Post on 02-May-2022
Research ArticleSupply Chain Inventories of Engineered Shipping Containers
David A. Field1 and Dennis E. Blumenfeld2
1Birmingham, MI, USA2Department of Industrial and Operations Engineering, University of Michigan, Ann Arbor, MI, USA
Correspondence should be addressed to David A. Field; dfield08@gmail.com
Received 18 July 2016; Accepted 8 September 2016
Academic Editor: Fu-Shiung Hsieh
Copyright Β© 2016 D. A. Field and D. E. Blumenfeld. This is an open access article distributed under the Creative CommonsAttribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work isproperly cited.
Manufacturing operations that assemble parts often receive components in expensive highly engineered shipping containers. Asthese containers circulate among suppliers, assembly operations, and logistic providers, they require inspections and repairs. Thispaper presents mathematical models that predict the number of available containers as functions of damage, repair times, andscheduled daily production. The models allow making complex decisions with a few basic parameters. Results not only show aminimal investment in the number of containers and safety stock but also quantify the dependence on damage rates and repairtimes for ordering additional containers.
1. Introduction
Many studies on shipping containers emphasize reposition-ing empty shipping containers [1β4]. However these andother studies, for example [5β7], that stress other aspectsof supply chains and include empty shipping containersrefer to large box containers, or twenty-foot equivalent units(TEUs) convenient for multitransportation modes, ports,fleet terminals, modal transfers, and so forth.
While many articles on large box containers have beenpublished, very little if any literature addresses the problemof maintaining the availability of highly engineered reusablecontainers required for shipping easily damaged parts in amanufacturing operation. Manufacturing each container cancost considerably more than $1,000 USD.Whenmanufactur-ing operations such as the assembly of automobiles requiresthousands of such containers, millimeter differences in theproduct and container dimensions can cost many millions ofdollars for extra containers. Moreover, expensive engineeringanalyses for stresses, strength, packaging, and transportationrobustness in preventing damage to sensitive parts becomeimportant because the cost and time for repairing containerscan become serious issues. Predictive mathematical modelsfor the availability of containers can savemillions of dollars inup front and ongoing cost formanufacturing a product whoselimited production lasts a few years.
The main contribution of this paper analyses man-agement of highly engineered containers. Our paper firstpresents a mathematical model for a supply chain consistingof a supplier, a general assembly plant, and a containermaintenance center (CMC); see Figure 1. Typically the assem-bly plant and a CMC are situated closer together than thesupplier.
At each node in the loop the number of containers thatexceed the daily requirement constitutes a safety stock. Thequantities of containers at each node and the number ofcontainers required for a steady state production have beenchosen for easy scalability when used with proprietary data.The paper then creates predictive mathematical models thatexplore and highlight the consequences of different assump-tions on the availability of highly engineered containers.These models can also show where to optimize cash flow. Aminimal number of containers plus a small safety stock canreduce the initial investment. Only as the number of damagedcontainers under repair grows should additional containersbe purchased. The models allow making complex decisionswith a few basic parameters.
Recent research has found that the excess of containersin manufacturersβ supply chains can be as high as 30%[8]. As an example of the importance of this paper, in theautomotive industry, easily damaged parts such as doors,
Hindawi Publishing CorporationInternational Journal of Manufacturing EngineeringVolume 2016, Article ID 2021395, 8 pageshttp://dx.doi.org/10.1155/2016/2021395
2 International Journal of Manufacturing Engineering
Supplier
Assemblyplant
Containermaintenance
center
Figure 1: Depiction of the logistical flow of containers from a CMCto a supplier to an assembly plant and back to a CMC.
hoods, roofs, and exhaust assemblies require containers.Consequently, each car model under current productionrequires many thousands of expensive highly engineeredreusable containers.
The second section of this paper develops a basic mathe-matical model and a simulation to demonstrate the predictedbehaviors. Explicit and implicit assumptions and their con-sequences are clarified. The third section establishes a moregeneral model that can accommodate a variety of probabilitydistributions. Two distributions illustrate the model. A sec-tion containing remarks and a summary section conclude thispaper.
