KOREAN ASPHALT INSTITUTE Vol. 8, No. 1, 14-20, 2018
*Corresponding author : Ki-Hoon Moon [Tel: 82-10-9464-1665,
E-mail: [email protected], [email protected]]
Comparison of Inter-conversion and Analytical-numerical Approaches on Thermal Stress Computation of Asphalt binder
Ki-Hoon Moon 1, * , Augusto Cannone Falchetto 2 , and Oh Sun Kwon 3
1 Senior Researcher, Korea Expressway Corporation (KEC) pavement research division, 208-96, Dong Budaero 922, Dong Tan, Hwa Sung city, Gyung Gi do, South Korea 18489, Tel:+82-31-8098-6294
2 Technische Universität Braunschweig, Department of Civil Engineering (ISBS), Germany
3 Senior Researcher, Korea Expressway Corporation (KEC) pavement research division, 208-96, Dong Budaero 922, Dong Tan, Hwa Sung city, Gyung Gi do, South Korea 18489
(Received April 14, 2018; Revised April 30, 2018; Accepted May 3, 2018)
Abstract: In the northern U.S. and Canada, low temperature cracking represents one of crucial distresses on asphalt pavements. As temperature drops thermal stress develops in asphalt layer then cracking occurs when it reaches at crucial level. Because of that reason, thermal stress is recognized as an important parameter for evaluating the low temperature performance of asphalt pavement. Conventionally, thermal stress is computed by converting creep compliance into its corresponding relaxation modulus based on different techniques. In this paper, analytical and approximate inter-conversion methods are applied to compute the asphalt binder thermal stress and then the results are compared. Clear differences in thermal stress are found when using power-law based approach compared to the numerical solution of the convolution integral according to Hopkins and Hamming’s algorithm. Based on the findings it can be said that both two different method can successfully be applied for providing lower and upper boundaries on estimating thermal stress of given asphalt binders.
Keywords: Thermal stress, Analytical numerical method, Inter conversion method, Hopkins and Hamming’s algorithm
1. Introduction
Thermal cracking is a serious distress for asphalt pavements built in northern U.S. and Canada. As temperature drops below 0°C, an increase in thermal stress can be observed in the restricted asphalt layers and when the temperature drops below a limit value known as critical temperature, cracking occurs. Low temperature cracking is source of significant negative effects on asphalt pavement performance; this is due the development of surface cracks and, then, to the infiltration of water in the pavement structures, potentially leading to a rapid deterioration process when associated to severe freeze-thaw cycles and traffic
loading (Marasteanu et al., 2007; Moon, 2010;
Moon et al., 2013; Moon et al., 2014; Cannone Falchetto et al., 2014).
Currently, many road agencies are devoting considerable efforts to maintain asphalt pavement performances and network systems at acceptable service levels. Along with these actions, various experimental methods have been developed during the past years to characterize the low temperature behavior of bituminous materials (Marasteanu et al., 2009; Moon, 2010; Moon et al., 2014; Cannone Falchetto et al., 2014). One fundamental test method to characterize viscoelastic materials is the creep test (Marasteanu et al., 2009; Moon et al., 2010;
AASHTO T313-12, 2012). Through this test, creep
compliance: D(t), can be obtained then from D(t),
relaxation modulus: E(t), can be calculated based
on different mathematical methods. Finally,
thermal stress: s(T), can be calculated with
different temperature level (Marasteanu et al., 2007; Moon et al., 2010; Moon et al., 2014). Even though various thermal stress computation approaches were introduced in many studies, not many studies considered and compared the effectiveness of each different thermal stress computation approach.
In the present paper, two different methods (i.e.
numerical method for computing the exact solution and approximate computation techniques) are used to convert BBR creep compliance data:
D(t), to relaxation modulus: E(t). Christensen, Anderson and Marasteanu (CAM) model (Marasteanu and Anderson, 1999) and Gauss quadrature method (Marasteanu et al., 2007; Moon et al.
2013) are then applied to compute thermal stress and the results are graphically compared. At the end of this paper, the recommendations for thermal stress computing approach are identified.
2. Inter-conversion Techniques for Converting Creep Compliance Into
Relaxation Modulus
2.1. Inter-conversion based on exact solutions - numerical algorithm
The creep compliance, D(t), and the relaxation modulus, E(t) of a viscoelastic material are correlated by the convolution integral as (Findley et al., 1976; Ferry, 1980):
(1)
(2)
where t > 0
Based on Laplace transformation, Equations (1) or (2) can be rewritten as:
(3)
where s > 0 is the Laplace transformation parameter.
However, in some cases the expressions for E(t) and D(t) cannot be easily manipulated in the Laplace domain. Therefore, a numerical approach can be used as an alternative and convenient method for solving the integral in Equation (1) and (2).
