یک روش برنامه ریزی عملیات تولید مبتنی بر قابلیت اطمینان برای اجزای یک سیستم پیچیده مکاترونیکی

Uncertainties in the values of the parameters of a system can originate from the manufacturing tolerances of the system components, which can produce a degree of unreliability in the performance of the system. A systematic framework for realistic reliability assessment of an electro-hydraulic servo system has been presented in this paper with the objective of providing adequate information for the selection of the best manufacturing process for each of the servo valve components. Monte Carlo simulation has been employed to evaluate the effect of these uncertainties of the servo valve parameters on the statistical performance of the system. Possible manufacturing processes have been introduced for each component and the justifiability of using each one has been discussed based on the estimated reliability of the system.

Selection of a suitable manufacturing process for a component is not generally a straightforward task [1] and [2]. Many factors and criteria have to be considered at the design stage, such as the dimensions and the general shape of the component, the material to be processed, the number of components that are going to be produced, costs and accessibility of the processes, and the allowed ranges for dimensional tolerances. For the fabrication of some components, several manufacturing processes are often used. The procedure of selecting the most suitable manufacturing process for a component is hence a trade-off among all these factors. With the proliferation of CAD–CAM tools, significant methods and software tools have been developed to simplify this selection process. Boothroyd et al. [3] were among the first researchers who addressed the integrated material and manufacturing process selection problem. They performed the process selection using production rules and pattern matching. This approach is quite effective; however, it does not either have the capability of weighting of the criteria or making comparisons among the alternative processes. Farris and Knight [4] and [5] developed a new method by mapping the component geometry and material requirements onto sequences of manufacturing processes and suitable materials. In this method, the geometrical and material constraints of the component are the inputs to the system. The output of the system is a list of practical manufacturing processes and materials appropriate for the component. Based on the difficulty of the process sequences and the agreement of the material specifications with the design requirements, the processes and materials are ranked. Dixon and Poli [6] developed a new method which used a guided iterative searching procedure for the evaluation of material and manufacturing processes. This was a handbook approach in which tables and charts were used for the evaluation and comparison of processes and materials. Kunchithapatham [7] developed a material-process advisory system which included 42 materials and 17 processes in its database. The system included three databases: materials database, process database and material-process compatibility database. This system was not able to consider the shape of the component for the selection of the materials and processes. Giachetti [8] introduced a new material and process selection system called MAMPS that integrates a formal multi-attribute decision model with a relational database. This system produces a ranked set of compatible material and manufacturing process options. In addition to the reviewed literature, some other research on material and process selection has been performed, which can be found in [9], [10] and [11]. The reviewed techniques can be helpful for the selection of a fabrication process for a single component. However, none of them considers the interaction of components with each other in a complex system. Consider that we have a complex system like a mechatronic device which consists of a set of different mechanical and electrical components. In such a system, the components interact with each other and a small change in the parameters of one component can change the performance of the system completely. In view of such dynamic interactions, tools for modeling, analysis, and design of a complex mixed system (e.g., a mechatronic system) need to have an integrated and system-based approach [12] and [13]. In particular, the reliability of a complex system can be significantly influenced by the selection of the manufacturing process of its components. This important process planning criterion has been overlooked in prior work, particularly for complex mechatronic systems. Since the fabrication process influences the final value of the parameters of each component, a suitable fabrication process cannot be selected without taking the interaction of all system components into account. The main objective of this research is to introduce an advisory method for evaluation and comparison of the manufacturing processes available for components of complex systems. In this study, a two-stage hydraulic servo valve has been used for the demonstration of the proposed procedure; however, it should be noted that the underlying concepts of the developed method are general and can be used for any other complex system. The servo valve is the most critical part of an electro-hydraulic servo system, which consists of very precise and sensitive components. A small change in the dimensions, metallurgical characteristics, or other parameters of these components can produce instability and error in the performance of the system. In the design phase, the values of these parameters are designated as crisp and certain numbers. Nevertheless, due to manufacturing limitations, the final values of these parameters always contain some tolerances and uncertainties. These uncertainties in the system parameters create a variance in the performance of the system from the ideal circumstances. In practice, no two produced valves would be exactly identical. It is possible that among the tens of thousands of produced valves, some cannot satisfy the highly restricted specifications to pass the quality control. The achievable range and the other statistical measures for a particular parameter value of a component directly depend on the manufacturing process selected for its fabrication. The number of unqualified valves produced could possibly represent a suitable criterion for comparing the available fabrication methods for valve components. However, in practice, quality control of all products is not possible. Therefore, simulation tools along with statistical techniques are proposed here to estimate this number. In the proposed method, the number of unqualified valves after production is predicted using the probability of failure estimation or reliability techniques. One of the most common techniques of estimating the probability of failure is Monte Carlo simulation (MCS). In this method, every parameter that has an inherent uncertainty is defined by a probability distribution. By using a random number generator, for each simulation, a set of samples is selected from the probability distributions and used for the parameters of the model. Simulations are then performed with those parameter values. Depending upon the number of uncertainties and the ranges specified for them, a Monte Carlo simulation could involve thousands or tens of thousands of recalculations before it is complete. The probability of failure can then be calculated from the results of simulations [14]. MCS is simple to implement, robust, and accurate with sufficiently large samples. The remaining parts of this paper are organized as follows. Section 2 discusses the modeling of the electro-hydraulic servo system. In the third section, possible manufacturing methods for servo valve components are proposed. Based on the achievable tolerances for dimensions and other characteristics of the servo valve parts, the uncertainty of the related parameters is estimated. In Section 4, these manufacturing originated uncertainties and parameter variations are applied to the model and the statistical performance and the reliability of the system under various circumstances are estimated using MCS. In the final section, based on the acquired results, the suitability of using each process is discussed.

In this paper, a novel advisory method was proposed for evaluation and comparison of the manufacturing processes available for components of a complex system. A two-stage hydraulic servo valve was used as an example for the demonstration of the proposed procedure. In order to compare different methods of manufacturing, the influence of manufacturing originated tolerances of servo valve parameters on its statistical performance was investigated and the number of failures of the electro-hydraulic system was estimated and compared in each case. For detecting a failure condition, two criteria recommended in SAE ARP490 standard were taken into account, including the frequency response of the system and the internal leakage of the servo valve. The result of simulations showed that the most important parameters in the frequency response of the system are the length of the air gaps in the torque motor, xoxo, and the diameter of the flapper nozzles, dfdf. By changing the manufacturing process of the valve components which affect the value of these two parameters and improving their normal distributions, the number of system failures was decreased significantly. On the other hand, by changing the manufacturing processes of some parts such as spool and sleeve and improving their dimensional precision, no significant improvement was observed in the dynamic performance of the system. The simulations also revealed that the width of spool edges and sleeve slots play a principal role in the internal leakage of the servo valve. By changing the machining technique and improving the dimensional tolerances of these two items, the number of system failures drops significantly. It was also found that the clearance between the spool and the sleeve is not a dominant factor in the leakage of the valve. The servo valve example showed how important it is to consider the interaction of components of a complex system for selecting suitable manufacturing processes. In some cases, by changing the manufacturing processes of two components separately, no significant change is observed in the statistical performance of the system; however, if those changes are made together, the statistical performance may improve significantly. On the other hand, it is possible that this effect becomes neutral by a simultaneous change in the manufacturing processes of some components. Hence, it is necessary to consider the interaction of all components with a structured method and include all possibilities in the analyses in order to be able to select the most suitable manufacturing processes for all the system components.