2. A Basic Mathematical Model
This section develops a basic mathematical model fromwhich extensions will lead to increasingly realistic models.This model clarifies explicit and implicit assumptions whosereassessment in Section 3 will guide the generation of amore generalmathematicalmodel.The following subsectionsdescribe the basic model, give its assumptions, supply nota-tion, derive a steady state formulation, and add a simulation.
2.1. Description and Assumptions. Although the motivationfor modeling a supply chain loop comes from supplying anautomotive assembly plant, the basic aspects of the modelingapply to any manufacturing that relies on reusable engi-neered containers supplying components for a manufacturedproduct. Assuming that the supply chain does not losecontainers due to theft, damage to containers and their repairimpact the availability of containers most severely. Our firstmodel represents a basic supply chain having three essentialelements; a supplier, an assembly plant, and a containermaintenance center (CMC) for inspection and repair ofdamaged containers.
Let π΄DP denote the number of containers required tomeet the scheduled Daily Production on an average day. Themodel assumes that the assembly plant initially has 3π΄DPcontainers of parts, 2π΄DP in the safety stock, and π΄DP for anaverage dayβs production. With 2π΄DP more containers at thesupplier, the plant can operate for one week. The CMC has asafety stock of 5π΄DP containers, enough containers neededfor a week of production. The total number of containers
equaling 10π΄DP provides a convenient scalable quantity formathematical modeling.
Note that the mathematical models in this paper do notinclude times for transporting containers between locations.Themodels assume that transportation of containers starts atthe end of a workday and continues overnight if necessary sothat containers arrive at their destination before an 8-hourshift of work begins. The models also assume that emptycontainers arriving at the supplier in the morning will befilled and ready for shipping by the end of the same day.Similarly, the assembly plant sends all the emptied containersto the CMC for inspection. After inspection all undamagedcontainers will be available for shipment to the supplier.
A steady state loop without any damage to containersand guaranteed overnight deliveries would not require safetystocks. Nonetheless containers often sustain damage andmust be repaired. The following model assumes a geometricdistribution for the probability π of a container sustainingdamage. The discreet geometric distribution has a basis onsequences of independent Bernoulli trials [9, p. 367]. We canmodel the number of times a container arrives at the CMCbefore the first damage is observed as the number of trialsfollowed by the first βsuccess.β If π, 0 β€ π β€ 1, denotesthe probability of a βsuccessβ and π denotes the number oftrials before the first βsuccess,β then the discrete geometricprobability mass function distribution π(π),
π (π) = π (1 β π)πβ1 , π β {1, 2, . . .} , (1)
has the cumulative distribution function πΉ(π),πΉ (π) = 1 β (1 β π)π , 1 β€ π. (2)
This very basicmodel also includes the following assump-tions. Containers are inspected upon arrival at the CMC.Theavailable repair time at the CMC equals 480 minutes (an 8-hour day). Repair time per container takes a finite timewithinminimum and maximum values. No damage occurs at theCMC. The total number of containers in the model remainsfixed, that is, no purchases nor losses. Furthermore it willbe shown in the next section that the model lacks memory,that managing undamaged and damaged containers at theCMCwith a first in first out (FIFO) and other containerman-agement strategies are equivalent, that containers repairedas good as new (RAN) or to the condition prior to damage(RTPC) are also equivalent, and that identifying a damagedcontainer has a fixed probability. These assumptions capturethe essential nature of the problem while keeping the modelsimple to provide insight.
2.2. Formulation and Notation. At the initialization of themodel the first containers shipped to the CMC from theassembly plant have been away from theCMC for 5workdays;2 days at the supplier and 3 days at the assembly plant. In aFIFO system the undamaged containers will spend another 5days before leaving the CMC again. Measure time π‘ in daysand let a circuit denote the 10 days it takes an undamagedcontainer to travel a complete loop leaving and returning tothe CMC. Let π’ count the circuits traveled by a container.
