Discrete Event Simulation to Improve Aircraft Availability and Maintainability

Discrete Event Simulation to Improve Aircraft Availability andMaintainabilityMassoud Bazargan ● Embry-Riddle Aeronautical University ● Daytona BeachRobert N. McGrath ● Embry-Riddle Aeronautical University ● Daytona BeachKey Words: Simulation, aircraft availability, maintainability, utilizationSUMMARY & CONCLUSIONSThis paper presents a study of maintenance operations at the Daytona Beach, Florida campus of Embry–Riddle Aeronautical University. Embry-Riddle is well-known for its large flight training programs. The Flight Training Department also maintains the school’s aircraft on-site at the Daytona Beach campus. There, overall system availability at the operational level has been a chronic problem. The number of aircraft grounded for maintenance often reaches a quarter of the fleet. To study the situation, discrete-event simulation modeling was used to examine performance measures such as aircraft cycle times and mechanic labor utilization. Also examined was the impact of adding new aircraft to the fleet. As a result, a new working schedule for the allocation of mechanics to various shifts was proposed and accepted by management, resulting in reduced aircraft downtime and improved labor utilization.1. INTRODUCTION“Simulation” refers to a family of computer-based techniques whereby the basic features of a system may be analyzed and simulated (Refs 1-3). Simulation modeling has become a very important tool for complex system analysis and decision-making. It has also been argued that simulation of maintenance functions is superior in approach than more traditional modes of analytical modeling and analysis (Ref 4.) This is because of the complexities of maintenance operations, and the intrinsic uncertainties about parameters that describe arrivals, task sequences and job content, and the availability and deployment of resources.2. THE STUDY2.1 Problem StatementSince 1926 ERAU has offered undergraduate and graduate programs in several areas of aviation operations, engineering, manufacturing, marketing, and management, through its three campuses. The Flight Training Department at the Daytona Beach campus has provided Federal Aviation Administration (FAA) approved flight and ground training necessary for pilot certification. It has also been an FAA-certified Repair Station, awarding the Airframe and Powerplant Certificate, the FAA’s license to perform maintenance on civilian aircraft.The Flight Training Department operated several types of aircraft, as follows:Cessna C172 - 53 aircraftMooney M20J – 8 aircraftPiper PA 28 - 2 aircraftPiper PA 44 - 7 aircraftAll aircraft were subject to a maintenance program that met and sometimes exceeded FAA standards. Due to scheduled maintenance requirements as well as normal levels of unscheduled maintenance, about 20% of the fleet was unavailable for flying on any given day. However, this number could fluctuate widely, causing the Flight Department great difficulties when scheduling and assigning aircraft to instructors and their students. When the operational availability was particularly poor, mission cancellations were not uncommon.Management of the Flight Training Department was interested in exploring the overall efficacy of its maintenance practices. The Business Administration Department was approached to study the problem and make recommendations. One reason for formal investigation was a foreseeable increase in the size of the fleet. By association, other issues were: The efficiency of the extant maintenance labor force. In particular, cycle times (the time fromdispatching an aircraft to the maintenancehangar, until it rejoined the fleet fullyoperational) seemed too high.How much additional labor would be needed to handle the increased number of aircraft.How adversely the current waiting and cycle times would be affected by the increased flow.As is typical in aviation, the Fleet Maintenance Division performed two major types of maintenance, scheduled (preventive) and unscheduled (corrective). Scheduled maintenance was performed after pre-determined numbers of flight hours were accumulated; this was standard practise in general aviation operations. Scheduled maintenance was designed to routinely control and maintain aircraft acording toFAA standards of airworthiness. Unscheduled maintenance 0-7803-7717-6/03/$17.00 © 2003 IEEEwas performed when an aircraft malfunctioned. Historical data at ERAU indicated that 85% (6 aircraft out of every 7) of all maintenance events were unscheduled. This was partly due to the infant mortality problems associated with the rejuventation of the Cessna 172 after a decade of manufacturing moratorium.At the time of this study (Winter 2000) there were 24 mechanics employed by the Maintenance Division. All mechanics were certified to work on all of the types of aircraft noted above. Nineteen full-time mechanics worked 5 days a week in three main shifts while 5 mechanics worked part-time. Figure 1 summarizes the maintenanace operaation. Scheduled maintenace for the C172, called A, B and C checks, was performed after 50, 100 and 600 of flight hours, respectively. Data indicated that 50% of all sheduled maintenance for the C172 was spent on the A check, 41% on the B check, and 9% on the C check. For other aircraft, the scheduled maintenace consisted only of A and B checks, 50% each.Unscheduled maintenance was managed in three categories, based on the amount of time required to fix the problem. Minor unscheduled maintenance took less than two hours, and accounted for 70% of all corrective maintenance events. Medium unscheduled maintenance took between two and four hours, and accounted for 20% of all corrective maintenance events. M ajor unscheduled maintenance