Development of Regression Models for Predicting Properties of High Strength Concrete Using Nondestructive TestsMohiuddin Khan, Shibli Russel (2007) Development of Regression Models for Predicting Properties of High Strength Concrete Using Nondestructive Tests. PhD thesis, Universiti Putra Malaysia.
AbstractHigh strength concrete (HSC) is a relatively recent development in concrete technology. It is being used increasingly in major civil engineering and building projects. This leads to the need for quality assurance of the insitu concrete. Testing of concrete traditionally involved compression testing of cylinders or cubes to obtain the properties and these may not adequately represent the insitu properties of concrete. This necessitates the use of nondestructive test (NDT). There are no standard relationships that had been established for high strength concrete physical and mechanical properties using Sclerometer test, Ultrasonic Pulse Velocity (UPV) methods and Pullout test. Prediction models need to be developed for concrete strength, density and static elastic modulus estimation. They are normally required in building or structural assessment, especially with the present trend of constructing modern structures using high strength concrete. Eight different mix proportions of HSC containing sandstone aggregate of nominal sizes of 10mm and 19mm and silica fume content were investigated in this study.The silica fume contents were varied at 0%, 5%, 10% and 15%. These mixes produced concrete at 28day strength between 40 MPa to 100 MPa. A total of 360 standard cubes (150mm), 144 cylinders (150 x 300mm) and 16 reinforced beams were cast for this study. A total of fortyfive standard cube specimens for each mix were tested at the age of 3, 7, 14, 28 and 56 days in both, nondestructive and destructive manner. On the other hand, eighteen cylinder specimens for each mix were tested at the age of 28 and 56 days in both, nondestructive and destructive manner. As for the pullout test some fortyfive inserts were prepared for each mix at the age of 3, 7, 14, 28 and 56 days. For each destructive test, an average of 45 values of nondestructive tests was obtained, which depends on the type of NDT techniques used. The results were analyzed using statistical tools (SPSS ver.13). The prediction models for each NDT technique were developed based on the obtained experimental results. Statistical tests of significance on the predicted models were performed to ascertain their reliability in estimating the concrete properties. Predicted models were also further validated using data from other researchers. The models developed in this study are expected to be used to estimate strength, density and static elastic modulus parameters using Sclerometer test, UPV method and Pullout test. The generalized power models for strength, density and modulus of elasticity prediction using Sclerometer and Pullout test were found to be unaffected by the aggregate sizes. The maximum error of these models were found to be ±12.5% for strengthSclerometer test, ±25% for strengthPullout test, ±3% for densitySclerometer test, ±2% for densityPullout test and ±5% for static elastic modulusSclerometer test.Strength, density and static modulus of elasticity prediction for direct and indirect UPV methods indicated that aggregate sizes should be known in advance. Generalized quadratic models were proposed for concrete mix with nominal aggregate size 10mm (series A10) for strength, density and modulus of elasticity prediction using UPV direct method. The maximum error of these models was found to be ±20% for strength, ±3% and ±5% for density and static modulus of elasticity respectively. A linear model for strength, a power model for density and a logarithmic model for static elastic modulus was proposed for 19mm maximum aggregate size. The quadratic models are valid for pulse velocity range between 4.7 to 6.1 km/sec and the other models are 4.3 to 5.5 km/sec. All of these models are found to be capable of predicting strength between 30 to 110 MPa, density between 2320 to 2525 kg/m3 and static elastic modulus between 28 to 40 GPa. Combined NDT methods were found to improve some of strength prediction. Statistical significant tests on the prediction models have been carried out to ascertain their reliability in estimating strength, density and static elastic modulus properties of concrete. Moreover, validation of the predicted models with other researchers further enhances reliability of each model. Thus, the proposed models for different NDT techniques can be used as a practical guide in the assessment of insitu concrete properties.
Repository Staff Only: Edit item detail
