Mixed Integer Least Squares Optimization for Flight and Maintenance Planning of Mission Aircraft George Kozanidis,! Andreas Gavra .? Eftychia Kostarelout | systems Optimization Laboratory, Department of Mechanical Engineering, University of Thessaly, Volos, Greece 2111 Combat Wing, Hellenic Air Force, N. Aghialos, Greece Received 14 September 2010; accepted 25 January 2012 DOI 10.1002/nav.21483, Published online 2 March 2012 in Wiley Online Library (wileyontinelibrary.com), Abstracts We address the problem of generating a joint flight and maintenance plan for a unit of mission ater. The objective is to establish a balanced allocation ofthe fight load and the maintenance capacity to the individual airraft ofthe unit, so that its Tong-term availabilty is Kept ata high and steady level. We propose a mixed integer nonlinear model to formulate the problem, the objective function of which minimizes a least squares index expressing the total deviation of the individual aircraft tight and ‘maintenance times from their comresponding target values, Using the models special structure and properties, we develop an exsct seach algorithm for its solution, We analyze the computational complexity of ths algoritam, and we preset computational results comparing its performance against that of a commercial optimization package. Besides demonstrating the superiority of the pro- posed algorithm, these results eveal that the total computational effrt required forthe solution ofthe problem depends mainl¥ on "wo crucial parameters: the sizeof the unit (ie. the numberof sirraft that comprise it) and the space capacity ofthe maintenance station, © 2012 Wiley Psiodias, ne. Naval Rescatc Logistics 5% 212-229, 2012 Keywords: fleet availabilty: flight and maintenance planning: mission srcrat: mixed integer nonlinear prog squares optimization 1. INTRODUCTION In this article, we address the problem of deciding which available aireraft to fly and for how Tong, and which grounded aircraft to perform maintenance operations on, in & group of aireraft that comprise a unit. The objective is to generate a flight and maintenance plan for each individual aircraft, so that the long-term availability ofthe unit is kept ata high and steady level. The motivation for dealing with this problem originates in a practical application drawn from the operation ‘of a typical combat wing of the Hellenic Air Force (HAP). ‘Our interest in the problem was further stimulated by the fact that it also arises in the everyday operation of several types of units, such as fleets of fire-fighting airerat, rescue choppers, and so forth ‘The HAF is primarily responsible for Greece's national air defense, It is split into four divisions: Tactical Air Force, ‘Air Support, Air Training, and a division responsible for other units and services. All units involved in air operations and missions belong to the Division of Tactical Air Force. Correspondence to: George Kozanidis (gko7@mie.uth gt) (© 2012 Wiley Periodicals, ne Therefore, the combat wings, which are encountered fur- ther down the organizational structure of the HAF, belong to the Division of Tactical Air Force, too, To retain a high level of readiness, the areratof each wing must collectively fly a given total Hight time in each time period (typically. fone month). This time is als refered to as the related military literature [15]. The target value of the flight load is issued by the wing command, and only small deviations from it can be tolerate. The “residual flight timo” of each individual airera is defined as the total remaining time that this aireaft ean fly until it has to undergo a maintenance check. This time is also referred to as “bank time” in the related military litera- ture [15]. The residual fight time of an arcratis positive i and only ifthisaircratis available to fy. Similarly, the “resid- ual maintenance time" of each individual aircraft is defined asthe total remaining time that ths aireraft needs to complete its maintenance service. The residual maintenance time of sn aieraft is positive if and only if this aircraft is undergoing a ‘maintenance check (and is therefore not available to fy) For providing maintenance service to the aircraft of the unit, there exists a station with certain time and space (alsoKozanidis, Gavranis, and Kostaelou: Flight and Maintenance Planning of Mission Aircraft 213 referred to as “dock space” in the related military literature) capacity capabilities. Given the fight requirements of the ‘wing. and the physical constraints that stem from the capac- ity ofthe maintenance station, the objective is to issue a fight and maintenance plan for each individual aircraft, so that the unit fleet availability is kept ata high and steady level. In this article, we develop a mixed integer nonlinear pro- ‘gramming model for this problem and an exact search algo- rithm for obtaining its optimal solution. Our mathematical formulation modifies properly an existing graphical heuris- tic tool for addressing the problem. The remainder of this article is structured as follows. In Section 2, we review the literature related to the problem, and in Section 3, we present the mathematical model that we developed for its, formulation, In Section 4, we develop the exact solution algorithm, and in Section 5, we analyze its computational complexity and we present results illustrating its computa- tional performance. Finally, in Section 6, we summarize the conclusions of this article and we suggest possible future extensions. 2, BACKGROUND INFORMATION Although the problem that we address in this article fs an important decision making problem encountered in several diversified areas, the relevant published research is rather limited. Steiner [14] presents a methodology for preven- tive aieraft maintenance scheduling in the Swiss Air Force, Which considers both operational constraints and! mainte nance requirements. This methodology creates an initial ma ter plan by combining maintenance actions of different types (levels), and issues a final maintenance schedule through a heuristic optimization procedure. Safaei et al. [12] develop a mixed integer optimization model to formulate the problem of workForce-constraned maintenance scheduling fora flet of military aircraft and use generic optimization software to solve it. The model utilizes a network flow structure to sim- ‘ulate the flow of aircraft between missions, the hangar. and the repair shop, Sgaslik[13] introduces a decision support system formain- tenance planning and mission assignment of ahelicopter leet. which partitions the master problem into two subproblems. ‘The author develops two elastic mixed integer programs to formulate these two subproiblems and solves them separately using standard optimization software. The frst subproblem is called the yearly planning model (YPM); it provides the maintenance schedule and flight hour distribution of the helicopters to assign them to inspections and exercises. The second subproblem is called the short-term planning model (STPM); it takes as input the maintenance schedule pro- duced by the YPM and returns the mission assignment of the helicopters. The YPM minimizes the cost associated with violating some of the problem’s constrains (e., those referring to the required flight load, the maintenance capacity. and the flight time of each individual aircraft, for a given lower bound on fleet availability. Onthe otherhand, the model that we develop inthis article does notallow any violation of these constraints. ‘The STPM goes one step further by assigning helicopters 10 specific missions, based on their technical characteristies and. the requirements of each mission. In the present mod the other hand, we express the flight load solely in terms of the total flight time that must be flown collectively by the aircraft of the wing. Pippin [11] adopts an existing heuristic methodology that utilizesa so-called “aircraft flowchart” graphical tool [15] and develops a mixed integer linear programming and a quadratic programming formulation to model the light hour allocation problem. Both models try to establish a flight time allocation that ensures a steady-state sequence of aireraft into phase maintenance. The first model minimizes the cost associated With the deviations of the individual aireraft residual fight times from their diagonal line target values but dees not incor- porate the apparent difficulties