James h Bookbinder

JAMES H. BOOKBINDERCAROLYNN I. BARKHOUSE

An Information System forSimultaneous Consolidation of

Inbound and Outbound Shipments

The day-to-day operations of a modern busi-ness are becoming more and more dependenton computers and information systems. Avail-able information must be efficiently and effec-tively managed in order for a company to sur-vive. Information systems, databases, andfourth-generation computer and simulation lan-guages have all played an integral part in theutilization and controi of infonnation.

The transportation and logistics fields are noexception. A logi.stics information .system is averitable necessity for competent materialsmanagement and distribution. A large numberof organizations implementing such systemshave noted significant cost savings by the logis-tics departments. One area with great potentialfor cost savings is the consolidation of inboundand outbound shipments. A system that canaccommodate pickups and deliveries simulta-neously could produce substantial market ad-vantage for the company.

This article discusses the nature and capabili-ties of a conceptual system, denoted by LIS.whose main thrust is the consolidation ofinbound/outbound shipments. The article willpresent a history of information systems in thelogistics field, review the logistic cycle, set thebasis for this LIS model, and consider therequirements for simultaneous consolidation.outlining the capabilities available through thisLIS.

The evolution of just-in-time manufacturingis rapidly becoming a way of doing businessfor many organizations. The article presents a

Mr. Bookbinder is professor of management sciences,University of Watertoo. Waterloo, Ontario, Canada N2L3GI; A/s. Barkhouse is systems speciatist, CanadianPacific Railway, Toronto, Ontario, Canada.

The authors are grateful to the Natural Sciences andEngineering Research Council of Canada (Grant OGP05292) and to the Manufacturing Research Corporationof Ontario, respectively, for support of their work.

brief description of that methodology, and thevarious unique requirements for LIS in such anenvironment. The introduction of electronicdata interchange and its impact on logisticsfunctions is examined, and an illustrative exam-ple is given which demonstrates how LISwould work. Later sections compare the jointconsolidation schedule, i.e., that resulting fromLTS. to the pertbrmance under separate consoli-dation, discuss potential future enhancements,and offer conclusions.

LOGISTICS INFORMATION SYSTEMS:

PAST AND PRESENT

Much research has been done concerning theopportunities, techniques, and advantages ofshipment consolidation. For example, Buffa'has written a series of articles detailing variousapproaches to inbound consolidation methods.A number of other authors have treated out-bound consolidation. Vollmann, Berry, andWhybark- have discussed Distribution Re-quirements Planning (DRP), a modification ofMaterials Requirements Planning to handle dis-tribution planning. Tyworth, Cavinato, andLangiey' describe general forms and frame-works for outbound shipment consolidation.However, at the time of writing, only onepaper" could be found which considered asimultaneous consolidation of inbound and out-bound shipments. In that work, consolidationoccurs at a transportation terminal, rather thanon routes which mix inbound and outboundmovements of goods.

Various authors have examined the growinguses of information systems and computerapplications in logistics. Bushnell. Low. andWiley' summarize examples of network models.Gustin" studies implementation trends of U.S.manufacturing and merchandising companies intransportation and distribution management,

TRANSPORTATION JOURNAL Summer

discovering numerous areas of increased com-puter utilization. Such systems, besides bring-ing efficient and effective methods of informa-tion control, can provide financial benefits.Calantone and Morris.' through examination ofcompanies implementing decision support sys-tems, report 43 percent of companies realizeddecreased costs and 78 percent realizedincreased cash flow due to computertechnology.

Other authors have recognized the need forand potential advantages of logistics informa-tion systems. Tyworth, Cavinato, and Langley"acknowledge information systems as an inte-gral part of the control process, presenting acomprehensive view of the requirements ofsuch systems for transportation. Langley" con-siders how the logistics area can make use ofthe information available through the latesttechnology. Haley and Krishnan'" propose asystem approach which integrates all aspects ofthe logistics cycle from marketing to customerservice. Langevin and Saint-Mlcux" describean interactive decision support system forphysical distribution planning through whichthe user designs delivery routes based on per-sonal knowledge.

The majority of these systems contain mod-ules which perform some form of consolida-tion, either inbound or outbound, using stan-dard operations research techniques and meth-ods. Casco. Golden, and Wasi!.'- on the otherhand, examine vehicle routing with backhauls.The algorithms and case studies presented con-centrate on adding backhaul shipments to pre-dominantly "delivery" routes. These methodseither attempt to delay pickups as much as pos-sible, placing them at the end of the route, orconsider deliveries a higher priority than pick-ups. No system or methodology could be foundwhich treated inbound and outbound shipmentsas equally important for vehicle routing.