International Journal of Manufacturing Engineering 3
The conditional probability of a random variableπProb (π β€ π’ + Ξπ’ | π β₯ π’) , (3)
where Ξπ’ equals one circuit, expresses the probability that acontainer needs repair after π’ circuits given that it has neverbeen repaired. Applying the geometric distribution in (1) toevaluate the probability in (3) yields
Prob (π β€ π’ + Ξπ’ | π β₯ π’) = Prob (π’ β€ π β€ π’ + Ξπ’)Prob (π’ β₯ π)
= 1 β (1 β π)π’+1 β (1 β (1 β π)π’)(1 β π)π’ = π.(4)
Equation (4) implies that a container entering the CMChas fixed a probability π of requiring repair regardless ofmany circuits the container has traveled.Therefore thismodellacks memory. Consequently a simulation of this model neednot identify individual containers nor record the numberof circuits they have traveled; also see [10]. Furthermore,whether the CMC uses FIFO or any other order for shippingand repairing containers becomes irrelevant. The lack ofmemory also implies that using RAN or RTPC for repairinga damaged container does not influence its probability ofneeding repair.
For sufficiently small values of π the daily average numberof damaged containers arriving at the CMC will be less thanthe average number of containers that can be repaired daily.Since these values negate the need of a safety stock at theCMC, the models formulated in this section will assume πhas a value that will produce more damaged containers eachday than the average daily repair rate. The following notationwill be used in our first mathematical model representing thesupply chain described in the previous subsection.
N(π‘) is the number of containers not in the repairqueue at the start of the tth day of production.π‘ is time measured in days.π is the probability of a container requiring repair onarrival at the CMC.π is the number of hours in a workday at the CMC.π΄RT is the average repair time per damaged container.π΄DP is the average number of containers used in dailyproduction and sent to the CMC.
Represent the daily number of available containers by
N (π‘)= 10π΄DP
+{{{{{{{{{
( ππ΄RTβ ππ΄DP) (π‘ β 1) , if ππ΄DP > ππ΄RT
,0, if ππ΄DP β€ ππ΄RT
.(5)
This first mathematical model assumes steady states in dailyproduction, repair rates, and no delays due to transportation.
This model remains a valid continuous approximation ofa discreet process as long as the assembly plant has π΄DPcontainers of parts for production.
When ππ΄DP > π/π΄RT, the number of containers avail-able for shipment from the CMC to the supplier decreases byππ΄DP β π/π΄RT each day until the CMC ships a steady stateof (1 β π)π΄DP + π/π΄RT containers. Since the CMC initiallyhas 5π΄DP new containers, this steady state number takes π‘CMCdays to be reached.
π‘CMC = 2 + 4π΄DPππ΄DP βπ/π΄RT. (6)
On the (π‘CMC + 1)th day and thereafter the supplierreceives (1 β π)π΄DP + π/π΄RT containers. Consequently, thenumber of containers at the supplier also decreases each dayafter the π‘CMCth day to a steady state of (1 β π)π΄DP +π/π΄RTcontainers available for shipping to the assembly plant. Thisdepletion takes another π‘π days.
π‘π = 1 + π΄DPππ΄DP βπ/π΄RT. (7)
A similar analysis of the model shows that the assemblyplant no longer maintains a steady average rate of using π΄DPcontainers on the morning when the assembly plant has lessthan π΄DP containers. Consequently the basic model endsafter another π‘DP days.
π‘DP = 1 + 2π΄DPππ΄DP βπ/π΄RT. (8)
Therefore the basic linear model ends after π‘π days whereπ‘π = π‘CMC + π‘π + π‘DP = 4 + 7π΄DPππ΄DP βπ/π΄RT
. (9)
Equations (5)β(9) provide an opportunity to plot thenumber of available containers at the CMC, supplier, andassembly plant. For example, suppose the manufacturing lifeof the product requiring shipments in containers lasts oneyear with 240 manufacturing days, a typical number thatincludes a plant shut down for plant repairs, productionenhancements, andwork holidays. Letπ΄DP = 100 containers,π = 0.05,π = 8 hours, and π΄RT = 4 hours. The dark brokenline segments in Figure 2 display the number of containersin the morning at the CMC, assembly plant, and the supplier.The gray line indicates the average number of containers usedin daily production.