introduced by the maintenance. aspect of the problem. The second model is based on a qua 0 (9) The objective function (1) minimizes the total deviation ndex that will be realized atthe beginning of the next time period, that is, the sum of squares of the deviations of the residual fight times ofthe available aireraft and the residual ‘maintenance times ofthe grounded airertt from ther core sponding target values on the associated flowcharts. To see ‘why this is true, note that this objective comprises two sum- mations. The frst one refers to the aircraft that are avaiable atthe beginning of the current period. Consider a particular of these aircraft with index i. Iz, <, then ths aieraft will wailable in the next time period, too. As the aireraft ‘with the smallest residual flight times will enter the main- tenance station for service, the index of this aireraft on the flowehart of available aireraftat the beginning of the next time period will become equal toi ~ z,. Additionally, as <, aireraft will enter and zg aircraft will exit the maintenance Station the new diagonal slope of this Howat will become equal to ‘Thus, the corresponding target value ofthe residual fight ime ofthis irra the beginning ofthe next time period will be equal to ( — On the other hand, if zp > 4, then this irrat wil join the set of grounded aircraft atthe beginning of the next time period with residual maintenance time equal to G, assuming an index j instead of j. Considering that airerft will exit the mainienance station at the beginning of the next period and that this willbe the ith in order aircraft to enter the main- tenance station, the index ofthis aircraft on the flowchart of rounded aireraft tthe beginning ofthe next time period will ‘become equal to (NA +i — za)- Moreover, the new diagonal lope of this flowchart will become equal to yq-2—. Thus the corresponding target value of the residual maintenance ‘ime ofthis aiterft atthe besinning of the next time period will be equal to (NA+ — The terms (1 —b,) and b; ae binary indicators denoting Whether aircraft / will be grounded or not in the next time period. More specifically, if by = 0, then aircraft will retain its availability in the next periods therefore, the square of its [Naval Research Logistics DOT 10.1002/nav residual flight time deviation on the lowchart of available air- craft will be taken into account in the objective function. On the other hand, if; = 1, then aircraft will be grounded inthe next period and the square of its residual maintenance time deviation on the flowchart of grounded aircraft will be taken nto. account in the objective function instead. The second ‘summation of the objective function pertains, in an identical way, fo the aireraft that are grounded atthe beginning of the current period. Constraint set (2) updates the residual flight time of each available areraft based on its initial status and the flight time ‘that will be assigned to it in the current time period. Similarly constraint set (3) updates the residual maintenance time of ceach grounded aircraft based on its initial status and the main- tenance time that it will receive in the current time period. Constraints (4) and (5) ensure that the order of available a= craft and the order of grounded aircraft, respectively, will be preserved. Constraints (6) and (7) compute the number of aireraft that wil enter and exit, respectively, the maintenance station at the beginning of the next time period, based on the values of binary variables b, and c, Constraint (8) ensures that the space capacity of the main- tenance station will not be violated, whereas constraint (9) ‘ensures thatthe total maintenance time that will be provided, by the station will be equal either to its total time eapac: ity or to the total maintenance requirements ofthe grounded aircraft, whichever of these two quantities is smaller. This, constraint serves to ensure both thatthe time capacity of the maintenance station will not be violated and that no part of this capacity will remain unused, whenever additional main- tenance requirements exist. Constraint set (10) ensures that the required flight load will be satisfied, within a tolerance defined by variables Land U. Por example, when L = 0.95 and U = 1,05, 1 maximum deviation of 5% from the target ‘value of the flight load can be tolerated Constraint set (11) imposes a lower bound equal t0 Yain fon the residual flight time of each available aircraft that ‘will remain available in the next period, too. This bound is, imposed when 4, = 0; otherwise, the comesponding con- straint becomes redundant, as by = | implies that this aircraft ‘will be grounded in the next period. This modeling technique prevents an available aircraft from ending up with positive but negligible residual flight time at the end of the current time period. Similarly, constraint set (12) imposes a lower bound of Guia on gj¢ When ¢; = O and becomes redundant when ¢; = I. This prevents a grounded aircraft from ending. up with positive but negligible residual maintenance time at the end of the current time period, We introduce constraints (11) and (12), because itis odd ‘and unrealistic to have a grounded (available) aircraft whose residual maintenance (fight) time is positive but arbitrarily small. This does not imply that a grounded aireraft whoseKozanidis, Gavranis, and Kostaelou: Flight and Maintenance Planning of Mission Aircraft 217 residual maintenance time drops below G ican be declared ft for fight and released from the maintenance station, because that would violate the safety standards. If an aircraft could indeed finish its service in atime frame that i tilly Sina than G, this would signify that the actual service dur tion is in fact less than Gand that G has been erroneously used. Moreover if we allowed the solver to release an areraft from the maintenance station before tis area finishes is service, then the solver would take advantage of this when- veri was preferable and would apply this smaller servi time instead of G.timprove the quality ofthe returned sol tion Similarly, imposing the lower bound Yaa on the residual Aight ime of each avilable airraft doesnot imply that an available ireraft can be grounded, a8 soon as is residual Aight time drops below Yn, because this would increase the average maintenance cost per hour flown, decreasing in this ‘way the long-term efficencyof the uni Constraint set (13) states that the residual Might time of an available aircraft must aopt00, befor this icra enters the migintenance station for service. This constrain is redundant when by = 0 and forces sgt O-alue when by = 1, asi implies, oper with consrint (2), that il be egal 10 pipe Similarly, consiraint set (Id) states that the resi ual maintenance ime of a grounded aircraft must drop to 0, before this srraft exits the station and becomes avail able. This constraint is redundant when cj =O and forces fn 10 Orvale when cj — I, a8 it impli, together with Constant (3), that will be equal t gjp. Constraint st (15) imposes an upper bound on the Bight time of each available airrat, This restition is usualy present due to technical reasons. Finally, constraints (16) (17) and (18) (19) are the non-negatvity andthe integrity constraint respectively. There exists a rae special casein which constraints (9), (12) and (14) cannot be satisfied simultaneously. This hap- pens when both B < °M, ¢,, and, at the same time, uuizing B fully by forsing 39% to be esual to B (as required by constraint (9) inevitably ests in one oF more rounded sitealt with residual maintenance time positive butstrcly smaller han Ga athe end othe curent period For example, if Sip = 2p = Guin B = 005, then the problem is infeasible. This is because there is no feasible way to utilize B fully without violating the restriction that the residual maintenance time of every grounded sreraft must be greater or equal 10 Gin Tn practice typical way to aveid leaving an sreaft with very small but postive residual maintenance time a the end ofthe euret time periods by forcing the station to provide the ext Gime that this aireraft needs to finish ts service ‘This implies thatthe time capacity ofthe station will be exogenotsly increased, OF course, if this extra time is neg- lgible, this increase will also be negligible. In the small numerical example introduced above, for example, setting B = 0.1 would make the problem feasible with no