Article Preview
Top1. Introduction
In an aviation world, an aircraft engine in good condition is needed. One way to ensure the good condition of an aircraft engine is through maintenance. Moreover, the maintenance process should be completed on time, because the late completion is highly costly. If an airline does not have high on-time performance (OTP), it will be abandoned by the customer. OTP is a measure of readiness of aircraft, crew, and other equipment to fly passengers to certain destinations. Therefore, to maintain aircraft availability and aviation safety, airlines must perform aircraft maintenance following good maintenance standards. To maintain the aircraft engines, the aviation industry needs an aircraft engine maintenance service from a company called Maintenance, Repair, and Overhaul (MRO) company.
Based on historical data of 30 engine overhaul jobs starting from November 30, 2018 to March 9, 2020, all engines have late completion from 1 day to 182 days. It is initially expected that this MRO company has low OTP since its existing planning and control system is a conventional project management system that considers constant activity times and unlimited resources. In fact, the activity times of engine maintenance are uncertain and cannot be easily estimated as constants. Additionally, each maintenance activity needs some resources (manpower and tools). The current planning system assumes that the available resources are unlimited, which is unrealistic. Therefore, it cannot predict accurate completion times of the engine maintenance jobs and cannot control maintenance jobs to be completed on time.
This paper proposes to handle the uncertain activity times using fuzzy numbers. It is estimated from real activity times as triangular fuzzy numbers (TFNs), including optimistic, most likely, and pessimistic times. The TFNs are selected since they have only three parameters, which is easier to be estimated than trapezoidal fuzzy numbers that have four parameters (Princy & Dhenakaran, 2016). The application of TFNs in this paper is motivated by a research work of Tansakul & Yenradee (2020) that successfully applies TFNs to represent uncertain durations of improvement projects in banking industry. Three scheduling methods are evaluated to know that which one is the most accurate to predict the completion time of engine maintenance. PERT, fuzzy PERT, and fuzzy CPM methods are selected since they are simple, well-known, and use the same input data of TFNs for activity times. A fuzzy resource requirement (FRR) method is newly developed to determine the amount of required resources in each time period from the schedule of maintenance activities. FRR method is proposed in this research because there is a previous study that solves late completion due to limited resource capacity in scheduling problem using similar method (Srizongkhram et al., 2020).
This paper has the following objectives.
- 1.
To estimate triangular fuzzy numbers (TFNs) that represent uncertain durations of maintenance activities based on real historical durations of maintenance activities.
- 2.
To compare scheduling performances among PERT, fuzzy PERT, and fuzzy CPM for selecting a method that accurately predicts the completion time of engine maintenance.
- 3.
To suggest corrective actions that can significantly reduce the late completion.
- 4.
To develop the FRR method to calculate the resource requirement to determine whether the available resources are sufficient to conduct the maintenance activities according to the schedule.
The scope of this paper is to consider only engine overhaul jobs, not minor repair and maintenance jobs since the latter does not face serious problems of late completion. The fuzzy activity times are estimated based on real activity times of 30 engines that are overhauled from November 30, 2018 to March 9, 2020 because data before November 2018 is too old and may not represent the current situation. Note that the real data after March 2020 is not available when this research is started. This paper applies project scheduling methods (PERT, fuzzy PERT, and fuzzy CPM) which are available. It does not intend to develop a new project scheduling method.