With the same values for π΄DP and π, the area abovethe gray curve in Figure 3 captures the pairs of values fordamage rates, π, and average repair times, π΄RT, in (9), thatproduce positive values of π‘π for the basic linear model. Thegray curve indicates the values of π andπ΄RT in (9) that renderthe denominator equal to zero. The values of π and π΄RT onthe black curve produce a value of π‘π = 240. In Figure 2 π‘π =237(1/3). The arrow in Figure 3 points to the location of πand π΄RT used in Figure 2. The arrow shows that a damagerate of 5% requires an average repair time of slightly less than4 hours. Damage rates and repair times are key performanceindicators [11].
4 International Journal of Manufacturing EngineeringN
umbe
r of c
onta
iner
s
120 160 20080 24040Production days
1
2
3
4
5CMC
ASP
SUP
ADP
Figure 2: The continuous approximation of the basic mathematicalmodel in (5). The curves πΆMC, π΄SP, and πUP starting on thevertical axis denote the number of containers at the CMC, theassembly plant, and the supplier. π΄DP denotes the average numberof containers used in daily production.
Figure 2
π
0.1
0.2
0.3
2 3 4 5 6 7 81ART
Figure 3: Values for π and π΄DP above the gray plot produce a validbasic model in (5). On the black curve the model terminates at π‘π =240. The arrow points to values used in Figure 2.
Scheduled production and repair times vary.Nonetheless,the basic linearmodel in (5) can bemodified to accommodatethese fluctuations. The inequalities in (10) revise (5) toincorporate standard deviations πDP and πRT in productionand repair times, respectively.
10π΄DP + ( ππ΄RT + πRT β π (π΄DP β πDP)) π‘ β€ N (π‘)β€ 10π΄DP + ( ππ΄RT β πRT β π (π΄DP + πDP)) π‘.
(10)
Assume that the number of containers used for dailyproduction varies uniformly by 5% and the repair timesvary uniformly from 2 to 6 hours. Figure 4 shows the resultof simulating the daily total number of available containersin the basic model and the number of containers at theCMC, the assembly plant, and the supplier. The dashedlines above and below the line indicate the total number ofavailable containers computed from standard deviations in
Num
ber o
f con
tain
ers
80 120 160 200 24040Production days
1
2
3
4
5
6
7
8
9
10
CMC
ASP
SUP
ADP
NUM
Figure 4: A simulation of the basic model in (5). The dashed linesindicate the total number of available containers computed withstandard deviations in the probability distributions ofπ΄DP andπ΄RT.
the probability distributions of π΄DP and π΄RT. The distancesabove and below the linearly decreasing total reveal thestronger influence ofπ΄RT in the denominator of (5). Note thatin Figure 4 the number of containers at the assembly plant atthe end of each day can be less than 0.95π΄DP because each daythe morning shipment of containers will have arrived fromthe supplier.
Remarks. Overnight deliveries imply that transportationtimes do not appear in the basic model. To accommodatetransportation times and have the same safety stocks at thethree nodes in the basic model, initialize themodel by addingmultiples of π΄DP containers based on transportation times,measured to the nearest average whole day, to the recipientsin the respective shipments between nodes. On the otherhand, depending on the transportation times, consideringcontainers on route as rolling safety stock may not require anincrease in safety stocks at the shipment recipients.