significant clfect. From a mathematical point of view, on the other hand, 4 suitable modification of the problem formulation can be adopted to model this, as described next. Letr be an auxiliary binary decision variable that takes the value 1 if constraint (9) is relaxed and 0 otherwise. If B-< Y2*4, ej», substitute consrtnt (9) with the following two constraint Dh so+re 20) Bed, ap where Fis a sufficiently large number, introduce the folo- ing constraint FU) + Shy = Dep — (NA = za)Gnin, (22) fat and add the term Fr tothe objective function, to ensure that Variable 1 will always be set to 0 whenever a feasible solution to the original formulation exists. The insight ofthis mod- ification is the following. If B < 3%, gj» and a feasible solution such that "4, = B exists, then willbe set 100 and thisequality will be denoted by constraints (20) and (21), ‘whereas constraint (22) will become redundant. Otherwise, + willbe setto 1, constraint (20) will become redundant, and constraint (22) wil set the total maintenance time that will be provided by the station equal t0 32", gj — (NA ~ 2a) Gri (ead of B. This is tru. though constraint (22) is expressed 4 inequality instead of equality, as no Feasible solution such that "4, fi; > D8, gig — (NA — 24) Gain Will exist. Thus, it will be acceptable to Slightly inerease the station's time capacity inthat case, under the restriction that every grounded aireratt that will not finish its service will end up with resid ual maintenance time equal to Gis atthe end of the current time period. The user should keep in mind that if 1 = 1 the correct value ofthe total deviation index results after the term Fr is subtracted from the optimal objective function value, When this modification in the problem formulation is adopted, the problem may be satisfied for several values of variable ca. To ensure that the slight increase on the time capacity of the maintenance station willbe the minimum pos- sible, the user i advised to select the solution with the mini ‘mum feasible value for variable <9. It seems logical that this ull be the most desirable solution, because it will result inthe smallest violation ofthe original constraint 2" hj = B. In {he small numerical example introduced above, for example feasible solution can be obtained by setting "M4, hy = 0.1 Naval Research Logistics DOT 10.1002/nawais Naval Research Logistics, Vol. 59 (2012) and z_ = 1, orby seting Sh, = 02 and zy = 2. The former out of these wo solufons seems preferable, because it results in the smallest violation ofthe original constraint ye hy = 0.08. ‘similar situation can arise when the satisfaction of the Aigh load inevitably results in one or more wvalable aireralt withresidvl ight time strictly smaller than Fig athe end of the curent period For example, i A = I. ip $= 005, L = 09, and U = 1.1 then the problem does not have a feasible solution. A modeling technique similar to the one presented shove canbe adopted in this case, to. Nevertheless, these two are rare extreme cases with prac cally negligible effec therefore, toavoid complicating things unnecessarily, in what follows we assume thatthe prob- tem defined by (1)-(19) i always feasible, and we suppress constants (20)-22) The progressively worse deviation index forthe subsequent aircraft thal ext or enter the station besides te first one, and the fact thatthe model does not allow an aierft to act both as available and as grounded within the same time period are main consequences of the planning horizon’ dsertzaion Which doesnot allow an ateraft to both ly and be serviced within the same time perio. The minimization ofthe units total deviation index in subsequent time periods will bal ance the allocation ofthe flight and ma smoothen things ott In practice, the ight load requirements refer to I-month Lime periods, the time capacity ofthe maintenance station s ‘eastred in monthly labor hours and the fighviaintenance Plans ofthe unit are typically reviewed and updated tthe beginning of each month, Tis setting not only hints toward the development of a discrete model but also implies that it is reasonable to choose an optimization period of | month, though it might seem preferable to make this choice in such a way that only one alrraft enters the station and only one aireraf exits the station at exch ime period This is nota very str limitation forthe present mode, because the fight load issued by the unit command is pra tically small on the one hand, and the Toad factor of the ‘maintenance station is practically equal o-1 on the other hand. In turn, the fight load ofthe nit can be satisfied using only some of the unit’ aircraft, whereas the time capaci ofthe maintenance station is always fully wilized, Hence, in terms ofthe units long-term aggregate fleet availability itis hardly everan ase, the gh time of one or more area that are it for fight is forced to zero value fr an extra time period. Similarly, delaying the begining ofthe maintenance craft for an extra time period is no eiical ether, because other grounded areraft that are present can always utilize the station's eapaciy fully. If iis Shsoliely necessary, the consequences of the progressively ‘worse deviation index forthe subsequent (besides the fist fone) areraft that change stats can be lessened by only mienance times and service of one oF more ai [Naval Research Logistics DOT 10.1002/nav including in the computation of the total deviation index the first aircraft that will become grounded and the fist aircraft that will become available (i... by assuming a devi- ation of 0 for every subsequent aircraft that will change status). ‘The formulation that we propose adopts the aircraft flow chart methodology both for the available and for the grounded aircraft is novelty lies in that it gives the user additional flex- ibility by allowing the slopes of the two diagonals to vary based on the actual number of aircraft that enter and exit the maintenance station, This seems far more rational than these numbers in advance and generating the flight and ‘maintenance plans afterward, based on the resulting diagonal slopes. We can essentially view this formulation as the composi- tion of two individual ones, one that pertains to the available and one that pertains to the grounded aircraft. The main dif- ference between them s that small deviation, determined by variables L and U., can be tolerated forthe satisfaction of the flight load by the available airerat, whereas no such tolerance exists for the total maintenance time that will be provided. by the maintenance station. Additionally, an upper bound is imposed on the maximum flight time of each available air- craft, whereas no such bound is imposed on the maintenance time of each grounded aircraft, Acirst glance, the problem formulation seems too restriet- ing, as it imposes a steady rotation of the aireraft in and ‘out of the maintenance station in nondecreasing order of theirresidualflight/maintenance times, without allowing any contravention of this order. For example, according to this, formulation, an aircraft that i available in the current period cannot be grounded at the beginning of the next one, unless all the available aircraft with smaller residual flight times are ‘grounded, too, Similarly, the grounded aircraft must always, exit the maintenance station in the exact same order in whieh they entered it in the first place. In the actual application that we study, no such restriction is present, Aircraft are allowed to enter and exit the main- tenance in any feasible order, and their indices are updated accordingly to represent their relative order in terms of their residual flight and maintenance times, With this in mind, the index of each aircraft at the beginning of the next should be a decision variable allowed to take any value and should not be determined by the exact number of aircraft that will enter and exit the maintenance station Nevertheless, we prove next that there always exists at least ‘one optimal solution to the problem that results when this restriction is relaxed that satisfies it. Besides establishing the. validity of the proposed formulation, this proof also reveals some crucial properties ofthe problem, which are utilized in the next section for the development ofthe problem's solution algorithm,Kozanidis, Gavranis, and Kostaelou: Flight and Maintenance Planning of Mission Aircraft 219 PROPOSITION |: Given the optimal