Prediction method for low-temperature performance of aged asphalt

Abstract

The invention discloses a method for predicting low-temperature performance of aged asphalt, and belongs to the technical field of performance evaluation of road materials. According to the method, firstly, an asphalt sample is subjected to simulated aging treatment, then improved column chromatography is adopted for separating and measuring the content of saturates, aromatics, colloids and asphaltene of aged asphalt, a chromatographic column is of an activated aluminum oxide and silica gel double-layer adsorption structure, gradient elution is combined, and the separation efficiency and reproducibility are remarkably improved. The method comprises the following steps: firstly, measuring a creep rate m value of aged asphalt, analyzing the correlation degree between the m value and components of the aged asphalt by using grey correlation, and establishing a multiple linear prediction model based on ridge regression by taking the contents of four components as independent variables and the m value as a dependent variable; the model effectively overcomes the multicollinearity problem among components, component analysis data are input into the model for unknown samples, the low-temperature performance of the samples can be predicted, and the research and development efficiency and the engineering quality control level are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of performance evaluation of road materials, and particularly relates to a method for predicting low-temperature performance of aged asphalt. BACKGROUND

[0002] Asphalt pavement is widely used due to its excellent driving comfort and easy maintenance. However, in high-cold and high-altitude regions, asphalt pavement is subjected to severe tests of low temperature, strong ultraviolet radiation and frequent freeze-thaw cycles all year round. As an organic cementitious material, asphalt undergoes irreversible changes in chemical composition and microstructure under the combined action of heat, oxygen and solar radiation, which is known as aging. Aging leads to the transformation of light components (such as saturates and aromatics) in asphalt to heavy components (such as resins and asphaltenes), and is accompanied by the generation of oxygen-containing functional groups (such as carbonyl and sulfoxide groups). Macroscopically, aging is manifested as an increase in hardness, an increase in brittleness and a decrease in ductility of asphalt materials, thereby significantly weakening their anti-cracking performance in low-temperature environments, ultimately leading to premature damage to the pavement and shortening the service life.

[0003] In order to evaluate the low-temperature performance of asphalt, the existing technology usually adopts macroscopic physical performance test methods, such as bending beam rheological test (BBR). BBR test evaluates the ability of asphalt to resist deformation and relaxation stress by measuring the creep stiffness modulus (S value) and creep rate (m value) at low temperature. The lower the S value and the higher the m value, the better the low-temperature anti-cracking performance of asphalt. However, such methods have significant limitations: they are "reactive" tests that can only characterize the changes in material performance after they have occurred, and cannot make performance predictions from the fundamental chemical composition of the material. This makes the formulation design and quality control of materials still rely on a large number of trial-and-error experiments, which is time-consuming and labor-intensive, and makes it difficult to reveal the inherent mechanism of performance degradation. In order to understand the aging behavior of asphalt from the chemical aspect, the SARA four-component (Saturates, Aromatics, Resins, Asphaltenes) analysis method is widely used. This method separates the complex asphalt components into four parts, saturates, aromatics, resins and asphaltenes, according to the differences in polarity and solubility. In theory, the relative content of these four components directly determines the colloidal structure and macroscopic properties of asphalt. Therefore, establishing the correlation model between SARA components and macroscopic properties is the key to realize the performance prediction of asphalt. However, the existing standard SARA analysis method has many serious defects, which hinder its application in accurate prediction model. First of all, the traditional method (usually gravity-driven open column chromatography) is tedious and time-consuming, and it usually takes 6-8 hours or even longer to complete the analysis of a single sample, and a large amount of toxic and harmful organic solvents are consumed. Secondly, the reproducibility is poor, and the results (especially the content of aromatics and resins) may vary greatly when the same sample is tested by different laboratories or even different operators in the same laboratory, with an error sometimes exceeding 20%. The unreliability of such data makes any performance prediction model based on it lack of robustness and credibility. In addition, the traditional method may also cause the loss of light components due to solvent evaporation, further affecting the accuracy of quantitative analysis. On the other hand, even if accurate and reliable SARA component data can be obtained, establishing an effective prediction model also faces statistical challenges. Since the contents of SARA four components are expressed as component data in percentage, there is an inherent linear correlation between them, i.e. there is a serious multicollinearity problem. If the traditional ordinary least squares (OLS) method is used for multiple linear regression modeling, the multicollinearity will cause the estimated value of the model coefficient to be extremely unstable and very sensitive to small changes in sample data, thus greatly reducing the generalization ability and prediction accuracy of the model, and even leading to conclusions that are contrary to physical reality. In summary, there is a lack of a complete and systematic solution in the existing technology that can combine fast and reliable chemical component analysis with advanced modeling techniques that can overcome the inherent statistical difficulties of data, thereby achieving accurate prediction of the low-temperature performance of aged asphalt. Therefore, there is an urgent need in the art to solve the problems of low efficiency and poor reproducibility of SARA analysis, and then based on this, to establish a stable quantitative relationship between chemical components and macroscopic properties by using appropriate modeling strategies, to provide a scientific basis for performance-oriented design and rapid quality evaluation of asphalt materials. SUMMARY

