It is known that magnetic drum separators are the most widely used type of magnetic equipment and separates magnetic material from its non-magnetic counterpart with consideration of magnetic field. The simulation of the arrangement of permanent magnet elements in a magnetic drum separator is the subject of this study to understand the magnetic flux density of this type of separator. The production cost of this equipment is high and big proportion of this cost is allocated by preparation and procurement of bar magnets. Therefore, the simulation of magnetic flux density of these magnet elements and the arrangement of permanent magnet elements become significant and play an important role in the design of drum separator. In this thesis, the commercial finite element software Maxwell is employed to simulate 2D modeling of magnetic drum separator. Then, the flux density of drum separator is simulated and analyzed via finite element method and the obtained values were compared with experimental data on magnetic drum separator. The results show that the maximum and minimum flux densities on third layer from the drum surface are 3698 and 3244 Gauss, respectively. In addition, it is concluded that the result of experimental tests and simulations are close to each other and maximum difference between experimental and simulation results in third layer is equal to 0.92% which is highly in good agreement. On the other hand, it is shown that 2D magnetostatic analysis could be used for modeling the magnetic field of drum separator. It should be noted that parameters such as length of bar magnet, number × height of bar magnet, poles distance and intermediate material which are used between the poles, are variable. The outputs of the simulation are normal component of magnetic flux density in third and fourth layer and amplitude of magnetic flux density in pole pitch and total prices of magnet elements. By using general full factorial experimental design, the variable parameters are determined based on different level of input parameters. By implementation of ANOVA analysis, the effects of essential factors and interaction on each response are obtained and discussed. Then, the response function of each element is derived by using regression analysis. The results show that using one bar magnet with double height decreases the cost of magnet elements as compared with two bar magnet in each pole. In addition, using magnet element with higher length leads to increase magnetic flux density but the cost of magnet bars increase. Nevertheless, both the using intermediate magnet element instead of air between poles and increasing the poles distance leads to reduce the cost of magnets. It is also founded that using magnet elements with double height, intermediate magnet element between poles and increasing poles distance from 15mm to 32mm decreases the costs of the magnets up to 26% and saving around 23500 USD. Keywords: Magnetic drum separator; Finite element analysis; Maxwell software; Full factorial experimental design, ANOVA analysis; Regression analysis.