solution tothe prob- lem formed by (1)-(19), there does not exist another solution that satisfies all the constraints except possibly (4) and/or (5), in which the two flowcharts of available and grounded air- craft at the beginning of the next time period result in lower total deviation index. PROOF: Suppose that thee exists another feasible solu- tion (et the supersriptq refer tothe values of the decision variables of that solution) that contains evo intially avail- able alrraft with indies (and m, sich that sp > Yap and yf, < ify. and results in lower total deviation index Consider another solution in which all the decision var ables take the same values as those of solution 4, except that 3 = (0p — Yap) ha tn = 97 — a = Yh Yan = yf 1 = Bh, and iy Bf. Is easy to verify that Aieraft and m no longer violate constraint set (4). More- over, if there isa one to one interchange ofthese aireraft and the relative order of the remaining aircraft onthe associated flowcharts is kept the same, the otal deviation index value ofthis solution i the same as that of solution g. Repeating this procedure for any (wo initially available aircraft that vio- Tate constraint set (4) in solution , we can eventually get a solution that has the same total deviation index value and sat- isfies constraint set (4) entirely. An identical procedure Gust exchange variables x,y, and Din the above substitution with the corresponding variables hg. and, respectively) can also be applied for any two initially grounded aircraft that violate constraint set (5), leading eventually to solution in which constraint set (5) i entirely satisfied, 00. Moreover, the total deviation index value of this solution will be the same as that of solution . However, this implies that this solution provides an improvement tothe objective function value of the optimal solution to the problem formed by expressions (1)-(19), contradicting its optimality. 4, SOLUTION METHODOLOGY 4.1. Solving for a Particular Combination of z, and z, Suppose that the optimal values of variables zp and za in the formulation (1)-(19) are known. Then, the optimal values of variables b, for Aande, for j = 1,....NA,the slopes ofthe flowchart diagonals at the beginning of the next period and the two sets of aircraft that will be available and rounded at the beginning of the next period are known, 00. In this case, obtaining the optimal values of the remaining decision variables reduces to solving two independent trivial ssubproblems, as explained next. The first of these subprob- lems pertains to the set of aircraft that will be available atthe beginning of the next period, whereas the second one pertains to the set of aircraft that will be grounded. Let K denote the former of these sets, indexed by k. Set K is the union of set K1, which consists of the aircraft that are grounded at the beginning of the current period and will exit the maintenance station at the beginning ofthe next one, and set K2, which consists ofthe aircraft that are available at the beginning of the current period and will remain available in the next period, too, Clearly the maintenance time of each aircraftin set K, must be set equal to its residual maintenance time at the beginning of the current peri. Similarly. let Af denote the latter ofthese sets, indexed by ‘m. Set M is the union of set Mj, which consists ofthe aircraft that are available at the beginning of the eurrent period and will be grounded at the beginning of the next one, and set Ms, which consists of the aircraft that are grounded at the beginning of the current period and will remain grounded in the next period, too. Clearly, the flight time of each aircraft in set My must be set equal (0 its residual flight time at the beginning ofthe current period, We update the indices of the aircraft of set K as fol lows. The aircraft of set Kz are indexed first with indices k= 1.....|Ko| according to their residual flight time order. followed by the aircraft of set &' with indices k = |Ka| + I... 1Kal + [Kil according to their residual maintenance time order. Noting that [A | = [Ko] + [A | the following qua- dratic optimization problem can be used to find the optimal flight times of the aircraft that comprise set K ba E(u) St. y= pa, k= 1... Kal Yin = Yk = [Kal +. IAL vl ts ¥out D vay SUS Min 2 Yoins KLIK BS Xa k= Leesa] m20, & 1Kal In this formulation, the objective function minimizes the total deviation index that will be realized on the flowchart of available aircraft atthe beginning of the next period. The first ‘wo sets of constraints update the residual flight times of the aircraft of sets K and K, respectively, at the beginning of the rnext time period, The next constraint ensures thatthe flight requirements ofthe current period will be satisfied (index m scans the available aircraft that will se up theirentireresidual flight time and will enter the maintenance station for service at the beginning of the next period), The next two sets of con- straints impose a lower bound on the residual flight time of each aircraf in set K atthe beginning of the next period and ‘an upper bound on the flight time of each aircraft in set Ko, respectively. Finally, the last set of constraints accounts for, the non-negativity ofthe flight times. Note that the yaq's and Naval Research Logistics DOT 10.1002/naw20 Naval Research Logistics, Vol. 59 (2012) the a's fork = 1....,|Ka] are decision variables inthis for- mulation, the yap'sfork = 1,.... [Ka] areknown parameters, and the ys fork = |&3|+1,..-.|K|are auxiliary decision ‘variables (they do not appear in the original formulation) with known values. Adding the auxiliary decision variables.xg for K = [Ka] + 1,-..s1K| and setting s = =. sap = ¥ [Kal + ss 1K]. Xe = min(Xwac. Sap — Yain) for 1K 3|e and Xe = O for k = [Ka] + Le. A]. we cobain the Following equivalent formulation x Min DG —9) 9)? @) - vl st 1S— Dw <3 5 ow) x Dons US= DO np 25) OsmSX. Keke nIKL 26) The problem defined by (23)-(26) is a quadratic program ‘The Hessian of the objective function is diagonal with al its diagonal elements equal 10 2: therefor, its postive definite ‘which implies tha the objective function is strictly convex Hence, as all the constraints are linear, the KKT conditions (see Bavaraa et al [1]) are necessary and sufficient for opt malty. We give next a simple procedure called “Sweep” that can be utilized to obtain the optimal solution, On the flowchart that results from the known values of Parameters yip i 23)-(26), considera line parallel to the agonal, which is inital placed far enough tothe top. so ‘hat all the aircraft e helo i asshov in Fig. 2 Gn what fol- lows, we do not distinguish between a point on the graph and ‘the area that this point depts) Assume now that his line starts moving toward the diagonal (and past it, while always remaining parallel oi, sweeping along "vertcally”eachair= craft that comes across. Throughout ths move, fight times {denoted by variables) ar accordingly assigned tothe a= craft in the order that they are swept by the line If the Hight time of an aireratk reaches its maximtim possible vale, X,, H. then the problem defined by (27)-(29) is infeasible. PROOF: The validity of this proposition results from the fact that the problem defined by (27)-(29) isa special case. of the problem defined by (23)-(26) with L = = | and US —Yonem Yap = US— Yves, Ymp- a 4.2. The General Case The above discussion implies that when the optimal values of variables z~ and 24 are known, obtaining the optimal val- ues of the remaining decision variables of problem (1)-(19) reduces to solving two independent trivial subproblems. The first ofthese subproblems is associated with the flowchart of the aircraft that will be available atthe beginning of the next time period, and the second one is associated with the flow= chart of the airerat that will be grounded atthe beginning of the next time period. Moreover, the optimal total index devi ation value of the problem for a particular value combination (fz and zy is equal to the sum ofthe optimal deviation index values of these two subproblems. Therefore, among all the feasible value combinations of variables z_ and za, the prob- Tem’s global optimal solution is associated with the one for, Which the total deviation index assumes its lowest value. ‘To exclude in advance infeasible value combinations of variables cy and zy, we consider variable zy first. Assume that the grounded