In this paper a simple and widely used
numerical algorithm developed by Hopkins and Hamming (1967) is applied to compute E(t) from D(t). The main computational steps for obtaining E(t) are as follows:
a. Select a time interval as:
(4) b. Express the integral form of D(t) as:
(5)
c. From Equation (5), f(t) can be calculated with the trapezoid integration rule:
(6)
d. Equation (1~2) can be modified using Equation (6) as:
(7)
From Equation (7), each element of the integral can be approximated as:
(8)
Where, (9)
Equation (8) can be then expressed as:
(10)
Solving for , Equation (10) becomes:
0
( ) ( )
t
E t − ⋅ τ D τ τ d = t
∫
0
( ) ( )
t
E τ ⋅ D t − τ τ d = t
∫
2
( ) ( ) 1 E s D s
⋅ = s
0 0, 1 1, 2 2,... n t = t = t = t = n
0
( ) t ( ) f t = ∫ D t dt
1 1 1 1
( ) ( ) ( ( ) ( )) ( )
n n 2 n n n n
f t + = f t + ⋅ D t + + D t ⋅ t + − t
1
1 1
0
( ) ( )
t n
n n
t + = ∫ + E τ ⋅ D t + − τ τ d
1
0 i ( ) ( 1 )
i n t i t n
E D t d
τ + τ + τ τ
=
= ∑ ∫ ⋅ −
1
( ) ( 1 )
i i t
t n
E τ D t τ τ d
+
⋅ + − =
∫
1/2 1 1 1
( i ) [ ( n i ) ( n i )]
E t + f t + t + f t + t
− ⋅ − − −
1 1
2
1 ( )
2 i i
t i t + t
+ = ⋅ +
1
1 1/2 1 1 1
0 ( ) [ ( ) ( )]
n
n i n i n i
t + i − E t + f t + t + f t + t
= − ∑ = ⋅ − − −
1/2 1
( n ) ( n n ) E t + f t + t
+ ⋅ −
) ( t n + 1 / 2 E
1
1 1/2 1 1 1
1/2 0
1
( ) [ ( ) ( )]
( )
( )
n
n i n i n i
n i
n n
t E t f t t f t t
E t f t t
−
+ + + + +
+ =
+
− ⋅ − − −
= −
∑
(11)
where, , ,
Then, the converted E(t) can be used to compute thermal stress based on the CAM model (Marasteanu and Anderson, 1999) and on Gaussian quadrature.
2.2. Approximated inter-conversion techniques 2.2.1. A simple Power-law based inter-conversion technique
In previous studies, it was found that values of D(t) and E(t) can be interrelated through a simple power-law functions as (Park and Kim, 1999;
Ebrahimi et al., 2014):
(12)
(13)
where E
1, D
1and n are positive constants. By taking the Laplace transformation of Equations (12) and (13), the following result can be derived as can be seen in Equation (14):
(14)
where parameter n can be expressed as:
or (15)
These power-law approximation methods provide very good results when D(t), and/or E(t), is close to a straight line in a log-log scale plot against time.
3. Materials and Testing
A set of two regular asphalt binders for HMA
pavement construction were prepared in this research. The given Performance Grade (PG) (AASHTO M 320-10-UL, 2010) was PG58-28 (unmodified binder), PG58-34 (Styrene Butadiene Styrene - SBS modified binder), respectively; both asphalt binders were provided from Korea Expressway Corporation Research Division (KECRD).
To compute creep stiffness, relaxation modulus and thermal stress, BBR tests (AASHTO T313-12, 2012) were performed. All the binders were tested in long term aged condition according to the Pressurized Aging Vessel (PAV) (AASHTO R028-12, 2012) procedure, respectively. Table 1 presents a summary of the prepared asphalt binders used in this study.
BBR creep tests; as can be seen in Figure 1, were performed on thin asphalt binder beams (102.0 ±5 mm×12.7±0.5 mm×6.25±0.5 mm) according to the current AASHTO specification (AASHTO T 313-12, 2012).
In this test, the mid-span deflection, d(t), of an asphalt binder specimen is measured for the entire duration of the test (total = 240 s), and then used to calculate the creep stiffness, S(t) and the m-value (slope of logarithmic S(t) vs time curve) as:
(16)
(17)
where S(t) is the flexural creep stiffness; D(t) is the creep compliance; σ is the maximum bending stress in the beam; ε(t) is the bending strain; P is the constant applied load (980±50 mN); l is the length of specimen (102±5 mm); b is the width of specimen (12.7 ±0.5 mm); h is height of specimen (6.25±0.5 mm); δ(t) is the deflection at the mid-span of the beam, and t is testing time (0~240 s).
1
1 1/2 1
0 1
( ) [ ( ) ( )]
( ) ( )
n
n i i i
i
n n
t E t f t f t
f t f t
−
+ + +
= +
− ⋅ −
= −
∑
( ) 0
0
f t = E t =( ) 0