[0004] In view of the above prior art, the present application discloses a method for predicting the low-temperature performance of aged asphalt, to solve the technical problems of low efficiency and poor reproducibility of SARA analysis in the prior art, and lack of appropriate modeling strategies to establish a quantitative relationship between the chemical components and macroscopic properties of aged asphalt.

[0005] In order to achieve the above object, the technical scheme adopted by the present application is as follows: a method for predicting low-temperature performance of aged asphalt is provided, which comprises the following steps: S1: aging treatment is performed on asphalt to obtain aged asphalt samples simulating different service life; S2: column chromatography is used to separate the aged asphalt, and then the saturated component content C S , the aromatic component content C A , the resin content C R and the asphaltene content C As are calculated by using the weight method and the difference method; S3: BBR test is performed on the aged asphalt to obtain the creep rate m value, and the grey correlation coefficient and the correlation degree between each chemical component and the low-temperature performance are calculated, so as to quantitatively identify the key chemical component which has the most significant influence on the low-temperature performance change; S4: the quantitative correlation between the m value and C S , C A , C R and C As is analyzed by using the grey correlation analysis method, then a multiple linear regression model is established by taking C S , C A , C R and C As as independent variables and the m value of the aged asphalt sample as a dependent variable, the model effectively overcomes the problem of multiple collinearity among the SARA component data by introducing an L2 regularization penalty term, and stable and reliable regression coefficients are obtained; then the C S , C A , C R and C As of the unknown aged asphalt sample are input, the m value of the unknown aged asphalt sample is calculated and the result is taken as an evaluation index of the low-temperature performance of the asphalt; In step S2, the column chromatography uses a chromatography column filled with double-layer adsorbents, the bottom adsorbent is activated alumina, and the upper adsorbent is activated column chromatography silica gel; the activation mode of the alumina is to heat the alumina at 360 DEG C for 5h, and the activation mode of the column chromatography silica gel is to heat the column chromatography silica gel at 150 DEG C for 18h; On the basis of the above technical scheme, the present application can also be improved as follows.

[0006] Further, the asphalt is rubber modified asphalt.

[0007] Further, the aging treatment mode is at least one of thermal oxidative aging, oxidative aging and low-temperature-ultraviolet coupling aging.

[0008] Further, gradient elution is used in column chromatography to separate each component, and the elution sequence is as follows: first, non-polar solvent is used to elute to obtain the saturate fraction, then medium-polar aromatic hydrocarbon solvent is used to elute to obtain the aromatic fraction, and finally strong-polar solvent is used to elute to obtain the gum.

[0009] Further, the non-polar solvent is n-heptane; the medium-polar aromatic hydrocarbon solvent is toluene; and the strong-polar solvent is a mixed solvent of dichloromethane and methanol, and the volume ratio of dichloromethane to methanol is 95:5.