aircraft are already arranged in nondecreas- ing order of their residual maintenance times, with indices J = 1,...,NA. Let Sum be a non-negative auxiliary vari able. Then, the following pseudocode provides a valid upper Sum + gp! if Sum = Bdo dsith else print j ~ 1 and exit; end if fend while print j — Naval Research Logistics DOT 10.1002/nawPROPOSITION 4: The value of variable :, in any feasible solution to problem (1}-(19) cannot be larger than the value printed by the above pseudocode. PROOF: We use variable Sum to store the sum of the resid ual maintenance times ofthe grounded aircraft that can finish their service in the current time period. Every time that the residual maintenance time of an aircraft is added to Sum and ‘Sum still remains less than or equal to B, we conclude that the time capacity of the maintenance station suffices to finish the service of that aircraft, too; therefore, we increase the upper bound on variable zy by 1. The procedure terminates either ‘when Sum becomes strictly greater than B, implying that the station's time capacity is not sufficient to finish the service of the last (or any subsequent) considered aircraft, or when the full list of grounded aircraft has been scanned, implying that NA is the maximum (and unigue, due to constraint (9)) feasible value of variable o ‘The following pseudo-code provides a valid lower bound on variable eq: Sum = S14) jp — NAGain) while j =NA do if Sum > Bdo print j and exit else Sum Sum + Goin: endif cond while print NA: PROPOSITION 5: The value of variable :y in any feasible solution to problem (1)-(19) cannot be smaller than the value printed by the above pseudocode. PROOF: The validity of the proposition results from the fact that if k is a feasible value for variable 24, then Dies sin + Di esilein — Gnin) 2 B, of equivalently DM, cup -NA(Gnin)-+4(Gin) = B holds. Thus, the lowest nwotenegative integer value of for which this inequality holds isa valid lower bound on variable cp. Iino value of k between 0 and NA satisfies this condition, then "4, gjp < B, and constraint (9) restricts 3") hy to be equal to D\, ej»; therefore, the only Feasible value of variable zy is NAL In the presence of the special case discussed in Section 3 that we resolve through the introduction of constraints (20) (22), the lower hound on variable zy will be larger than its upper bound, rendering the entire problem infeasible, Next. ‘we exclude infeasible values of variable ¢,. Assume that the available aircraft are already arranged in nondecreasing [Naval Research Logistics DOT 10.1002/nav Naval Research Logistics, Vol. 59 (2012) ‘order oftheir residual fight times, with indices # = 1....¥A. ‘The following pseudocode provides a valid upper bound on variable Sum = 0:1 =1; while i < min(C, A) and yip ¢ Xmax do Sum = Sum + Yips if Sum = US do +1: else print i — 1 and exit: endif fend while print — PROPOSITION 6: The value of variable z, in any feasible solution t problem (1)-(19) cannot be larger than the value printed by the above pseudocode. PROOF: Weuuse variable Sum to store the sum of the resid- ual flight times ofthe available aircraft that can be grounded, in the current time period. Every time that the residual fight time of an aircraft is added to Sum, and Sunt still remains less than or equal to US, we conclude that the fight load suf fices to ground this aireraft, too: therefore, we increase the ‘upper bound on variable z, by 1. The procedure terminates either when the upper bound on variable z_ assumesits largest possible value (=min(C, A)), or when the first aieraft with residual flight time strictly greater than Xa is encountered. ‘or when Sun becomes strictly greater than US. o ‘The following pseudocode provides a valid lower bound on variable sum ip — Ain while j 2A do if Sum = ES do print j and exit else Sum = Sum + Yaa: j= 7-415 end if end while print “problem is infeasible”; PROPOSITION 7: The value of variable zy in any feasible solution to problem (1)-(19) cannot be smaller than the value printed by the above pseudocode. PROOF: Te vality ofthe proposition results from the fact that if& isa feasible value for variable zy then 7%) yp DhisiWin — Yin) > LS, or equivalently, D7) vip — ‘AVoua)-+£ nu) © LSholds, Thus, the lowest non-negative integer value of k For which this inequality holds isa validKozanidis, Gavranis, and Kostaelou: Flight and Maintenance Planning of Mission Aircraft 23 lower bound on variable <,. If this inequality does not hold even when k = Athen the problem is clearly infeasible. 0 ‘Once we determine the individual upper and lower bounds, fon variables 2 and <4, we utilize constraint (8) to eliminate ‘those value pairs that violate the space capacity of the main- tenance station, After that, the proposed algorithm searches, among the remaining value combinations of variables =, and, 4 to find the one that leads to the lowest total deviation index, 4.3. A Small Numerical Example In this section, we illustrate the application of the above algorithm on a small numerical example. Consider a unit comprising six aireraft, four of which are available and {wo of which are grounded at the beginning of the current time period. Table I presents the residual fight times of the available aircraft and the residual maintenance times of the rounded sireraft. The values of the other problem parameters are $ B = 325h,G = 320h, ¥ = 00h, C =3, Xe Yoin = 0.18, Gryig = 0-1 h, L = 0.95, and U= 1.05. The available/grounded aireraft are already sorted in non- decreasing order of thei residual fight/maintenanece times. If this were not the ease, the user Would have to rearrange them and update their indices accordingly. Applying the pro- cedure for obtaining the bounds on variables 10 <2, <2and 0 < zg < I. Only five of the resulting six Value combinations of zy and zy are feasible, because e straint (8) is violated for zy = 2 and z_ =O. Table 2 presents the total deviation index value of each combination. Thus, the optimal solution is the one wit 1 and Sa = 1 and objective function value 9864.06" This total deviation index value results as follows. AS zy = 1 and te = 1. the available aireraft with index i = 1 will be {grounded and the grounded aircraft with index j = 1 will 125h, 50h, “Table A, Residual fighvmaintenance times (sp /gy) (h) 3 Bold-sige entries denote and plain-style entries denote fright times of available air erat ‘Tuble2. Optimal total deviation index value for each combination of zy and 0 1 0 87,652.08 23.098.56 It 2721581 9864.06 2 F 37819.62 become available at the beginning of the next time period, ‘Therefore, x1 = 38, yig = 0. hy = 130. and gin Set K comprises the available aircraft with intial indices i = 2-4 (these aircraft comprise set A) and the grounded, aircraft with initial index j = 1 (this aircraft comprises set K;)-Set M comprises the grounded aircraft with initial index Jj =2 (this aircraft comprises set Mz) and the available air- craft with intial index 7 = 1 (this aircraft comprises set M}). ‘To compute the values of the remaining decision variables, ‘we update the aircraft indices first, as shown in Table 3 Using the updated indices, the following quacratic opti- mization problem can be used to compute the flight times of the aircraft that comprise set Kx we (rm aq = 300 — 4 0.95(125) = xy +49 +n + ay +38 = 1.05(125) Ye ZOL, P= Med 11 £50, 20, F ‘After some basic manipulation, we obtain the following. ‘equivalent formulation: Min ((30—.1) — 75)? + (Q73 = x2) — 150 + ((298 = 15) = 225)? + ((300 ~ 24) — 300) st 80.75 < xy tar tas ay $93.25 Osx £499; OSm 550; OSH <50: Osu <0 ‘Arranging the quantities LL, UL, X, and D of Proposition 2 in nondecreasing order, we have LL = 80.75, UL = 93.25, D = 100, and X = 149.9, Therefore, the optimal solution of the problem is the one obtained by Procedure Sweep, when the sum of the assigned aircraft flight times becomes equal to 93.25, Using the updated indices, the flight times of the sircraft indices. 7 3 Naval Research Logistics DOT 10.1002/naw24 Naval Research Logistics, Vol. 59 (2012) aircraft of set Ky in that solution are x; = 0. x2 = 50, and 43.25, or using the initial indi xs = 50, and xy = 43.25. After substituting these values in the objective function of the above formulation, the deviation index value fon the flowchart of the aircraft that will be available at the beginning of the next time period is equal to 6839.06, Simi larly, the optimal maintenance times are 130 for the grounded aircraft with initial index j = 1 and 195 for the grounded air- craft with initial index j= 2, leading to a deviation index value of 3025 on the flowchart of the aircraft that will be ‘grounded at the beginning of the next time period. Therefore, the total deviation index of the solution with 2, = 1 and [is equal to 6839.06 + 3025 = 9864.06. 