PROBLEM DEFINITION AND ASSUMPTIONS

Many businesses have utilized consolidationopportunities to their advantage. Since consoli-dation of inbound or outbound shipments canresult in significant savings, a company thateffectively combined both types of shipmentson a common vehicle could potentially savemoney through greater efficiency in routingtrucks.

With increasing frequency, recognized tech-niques and heuristics for complex operationsresearch methodologies are being programmedfor computer manipulation. Consolidation is noexception. The vast amounts of information,multiple routes, and number of shipments canbe handled more efficiently by a computer.Therefore, when considering simultaneousinbound and outbound shipment consolidation,a computer algorithm is the natural way to go.This article attempts to determine the varioussteps and modules that would be needed to per-form this unique form of consolidation.Throughout, we emphasize the businessaspects and omit the mathematical details.

Overview ofthe Logistics Cycle

Contrary to what may be popular opinion ingeneral management circles, logistics coversmore than the physical distribution of goodsfrom point of origin to point of destination.Logistics has been defined as:

Ihc process of planning, implenienling and control-ling the etTicicnt tlow and storage of raw maierials,in-process inventory, finished goods, and informa-tion from the point of origin to point of consumption,fur the purpose of meeting the customer servicegoals of the organisation at an acceptable cost.' *Other authors have given comparable defini-

tions, covering a large number of businessfunctions, most outside of what is traditionallyconsidered to be distribution.

The logistics cycle begins with the marketingdepartment, which advertises the goods avail-able. This, triggers an order from a potentialcustomer, perhaps through contact with a sales-person. In order to fill this request, purchases ofraw materials must be made, the productionprocess started, and transportation of the goodsarranged. This, hopefully, puts in motion a cyclewhich never ends until the business ceases toexist. The effectiveness of the cycle is measuredprimarily by a customer service goal set by thecompany. Each of the various functions playsan important part of the logistics efforts of thefirm. Each must therefore be integrated toachieve a productive system.

LIS RequirementsThe above-mentioned functions can all be

adapted to a computer environment through theuse of information systems. Order processing,production planning, and inventory control

1993 IN FORMA TION S YSTEM

have all been transferred to computer systemsby successful companies. Today, many vehiclerouting programs are available to aid in deter-mining distribution paths and networks.Unfortunately, all of these modules are. for allintents and purposes, designed independentlyof one another. Interface between the variouscomponents is essentially a last-minute consid-eration instead of an initial element in thedesign. Development of an efficient logisticsinformation system requires a coordinatedeffort among all areas.

Although a thorough system definitionwould inciude specifications for all elements inthe cycle, their inclusion would exceed thescope of this report. Therefore, all future refer-ence to this information .system will concernthose functions directly related to shipmentconsolidation. It will be assumed that this .sys-tem is simply a sub-module of a larger system.

Model Assumptions

In order to determine the specifications lorour information system, certain assumptionsare made to keep the system manageable inscope;1) The company is a manufacturing firm. The

terminology used in the remainder of thisarticle will be that of a manufacturing firm.

2) Transportation is by truck. Shipmentsrequiring other types of transportation willbe excluded from consideration.

3) The company manages its own truck fleet,which is large enough to handle the flow. Inorder to consider joint con.solidation, thecompany must have control over a reason-able number of pickups and deliveries, andtherefore must be able to perform theseduties on its own.

4) All trucks (trailers) arc identical in size,shape, and capacity, and contain multipledoors. Multiple doors are required to alloweasier access to the interior of the trailer.increasing the tlexibility to mix pickups anddeliveries.

5) There is sufficient compatibility betweenproducts that they can be shipped together.These goods have identical storage require-ments, such as those concerning tempera-ture. Incompatible items will be eliminatedfrom consideration.

6) All suppliers and customers are within a

given radius of the plant. This is to facilitatethe assignment module.

7) Access to vehicle-routing software is assumed.8) The objective is to minimize cost while

maintaining a desired customer service level.Note that in most cases, these assumptions willnot appreciably limit the applicability of oursystem to a manufacturing firm which operatesa private fleet.

LISThe proposed system, denoted by US. will

require certain facilities and capabilities inorder to provide meaningful and reliable data.The following presents a description ofthe var-ious modules and their respective components.

