© Faculty of Mechanical Engineering, Belgrade. All rights reserved FME Transactions (2012) 40, 153-164 153 Eugeniusz Rusiński Professor Wroclaw University of Technology, Poland Faculty of Mechanical Engineering Jerzy Czmochowski Associate Professor Wroclaw University of Technology, Poland Faculty of Mechanical Engineering Damian Pietrusiak Research Assistant Wroclaw University of Technology, Poland Faculty of Mechanical Engineering Selected Problems in Designing and Constructing Surface Mining Machinery Contemporary machine design not only needs to meet the applicable requirements, norms and regulations but also requires a novel approach to the designing process. The article presents the essence and role of experimental and numerical analyses, which are becoming increasingly popular in the designing process and in the testing of dynamically loaded structures. What is particularly important in machine modeling is the accuracy of representation and thus the complexity of the model. In order to fully and accurately represent the reality it is necessary to verify and update the models based on experimental tests. Keywords: surface mining, bucket wheel excavator, FEM, experimental techniques 1. INTRODUCTION Contemporary designs not only have to meet the requirements and standards for surface mining machinery but also should include the results of frequency tests of free and forced vibrations for individual elements and sections. Owing to the rapidly increasing calculating power of computers it is now possible to solve very complex numerical problems. However, it is also important not to overcomplicate the task if it is possible to obtain sufficiently accurate results using a simpler model. In order to make such a decision one needs expert knowledge of the tested machine. It is important to anticipate the range of potential free vibration frequencies which can influence the use of the machine, and to understand where and how simplifications can be introduced. The discreet models of steel structures used today are built using either beam elements or shell elements. Using beam elements is not overly complicated and makes it possible to represent the stiffness of the structure. The drawback of this method is that it does not take into consideration the joint stiffness of the superstructure. Most often it is much higher than the stiffness of the beam itself. Neither of these modeling methods includes the influence of the type of joints on the stiffness of the superstructure. In both cases it is crucial that the mass distribution corresponds to the distribution in the stability proof or the one based on machine tests. When analyzing the machine's vibration it is also important to include stress of elements such as elastic and steel lines. The stress level can influence the stiffness change and thus the frequency of vibrations. It is possible to update the numerical models so that they most accurately represent the studied machine. This is initially done already at the stage of mass distribution and introduction of load to the elements of the machine. If there are still large discrepancies with respect to the values obtained during tests, the updating of mass and stress should be verified and repeated. Due to the use of the said simplifications, the updating process may be complicated. It is also possible to change the values of physical parameters such as Young's modulus, stiffness or mass density. Another approach is updating based on purely mathematical methods. Mathematical algorithms directly modify the values of stiffness and mass matrices [1]. Yet such a change in parameters does not provide any physical interpretation. By applying the above methods of updating it is possible to obtain a model which precisely represents the values measured on the machine. However, it is important to realize that a model created in this manner may differ from the real machine with respect to other parameters and in fact does not provide an accurate representation of the real structure. Models built for the purpose of calculating ultimate and fatigue strength require a more accurate discretization than those built for the purpose of numerical modal analysis. Before a model is created it is important to know what purpose it will serve. Although beam and shell elements are used, which reduces the size of the model, the number of nodes often exceeds 300 000. This may be less problematic in the case of static calculations but hardware and software requirements for eigenvalues solver algorithms are much higher. However, numerical calculations are based on experimental tests on real machines. It is therefore necessary to verify the assumptions of numerical models. Such verification can justify the chosen calculation method and, most importantly, the boundary conditions for the model. Only after such validation the models may be used by experienced users to only perform numerical analyses in design, to assess the tension level or to perform modal analysis. Received: June 2012, Accepted: November 2012 Correspondence to: MSc Damian Pietrusiak Wroclaw University of Technology, Lukasiewicza 7/9, 50-371 Wroclaw, Poland E-mail: damian.pietrusiak@pwr.wroc.pl