5. COMPUTATIONAL IMPLEMENTATION In this section, we analyze the worst-case computa- tional complexity of the algorithm that we propose, and we present computational results evaluating its performance on randomly generated instances. A. Computational Compl ity Problems similarto the one defined by (23)-(26) have been studied extensively inthe past (Helgason etal [4]: Brocker {2}; Calamai and Moré [3]; and Pardalos and Kovoor [10]} and several exact solution algorithms have been proposed, some of which are asymptotically optimal in terms of com- putational performance. Next, we utilize one of these papers to analyze the computational effort required to solve the problem defined by (23)-(26). LEMMA 1: The problem defined by (23)-(26) can be solved in time O(|). PROOF: The values of LL and UL are known, We can compute the values of D and X in time OCIK'). as D SIE, min Xe. [ap — AST) and X= SUE) Xy. Finding if there exists an arrangement of LL, UL. D, and X in which LL precedes X requires time O(1). If such an arrangement exists and Q is the second quantity in order, we can equiv- alently transform the problem defined by (23)-(26) into the following problem in time O(K ): se E (Gotan) su Doherty keusm. k IKI. whe 2, ae = 20xp — ks) by = 1s bo =O. te = 0, and uy = X¢, for -IA |. We have suppressed the [Naval Research Logistics DOT 10.1002/nav term (yap — ks)? in this formulation, because it is constant and does not affect the optimization. This problem ean be solved in time O(| | (see Brucker [2)). Therefore. the prob- lem defined by (23)-(26) can be solved in total time O((K ). 100, o The following proposition utilizes Lemma 1 ana- Iyze the computational complexity of the proposed solution algorithm, PROPOSITION 8: The computational complexity of the solution algorithm that we propose is OA log(A)) + OW (NA) mIN(C. AY). PROOF: The total time required to arrange the available aircraft in nondecreasing order of ther residual flight timesis, O(A log(A)). The total time required to arrange the grounded aireraft in nondecreasing order of their residual maintenance. times is OVA log(NA)). The total time required to find the upper and lower hound on variable z, is O(A). The total time required to find the upper and lower bound on vari- able 250. Columns 8 and 9 of Table 4 present our proposed solution algorithm's average and maximum computational times. We explain the results shown in the last to columns of Table 4 in the next subsection Columns 2 and 3 of Table 5 show the average and max- imum percentage difference of the objective value of the solution returned by LINGO’s local solver from the prob- lem’s global optimal solution objective value. The next two columns of the same table show the average (rounded to the nearest integer) and the maximum number of Feasible value combinations of variables z_ and zy. OF course, these results are always the same for both the solution algorithm that we propose and the decoupling LINGO model. We explain the results shown in the last two columns of Tables 5 in the next subsection The superiority of the solution algorithm that we propose ‘becomes immediately clear, because its computational times, are significantly lower than those of both LINGO models. As the results of Table 4 demonstrate, the computational savings increase considerably for large scale problem instances, for Which the application of LINGO appears impracticable. This [Naval Research Logistics DOT 10.1002/nav is partially due to the fact that the increase in the number of feasible value combinations of variables zy and zy is quite ‘moderate as the problem size increases, and partially due 10 the eficiency of Procedure Sweep. The variability ofthe solu- tion times appears higher in the case of LINGO than in the ccase ofthe algorithm that we propose. Additionally, the o inal LINGO model with the local solver invoked has a rather ‘unusual behavior, because its average computational require- ments for N= 250 are lower than those for N = 150 and N = 200, This is an indication thatthe total computational effort also depends on the specific characteristics of each problem instance besides its size. With the exception of the problems with = 50, the average computational require- meats of the decoupling LINGO model are higher than those of the original LINGO model with the global solver invoked. ‘This is an indication that the significant computational sav- ings of our solution algorithm should be attributed more to the efficiency of Procedure Sweep than to the decoupling of the original problem into smaller subproblems alone. Asexpected, the computational requirements of LINGO's local solver are significantly lower than those of the global solver. This comes at a price, however, as columns 2 and 3 of Table 5 verify, because the objective value of the solution retumed by the local solver is on the average approximately between 100 and 550% higher than that of the global opti mal solution. In the worst case, this percentage difference increases to ~2000%. One way to improve the quality of the solutions retumed by LINGO’ local solver isto increase the ‘number of “multstart solver attempts” on the “Global Solver” tab, but thisalso increases the computational time required for termination. In general, as this number increases, the results, retumed by the local solver resemble those retumed by the. ‘global solver in terms of solution quality and! computational requirements 5.3. Algorithmic Enhancements ur extensive computational experience with the proposed solution algorithm and LINGO has provided considerableKozanidis, Gavranis, and Kostaelou: Flight and Maintenance Planning of Mission Aircraft 21 evidence suggesting that the two-dimensional “cost-matrix” ‘with rows the feasible values of variable z,, columns the fea- sible values of variable zp. and elements the optimal total deviation index values for each particular combination of sq and zg, may possess a special type of convexity called row and column convexity, More specifically, let TC(g.) be the problem's optimal total deviation index value for the combination wit g and z, = a. The corresponding matrix is row convex if TC(g,a) < TC(g,a + 1) implies TC(g,a +1) = TC(g.a +1) for every feasible / > 2 and if TC(g,a) < TC(g,a—1) implies TC(g.a—1) < TC(g.a—i) for every feasible / > 2. Similarly, the corresponding table is column convex if 7C(g.a) < TC(g + l.a) implies TC(g + la) = TC(x +/,a) for every feasible / > 2 and if TC(g.a) < TC(g—l,a) implies TC(g—1.a) = TC(g—i.a) For every feasible i > 2. Despite numerous and tedious attempts, we have not been able to develop a formal mathematical proof tI ishes the validity of this property. On the other hand, despite exten- sive experimentation, we have not been able to discover a single counterexample that disprovesit either. IFthis property is indeed valid, then we can exploit it to improve the com- putational performance of the proposed solution algorithm considerably. To show this, we developed a simple modifica- tion of this algorithm, which does not compute the optimal total deviation index value for every feasible combination of q and zy; instead, this algorithm computes this value for the middle element of each row (or column using a specific sim- plisticrute) of the cost-matrix and terminates its search within the Same row (or column) as soon ast has established, assum- 1 that the cost-matrix is indeed row and column convex, that ‘no further improvement on the objective ean be accomplished. inthe same row (or column). After this procedure is repeated for all rows (or columns) of the matrix, the best incumbent solution is returned by the algorithm. The last wo columns, of Table 4 present the average and maximum computational times of this modified algorithm, and the last two columns of ‘Table 5 present the average and maximum number of value ‘combinations of variables z, and z, for which this algorithm computed the total deviation index. These results show that if the cost-matrix is indeed row ‘and column convex, then the solution algorithm that we pro- pose