General Overview

LIS is a conceptual model intended for useby organizations with inbound and outboundshipment requirements. It is designed to coordi-nate a company's pickups and deliveries inorder to exploit potential savings due to consol-idation: fewer trucks and drivers, improvedcustomer service, and more reliable receipt ofraw materials. LIS maintains and controls datathat is relevant and up-to-date, invoking a sortalgorithm to assign shipments to trucks andprovide improved routing schemes. The systemalso allows overrides where and when the userdeems necessary. Post-analysis capabilities andperformance measurements are provided.

LIS is a menu-driven system which facili-tates access to the various modules. Each mod-ule is self-contained while maintaining overallsystem consistency and integration. The fol-lowing sections present more detailed descrip-tions ofthe relevant components.

Inbound Shipments

Inbound shipments refer to those items con-sidered as inputs to manufacturing, such as rawmaterials and components. These items, nor-mally ordered by the purchasing department,will be required at specific times and locationsand in exact quantities throughout the produc-tion process. More timely receipt of goods willenhance the reliability of manufacturing, andshould help improve customer service.However, to achieve this goai, certain elementsof information will be required:

TRANSPORTATION JOURNAL Summer

1) Date and time that goods are available forpickup

2) Date and time tbat goods are required forproduction

3) Time window within a given day forpickup at the supplier

4) Type and location of tbe supplier's load-ing facility

Outbound ShipmentsThe company's outbound shipments are of its

finished goods. These items, generally soldthrough the joint efforts of marketing and sales,will be required at specific times and locations,and in tbe quantities ordered, either as goods forresale or as inputs to another production proc-ess. Similar to the inbound case, new informa-tion will be needed to facilitate this endeavor:

1) Date span for delivery at customer2) Date and time that goods are available for

shipment3) The customer's time window for drop-off

within a given day4) Unloading facility type and location at

customer

Sort AlgorithmThe sort algorithm is employed to develop a

merged list of pickups and deliveries, therebysetting priorities for vehicle routing. A sortcould be invoked whenever desired. However.a regular time interval would be most efficient,with emergency runs performed as necessary.Tbe default interval for sorting is one week.This is reasonable since the company will con-tinually accumulate information and shouldbave detailed knowledge of required inboundand outbound shipments for the coming week.That initial plan, although subject to futurechanges, provides the company with a basis forpersonnel and maintenance scheduling and thelike. It can also alert the user to possible prob-lems or anomalies which, if remedied, couldpotentially result in additional savings.

Using this weekly schedule as a base, newruns of the algorithm would be initiated on atwo- to three-day cycle for further refinementas new data are made available. The last stepwould be a final daily schedule, kept firmexcept in situations deemed emergencies by thelogistics department.

Initial information is required before the

actual algorithm is invoked. These data concernany environmental changes related to trucksand road conditions. Examples include:

1) New truck acquisition2) Truck disposal3) Scheduled truck maintenance

(These elements could appear on the sort-rou-tine screen as modifiable inputs concerning thenumber of available vehicles and maintenance.Any changes made at this time could auto-matically update tbe data for future runs ifappropriate.)

4) Scheduled road work by city repaircrews, wbere tbis would affect travel time

5) Scheduled new road construction andexpected completion dates(Tbe vehicle-routing software would accommo-date the above elements in some fashion,assuming a geographical database were avail-able. Should such a database be unavailable,the above information would have to be incor-porated tbrough projected travel times for givenroutes.)

The sort algorithm can be divided intounique phases as outlined below.

Phase I: Elimination Phase

Tbis phase deletes tbose shipments whichcannot be handled under the stated assump-tions:1) Shipments for which there already exists

alternative transportation, determinedthrough contracts and the other previousarrangements

2) Shipments which require different modes oftransportation, such as air freight

3) Shipments to or from destinations outsidethe given radius of consideration

4) Shipments which cannot be carried on tbetrucks due to differing storage and/or trans-portation requirements

Elimination of the above will yield a list ofshipments suitable for assignment and routing.

Phase II: Sort Procedure

This list resulting from Phase I will now besorted. The user can determine which relativepriorities are to be applied for routing. Sortingcan thus be based on any of the following:I) Preferred vendors/customers. Certain ven-

dors or customers may be deemed moreimportant tban others, and therefore deserve

INFORMATION SYSTEM

initial consideration in order to maintaindesired service parameters.

2) Vendor/customer location3) Dateand time of pickup or delivery4) Sizeof shipment (dimensions)5) Weight of shipment6) Shape of shipment7) Time window for delivery or pickup

Phase III: Vehicle-Routing

The sorted list from Phase II is used as inputto the vehicle-routing software. Output will betbe recommended routing. That output shouldalso report any problems such as unassignedshipments. This routing can be saved and re-trieved at a later date, if desired, for study orevaluation of user-determined changes.