The basic model yields linear depletion rates of availablecontainers at the three locations in the loop. With thevariables, lengths of a production run, hours in a workday,a steady average daily production schedule, and a set of safetystocks, the basic model yields easily computable values fora fixed probability of damage to a container and averagerepair times that produce valid mathematical models. Thegray plot in Figure 3 shows that the asymptotic damage rateslightly less than 3% keeps themodel going to term (240 days)regardless of the average repair rate. The black plot showsthe pair of probabilities and repair times which the modelterminates at 240 days. The simulations of the basic modeland other models in this paper use SRAND, the randomnumber generator supplied in UNIX operating systems andsuggested in [12].
International Journal of Manufacturing Engineering 5
The basic model does not address stolen containerscoveted for their lucrative scrap steel. Nevertheless the modelcan accommodate lost and stolen containers next to 10π΄DPin (5) by adding terms with negative values for the numberof containers at supplier, the assembly plant, and the CMC.Attribute losing containers during transportation betweenlocations to the point of departure.
3. A Length of Service Model
The punishment endured by engineered containers duringshipments and in plant mishaps requires inspections andrepairs. Unlike the basic model in the previous section, thissections assumes that the probability of damage requiringrepair increases over time. The probabilities for requiringrepair depend on the number of circuits each container hastraveled since its delivery from its manufacturer or since itsmost recent repair. In this section containers that do notrequire repair upon arrival and inspection at the CMCwill besent to the supplier at the end of the day. Repaired containerswill be considered new containers.
With a steady average daily production requiring π΄DPfull containers at the assembly plant, initialize the model asfollows. Before the first day the CMC has shippedπ΄DP emptynew containers to the supplier, the supplier has shippedπ΄DP full containers to the assembly plant, and the assemblyplant has shipped π΄DP used empty containers to the CMC.Consequently at the beginning of the first day the containersfrom the assembly plant arrive at the CMC after having madeone circuit around the loop from the CMC to the supplier tothe assembly plant and returned to the CMC.
Each container arriving at the CMC has a probabilityππ of requiring repair, where π denotes that the containerhas completed π circuits. ππ, a decreasing function of π, willeventually produce more damaged containers than can berepaired each workday. Along with the containers that didnot require repair upon arrival and inspection at the CMC,the CMC sends to the supplier the containers repaired duringthat day. If these containers do not total π΄DP containers, theCMC compensates by adding containers from its safety stock.
Represent the daily shipments to the supplier with thefollowing additional notation.
ππ is the probability that a container requires repairgiven that it has traveled π circuits since being new orsince its repair to be new.
ππ is the number of containers sent to the supplierfrom the CMC on the nth day.
ππ is the number of new containers sent to thesupplier on the nth day.
π·π is the number of containers deducted from thesafety stock at the CMC on the nth day.
The following equations assume that the daily shipmentsof containers leaving the CMC remain together each day astheymove from theCMC to the supplier to the assembly plant
and return to the CMC. For simplicity assume that π1π΄DP >π/π΄RT.
π1 = (1 β π1) π΄DP + ππ΄RT+ π·1, (11)
where
π·1 = π1π΄DP β ππ΄RT,
π1 = π1π΄DP.(12)
The shipments to the supplier emulate (11) and (13) anduntil the 7th day when, at the end of the day, the CMC shipsthe used containers requiring no repairs that arrived on the1st day at the CMC. On the 7th day
π7 = (1 β π1) (1 β π2) π΄DP + (1 β π1) ( ππ΄RT+ π·1)
+ π·7,(13)
where
π·7 = π2 (1 β π1) π΄DP + π1 ( ππ΄RT+ π·1) β ππ΄RT
,π2 = (π2 (1 β π1) + π21)π΄DP.
(14)
Every 6 days the composition of the containers sent the to thesupplier changes. In general,
ππ = (βπ/6ββπ=0
(βπ/6β+1βπβπ=1
(1 β ππ))π6πβ5)π΄DP
+ πβ(πβ1)/6β+1π΄DP,(15)
where βπ₯β equals the largest integer less than or equal to x, andthe number of used containers sent to the supplier satisfies
ππ= (βπ/6ββπ=0
πβπ/6β+1βπ(βπ/6β+1βπβπ=1
(1 β ππ))π6πβ5)π΄DP
+ πβ(πβ1)/6β+1π΄DP, π β₯ 1, πβ5 = 1.(16)
ππ includes used containers, repaired containers, and contain-ers taken for the safety stock. From generalizing (12) and (14)the daily number of containers depleted from the safety stockcan be expressed as
π·π = πβ(πβ1)/6β+1 β ππ΄RT. (17)
For some probability distributions πππ΄DP < π/π΄RT forinitial values of π. For these values of π, π·π and ππ will bezero.