can be significantly expedited by cleverly incorporating this property into its original design. The modified algo- rithm returned the global optimal solution in every problem instance out ofthe 200 on which it was applied, giving us. for ‘one thing, strong evidence that this is indeed true, Given this, intuition, we believe that future research should investigate ‘whether this property is actually valid or not, Ifit turns out 10 be, asuitable modification ofthe proposed solution algorithm that exploits this result to the greatest extent should be devel- ‘oped, so that additional computational benefits can be gained. Given the deficiency that stems from the lack ofa formal proof ate for the validity of this property, we did not develop a sophis- ticated design for the cost-matrix search, which explains the rather simplistic design that we present above, 6, SUMMARY AND FUTURE WORK In this article, we developed 1 mixed integer nonlinear ‘model for flight and maintenance planning of a group of aircraft that comprise a unit. The objective is to provide a balanced allocation of the flight load and the maintenance capacity to each individual aircraf, so that the long-term availablity of the unit is kept at a high and steady level ‘The formulation that we propose is based on a suitable mod- ification of an existing graphical heuristic too! for addressing this problem. Utilizing the problem's special structure, we also developed an exact search algorithm for its solution. ‘Our computational results demonstrate the superiority of the proposed algorithm over acommercial optimization package. Thisarticle points to several promising directions for future research. The question of whether the computational perfor- ‘mance of the proposed algorithm can be further improved through appropriate enhancements remains open. Therefore, investigating whether the special convexity result discussed in the previous section is valid stems as a very interesting task, Another interesting direction for future research is 10 apply the proposed medel on a time horizon of several time periods, to compare its long-term performance and behavior. ‘This will also render this model directly comparable to other, models from the existing literature, which are targeted toward longer planning horizons. Future articles should also be directed toward the devel- ‘opment ofa stochastic model that will incorporate the uncer- tainty that some of the problem's parameters might exhibit ‘Weedeliberately decided to deal withthe deterministic version Of this problem in this article, because itis novel, complex, interesting, and realistic enough to constitute a contribution in itself, which is useful for researchers and practitioners Moreover, the deterministic model that we address in this, article is a basic building block toward developing a more complicated stochastic model that will take into consider- ation random events, such as unforeseen failures. A recent example in that direction is the article by Mattila etal. [9] ‘who use diserete event simulation to model the maintenance of military aireratt in the Finnish Air Force and study its impact on aircrat availability bth under normal and conflict ‘operational conditions. Two factors of the problem under consideration exhibit significant uncertainty in practice: the fact that the actual duration of the maintenance service may turn out to be longer than its nominal value (expressed by parameter G), and the fact that a mission aircraft may fail in fulfilling its entire flight load (expressed by decision variable x,), The grounding ‘of an aiveraft for a longer time period than the one deter- ‘mined by parameter G may be dictated due to the detection Naval Research Logistics DOT 10.1002/naw28 Naval Research Logistics, Vol. 59 (2012) of serious findings during the regularly scheduled mainte- nance inspection or due to the lack of specific spare parts or Stalf expertise, which are needed to complete the service of the aircraft according to the prescribed safety standards. On the other hand, an aircraft may not beable ofl the entire time that has been assigned (o it, due to an unexpected event such san unforeseen failure. OF course, in case of such undesi able events, the user always has the option of reapplying the ‘proposed model after updating the state of the system, but a ‘Stochastic model would clearly incorporate such uncertainties ‘more accurately, leading to a better long-term performance for the entire system. APPENDIX. PROOF OF PROPOSITION 2: Let Ay, Ap. and (l= J...) be ‘te non-negative dal mokipios of constraints (24)-(26, eepctsel In ‘xkltion tothe original constants ofthe problem, the KK conditions a 2p — ks 4) — Ha 4a be > Iki G0) IKI Bb, aL-24p — Hg) Ay 2 Had Mu (= Ss) ° om he (Es -05+ =) =0 3) alte K=O =D oH dy20 dp20 m20, & x as) We cal the quantity (4p As 14) the “perpendicular distance of arcralt fom the diagonal” Despite he wo "itance” inthis inion, not that this quunty negative when arate Blow the diagonal tthe end ofthe eurent period. As LL. = UL always, there are 12 posible distin rangement ofthe quantities LL, UL, D. and X. When there does et exist fan arangement in which LL precedes X. the problem i clea infeasible [ecause the Might requirements (constant (2) cannot he satisfied. even ‘when every aircraft is asignod its manitum possible ight time, tn each ofthe romain eit cases, is clear hat the solution obained feo the application ofthe Procedure Sweep wes the sum ofthe assigned ical Tight dimes becomes equal othe second dusty in the reangemet,sats- fas (24) (26) and is there feasible. We show next tha ths solution also satisfies coulitons (30)-(3), and is thevefore optimal 00. (CASE 1: The arangomeat is (LL, UE, D.X} o (LL, UL.) In this ase, the sum ofthe assigned scat Hight tes inthe obained solution sequal othe second quantity inthe arangement, UL. We partion ‘he indices of the decison variables ofthis sclton ino four sts 4 st Sy contains the indices ofthe variables suc that xq = 0 Xs ‘set Sy coms the indices ofthe vaiales x sch hat xg = 0 X ‘set $5 comin the indices of he vaiales a such hat Oe = re 4. Se 4 contains he indices ofthe variables 2 such hat = re © Xs Westar 24onp — bs — nu) forsome k € 84 Noe that he valu of isthe same fr any & Si a eS contans the indies of the [Naval Research Logistics DOT 10.1002/nav ‘varales that ion the sweeping line atthe cute solution, anda result heir perpendicular distance from the diagonal isthe same. Additionally, each ofthese distances non-nogatve, because the ae hat UL appears before D inthe avagement imps thatthe sweeping Tin His above the Aiagoal at the caren solution; therefore 23 i non-negative, too, We aso sey = max(2(qny —k8 44) ~ 33.0) fore Siu, = Olork © S3USy date = 20iy bs — 4) hy Fork © Sy. Tho mulipio uy fore Sy re always non-negative, a et 3) conan the indices of the vail tha ‘ser inly sept and ater disengaged by the sweeping line, cause they reached thet upper bound thesfors the perpendicular distance ofeach of ‘hess points from he diagonal cannot he salle th he perpendicular i= tance from the diagonal of any point that ex on the sweeping line 3 he cureat solution. Fork 5; and € 5. constrains (30) and (31) ae clearly Satisfied. Constraints (30) ace clearly satisfied as an equality for Sx ‘hereon, constraints (31) re sais, oo, For 8, comstaits 31) ae ‘leuly sted and constrains (30) ae sisi if 2 > 24g — Rs — 0, ‘which i trac a st Ss contain the indices of the arabes that have ot thon swept by te line yet; therefore, their perpendicular distance for the iagonal cannot he ager than the perpendicular distance rom te dgonal ‘oF any point hat lies onthe sweeping Tine at he curent solution, Finally constants (5238) ae clay said, too. Hence, the cueeat solution Cogether with, 22.andug (k= I,K} asthe dual mulpis sashes the KKT conditions and is therefore optima Tse Sy iscampt.thon the above analysis remains the same, but ip neds tw be set equal to max(mases, 2104p —A8 —).0) [bath 83 and Si ate cmp then every deinen arable hs akon is maNimum possible ale ‘This implies dat UL = X, and this case redices 9 Case 4 through a8 appropiate earangeneat (CASE 2: The arangomeat is (LL, D, UX] 0 [LL D.X, UL) In this cae, de sn ofthe assigned sera fight ses in the sition ‘blaine qa tothesecond quant inthe arrangement, Dwhichimplics thatthe sweeping line coincide with the diagonal. We partition the indies ‘ofthe dino arial of his solution nth