Performance Measures

Key to the evaluation of any system is theanalysis of performance. Feedback on day-to-day operations, as well as on the extent towhich company goals and objectives are beingmet. is essential if an organization is to exploitall areas with potential benefits. To this end,LIS will provide the ability to analyze and eval-uate many aspects ofthe logistics system.

The main ingredient to successful implemen-tation is meeting the company's service levels.Care must be taken to determine if these pre-specified levels are being attained. Rushton andOxley'" discuss customer service and enumeratesome of tbe more popular elements whichcould be available through LIS. Distributionmanagement provides many other areas wheremeasurement of performance is advantageous.Tyworlh. Cavinato, and Langley'' furnish a listof control measures which could be used tomonitor traffic. Continual evaluation of .servicelevels will aid in early detection of anydecrease in effectiveness, and hence lead tomore timely correction.

Private carriage requires attention to certainpertbrmance aspects to ensure efficient opera-tion. Parameters such as tieet utilization andload factor are important indicators regardingtruck usage. Driver performance can be evalu-ated through route sheets or tachographs. LISwill contain methods for analyzing such data.

Because of the unique shipping consolidationperformed by LIS, a valuable measure would

be the comparison of these costs to those thatwould have been incurred using the previoussystem. Tbis will provide concrete evidence ofsavings incurred.

Post-Analysis Module

Several opportunities, such as those outlinedby Firth et al,'" although detectable throughbasic methods, may be quite difficult to incor-porate. Eor example, suppose stops were tooconcentrated in one portion of the day.Although LIS could heip in recognizing thisproblem, its resolution would require a morein-depth approach. Specific action by the userwill be required, but LIS can help a great dealin this regard.

LIS will thus provide a somewhat innovativegraphical analysis for those organizations whowish it. Each suggested routing will be avail-able for terminal display via the geographicdatabase. The user will be allowed to inputchanges to evaluate their impacts on mileage,service levels, and times from origin to destina-tion. The magnitude of these changes willdetermine the difficulty of estimating theeffects; certain differences will require invok-ing the sort algorithm for a test run. while oth-ers will simply need a few minor calculations.This facility can be used as well to input last-minute alterations when time does not permit arun ofthe entire system.

By employing the graphical output, the usercan also evaluate various routings for robustnessand ease of modification or addition. This isparticularly valuable if the week-to-week or day-to-day shipping requirements are fairly consis-tent. The organization may be able to select analternate route which on the surface may costmore, but may be robust enough to handlechanges at lower cost than the suggested routing.

JIT-LIS INTEGRATIONPioneered in Japan, just-in-time (JIT) manu-

facturing ideology has progressively beenadopted by many North American firms. Anincreasing number of companies are recogniz-ing the benefits of lower inventory and moretimely delivery of higher-quality goods. Thestringent demands of tbe methodology are moreeasily and accurately met by computer utiliza-tion and information system development. Forthe company implementing JTT principles, LiS

10 TRANSPORTATION JOURNAL Summer

will prove a valuable investment.

JIT OverviewOver tbe past twenty years, the Japanese

have developed a system designed to meet twomajor goals: reduction of inventory and 100percent quality of all parts, subassemblies, andproducts. At each stage in the manufacturingprocess, each component or subassemblyarrives just in time for addition or assembly, tbefinal product being completed just in time to besold. Consequently each shipment, inbound andoutbound, must be of tbe highest quality; nodefects or missing items can be tolerated.However, such stringent demands cannot bemet absolutely; they merely represent an idealwhich one strives to attain. Schonberger'' hasdefined tbis ideal as:

all materials IbeingI in active use as elements ofwork in process, never at rest collecting carryingcharges. 1JIT| is a hand-to-mouth mode of operation,with production and delivery quantities approachingone single unit — piece-by-piece production andmaterial movement.

Each refinement of a production process leads acompany one step closer to achieving this ideal.

Tbe effectiveness of tbe JIT methodologyhas been documented by many companies andauthors. For instance, Toyota has shown greatsuccess through JIT. As described byMonden,'" that firm's goal is to "produce thenecessary units in the necessary quantities attbe necessary time." Toyota's phenomenal suc-cess bas prompted other organizations to inves-tigate the principles and to change their percep-tion of the manufacturing process. NorthAmerican companies are attempting to emulatetheir Japanese counterparts in the bope ofreducing costs und hence increasing protlts.

JIT requires cooperation between supplierand customer. Each must be committed if theimplementation is to be a profitable success.An effective JIT approacb requires delivery ofsmaller but more frequent shipments, and theseshipments are to be bigh-quality merchandise.More accurate and timely information must beexchanged if tbe system is to benefit both par-ties. Suppliers and customers must be in con-stant contact to ensure the production processdoes not stop because of incorrect shipments.