took more than 4 hours, and accounted for the remaining 10% of all corrective miantence events.Labor estimates were based on FAA safety regulations, manufacturers’ recommendations, and the experience in the maintenance division. The labor requirements for unscheduled maintenance were assumed to be normally distributed, with means 2, 4 and 6 hours, and standard deviations equal to 10% of their means.Data was used to estimate the average number of aircraft arriving at the maintenance hanger daily. The number of aircraft requiring maintenance was smaller on weekends because the Flight-Training Department scheduled fewer activities then. Utilizing a data-fitting package, the arrival rates mostly followed Poisson and Gamma distributions. As the fleet was relatively new, it was expected that rates would remain relatively stable for the next 3 years.2.2 AnalysisInitial efforts were focused on developing discrete event simulation models to study the maintenance operations, assuming current practices. All simulation models reported here were developed using the ARENA modeling system (Ref. 5) These models were run for 25 replications, each simulating a 15-week training period (i.e., the duration of one semester).The initial analysis showed that on average, of the time that an aircraft was “down,” or in the maintenance cycle, only 20% of the time was value-adding (direct labor,) while the other 80% was non-value-adding (i.e., waiting or indirect labor.) The maintenance cycle was defined as the total time that it took from delivering the aircraft to the maintenance operation until it re-joined the operational fleet. This time was obtained by adding value adding and waiting times.At first, the maintenance division entertained doubts about these results. They noted, for example, that often an aircraft that entered the maintenance cycle on a weekday morning would exit it that afternoon. Therefore, those total (cycle) times looked somewhat excessive to them. These doubts started to disappear when they were shown the scheduled and unscheduled maintenance cycle times for each aircraft during various shifts and weekends.Considering that 80% of the maintenance cycle time was non-value-adding, one should expect to see long lines of aircraft that could be viewed in a traditional sense as being Work In Process (WIP). At any given time, about 20% of the fleet was down for maintenance. Confidence strengthened through efforts to frequently and randomly visit the hangar, count the WIP, and measure cycles times based on individual aircraft log books.A major concern of the maintenance division was the number of mechanics and their utilization rates. Figure 2 presents the average utilization of each mechanic during one semester. In this figure M01-M09 represent the mechanics working in shift 1, M10-M16 mechanics working in shift 2, M17-M19 Mechanics in shift 3, and M20-M24 are part-time mechanics. The average utilization of all mechanics was 59%. Utilization rates were calculated as the times a mechanic was actually performing maintenance (hands-on) divided by the total time the mechanic was available according to his/her schedule. The utilization rates do not include the times that the mechanics were consulting with their supervisors, waiting for spare parts, lack of space in the hangar for aircraft to be maintained, data entry, writing in log books, etc.Data were received with significant interest within the flight-training department. Figure 2 shows uneven utilization rates of mechanics across various shifts. The first 10 mechanics worked in the first shift. One can plainly see a downward trend for the first 7 mechanics. Similarly, a downward trend can be observed for mechanics M11-M17 (during the second shift). These downward trends simply indicate that there were too many mechanics in shifts one and two (all full time mechanics). In fact, the utilization rates of part-time mechanics (M19-M24) were higher than utilization rates of some of the full-time mechanics in shifts 1 and 2. Management’s next request was to determine how best the full time mechanics could be assigned to shifts, so that the cycle times could be improved. On the other hand, the reality of the situation was that many first-shift mechanics would object to being reassigned. The Flight Training department had to offer costly incentives in the form of bonuses and extra pay to persuade the mechanics to change their working days/hours. Various scenarios were analyzed and presented to the flight training management with different number of mechanics in each shift. Based on the expected improvements and the budget constraints, management agreed to the reallocation of only 4 mechanics. Two mechanics from each of shift 1 and 2 were removed and assigned to shift 3. A new working day for shift 3 (Sat- Wed) with 2 mechanics workingin this shift was also proposed.To evaluate the impact of this change on existing maintenance operations, the simulation model was modified to incorporate the noted changes. All the parameters in the revised simulation model such as arrival rates, maintenance times and number of mechanics were left unchanged. The only modification that was made was changing the working schedule of the mechanics according to the new proposed timetable.The value adding times remained the same (the work content did not change.) However, a major improvement in waiting times did result. Average waiting times were reduced by 31%, 34%, 40% and 35% among the four types of aircraft. These reductions in waiting times translated into cycles times shortened by 25%, 27%, 33% and 27% respectively. Naturally, shorter waiting times imply a reduced WIP. A reduction of 28% in WIP was found.Figure 3 shows the average utilization rates of mechanics during a semester. The