same four sets sin Case We set 4 = 2 = O.nk = maxi2(rip As ~14).0) fork € Six = Ofer S20 Se and ag = 204p ~ As — ra) fore Sy The multipin a foe [ee Ssareaiwaysnon-negalve ast Ss contains th indices othe arabes ‘hat wer inal swept and later disengaged bythe swecping line, because they reached ther upper bound thersfrs asthe sweeping ine coincides ‘withthe diagonal a the current solution, the perpendicular distance ofeach, ofthese ins mm he digonal cant he nepative. Fork © Sj and € 5s, ‘consrams (30) and 31) areal aise, Fork ¢ Sp, contains (31) are leu sted and constraints (30) te satsied if —2(j4p— As — 4) 20, ‘which is true as et Sp contain the indices of the arabes thar have ot ton swept hy the line ys therfore, as the swscing Fine coincides with the diagonal a the curea solution, cach ofthese points has now postive petpendculr distance fom the diagonal. Set, contains the indies of the ‘acabls tha eon the seeping line atte curentsalition. As the sweep ing ine coincides with the dagen, he perpendicular distance ofeach of ‘hese variables from the diagonal is qual oO. Asa result coms (30) sind GI) are also staid foe K © Sy. Finally, consents (32)435) ate ‘leuly stised, too. Hence, the cumen solution together with 4,33, and 14 =... [Aas the dal muller satis the KKT condions and is therefor opin. (CASE 3: The arangsment is (D: LL, UX] 01D. UL X,UL) In this case, the sm ofthe assigned aircraft gh times in the sation obtine i xual the second anti inthe serangement, LL. We pai jon the nos othe dessin vrs his sTaion no he sa Four sets a5 in Cases I and 2. We set ig = Oand 24 = ~20p ~ ks — m0). for & © Sg, Note thatthe valve of isthe same For any E € Su as se ‘St comin the ince ofthe variables that lie onthe sweeping line athe ‘urent soetion, and sa es their perpendicular distance from te dg ‘nals the same Addons ach of hese distances is nonspositve 35 the fact hat appears ist i the arrangement implies tat the sweepingKozaniis, Gavranis, and Kostaelou: Flight and Maintenance Planning of Mission Aircraft 29 line does nt leave the diagonal athe cuseat lution: therefore ison-negatve, oo, We also Seay = MC — AB — 4) + 1,0) foe Ee Shug = Ofork © 810 Sq and uy = 2p —ks —m4)-+2) fork © S ‘The muliplics ug fore S) are always now-acgative, asst Sy contains indices ofthe triable that were nally wey and itr disengage by ‘seeping line, because they reacad thee upper bound: therefore, the ‘perpendicular distance ofeach of hese pins om the diag cannot bo ‘srllr than the perpendicular distance frm the diagonal of any pin hat Tas om the sweeping line athe curen solution. For ke Sy aml K€ Ss, onsen (30) an (31) ae clay satis. Constraints (30) ae scaly ‘tnd an equality fork € Se therefore, constrains (81) ar aie, too. For ke 8 constraints (31) ate larly sts and constrains (30) ae satisiod if “21 2 pap ~ ks — 1). which i te as et S2 conan the indies ofthe wriabls that vena vn swept th ine yt therefore the ‘erpenicule distance ofeach of these pint rom te diag cannot Be Taygerthan the perpendicular distance from tb diagonal of any poi hat is fon the sweeping Tie atthe current solution. Filly, constrains (32)-35) fat clearly satisfied, to, Hence, the cute solution Loge wilh 20, andy (= I, [A asthe dal multiplies ass the KKT conditions fn is therefore opi set Ss isempy then the above analysis emis the sane, but A aes tobeset equal to minjess(~20p —£5—4)- which always aon-pegatve {st $y contuns the indies othe vriahes tht have not heen swept {he line ye: therefore asthe sweeping ine does ot is above the digoaal athe curet solton, ach ofthese pots has non-positive perpendicular ‘isuance Tom the diagonal I ah Soa tt emp. thn very decision ‘ariale has taken is maximum possible value. ID = LZ = this ase seduces to Case trough an appropeat rearrangement. WD = LL = X, he seeping ines helo the diagonal ath curent solaton Ih me, we partion the indies of the decision variables ofthe eurensliaion 1 st Sy contains the indices ofthe variables such thal Xs 1 st) comtans the indices ofthe variables sto that O < 39 = Xt Weset2y = —26y4p—fs—x)-whorok € Sp sthe index ofa vail thats ciety onthe sweeping ine, max Q(vip K-00) $21.0), Fork © Sy and ag = 2004p Ks) +, fork Sp. Ashe sweeping line lies below the diagonal is sel pine. Additonal. the muller 1 for F © Sp ae alays non-negative, Bocase the perpendicular distance fiom the diagonal of ay pot that es onthe swecping Hine tthe eueat soluion cannot be lager than th perpendicular distance from the diagonal ‘fang other point Constaits(30)-(38) ar clay satisied fork © S15. ence the erent solution together with 2.42, and ap (E= Ty) a8 ‘he dil multipass the KKT conditions ai is therefore opin (CASE 4: The aangomeat is (LEX, UL, D] o (HX. DUE), In this ease, te sum of dhe assign aircraft times in the solution ‘btn seal tothe second quantity ia the arangement,X. We partion rcs ofthe decision variables of his ston ina wo ss 1 st 5, comtins the inices ofthe variables sich that; = 0 Xi ‘sot 5 comin the infices ofthe vaisles x, sich hat <= x, = Xs We set Ay = 2 = 0, and ay = maxQGyy — ke — 0) 0) fork € 81 andy = 20 4p — &8 — 4) fork © SThe multiplies for ke Sp are always nan-aepative a6 et contin the indices ofthe variables ha have tea been swept bythe sweeping ine o tsi upp bound: here 8 {he swecping line docs aoe low the diagonal a he cureat sluon, ‘ei perpendicular distance from the digonal non-negative. Consrais (20-35 are cleat sted for 5) U Sy Hence the creat Solution, together with. 22.ande (& = I...) asthe dual multipliers, sass the KKT conditions andi therefore opti. a ACKNOWLEDGMENTS ‘The authors thank Prof. Liberopoulos of the Department of Mechanical Engineering of the University of Thessaly for several fruitful discussions that led to the problem for- ‘ulation of Section 3, Nikolas Kanonis, an undergraduate student of the same department, for helping with the numer- ical results, and an anonymous referee for his/her valuable comments. REFERENCES U1] MS. Bazarss, HLD. Sherali, and CM. Shetty, Noafinear programming: Theory and algorithms, Wiley-Interscieace, New York, 2006, [2] P. Brucker, An O(n) algorithm for quadratic knapsack prob- lems, Oper Res Lett 3 (1984), 163-166. [3] PH. Calamai and JJ. Moré, Quasi-newton updates with hounds, SIAM J Numer Anal 41987), 1434-144 [4] Ro Helgason, J. Kennington, and H. Lall, A. polynomially bounded algorithm fora singly constrained quadratic progrun, “Math Program 18 (1980), 338-343. [5] G. Kozanidis, A multiobjective model for maximizing Neet availability under the presence of flight and maintenance requirements, J Adw Transport 43 (2008), 155-182. [6] G.Kozanidis. A. Gavranis, and G. Liberapoulos, “Heuristics for maximizing feet availability subject to flight & mainte- nance requirements.” in: Proceedings ofthe 10th International Conterence on Applications of Advanced Technologies in ‘Transportation, Athens, Greece, 2008, 15 p. (71 G. Kozanidis, G. Liberopoulos, and C. Pitsitkas, Flight and maintenance planning of military aireraft for maximum feet availability, Miltary Oper Res 15 (2010), 53-73, {81 LINGO 1110, User's guide, LINDO Systoms, ine., Chicago, TL (2008), Available at: hiphvww lindo.comy. Accessed on 17 February, 2012. {9] V.Matila, K.Vitanen, and. Raivio, Improving maintenance decision making in the Finnish Air Force though simulation, Interfaces 38 2008), 187-201 [10] PM, Pardalos und N. Kovoor, An algorithm for singly con- strained class of quadruie programs subject to upper and lower bounds, Math Program 46 (1990), 321-228 11] BLW. Pippin, Allocating Hight hours to army helicopters, MSe ‘Thesis, Naval Postgraduate School, Monterey, CA, 1998. [12] N. Safsei, D. Banjevie. and A.KS. Jardine, Woskforce- ‘constrained maintenance scheduling for military sircraft let ‘Accase study, Ann Oper Res 186 2011), 295-316. [13] A.Sgasik, Planning german army helicopter maintenance and mmission assignment, MSe Thesis, Naval Postgraduate Schoo Monterey, CA, 1994. [14] A. Steiner," heuristic method forareraft maintenance sehed- tuling under various constraints,” in: Proceedings ofthe 6th Swiss Transport Research Conference, Ascons, Switzerland, 2006, 28 p. 115] US. DoA, Field manual No. 304.500: Army aviation maintenance (Appendix D: Maintenance management tool), US. Deparment of the Amy, Washington, DC, 2000, Available at: hiip/uwye globalsccurity orgmilitaryibracy! policyfarmy/inv3+04-5001. Accessed on 17 Febeuary, 2012. Naval Research Logisties DOT 10.1002/na