PURPOSES : For most local governments, including that of Gangwon-do, the establishment of an organized pavement management system is insufficient, resulting in problems such as inefficient distribution and use of maintenance budgets for deteriorated road pavements. In this study, we aimed to contribute to the establishment of a more reasonable road maintenance strategy by developing a model for predicting the annual international roughness index (IRI) change for national highway asphalt pavements in Gangwon-do based on big data analysis. METHODS : Data on independent and dependent variables used for model development were collected. The collected data were subjected to exploratory data analysis (EDA) and data preprocessing. Independent variable candidates were selected to reduce multicollinearity through correlation analysis and specific conditions. A final model was selected, and sensitivity analysis was performed. RESULTS : The final model that predicts annual IRI change uses independent variables such as annual temperature range, minimum temperature, freeze-thaw days, IRI, surface distress (SD), and freezing days. The sensitivity analysis confirmed that the annual IRI change was affected in the order of annual temperature range, minimum temperature, freeze-thaw days, IRI, SD, and freezing days. CONCLUSIONS : Road maintenance can be performed rationally by predicting future pavement conditions using the model developed in this study. The accuracy of the prediction model can be improved if additional data, such as material properties and pavement thickness, are obtained in future studies.

PURPOSES : The purpose of this study was to select recurring congested sections through an objective and reasonable method using intelligent transportation system (ITS) collection data as the first phase for controlling recurring congested sections. The selection is meant to enhance the utilization of national highway ITS collection data that are just managed as statistical data focusing on real-time traffic information or not satisfactorily used. METHODS : Hourly statistical data were obtained using travel speed with a 5-min frequency per information-providing section stored in the database server. The travel speed data were collected from field equipment in general national highway sections (793.8 km in 380 sections of 16 lines in 2017) in the Seoul Metropolitan Area through a literature review of congestion standards. Subsequently, the congested sections were drawn by dividing a month into weekdays and weekends, and the status of the monthly change and characteristics of the congested sections were analyzed. Based on the monthly data on the congested sections, monthly mean congested days and hours were calculated to select the significant recurring congested sections in 2017 and applied to the congestion frequency standard. RESULTS : Recurring congested sections occurred in seven sections (33.9 km) in three lines — in five sections (24.8 km) of two lines during the weekdays and in two sections (9.1 km) of one line during the weekends. The recurring congested sections selected based on the congested days and the recurring congested sections selected based on the congested hours were found to be the same. The recurring congested sections during the weekdays and weekends were not mutually duplicated and were divided. More congestion occurred during the weekdays than on weekends. CONCLUSIONS : In the seven sections selected as recurring congested sections, congestion occurred for more than six months during one year, indicating that congestion is concentrated in specific months, albeit continuously and repeatedly occurring over the long term. The recurring congestion section selection standard and method used in this study are considered to be objective and reasonable. If recurring congested sections are selected using the standard and method presented in this study using ITS collection data targeting general national highways nationwide in the Seoul Metropolitan Area, it is necessary to determine whether the standard and method reflect actual congestion situations satisfactorily. According to the result, a further study considering congestion speed and congestion frequency standards is needed.

PURPOSES : To efficiently manage pavements, a systematic pavement management system must be established based on regional characteristics. Suppose that the future conditions of a pavement section can be predicted based on data obtained at present. In this case, a more reasonable road maintenance strategy should be established. Hence, a prediction model of the annual surface distress (SD) change for national highway pavements in Gangwon-do, Korea is developed based on influencing factors. METHODS : To develop the model, pavement performance data and influencing factors were obtained. Exploratory data analysis was performed to analyze the data acquired, and the results show that the data were preprocessed. The variables used for model development were selected via correlation analysis, where variables such as surface distress, international roughness index, daily temperature range, and heat wave days were used. Best subset regression was performed, where the candidate model was selected from all possible subsets based on certain criteria. The final model was selected based on an algorithm developed for rational model selection. The sensitivity of the annual SD change was analyzed based on the variables of the final model. RESULTS : The result of the sensitivity analysis shows that the annual SD change is affected by the variables in the following order: surface distress ˃ heat wave days ˃ daily temperature range ˃ international roughness index. CONCLUSIONS : An annual SD change prediction model is developed by considering the present performance, traffic volume, and climatic conditions. The model can facilitate the establishment of a reasonable road maintenance strategy. The prediction accuracy can be improved by obtaining additional data, such as the construction quality, material properties, and pavement thickness.