[0010] Further, the multiple linear regression model is the following multiple linear regression equation: ; is the predicted creep rate m value, C S is the saturate content, C A is the aromatic content, C R is the gum content, C As is the asphaltene content.

[0011] The beneficial effects of the present application are: 1. The improved SARA analysis method shortens the analysis time of a single sample from 6-8h to less than 3h, and significantly improves the accuracy and reproducibility of the data through the standardized adsorbent activation and packing process, providing a high-quality data basis for subsequent modeling.

[0012] 2. The ridge regression model is first applied to the correlation analysis of SARA components and low-temperature performance, fundamentally solving the problem of failure of traditional regression methods due to multicollinearity, making the prediction model more robust and having stronger generalization ability.

[0013] 3. The present application provides a rapid and low-cost asphalt performance evaluation tool. Material manufacturers or engineering units do not need to perform time-consuming and long-term aging and low-temperature physical performance tests, but only need to perform a rapid chemical component analysis once to predict the long-term low-temperature performance of the material, greatly improving the research and development efficiency and the engineering quality control level. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 is a schematic diagram of a wet-packed chromatographic column; Figure 2 is the creep rate m value of the aged asphalt sample at-6℃; Figure 3 is a result graph of the four-component content data of the aged asphalt sample; Figure 4 is a correlation analysis result graph of the aged asphalt component comprehensive coefficient Ia and the corresponding measured m value; Figure 5 is a ridge regression fitting effect graph. DETAILED DESCRIPTION

[0015] The specific embodiments of the present application will be described in detail below with reference to the accompanying examples.

[0016] 1. Preparation and aging treatment of asphalt samples ① Preparation method: The base asphalt was heated to 177℃, and a predetermined amount of rubber powder was added under a stirring rate of 1000 rpm. After continuous stirring at this temperature and speed for 30 min, the rubber modified asphalt was obtained.

[0017] ② Short-term thermal oxidative aging: The Rolling Thin Film Oven (RTFO) test was performed according to the “Standard Test Methods of Bituminous Pitch and Asphalt Mixture for Highway Engineering” (JTG E20-2011). This step simulates the thermal oxidative aging of asphalt during the mixing, transportation and paving processes.

[0018] Long-term oxidative aging: The sample after RTFO treatment was subjected to accelerated aging in a Pressure Aging Vessel (PAV). This step simulates the long-term oxidative aging process of asphalt pavement during the 5 to 10 years of service using high temperature and high pressure air.

[0019] Strong radiation-low temperature coupling aging: To more realistically simulate the special environment of high-cold regions, the sample after PAV aging can be further subjected to low temperature-ultraviolet coupling cycle aging. A preferred cycle process is as follows: first, freeze the sample in a low temperature environment of -10℃ for 2 hours; then move it into a xenon lamp ultraviolet aging box, and perform ultraviolet light irradiation for 10 hours under the conditions of blackboard temperature 60℃, box humidity 60%, and light intensity 580W / m². Then, turn off the ultraviolet lamp for 1 hour, and repeat the cycle multiple times to simulate the cumulative aging effect of different service years.

[0020] 2. SARA four-component separation method ① This method uses a custom-made glass chromatography column with an inner diameter of 1.5 cm and a length of 60 cm. A polytetrafluoroethylene stopcock is provided at the lower end of the chromatography column to control the flow rate.

[0021] ② Activation of adsorbent: The basic alumina (Brockmann activity I grade, particle size 50-200 pm) was placed in a muffle furnace and heated at 360℃ for 5h to obtain, which can completely remove the adsorbed water and achieve the highest activity. The activated basic alumina serves as the bottom adsorbent. The column chromatography silica gel (70-230 mesh) was placed in an oven and heated at 150℃ for 18h to obtain. This condition is sufficient to remove physically adsorbed water while avoiding excessive changes in the surface structure of the silica gel. The activated column chromatography silica gel serves as the upper adsorbent.

