 3 days and periodically, as the specimens cured,
related to density of the compacted specimens. Thus, the tests carried out throughout 365 days at various
it was necessary to optimize the water content at periods. The specimens were tested at 25oC. Pulse
compaction to maximize the desired mixture period was 1000 ms and rise time was 40 ms.
properties. Optimum water content (as percent of dry Static and repeated load asphalt creep tests
aggregate mass) at compaction was found 3% (that is Uniaxial compression loading was used to
water content resulting in the highest density). determine the effect of cement on creep and
permanent deformation of cold-mix asphalt. Creep
Resilient modulus test
Specimens of cold mix asphalt were prepared for and permanent deformation tests were conducted in
different additions of Portland cement from 0% to 6% accordance with producers outlined in NCHRP
(as percent of dry aggregate mass) instead of mineral Report 33814. Specimens of the same design as those
filler using Marshall hammer compaction with 50 used in the resilient modulus tests were subjected to
blows to each end of the specimen. The specimens static load asphalt creep test for three levels of
were prepared for three levels of residual asphalt residual asphalt content; 4.2%, 5.2% and 6.2%. All
content of 4.2%, 5.2% and 6.2% (as percent of dry tests were performed at 25oC after 28 days.
aggregate mass). Besides, conventional hot mixtures, Moisture damage test
with aggregate type and grading, were prepared The effect of cement on the resistance to water
without cement as the reference samples. Bitumen damage of mixtures has been reported by a number of
content equivalent to the cold-mix with 5.2% residual authors including Ishai and Nesichi16 and Lottman17.
content was used. Both the emulsion and hot mixtures Moisture damage evaluation can be determined using
were prepared in sufficient quantity to allow three various methods, the most common of which are
1100 g specimens to be produced from each mixture. summarized in a recent publication by the National
The cold mixtures were prepared and compacted at Research Council in the United States of America18.
room temperature, whereas the hot mix specimens This method uses either the ratio of the resilient
were produced at a temperature of 140oC. Specimens modulus or the indirect tensile strength of wet
ORUC et al.: MATHEMATICAL MODEL OF COLD-MIX ASPHALT 515

(moisture conditioned) and of dry (unconditioned)


samples as indicators of moisture susceptibility. In
this study, the effect of water damage on durability of
cold mix and hot mix was evaluated by measuring the
resilient modulus of mixtures before and after
soaking. Specimens of the cold mixtures prepared at
5.2% of residual asphalt content were tested at 25oC
after 3 days curing time. Then, the specimens were
soaked 1 day under water at 50oC. The ratio of
resilient modulus (Mr) values for conditioned and
unconditioned specimens was determined.
Mathematical Model for Resilient Modulus
The object of model fitting is to obtain the simplest
model that adequately fits the data, so that it can be Fig. 1 — Effect of cement on resilient modulus at 4.2% of
useful for prediction purposes. Relationships between residual asphalt content
engineering variables are not always linear, or may
not always be adequately described by linear models.
Experimental data for such variables may show a non-
linear trend between the observed values of the
variables. Although a linear relation may be used to
describe the general trend between each pair of
variables, predictions based on such linear relation-
ships may overestimate (in certain ranges of the
variables) or underestimate (in other ranges of the
variables) the expected result. In such cases, a linear
relationship between the variables would be more
appropriate. The determination of such non-linear
relationships on the basis of observational data
involves non-linear regression analysis.
Non-linear regression is usually based on an
assumed non-linear (mean value) function with cer- Fig. 2 — Effect of cement on resilient modulus at 5.2% of
residual asphalt content
tain undetermined coefficients that would be evalua-
ted from the experimental data. The simplest type of
non-linear function for the regression of Y an X is
E(Y|x) = α + βg(x) ...(1)
where g(x) is a predetermined non-linear function of
x. Non-linear regression analysis is usually based on
the assumption of a constant variance Var. (Y|x), or a
variance that is a function of g(x)19.
By defining a new variable x' = g(x), Eq. (1)
becomes
E(Y|x') = α + βg(x') …(2)
Model was established according to the dependent
variable, which is the resilient modulus because it was
measured during the indirect tensile test. The model
established for resilient modulus is as follow:
Mr = α+(β1* X1+β2*X2)*LOG(X3)+β3*X1+β4*X2 Fig. 3 — Effect of cement on resilient modulus at 6.2% of
…(3) residual asphalt content
516 INDIAN J. ENG. MATER. SCI., DECEMBER 2006