As a mathematical model, (15)β(17) remain valid untilthe CMC no longer can send π΄DP containers to the supplier.
6 International Journal of Manufacturing Engineering
In order to extend the model, the supplier can make up theshortfall of sending π΄DP containers to the assembly plant byusing containers available from its safety stock. When thesafety stock at the supplier ends, the same strategic decisioncan be employed at the assembly plant until the extendedmodel fails.
To compare this model to the basic model we createprobabilities naturally related to the geometric distributionused in the simulation of basic model. Let ππ = 1 β ππ,π = 0.995 so that the probability of repair for a containerarriving at the CMC after π circuits increases with π. Thevalues of π΄DP, π΄RT, andπ and the safety stocks remain thesame for both models.
In Figure 5 the gray horizontal line benchmarks π΄DPwith respect to the other plots. The black plots show contin-uous approximations of the daily total number of availablecontainers and the available containers at the supplier, theassembly plant, and theCMC.The initialization of this servicedependent model starts on the 0th day and ends on the 227thday. The darker gray plots present the result of a simulationof this service dependent q-model.
Figures 4 and 5 show that the number of availablecontainers at the supplier and assembly plant behaves sim-ilarly with the lengths of their horizontal portions differingin length and the curves connecting them in the servicedependent model. The most noticeable difference occurs inthe graphs of available containers at the CMC where theirplot has a curved decrease between two linear portions. Thelengths of the horizontal portions of the various plots andthe life expectancy of the models depend highly on the repairtimes of damaged containers.
TheWeibull distribution [13, 14] has earned popularity inmodeling engineering reliability and failure analyses duringa productβs life. For this reason we examine the servicedependent model with the Weibull distribution. A randomvariable π has a Weibull distribution with shape and scaleparametersπΌ > 0 andπ½ > 0when the probability distributionfunction π(π‘; πΌ, π½) ofπ can be expressed as
π (π‘; πΌ, π½) = {{{{{πΌπ½ ( π‘π½)
πΌβ1 πβ(π‘/π½)πΌ , π‘ β₯ 00, π‘ β€ 0. (18)
Integrating π(π‘; πΌ, π½) yields its cumulative distribution func-tion πΉ(π‘; πΌ, π½),
πΉ (π‘; πΌ, π½) = {{{1 β πβ(π‘/π½)πΌ , π‘ β₯ 00, π‘ β€ 0. (19)
Applying the cumulative distribution πΉ(π‘; πΌ, π½) in (19) toevaluate the conditional probability in (4) yields
Prob (π β€ π’ + Ξπ‘ | π β₯ π’) = Prob (π’ β€ π β€ π’ + Ξπ‘)Prob (π’ β€ π)
= (1 β πβ((π’+Ξπ‘)/π½)πΌ) β (1 β πβ(π’/π½)πΌ)
1 β (1 β πβ(π’/π½)πΌ)= 1 β πβ((π’+Ξπ‘)/π½)πΌ+(π’/π½)πΌ .
(20)
Num
ber o
f con
tain
ers
240120 160 2008040Production days
1
2
3
4
5
6
7
8
9
10
CMC
ASP
SUP
ADP
NUM
Figure 5: The continuous approximation and a simulation of theservice dependent model in (11)β(17) for ππ = 1 β ππ and π = 0.995.The curves πΆMC, π΄SP, and πUP starting on the vertical axis denotethe number of containers at the CMC, the assembly plant, and thesupplier.π΄DP denotes the average number of containers used in dailyproduction.
