The effect of bedding errors on the accuracy of plate load tests

This dissertation addresses the effect of bedding errors on the accuracy of plate load tests on soil. For this research project the bedding errors were quantified by measuring the surface roughness by means of a high precision laser measuring system. The test area was scanned before and after each test in order to evaluate the change in surface roughness during each test. In addition, a modified plate load test was designed to eliminate the effect of bedding errors that occur during these tests. Telescopic probes were used to measure the relative displacement at two points below the centre of the plate.;The stiffness values, determined from the vertical displacement of the plate, were compared with the internal stiffness values determined by means of the telescopic probes. All stiffness values were compared with continuous surface wave (CSW) measurements performed on the same material.;The main objectives of this research project were, firstly, to quantify the bedding errors that occurs during plate load tests. Secondly, to study and evaluate three different surface preparation methods used in plate load tests to achieve a levelled and smooth test surface and lastly to evaluate the effectiveness of using telescopic probes in routine plate load tests to eliminate the effects of bedding errors.;The results of this dissertation have demonstrated that bedding errors could have a significant effect on soil stiffness values, which have been determined by the 'conventional' method of conducting plate load tests. Plaster of Paris resulted in the most accurate stiffness values determined by conventional plate load tests. The results have further demonstrated that telescopic probes are necessary to determine soil stiffness values with a high level of accuracy and bedding errors cannot be eliminated on a poorly prepared surface by applying initial load cycles.;A series of plate load tests was conducted at the experimental farm of the University of Pretoria. In total six precise plate load tests were performed. The test areas were levelled by means of three surface preparation methods namely, i) a thin layer of plaster of Paris; ii) a thin layer of well-graded sand and iii) by using only hand tools.