Trial Program Study of Dynamic Compaction Earth Work in Taiwan

A thermal power plant is planned to be built at Taichung Harbor area, and the construction site is located on a hydraulic fill. The medium-to-fine, loose [N value of the Standard Penetration Test (SPT), < 10] sand deposits on the hydraulic fill will be subject to liquefaction, and because of this, soil improvement is required. Dynamic compaction is thus far considered one of the most appropriate methods used to solve liquefaction, and therefore, a trial program was initiated to obtain ongoing construction details for the whole construction site. The trial area was divided into three sections (Sections A, B, and C) to carry out three different trials. For Section A, the drop weight is 18 Mg, the drop height is 22 m, the grid spacing is 4 m, and the drop number is 8. For Sections B and C, the drop weights are 25 Mg, the drop heights are 25 m, grid spacings are 5 and 6 m, respectively, and drop numbers are 10 and 40, respectively. To achieve a better result in densifying the ground surface, a post-compaction program is introduced into this project before roller compaction is performed. During the post compaction, all trials have drop weights of 13.88 Mg, drop heights of 10 m, and a grid spacing of 2 by 3 m for Section A and a grid spacing of 3 by 3 m for Sections B and C. Results of improvement are expressed as the N value of SPT (SPT-N) and cone resistance, q_c. In addition, the excess pore water pressure was also investigated during compaction. Findings indicated that either the SPT-N or the Cone Penetration Test q_c improved about 1.5 to 2.0 times. Satisfactory results were also detected for the post compaction. Comparison and analysis showed that the program used for the Section B trial, that is, the trial program with application of medium energy, is the most effective program for use in the whole project. In addition, the soil particle velocity caused by tamping is found to be less than 30 mm/s and will cause no tangible effect on adjacent structures.