This Weibull service dependent model applies the prob-ability in (20) to every container arriving at the CMC. Notethat Weibull models normally set repairs to βcondition priorto failureβ by subtracting one from the number of circuitstraveled by a repaired container. The RAN condition in thispaper resets the number of circuits to zero after repairing adamaged container.
The black plots in Figure 6 evaluate the continuousapproximations of service dependent mathematical modelwith the Weibull distribution for πΌ = 10.0 and π½ = 3.5recommended in [15]. An interval of [0, 7] for the Weibullcumulative distribution function was divided into 40 subin-tervals to accommodate the 6-day circuit that containers needto return to the CMC. The plots maintain the general shapeof the plots in Figures 4 and 5 except for the more precipitousdecreases. The service dependent mathematical model usingtheWeibull distribution starts on the 0th day and ends on the223th day. The lighter gray horizontal line benchmarks π΄DPwith respect to the other plots. The darker gray plots presentthe result of a simulation of this Weibull service dependentmodel.
4. Remarks
Other important factors also influence the availability ofengineered containers. For example, there may be more thanone supplier, transportation may take more than overnight,and, most importantly, there may be different probabilitydistributions in repair time and scheduled production.Thesefactors can be incorporated into the basic steady state model
International Journal of Manufacturing Engineering 7N
umbe
r of c
onta
iner
s
80 120 24040 160 200Production days
1
2
3
4
6
5
7
8
9
10
CMC
ASP
SUP
ADP
NUM
Figure 6: The continuous approximation and a simulation of theservice dependent model in (11)β(17) with the Weibull distributionfor πΌ = 10.0 and π½ = 3.5. The curves πΆMC, π΄SP, and πUP starting onthe vertical axis denote the number of containers at the CMC, theassembly plant, and the supplier. π΄DP denotes the average numberof containers used in daily production.
in a number of ways. The following paragraph provides anexample.
By averaging the contributions of the suppliers to theprobabilities of their containers requiring repairs, the basicmodel can still use one value for π while accommodatingmultiple suppliers. Suppose k suppliers send full containers tothe assembly plant and the ith supplier sends an πΌπth fractionof the total number of containers. If a container from the ithsupplier has a ππ probability of needing repair when observedat its arrival at the CMC then let
π = πβπ=1
πΌπππ, 0 < πΌπ, 0 < ππ < 1, πβπ=1
πΌπ = 1. (21)
Equation (21) which lumps multiple suppliers as one suppliersuggests adding safety stocks based on the same percentagesmultiplied by transportation days to morning arrivals.
5. Summary
This paper explored the availability of highly engineeredcontainers for shipping parts from suppliers to assemblyplants by taking into account that damaged containers requirerepair and therefore stay temporarily out of service.
The paper first developed a basicmodel for a supply chainconsisting of a supplier, an assembly plant, and a containermaintenance center (CMC) used for inspection and repairof damaged containers. This basic model assumed steadyaverage daily production and repair rates and a Bernoulliprobability distribution for requiring repair. It also assumed
next day deliveries of containers between each facility. Themodel provided feasible pairs of values for the probabilityof repair and average repair time. Plotting the continuousmodel illustrated the number of containers at each facilityas a function of production days, given the average dailyproduction, probability of repair, and average repair time.The model offered straight forward and easily calculatedparameters. The basic model also accommodated variabilityin scheduled production and repair times through theirstandard deviations. An illustrative simulation used uniformdistributions for the production and repair parameters. Thisbasic model exhibited linear behaviors.
Another model introduced the probability of a containerrequiring repair as an increasing function of circuits acontainer has traveled. The generality of this model enabledalternative probability distributions of circuit based damageand repair. Examples with a geometric based and a Weibullbased probability distribution featured linear and nonlinearcharacteristics. In particular, aWeibull distribution producedmore precipitous decreases in the availability of containers atthe three locations in the supply chain.
