The demand for High Performance Concrete (HPC) is steadily increasing with massive developments. Conventionally, it is possible to use industrial products such as silica fume (SF), fly ash, as supplementary cementitious materials (SCM), to enhance the attributes of HPC. In recent years, nano-silica (NS) is used as an additive in added mainly to fill up the deviation arises with the addition of SF for HPC. This study aims to optimize the proportion of NS (produced in Vietnam) in the mixture used for fabricating 70 MPa high-performance concrete. SiO2 powder with particle size from 10 to 15 nm were used for mixing. A series of compressive strength test of HPC with nano-SiO2 varied from 0 to 2.8 percent of total of all binders (0%, 1.2%, 2%, 2.8%), and the fixed percentage of silica fume at 8% were proposed. Results show compressive strength increases with the increase of nano-SiO2, but this increase stops after reaching 2%. And at day 28 of the curing period, only concrete mixture containing of 8% silica fume and 2% nano-SiO2, had the highest compressive strength.

The expressway network in Vietnam is developing strongly, playing the role of the backbone of the national road system, in which bridge construction accounts for a large proportion. With many specific characteristics and complex risks always hidden in all stages of the expressway project, risk assessment to have solutions and plans to prevent and respond to risks, limiting the impacts of quality assurance and operational safety of the works is essential. However, the current risk assessment and forecasting models still have many limitations. The application of Machine Learning to all aspects of life is getting more popular. This article develops the algorithms, models and program to assess the technical risks in the period of construction and service of expressway bridges in Vietnam using Machine Learning, in order to solve the current limitations in this work. The selection of key influencing factors is especially important in the field of risk assessment. It improves the classification model's performance by focusing only on the most important factors in the data. Via the applications of artificial neural networks and the Random Forest Algorithm in data processing, the performance risks for bridge management can be analyzed, and performed in more detail and exactly. The possible multiple and non-linear relationships of the risks can be investigated. Based on the results, the proposed model helps the managers to make optimum decisions on managing the risks in advance and to obtain sustainable solutions

No abstract