sequence of the mechanics in this figure is same as Figure 4. For example, mechanics 8 and 9 are presented in the same positions as figure 4, despite the fact that they were now working in shift 3. As expected, this figure implies a higher utilization across the workforce. Furthermore, this rate has become importantly smoothed, mechanic to mechanic.In an effort to accommodate a growing student body and demand for flight training, the Flight Training department intended to increase the size of the fleet. In fact, a major impetus for this decision was the constant high number of grounded aircraft at the maintenance hangar, frequently resulting in training cancellations. The Department was planning to acquire eight new Cessna C172 aircraft, which would impact the flow of aircraft in the maintenance division. The simulation model was used to help forecast the situation. First, several meetings with the management of the flight training and maintenance personnel were held to make sensible assumptions about the increased rate of arrivals. In other words, it was not certain that the scheduled and unscheduled maintenance rates for the new aircraft should be assumed to be the same as for the existing C172 aircraft. It was finally decided that because the existing fleet was young, that the new arrivals would not appear in any very different pattern. Accordingly, an increase of 8 new aircraft translated into an increased arrival rate of 15% (before 53, now 61) for the C172 subfleet. As such, the three performance measures -- cycle time(s,) WIP, and mechanic utilization rates, could be reconsidered.The simulation model was revised to incorporate the new arrival rates for C172s. All other parameters remained unchanged. A 23% reduction in cycle times occurred despite the increased number of aircraft. This time reduction, despite the increased number of aircraft, was seen as a major contribution of this project by the Flight Training Department, since they naturally anticipated longer cycle times.WIP showed a 15% reduction in the number of aircraft waiting for, or in the process of being, maintained.Figure 4 shows the average utilization rate of mechanics under the proposed working schedule. Again, the sequence of mechanics in this Figure is similar to that described earlier. The average utilization among all the mechanics increased to 72%.CONCLUSIONFrom the beginning of the project, the management of the Flight Training Department insisted that the Maintenance Division should be involved and informed of any changes or recommendations and the reasons behind them. The analysis and findings of the simulation models were presented to the mechanics involved in the project. Here, the software’s animation capability proved to be a helpful tool for understanding the outcomes. Mechanics requested various what-if scenarios, and could visually see the impact immediately. Interestingly, personnel seemed to instinctively know that shift 3 was undermanned. But they probably did not know the extent of the impact that such a timetable would make on the overall maintenance operation. The Flight Training Department therefore had little difficulty convincing the mechanics that a timetable re-schedule was necessary for the well-being of the department. They started implementing the proposed timetable in March of 2001.REFERENCES1. M. Pidd, Computer Simulation in Management Science, 4th edition, 1988, John Wiley & Sons, Chichester.2. A.M. Law, W.D. Kelton, Simulation Modeling and Analysis, 3rd edition, 2000, McGraw-Hill.3. W. Byrd, C. Ntuen, E. Park, “Computer-based model for system level reliability and maintainability allocation”, Comput-Ind-Eng, Vol 23 No 1-4, 1992, pp. 177-80.4. J. Crocker, “Effectiveness of maintenance”, Journal of Quality in Mainteancne Engineering, vol. 5 no. 4, 1999, pp.3-7-13.5. Kelton, D., R.P. Sadowski, & D.A. Sadowski ( 1998), Simulation with ArenaBIOGRAPHIESMassoud Bazargan, Ph.D.Department of Business AdministrationEmbry-Riddle Aeronautical University600 S. Clyde Morris Blvd.Daytona Beach, Floridabazargam@Dr. Massoud Bazargan is an Associate Professor in the Business Administration Department of Embry-Riddle’s Daytona Beach Campus. He teaches Production Operations Management, Operations Research, Simulation and project management at both graduate and undergraduate levels. Dr. Bazargan earned his Ph.D. in Manufacturing Engineering from University of New South Wales – Australia, an MA in Operations Research from University of Lancaster – UK, and a B.S. in Mathematics and Economics from University of Manchester – UK.Robert N. McGrath, Ph.D.Department of Business AdministrationEmbry-Riddle Aeronautical University600 S. Clyde Morris Blvd., Daytona Beach, Florida 32114-3900mcgrathr@Dr. Bob McGrath is an Associate Professor in the Department of Business Administration at Embry-Riddle Aeronautical University in Daytona Beach,Florida. Bob began his career by graduating with honors from the UnitedStates Air Force Academy, and serving five years as an Air Force Maintenance Officer. Subsequently he worked for Texas Instruments’ Defense Suppression Division as a logistic support analyst, General ElectricAircraft Engines as a Service Engineer, and for the Lockheed Aeronautical Systems Company as a Materiel Manager. He holds a Master of Arts inPublic Administration degree, a Master of Business Administration degree,and achieved his Ph.D. in Strategic Management and Technology Management from Louisiana State University.Figure 1: Breakdown of Maintenance Operations.Figure 2: Average Mechanic Utilization – Based on Existing Timetable .Figure 3: Average Mechanic Utilization – Based on Proposed Work Schedule.Figure 4: Average Mechanics’ Utilization – With Increased Number of Aircraft Based on Proposed Work Schedule.。