Search

In the Industry 4.0 era, Preventive Maintenance (PM) is still an attractive solution to prevent breakdowns and failures and to reduce maintenance and failure costs. A PM program is part of both the Total Productive Maintenance (TPM) philosophy and the Reliability Centered Maintenance (RCM) process. A prerequisite to carry out effective PM activities is the availability of a reliable estimate of the equipment failure rate. Assessing it may be a hard task, as it requires analysing a large set of maintenance data, which includes both quantitative and qualitative variables. To this aim, it is possible to exploit advanced data analysis techniques that permit extracting information and knowledge from big datasets. This paper presents an ensemble-learning model to estimate the failure rate of equipment subject to different operating conditions. At the same time, the method permits to identify the most important working parameters affecting the failure rate. An industrial application is considered to show the potentialities and the effectiveness of the proposed method. In particular, a sample of 143 centrifugal pumps installed in an oil refinery plant is analysed.

The single-machine jobs-planned maintenance tasks joint scheduling problem is faced.The machine is subjected to failures with a Weibull distributed time-to-failure.The optimum number of planned maintenance tasks is computed.A specific metaheuristic based on the harmony search algorithm is presented. This paper focuses on the single machine scheduling problem, with sequence dependent setup times. Both processing and setup times are deterministic and the objective is to minimize total earliness and tardiness penalties. The novelty of the model can be traced in the fact that the single machine is subjected to breakdowns and that, in order to increase its availability, planned maintenance tasks are also performed. Hence, jobs and maintenance tasks are jointly considered to find the optimal schedule. These features make the problem NP-hard and so, a quasi-optimal solution is searched using a recent metaheuristic, which integrates harmony search and genetic algorithms. In order to validate the proposed metaheuristic, a comprehensive set of scheduling problems was fully investigated. Obtained results, compared with those of exhaustive (for small problems) and standard metaheuristics, confirm both the robustness and the speed of the proposed approach.

This study considers the continuous-review ( r, q ) inventory policy with complete backordering, a deterministic and constant lead time, and a Gaussian lead-time demand. A closed-form solution to the first-order conditions of optimality is not available, so this system can only be optimized using an iterative method. However, an iterative optimization approach may still limit the application of the model itself in a practical context with numerous items and the need to frequently recalculate the control variables for the respective inventory policies. In fact, if a closed-form optimal solution is not available, the computational effort required by an iterative method may not be negligible and its implementation can actually be technically difficult. Thus, to overcome the characteristic limitations of an iterative optimization approach, we develop a closed-form near-optimal solution in this study, which is derived via an ad hoc logistic approximation of the standard normal loss function. We then provide an alternative formulation of the expected total cost function. In particular, by introducing a parameter for the number of admissible stockouts per time unit, we can ensure the functional dependence of the service level on the order quantity. This allows us to optimize the order quantity, where the safety stock and the expected shortage are considered by the constraint mentioned above. We present numerical tests to investigate the efficiency of the approximated solution method and to study the sensitivity of the new cost model with respect to some fundamental parameters.