[0022] ③ Wet packing of chromatographic columns: such as Figure 1 As shown, a small wad of defatted glass wool is placed above the stopcock at the bottom of the chromatography column to prevent adsorbent loss. 10g of activated basic alumina is mixed with 50mL of n-heptane in a beaker to form a homogeneous slurry. This slurry is poured into the chromatography column all at once, the stopcock is opened, and the n-heptane is allowed to flow out slowly. Simultaneously, the column wall is gently tapped with a rubber mallet to ensure the alumina particles settle tightly and uniformly. Once the alumina layer has completely settled and the upper liquid level has dropped to near the adsorbent surface, 30g of activated silica gel is mixed with 100mL of n-heptane to form a slurry, which is then carefully added above the alumina layer. The column wall is continued to be tapped gently until the silica gel layer has completely settled. This bilayer structure, with alumina at the bottom and silica gel at the top, utilizes the selectivity differences of the two adsorbents for components of different polarities, enabling clearer separation. Then, a layer of clean sea sand about 0.5 cm thick is placed on top of the silica gel layer to prevent the surface of the adsorbent from being disturbed when the eluent is added. Heptane is continuously added until the eluent is clear and transparent, and then the liquid level is lowered to the top of the sea sand layer, thus completing the preparation of the chromatographic column. Throughout the process, it is necessary to ensure that the adsorbent bed is always wetted by the solvent and does not dry out.

[0023] ④ Sample loading and rinsing separation (1) Sample loading: Accurately weigh about 1.0 g of aged asphalt sample, dissolve it completely in 3 mL of toluene, and carefully and evenly add the solution to the sea sand layer at the top of the chromatographic column using a dropper. Open the stopcock to allow the sample solution to completely enter the adsorbent bed.

[0024] (2) Washing steps: Part 1 (saturated fraction): Use 150 mL of n-heptane as the washing solution and wash at a flow rate of approximately 5 mL / min. Collect all the effluent. n-Heptane is a non-polar solvent that can elute the weakest polar saturated fractions (mainly chain and cyclic alkanes) in the asphalt.

[0025] Part Two (Aromatic Fractions): Replace the receiving bottle with 200 mL of toluene as the eluent. Toluene is more polar than n-heptane and can elute aromatic fractions (compounds containing aromatic structures such as benzene rings) with slightly stronger adsorption capacity from the adsorbent.

[0026] Part 3 (Colloids): Replace the receiving bottle again and use a 150 mL mixture of dichloromethane and methanol (95:5 volume ratio) as the eluent. This highly polar mixed solvent can elute highly polar colloids (large molecules containing many heteroatoms and polar functional groups).

[0027] (4) Quantification of components: The collected three parts of effluent were placed on a rotary evaporator, and the solvent was removed under water bath heating (temperature not more than 60°C) and reduced pressure. The residue of each component after removing the solvent was placed in an oven at 105°C for 30 min to remove residual solvent, then cooled to room temperature in a desiccator, accurately weighed, and the mass of saturated fraction m sat , aromatic fraction m aro and resin m res was obtained, respectively. Asphaltene is the most polar and most difficult to dissolve component in asphalt, which will be firmly adsorbed on the column head under the above elution conditions and will not be eluted, and its mass is calculated by difference method: m asp = m total − m sat − m aro − m res , where m total is the total mass of the initial sample. Finally, the mass percentage of each component was calculated: the content of saturated fraction C S = (m sat / m total ) × 100%, the content of aromatic fraction C A = (m aro / m total ) × 100%, the content of resin C R = (m res / m total ) × 100%, and the content of asphaltene C As = (m asp / m total ) × 100%.