PURPOSES : The surface distress of asphalt pavements is one of the major factors affecting the safety of road users. The aim of this study was to analyze the factors influencing the occurrence of surface distress and statistically predict its annual change to contribute to more reasonable asphalt pavement management using the data periodically collected by the national highway pavement data management system. METHODS : In this study, the factors that were expected to influence the surface distress were determined by reviewing the literature. The normality was secured by changing the forms of the variables to make the distribution of the variables got closer to normal distribution. In addition, min-max normalization was performed to minimize the effect of the unit and magnitude of the candidate independent variables on the dependent variable. The final candidate independent variables were determined by analyzing the correlation between the annual surface distress change and each candidate independent variable. In addition, a prediction model was developed by performing data grouping and multi-regression analysis. RESULTS : An annual surface distress change prediction model was developed using present surface distress, age, and below 0 ℃ days as the independent variables. As a result of sensitivity analysis, the surface distress affected the annual surface distress change the most. The positive correlation between the dependent variable and each independent variable demonstrated engineering and statistical meaningfulness of the prediction model. CONCLUSIONS : The surface distress in the future can be predicted by applying the annual surface distress prediction model to the national highway asphalt pavement sections with survey data. In addition, the prediction model can be applied to the national highway pavement condition index (NHPCI) evaluating the national highway asphalt pavement conditions to be used in the prediction of future NHPCI.

PURPOSES : Rut depth of asphalt pavements is a major factor that affects the maintenance of pavements as well as the safety of drivers. The purpose of this study was to analyze the factors influencing rut depth, using data collected periodically on national highways by the pavement management system and, consequently, predict annual rut depth change, to contribute to improved asphalt pavement management. METHODS : The factors expected to influence rut depth were determined by reviewing relevant literature, and collecting the related data. Further, the correlations between the annual rut depth change and the influencing factors were analyzed. Subsequently, the annual rut depth change model was developed by performing regression analysis using age, present rut depth, and annual average maximum temperature as independent variables. RESULTS : From the sensitivity analysis of the developed model, it was found that age affected the annual rut depth change the most. Additionally, the relationship between the dependent and independent variables was statistically significant. The model developed in this study could reasonably predict the change in the rut depth of the national highway asphalt pavements. CONCLUSIONS : In summary, it was verified that the model developed in this study could be used to predict the change in the National Highway Pavement Condition Index (NHPCI), which represents comprehensive conditions of national highway pavements. Development of other models that predict changes in surface distress as well as international roughness index is required to predict the change in NHPCI, as they are the independent variables of the NHPCI prediction model.

PURPOSES : Despite the availability of larger traffic data and more advanced data collection methods, the problem of missing data is yet to be solved. Imputing missing data to ensure data quality and reliability of statistics has always been challenging. Missing data are imputed via several existing methods, such as autoregressive integrated moving average, exponential smoothing, and interpolation. However, these methods are complicated and results in significant errors. METHODS : A deep-learning method was applied in this study to impute traffic volume data of the South Korean national highway. Traffic data were trained using the long short-term memory method, which is a suitable deep-learning method for time series analysis. RESULTS : Three cases were proposed to estimate the traffic volume. In the first case, which represented the general conditions, the mean absolute percentage error (MAPE) was 12.7%. The second estimation case, which was based on the opposite traffic flow, exhibited a MAPE of 17%~18%. The third case, which was estimated using adjacent-section data, had a MAPE of 18.2%. CONCLUSIONS : Deep learning may be a suitable alternative data imputation method based on the limited site and data. However, its application depends on the specific situation. Furthermore, deep-learning models can be improved using an ensemble method, batch-size, or through model-structure optimization.

PURPOSES : The study aims to predict the service life of national highway asphalt pavements through deep learning methods by using maintenance history data of the National Highway Pavement Management System. METHODS: For the configuration of a deep learning network, this study used Tensorflow 1.5, an open source program which has excellent usability among deep learning frameworks. For the analysis, nine variables of cumulative annual average daily traffic, cumulative equivalent single axle loads, maintenance layer, surface, base, subbase, anti-frost layer, structural number of pavement, and region were selected as input data, while service life was chosen to construct the input layer and output layers as output data. Additionally, for scenario analysis, in this study, a model was formed with four different numbers of 1, 2, 4, and 8 hidden layers and a simulation analysis was performed according to the applicability of the over fitting resolution algorithm. RESULTS: The results of the analysis have shown that regardless of the number of hidden layers, when an over fitting resolution algorithm, such as dropout, is applied, the prediction capability is improved as the coefficient of determination (R2) of the test data increases. Furthermore, the result of the sensitivity analysis of the applicability of region variables demonstrates that estimating service life requires sufficient consideration of regional characteristics as R2 had a maximum of between 0.73 and 0.84, when regional variables where taken into consideration. CONCLUSIONS : As a result, this study proposes that it is possible to precisely predict the service life of national highway pavement sections with the consideration of traffic, pavement thickness, and regional factors and concludes that the use of the prediction of service life is fundamental data in decision making within pavement management systems.