Results and Discussion From the water loss results in Figs 4-6, the
The results of the resilient modulus tests are shown following points are considered important: (i) The
in Figs 1-3 for each of residual asphalt content. majority of the water loss occurred in the initial weeks
Besides, specimens of cold-mix were used for water- after specimen production; (ii) Generally, it can be
loss measurements, which were simply taken as the said that the addition of cement slowed down and
loss in mass. The relative humidity measured was decreased the rate of water loss, but the results did not
73% during the testing period. The results of the tests truly follow a trend; (iii) Evaporate of the water
are shown in Figs 4-6. continued over several weeks, but evaporate stopped
Results of the resilient modulus tests shown in after about 8-10 weeks, and a remainder of water may
Figs 1-3 indicate that resilient modulus of the cold never disappear completely; and (iv) Increase of
mixtures increased steadily over several weeks and residual asphalt content slowed down and decreased
months time, on the contrary, the hot mix showed no the rate of water loss.
significant change. The rate of resilient modulus of Results of the uniaxial creep tests shown in
the cold mixtures increased as the cement content Figs 7-9 indicate that the sudden increment in
increases. On the contrary, the resilient modulus of deformation occurred by load applied. Then, the
the cold mixtures decreased as the residual asphalt increment in deformation slowed down. The rate of
content increases. deformation of the cold mixtures decreased with

Fig. 4 — Water loss from specimens with various levels of Fig. 6 — Water loss from specimens with various levels of
cement at 4.2% of residual asphalt cement at 6.2% of residual asphalt

Fig. 5 — Water loss from specimens with various levels of Fig. 7 — Effect of cement on creep strain at 4.2% of residual
cement at 5.2% of residual asphalt asphalt content
ORUC et al.: MATHEMATICAL MODEL OF COLD-MIX ASPHALT 517

Fig. 8 — Effect of cement on creep strain at 5.2% of residual Fig. 10 — Permanent strains at 25oC for various cement contents
asphalt content compared with hot mix

Fig. 9 — Effect of cement on creep strain at 6.2% of residual Fig. 11 — Effect of cement level on the resistance to water
asphalt content damage

cement content. On the contrary, the rate of From the permanent deformation results in Fig. 10,
deformation of the cold mixtures increased with the following fact can be indicated: (i) Without
residual asphalt content. Results given in Fig. 8 cement, clod-mix specimens failed after less than
indicate that deformation of the hot mix was higher 30,000 cycles, which indicates that the unmodified
than the cold mixtures added even 2% of cement. As cold-mix has rather poor resistance to permanent
expected, the creep modulus decreased with increased deformation; (ii) The resistance to permanent
loading time. The results of static creep tests showed deformation of cold-mix was increased by the
the enhancement of mixture creep resistance due to addition of cement; (iii) The cold-mix specimens with
the addition of the cement. cement offered better resistance to permanent
Specimens of the cold mixtures prepared at 5.2% deformation than hot mix (without cement);
of residual asphalt content were subjected to repeated (iv) These results suggest that cement acts as a
load asphalt creep tests to determine permanent secondary binder in cold-mix.
deformation characteristics of the cold mixtures15. Results of the moisture damage tests are shown in
The specimens were tested at 25oC after 28 days Fig. 11. From these results the following points can be
curing time. The tests were performed at a load stated: (i) Without cement, cold-mix specimens failed
frequency of 1 Hz with 0.5 second loading and 0.5 after sixth hour of conditioned; (ii) The resistance to
unloading for 45000 cycles. Fig. 10 shows the water damage of cold-mix was increased by the
averaged results obtained from the sets of three addition of cement; (iii) The ratio is 0.91 for hot
repeated load asphalt creep tests on specimens with mixture. The ratio obtained for cold mixture with 3%
different content of cement. cement is 0.85 even only 3 days of curing; and
518 INDIAN J. ENG. MATER. SCI., DECEMBER 2006

(iv) The results indicate that cement is an effective


adhesive for cold mixtures.
Correlation matrix and regression statistics of the
model established is shown in Table 4 and 5.
Frequency distribution of residuals, normal
probability plot of residuals and observed versus
predicted values of model established are shown in
Figs 12-14 respectively.
While developing the mathematical model, the
correlation matrix (Table 4) must first be calculated
between the dependent and the probable independent Fig. 12 — Frequency distribution of residuals of model
established
variables. Since the correlations among residual
asphalt content, cement level and curing time are
weak, it is reasonable to choose all the three as
independent variables and the resilient modulus as the
dependent variable. Cement content is the most
effective parameter on the resilient modulus. Table 5
shows the regression coefficients. The values are all
in the good range. In Table 5, since all p-values are
smaller than 0.05, the calculated coefficients are all
reasonable.
In order to determine whether or not any systematic
errors are associated with the experiments, the Fig. 13 — Normal probability plot of residuals of model
frequency distribution of the residuals is plotted and established
compared with normal distribution (Fig. 12). Since
frequency distribution of the residuals conforms to
normal distribution, no systematic error could be
Table 4 — Correlation matrix of the proposed model
Variables Resilient Asphalt Cement Curing
modulus content content time
Resilient modulus 1.00
Asphalt content -0.42 1.00
Cement content 0.83 0.00 1.00
Curing time 0.20 0.00 0.00 1.00 Fig. 14 — Observed and predicted values of model established