The originalmodel and its extensions which allowed vari-ations in production and damage rates provided analyticaltools to observe how the availability of engineered containerscan change over time at the supplier, assembly plant, andCMC. These models depended on easily scalable repair andproduction rates and gave insight into managerial decisionsregarding the fleet size of shipping containers needed in asupply chain.
These models can also show where to optimize cash flow.A minimal number of containers plus a small safety stockcan reduce the initial investment. Only as the number ofdamaged containers under repair grows should additionalcontainers be purchased. Thus the models allow makingimportant decisions with a few basic parameters.
Competing Interests
The authors declare that they have no competing interests.
Acknowledgments
The authors thank Michael Wincek for his advice and TaehoYang for his preliminary contributions.
References
[1] J.-X. Dong and D.-P. Song, βContainer fleet sizing andempty repositioning in liner shipping systems,β TransportationResearch Part E: Logistics and Transportation Review, vol. 45, no.6, pp. 860β877, 2009.
[2] S. T. Choong, M. H. Cole, and E. Kutanoglu, βEmpty containermanagement for intermodal transportation networks,β Trans-portation Research Part E: Logistics and Transportation Review,vol. 38, no. 6, pp. 423β438, 2002.
[3] Y. Long, L. H. Lee, and E. P. Chew, βThe sample averageapproximation method for empty container repositioning withuncertainties,β European Journal of Operational Research, vol.222, no. 1, pp. 65β75, 2012.
8 International Journal of Manufacturing Engineering
[4] K. Shintani, R. Konings, and A. Imai, βThe impact of foldablecontainers on container fleet management costs in hinterlandtransport,β Transportation Research Part E: Logistics and Trans-portation Review, vol. 46, no. 5, pp. 750β763, 2010.
[5] P. Jula and R. C. Leachman, βLong- and short-run supply-chainoptimizationmodels for the allocation and congestionmanage-ment of containerized imports from Asia to the United States,βTransportation Research Part E: Logistics and TransportationReview, vol. 47, no. 5, pp. 593β608, 2011.
[6] L. Coslovich, R. Pesenti, and W. Ukovich, βMinimizing fleetoperating costs for a container transportation company,β Euro-pean Journal of Operational Research, vol. 171, no. 3, pp. 776β786,2006.
[7] K. H. Kim and H. B. Kim, βSegregating space allocationmodels for container inventories in port container terminals,βInternational Journal of Production Economics, vol. 59, no. 1β3,pp. 415β423, 1999.
[8] C. Hanebeck and M. Lunani, RFID-Enabled Returnable Con-tainer Management: Solution to a Chronic and Wasteful Auto-motive Industry Problem, IBM Global Business Services, 2008.
[9] A. M. Law and W. D. Kelton, Simulation Modeling & Analysis,McGraw-Hill, New York, NY, USA, 2nd edition, 1991.
[10] P. G. Hoel, S. C. Port, and C. J. Stone, Introduction to ProbabilityTheory, Houghton Mifflin, Boston, Mass, USA, 1971.
[11] Savi Technology, βBest practices in improving container man-agement and increasing supply chain efficiencies,β White Paper,2014, http://www.savi.com.
[12] S. K. Park and K.W. Miller, βRandom number generators: goodones are hard to find,β Communications of the ACM, vol. 31, no.10, pp. 1192β1201, 1988.
[13] G. Bolch, S. Greiner, H. de Meer, and K. S. Trivedi, QueueingNetworks and Markov Chains, John Wiley & Sons, New York,NY, USA, 1998.
[14] A. M. Mood, F. A. Graybill, and D. C. Boes, Introduction tothe Theory of Statistics, McGraw-Hill, New York, NY, USA, 3rdedition, 1974.
[15] J. L. Devore, Probability and Statistics for Engineering and theSciences, Brooks/Cole, Belmont, Mass, USA, 4th edition, 2004.
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Journal ofEngineeringVolume 2014
Submit your manuscripts athttp://www.hindawi.com
VLSI Design
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Modelling & Simulation in EngineeringHindawi Publishing Corporation http://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
DistributedSensor Networks
International Journal of