One of the oldest systems for improvement of loose soils beneath structures foundation is using in-situ concrete piers. In recent years, the use of rammed aggregate piers (RAPs) has been considered as a modern method for retrofitting loose soils and a suitable alternative for deep and semi-deep foundations. However, the behavior comparison of these two types of piers as two types of semi-deep foundations has been less studied. For this purpose, this study deals with numerical study and comparison of the numerical and experimental results of these two semi-deep foundation systems in retrofitting of loose soils. The experimental results are mainly based on the research conducted by Dr. Niroumand. In this regard, two groups of five rammed aggregate and in-situ concrete piers with constant diameter and different lengths were built in the coastal soils in Bushehr Port and subjected to loading test by Dr. Niroumand. Numerical modelling of the piers has been performed in order to compare and identify the failure mechanism and determine their compressive bearing capacity using the finite element method as two-dimensional using the plaxis2D software. For numerical modelling of the RAPs and surrounding soils, the hardening soil model with the parameters obtained from the field tests has been used. According to the comparison done between the experimental and numerical analysis results, the results obtained from the numerical modelling had good agreement with the field results in small scale, so that the ratio of the design limit load of the concrete pier to the RAP is approximately equal to unit. Whereas, the ratio of the top settlement of the in-situ concrete pier in the design limit is 1.4 times the corresponding values in RAPs in average which indicates the better performance of the RAPs than the in-situ concrete piers in the same dimensional conditions (the same length and diameter). Furthermore, in order to investigate the effects of the piers dimensions (length and diam