Table 5 — Regression statistics


N=168 Coefficients Standard error t-statistic Prob.
Constant (α) 15509.05 796.9679 19.46007 0.0000
β1 84.43554 27.74877 3.042857 0.0027
β2 150.1182 40.51018 3.705689 0.0003
β3 -3181.904 174.4001 -18.24485 0.0000
β4 1829.078 148.6583 12.30390 0.0000
R-squared 0.925187
Adjusted R-squared 0.923351
Standard error 1560.662
F-statistic 503.9421
Prob. (F-significance) 0.000000
ORUC et al.: MATHEMATICAL MODEL OF COLD-MIX ASPHALT 519

associated. Fig. 13 shows the same from another X2 = cement content (%)
point. The good correlation is displayed between X3 = curing time (day)
expected normal value and residual and good α, = constant
correlation is obtained between observed and β1, β2, β3, β4 = coefficients
predicted values (Fig. 14).
Conclusions References
This experimental study has focused on the effect 1 Bituminous Materials in Road Construction, Transport and
of cement in enhancing the mechanical properties of Road Research Laboratory, Crowthorne, Berkshire,
England, UK, 1969.
dense graded cold-mix asphalt for structural layers in
2 Schmidt R J & Graf PE, Proc. AAPT, 41 (1972), 118-162.
highway pavement roads. A satisfactory solution has
3 Schmidt R J, Santucci L E & Coyne L D, Proc AAPT, 42
been introduced to the most important problems of (1973) 300-319.
cold-mix asphalt. Based on the results of this study 4 Terrel R L & Wang C K, Proc AAPT, 40 (1971) 108-125.
the following conclusions can be made: 5 Head R W, Proc AAPT, 43 (1974) 110-131.
(i) Critical early life properties of cold-mix 6 Dardak H, Performance of Different Mixes of Sand
asphalt based on the resilient modulus, Emulsion in Indonesia, First Congress on Emulsion, Paris,
1993, 4-12-042/01-05.
resistance to water damage, creep and
7 Uemura T & Nakamori Y, Stabilization Process of Cement
permanent deformation are all improved by
Asphalt Emulsion in Japan, First World Congress on
cement addition. Cement is an effective Emulsions, Paris, 1993, 4-13-16/01-06.
adhesive and acts as a secondary binder in 8 OECD, Road Binder and Energy Savings, A Road Transport
cold-mix asphalt. Research Report, Paris, 1984.
(ii) Water in dense graded emulsion mixtures is a 9 Needham D, Developments in Bitumen Emulsion Mixtures
for Roads, Ph D thesis, University of Nottingham,
vital ingredient of the process but becomes a Nottingham, 1996.
problem since inhibiting compaction and 10 Li G, Zhao Y, Pang S S & Huang W, Cem Concr Res, 28,
delaying strength gain. Most of the water loss (5) (1998) 635-641.
is achieved in the first weeks but some 11 Potti J J, Eckmann B, Brûlé B, Fabre J C, Such C, Leal-
remains in the longer term. Calderon F & Fuente J, Slow Setting Cationic Bituminous
Emulsions for Construction and Maintenance of Roads
(iii) The resilient modulus and resistance to creep (Optel), Eurasphalt & Eurobitume Congress, Barcelona,
and permanent deformations of the cold Spain, 2000.
mixtures decreases with increasing residual 12 The Asphalt Institute, Asphalt Cold Mix Manual, Asphalt
asphalt content. Institute Manual Series No. 14 (MS-14), 3rd Ed, USA, 1989.
(iv) Mathematical model describing the 13 Schmidt R J, Highway Res Board, 404 (1972) 22-32.
relationship between the resilient modulus 14 Von Quintus H, Scherocman J, Hughes C & Kennedy T,
Asphalt–Aggregate Mixture Analysis System (AAMAS),
and residual asphalt content, cement addition NCHRP Report 338, (TRB, Washington), 1991.
and curing time has been established. It was 15 Alderson A, UMATTA Testing Equipment Trial Course,
shown that the resilient modulus can be England, 1995.
predicted using such mathematical model. 16 Ishai I & Nesichi S, Transp Res Board, (1988) 12-17.
Acknowledgements 17 Lotman R P, Transp Res Board, (1982) 88-95.
The authors would like to thank Karadeniz 18 Stuart K D, Evaluation of Procedures Used to Predict
Technical University Research Fund for supporting Moisture Damage in Asphalt Mixtures, Report No. FHWA-
RD-86-090, (Federal Highway Administration, Washing-
the Project No: 20.112.001.7. ton), 1986.
Nomenclature 19 Alfredo H S & Wilson H T, Probability Concepts in
Mr = resilient modulus (MPa) Engineering Planning and Design Volume 1-Basic
X1 = residual asphalt content (%) Principles, (John Wiley & Sons, Inc., New York), 1975.

View publication stats

You might also like