|Title: ||Personnel scheduling for aircraft line maintenance-solution methods for the tactical and operational decision level.|
|Other Titles: ||Personnel scheduling for aircraft line maintenance-solution methods for the tactical and operational decision level.|
|Authors: ||Van den Bergh, Jorne|
|Issue Date: ||24-Feb-2015 |
|Abstract: ||Labor is one of the most significant costs for¨ companies in highly developed countries. In the la st decades, many companies have relocated their pr oduction facilities to low wage countries. Since s ervice needs to beprovided on site, service compan ies, however, cannot escape from the costly wage e xpenses. Therefore, it is very important that over staffing and idleness are avoided by creating good ¨personnel schedules. The personnel scheduling pro blem we will study in this dissertation originates ¨froma line maintenance department of an aircraft¨ maintenance company. A line maintenance program re presents the regular short inspections of aircraft ¨between their arrival and their departure at some ¨specific airport. Whereas the workload that stems ¨from the tasks that need to be done during these¨ inspections is rather stable, the main source of u ncertainty forthese inspections is the timing of t he demand. Delays can originate forinstance from e arlier flights, bad weather conditions, or air tra ffic management. In this dissertation, we propose¨ methods which should facilitate the scheduling in¨ practice, and offer thereby an alternative for the current practice of scheduling by paper and pen. |
Chapter 2provides a classification ¨of the personnel scheduling literature from 2004¨ onwards. The classification fields, such as applie d methods, decision delineation, and uncertainty i ncorporation, guide the reader in his/her search f or specific problem settings and the corresponding ¨approaches of the research community. The current ¨literature mainly addresses the staffing and/or s cheduling of workers considering fixed inputs. We¨ adviseresearchers to integrate multiple decisions¨ into the personnel scheduling problem such as dema nd forecasting, machine scheduling, or considering ¨multiple locations. The last decades, companies m ore and more consideremployee preferences (such as ¨requests for specific working days or shifts, ass ignments to a specific location or working partner , preferred durations or start times) in order to¨ satisfy the workforce and to allow them to flexibl y manage their personal lives. There are still great opportunities in finding algorithms that effici ently cope with those preferences. Lastly, most pa pers appear to feature a deterministic approach, w hile real-world personnel scheduling problems have ¨to deal with a variety of uncertainty sources. In ¨situations where uncertainty has a strong effect¨ on the personnel schedule, such as volatile demand ¨or last-minute changes, it could prove very benef icial to incorporate this uncertainty inthe decisi on-making process. Instead of integrating this unc ertainty, researchers could also test the robustne ss of their solutions, for instance by simulating¨ the stochastic behavior of demand, or worker avail abilities. In the literature, however, we hardly e ver encountered this type of analysis.
In Chapter 3 the different types of ai rcraft maintenance that appear in literature are d iscussed. We found the terminology consisting of m any overlapping definitions very confusing. To obt ain a better insight into the different types of a ircraft maintenance, we provide a taxonomy that cl assifies all types and we discuss the literature t hat addresses the combination of aircraft maintena nce and personnel scheduling. In the second sectio n of this chapter we provide the details of a give n personnel scheduling problem of an aircraft main tenancecompany. In this problem, personnel rosters ¨need to be created while minimizing the labor cos ts that stem from the provision of a line maintena nce program. This problem setting is the basis for ¨the problems studied in Chapters 4, 5 and 6.
Chapter 4 is dedicated to a three-ste p evaluation procedure for personnel rosters. Due¨ to the usage of heuristics and multi-objective per formance criteria, solution procedures typically g enerate multiple solutions, or in this case, perso nnel rosters. For the management, it is difficult¨ to create a ranking among these solutions. Therefo re, we present an approach in which the rosters, w hich arecreated based on a deterministic problem s etting, are used in a stochastic environment. This ¨results in performance criteria, such as service¨ rate and tardiness values, that are evaluated by means of a Data Envelopment Analysis (DEA). This ap proach should enable that management decisions are ¨no longer based on instinct but on objective crit eria.
In Chapter 5, we study two¨ variants of the resource loading problem (RLP), in ¨which a number of jobs with time windows need to¨ be scheduled with monotone execution patterns. Thi s latter characteristic ensures that a given job c an only be assigned a non-decreasing (resp. non-in creasing)number of workers over the periods in whi ch the job is processed. We distinguish betwe en RLP-selection in which no non-regular capacity¨ isavailable and selected jobs yield a reward, and¨ RLP-capacity in which all jobs need to be executed ¨and non-regular capacity is minimized. The result ing integer programming (IP) formulation for real- life problem instances tends to be too hard in ord er to be solved in limited time with a commercial¨ solver. Therefore, we carry out a Dantzig-Wolfe de compositionand try to solve the resulting problem¨ with column generation. Unfortunately, these two C G methods do not give good results. The CG method¨ thatuses execution schemes (i.e., columns which st ate in which period and with which intensity a job ¨should be executed) has the advantage that it is¨ the only solution method which is able to find res ults for the large instances (with 300 flights).
Finally, Chapter 6 follows w ith the presentation of a proactive/reactive polic y to deal with the maintenance job scheduling prob lems on the operational level which areinfluenced¨ by uncertainty in the arrival times of the jobs. T he proactive policy consists of creating a personn el roster and a baseline schedule which take into¨ account the delay distributions of the jobs and sc hedules each job as late as possible. In the react ive policy, we use a rolling horizon algorithm whi ch makes use of the most recent information withre spect to the arrival times of the jobs. The result s indicate that thequality of information is highl y important. Better estimations on the real releas e times of the jobs (i.e., the arrival delays of f light) and the earlier these are retrieved, the cheaper the reactive scheduling policy. Also the job ¨characteristics play an important role with respe ct to the costs which have to be made to create fe asible schedules.
|Publication status: ||published|
|KU Leuven publication type: ||TH|
|Appears in Collections:||Faculty of Business and Economics, Campus Kulak Kortrijk – miscellaneous |
Faculty of Economics and Business, Leuven - miscellaneous
Research Center for Operations Management, Leuven
Research Centre for Quantitative Business Processes, Campus Brussels (-)
Faculty of Economics and Business (FEB) - miscellaneous