[0028] 3. Grey correlation analysis method to analyze the close degree of chemical components and low temperature creep rate m value The creep rate m value measured by BBR test of a group of aged asphalt samples was taken as the reference sequence, because the m value directly reflects the advantages and disadvantages of low temperature performance. The corresponding SARA four-component content was taken as four comparison sequences, and since the dimensions and numerical ranges of each sequence are different, all data need to be normalized or initialized to convert them into dimensionless sequences that can be compared. The correlation coefficient of each comparison sequence with the reference sequence at each sample point was calculated. The calculation formula is:

[0029] where x0(k) is the value of the reference sequence at the kth sample point, x i (i, k) is the value of the ith comparison sequence at the kth sample point, and p is the resolution coefficient, usually taken as 0.5. The average of all correlation coefficients of each comparison sequence is taken to obtain the grey correlation degree of the sequence with the reference sequence:

[0030] n is the number of sample points, r i is the grey correlation degree of the ith comparison sequence and the reference sequence, i.e. the average of the correlation coefficients of all sample points of the sequence. The grey correlation degrees of the four components are compared. The greater the correlation degree value, the more similar the change trend of the component content and the low-temperature creep rate m value, i.e. the more significant the influence on the low-temperature performance. Through this step, the key chemical component affecting the low-temperature performance of asphalt can be determined.

[0031] 4. Establishing a prediction model As described in the background, the SARA four-component data has serious multicollinearity. Ridge regression is a biased estimation regression method specially used for processing the problem of multicollinearity. The basic idea is to add an L2 regularization term (penalty term) to the loss function of ordinary least squares (OLS), which is equal to the sum of squares of regression coefficients multiplied by a normal number λ.

[0032] The loss function of OLS is:

[0033] The loss function of ridge regression is:

[0034] By minimizing this new loss function, ridge regression not only fits the data, but also penalizes the size of the regression coefficient. When λ>0, the model tends to select regression coefficients with smaller absolute values. Although this "shrinkage" effect brings a certain bias to the coefficient estimates, it can significantly reduce their variance, thereby obtaining a more stable, reliable and powerful prediction model than OLS in the presence of multicollinearity.

[0035] The established prediction model is a multiple linear regression equation:

[0036] wherein, is the predicted creep rate m value, β0 is the intercept term, β1, β2, β3, β4 are the regression coefficients of saturates, aromatics, resins and asphaltenes content respectively. These coefficients are estimated by ridge regression algorithm. A part of samples (training set) are used to train the model with their SARA component data and corresponding measured m value. The key of the training process is to determine the optimal ridge parameter λ. This is usually done by cross-validation method, i.e. to select a λ value that can minimize the prediction error (such as mean square error MSE) of the model. Another part of samples (test set) that are not involved in the training are used to validate the prediction performance of the model. The SARA component data of the test set are input into the trained model to get the predicted m ̂ value, which is compared with the measured m value. The accuracy of the model is evaluated by calculating the coefficient of determination (R2), root mean square error (RMSE) and other indicators. Once the model is verified to be reliable, it can be put into practical application. For any unknown sample, only its four-component content is measured by the improved SARA analysis method of the present application, and then substituted into the model, the low-temperature creep rate m value can be quickly predicted.

[0037] Example 1 In this example, 90# base asphalt, 40 mesh rubber powder, and three different rubber powder contents (10%, 15%, 20%) are selected to prepare three different rubber modified asphalt samples (codes M7 to M9) by external mixing method as shown in Table 1. Preparation method: the base asphalt is heated to 177℃, the predetermined amount of rubber powder is added under the stirring speed of 1000 revolutions per minute, and after 30 minutes of continuous stirring at this temperature and speed, it is obtained. And carry out ultraviolet aging (0, 4, 6, 12 months) in a low temperature environment of-6℃.

[0038] Table 1 Rubber modified asphalt of various formulations

[0039] The above samples are subjected to BBR test and SARA analysis to obtain their creep rate m value at-6℃ and four-component content data, and the data results are shown in Figure 2 and Figure 3 .