PURPOSES: Case studies of an asphalt-overlay project with a performance-based contract method were conducted on a national highway in Korea to evaluate the effect of the method on asphalt pavement maintenance. This study evaluated the procedure of the performance-based contract method. METHODS: In this study, an asphalt-pavement maintenance project for a national highway was assessed with a performance-based contract to investigate the advantage of the new contract procedures. This is the first trial applying the performance-based contract to a pavementrehabilitation project in Korea. In the four case studies, the warranty period of the performance-based contract was designed for seven years. The research team monitored the construction site to compare the normal contract method with the performance-based contract method. The case studies’project sites were investigated after the end of the construction. RESULTS : Based on the limited case studies, the performance-based contract method could extend the service life of the asphalt pavement and reduce the pavement-maintenance budget because the quality control was well managed by the contractors. However, a few construction laws would be necessary to apply the performance-based contract method in the future. CONCLUSIONS : Using the performance-based contract, the construction company made great efforts to guarantee the warranty period and to apply the optimal maintenance method, based on the pavement distress condition. The contractor and the agency would need to understand the new performance-based contract system for it to be activated. Therefore, a proper education program for the performancebased contract system would be needed to educate the stakeholders regarding the procedures and their effects on the pavement management and maintenance.

PURPOSES: The occurrence of unexpected disasters, including fire events, increases as the road network becomes complicated and traffic volume increases. When a fire event occurs on and under bridges, the damage extensively influences direct damage to structures, vehicles, and human life and secondary socioeconomic issues owing to traffic blockage. This study investigated potential fire-hazard risks on bridges of the Korean national route roadMETHODS: The investigation was conducted using field investigation and analysis with satellite pictures and road views from commercial websites and the Bridge Management System (BMS). From the filed investigation, various potential fire resources were identified. The satellite pictures and road views were helpful in measuring and recognizing conditions underneath bridges, stowage areas, etc.RESULTS : There are various potential fire resources underneath bridges such as piled agricultural products, parked petroleum tanks, construction equipment, and attached high-voltage cables. A total of 94.6% of bridges have underneath clearances of less than 15 m. A bridge underneath volume that can stow a potential fire hazard resource was 7,332 m3 on average, and most bridges have about 4,000 m3 of space. Based on the BMS data, the amounts of PSC and steel girders were 29% and 25%, respectively.CONCLUSIONS : It was found that the amount of stowed potential fire hazard resources was proportional to the underneath space of bridges. Most bridges have less than 15 m of vertical clearance that can be considered as a critical value for a bridge fire. The fire risk investigation results should be helpful for developing bridge fire-protection tools.

PURPOSES : The purpose of this study is to analyze the performance life of hot central plant recycling (HCPR) and hot in-place recycling (HIR) pavements applied to the National Highway for the past 20 years and compare it with conventional hot-mix asphalt (HMA) pavement. METHODS: In order to analyze the performance life of recycling asphalt pavements, a comprehensive literature review was conducted to investigate the government law and official system for the use of recycling asphalt pavement in Korea and foreign countries. Next, the application information of using a hot central plant recycling and hot in-place recycling pavements in the national highway is collected from the database of pavement management system (PMS) and then their field condition is visually surveyed. Finally, the performance life of recycling asphalt pavements in the national highway is analyzed and compared with conventional hot-mix asphalt pavement. RESULTS: Institutions are encouraging the promotion of using recycled asphalt pavement through various legal systems in Korea as well as abroad. Based on analysis results for the average performance life of hot central plant recycling pavement applied to the national highway, the average performance life is estimated to be 10.2 years. However, the average performance life of in-place recycling pavement is estimated to be 6.5 years. However, it is expected to increase performance life after the HIR construction system is modified. CONCLUSIONS : Based on the limited data analysis of the performance life of recycled asphalt pavements, HCPR shows similar performance life to conventional asphalt pavement but HIR shows shorter performance life than conventional asphalt pavement. However, it is noted that all performance life data is very limited and it should be monitored and analyzed further.

우리나라 도로교통안전 수준은 선진국에 비해 하위권에 속한다. ‘17년까지 자동차 1만대당 교통사고사망자 2.4명, OECD 평균 1.2명 수준으로 OECD 수준으로 달성을 위해 국정과제로 교통안전 선진화 대책 추진 중이다. 도로의 안전성을 객관적·정량적으로 판단하여 효과적인 안전사업 계획 수립을 위해 국내 도로 여건을 반영한 도로안전편람을 개발하고자 한다. 미국의 경우 도로안전편람(2010, AASHTO) 발간을 통해 도로 안전성 향상 사업을 수행하는데 참고하는 기본체계를 확립한 바 있다. 도로안전편람 개발을 위해 이 연구는 일반국도 상 입체교차로를 대상으로 위험도평가모형을 개발하였다. 전국 243개 입체교차로에 대해서 3년간(2011년∼2013) 발생한 교통사고를 조사하였으며, 도로기하구조 특성(입체교차로 유형, 진출입로 길이 등) 및 교통량(주도로, 부도로 등)을 조사하였다. 모형은 음이항회귀모형을 이용하여 사고건수, 준도시/지방부, 본선/램프모형을 구축하였다. 구축된 모형의 정확성 분석을 위해 MAD와 MPB를 이용하였다. 모형구축 결과를 바탕으로 입체교차로의 안전성능함수(SPF)와 사고보정계수(CMF)를 개발하였다. 교통량(주도로, 부도로), 감속차로 최소길이 불만족 개수, 램프형태 등이 입체교차로의 교통사고발생에 주요한 영향을 미치는 것으로 나타났다.