The discrete element method (DEM) is the dominant numerical approach for simulating particles and granular materials. The technology readiness level of the method has now reached a high level in numerous industries ranging from coarse materials in e.g. the mining industry to powders in pharmaceutical, food and agricultural applications. Depending on the application area, different aspects are of key importance to reach valuable output from simulations. In some cases, the particles shape representation is paramount to success, and in others the sheer number of particles or the implementation of a specific advanced contact model will be the critical point. A frequently overlooked perspective, independent of the mentioned critical points, is how to actually work with and apply DEM in industrial and R&D contexts to reach sound solutions and industrial value. In this paper a methodology is presented on how to perform the material model calibration and engineering design process efficiently. A general rule for all problem solving is to avoid trial and error iterations. When using DEM this is critical as the computational costs are normally high and the challenge is to systematically extract the maximum information possible from a limited number of simulations. The first step in the proposed framework is to apply design of experiment (DOE) variable screening and analysis of variance (ANOVA) to screen out which variables that have a significant influence on the measured response. The second step is to apply response surface design matrices in order to build a robust surrogate model. The surrogate model should further on be statistically validated to find if the design matrix needs augmentation with more data points. When a robust surrogate model has been developed, a suitable optimization algorithm may be applied to find the optimum that represents a calibrated material model or an optimal design. The methodology is exemplified by the calibration procedure of two different granular materials (an XLPE polymer pellet and a granite rock) and the corresponding engineering design process for the rock material application. The aggregate application is the problem of optimal feeding of cone crushers. The DEM simulations are performed in an in-house developed state-of-the-art DEM code optimized for GPU parallelization. The results show that variable setting accessibility for simulation batch processing and automated post processing are critical in order to efficiently run numerous simulations with input from DOE design tables. The case study demonstrate how DEM can be applied using the proposed methodology to achieve significant performance improvements. The numerical predictions show the advantages of the method and how it through a systemic approach facilitates communication between the simulation engineer and the process engineer receiving the simulation output.