[0040] In order to visually show the internal relationship between chemical components and low temperature performance, the asphalt component comprehensive coefficient Ia is introduced, which is defined as the ratio of heavy component content to light component content: Ia = (C S +C A ) / (C As +C R ) The Ia value of all samples is correlated with the corresponding measured m value (-6℃), and the results are shown in Figure 4 From the figure, it can be seen that the creep rate m value and the Ia value show a strong negative correlation (coefficient of determination R2 The value is very high), that is, as the aging deepens, the proportion of heavy components (Ia value) increases, the stress relaxation ability (m value) of asphalt significantly decreases, and the low temperature performance deteriorates. This provides a solid physical and chemical basis for the subsequent establishment of a four-component prediction model.

[0041] The obtained SARA component data (C S ,C A ,C R ,C As ) were used as independent variables, and the measured m value at -24℃ was used as the dependent variable. A ridge regression model was established using statistical software. After determining the optimal ridge parameter λ by cross-validation, the following prediction equation was obtained:

[0042] In order to verify the prediction accuracy of the model, a part of the data was used as a test set. The SARA component content of the test set samples was substituted into the above model to calculate the predicted value. The predicted value was compared with the measured m value, and the results are shown in Table 2 and Figure 5 .

[0043] Table 2 Model predicted value and measured m value

[0044] As can be seen from Table 2 and Figure 5 , the model predicted value is highly consistent with the measured value, the relative error is basically controlled within ±3%, and the determination coefficient R 2 is close to 1.0. This fully proves that the method proposed in the present application can accurately and reliably predict the low temperature performance of aged rubber asphalt, and has extremely high industrial application value.

[0045] Although the specific embodiments of the present application are described in detail in combination with the embodiments, it should not be understood as limiting the protection scope of the present patent. Various modifications and variations made by those skilled in the art within the scope described in the claims are still within the protection scope of the present patent.

Claims

1. A method for predicting the low-temperature performance of aged asphalt, characterized in that, Includes the following steps: S1: Asphalt is aged to obtain aged asphalt; S2: The aged asphalt was separated using column chromatography, and the saturated fraction content C was determined. S Aromatic content C A Gluten content C R and asphaltene content C As ; S3: The creep rate m value was measured by BBR test on the aged asphalt; S4: Use grey relational analysis to analyze the relationship between m and C. S C A C R and C As The quantitative correlation between them, and then with C S C A C R and C As A multiple linear regression model is established with m as the independent variable and the value of the aged asphalt sample as the dependent variable. Then, the C-value of the unknown aged asphalt sample is input. S C A C R and C As The m value of unknown aged asphalt samples was calculated and the result was used as an evaluation index of the low-temperature performance of asphalt. In step S2, the column chromatography method uses a chromatography column packed with a double-layer adsorbent. The bottom adsorbent is activated alumina, and the top adsorbent is activated silica gel. The alumina is activated by heating it at 360°C for 5 hours, and the silica gel is activated by heating it at 150°C for 18 hours.

2. The method for predicting the low-temperature performance of aged asphalt according to claim 1, characterized in that: The asphalt is rubber-modified asphalt.

3. The method for predicting the low-temperature performance of aged asphalt according to claim 1, characterized in that: The aging treatment method is at least one of thermo-oxidative aging, oxidative aging, and low-temperature-ultraviolet coupling aging.

4. The method for predicting the low-temperature performance of aged asphalt according to claim 1, characterized in that: In column chromatography, a gradient elution method is used to separate the components. The elution order is as follows: first, a non-polar solvent is used to elute the saturated fraction, then a moderately polar aromatic hydrocarbon solvent is used to elute the aromatic fraction, and finally a strongly polar solvent is used to elute the colloid.

5. The method for predicting the low-temperature performance of aged asphalt according to claim 4, characterized in that: The non-polar solvent is n-heptane; the moderately polar aromatic hydrocarbon solvent is toluene; and the strongly polar solvent is a mixture of dichloromethane and methanol, with a volume ratio of dichloromethane to methanol of 95:

5.

6. The method for predicting the low-temperature performance of aged asphalt according to claim 1, characterized in that: The multiple linear regression model is the following multiple linear regression equation: ； It is the predicted creep rate m value, C S For saturated fraction content, C A For aromatic content, C R For gelatin content, C As This refers to the asphaltene content.

Images