LIS RequirementsThe integration of JIT principles into this LIS

requires special considerations. Bookbinder andDilts'" describe a conceptual model of a JITlogistics information system which examinessome of these issues. Although their model islimited to inbound shipments, the ideas are stillrelevant to the LIS presented in this article.While discussion of those concepts is tangentialto the present subject, namely inbound/out-bound consolidation, the following is a briefexamination of tbe changes necessary to adaptthe LIS of the previous section to a JIT environ-ment. We note tbat not all shipments need to beon a just-in-time basis.

As the first JIT modification lo LIS, datarequirements for inbound shipments will nowbe more precise. Particular dates and times thatitems are needed for production will be speci-fied. An exact drop-off location at the manufac-turing plant must be given for each incomingitem. The time necessary to move items fromthe unloading facility to this location should betaken into consideration.

Changes to outbound shipment informationwill concern the date and time that goods areavailable for delivery; JIT should more accu-rately pinpoint this time. It should be notedthat, due to tbe nature of JIT. it may be advis-able to invoke the sort procedure with relaxedpickup time constraints prior to contractualnegotiations with prospective suppliers. Thisless-constrained schedule may aid in determin-ing desirable times for pickups at the suppliers,so as to improve the vehicle routing as mucb as

possible.Because JIT operates on a more stringent

schedule, the sort procedure will have to be car-ried out at shorter time periods. The defaultinterval is two or three days, to allow tbescheduling of trucks, personnel, and mainte-nance. However, as last-minute informationconcerning the quantities and exact timerequirements for each item are relayed to tbesupplier, the sort procedure can be invoked bythe organization at moments closer to tbe dead-line. Performance measures and post-analysiscapabilities will still be available, and will con-tain those elements mentioned earlier. Thegraphical analysis may prove invaluable. Theability of this module to consider alterations tothe schedule could be perfectly suited to a JITenvironment, where last-minute changes andemergencies could be detrimental or fatal to theproduction process.

1993 INFORMATION SYSTEM

EDl-LIS INTEGRATION

Electronic Data Interchange (EDI) is modify-ing the way many companies do business, byeliminating verbal and written communication.EDI allows a direct communication linkbetween different computer systems, thusincreasing the speed and accuracy of informa-tion transmitted between companies. Datatransferred by a per.son takes time to reach theinterested party and hence can be out-of-dateand irrelevant. By lessening the need for humanintervention, organizations can exchangeimportant information on a more timely basis,making it both accurate and up-to-date.

The logi.stics and transportation fields haveled the way in introducing EDI to the everydaybusiness world. The vast amount of data trans-ferred between organizations involved in thelogistics cycle made EDI a virtual necessity forsuccess. Lambert and Stock-'* present a list ofEDI usage which demonstrates the degree towhicb it has been applied in the transportationarea.

EDI provides many benefits beyond tbeincreased speed and accuracy of information.

such as decreased transit times and the lowerchance of human error. The grocery businesshas estimated direct cost savings of $68 mil-lion, indirect savings of up to $250 million, andefficiencies in areas such as purchasing andinvoicing in excess of $500 million per year.^'"Eigures of this magnitude warrant considerationof the use of EDI in all areas of business wheredata must be relayed between two or morecompanies.

Tbe joint consolidation of inbound and out-bound shipments requires immense cooperationbetween the firms involved. Data must flowfreely and on a timely basis. EDI is thereforethe perfect companion to LIS. With EDI theflow of information is automatic and up-to-date. Parties can be notified immediately of anychanges in arrival pattems, shipment quantities.or other relevant matters. If these changes war-rant new routings, logistics personnel would benotified at once.

The demanding time requirements of JITprovide a perfect environment for the imple-mentation of EDI. Purchase orders placed with-in hours of the deadline require quick and accu-

Table la. Shipment List for Routing

Customer/Supplier

C,

C.

cc.c

s,S:

s.S4

Date forPickup/Delivery

Dayi

Dayi

Dayi

Day>

Day:

Day;

Dayi

Dayi

Dayi

Day:

TimeWindow

9:00-10:15

9:00-12:00

2:30- 5:00

9:00-11:00

10:30-12:30

11:00- 1:00

9:00-11:00

2:00- 3:30

11:00- 1:00

3:00- 5:00•

StopTime

(in mins)

51

52

30

51

47

42

30

20

32

32

Product

Prod.

Prod,

Prod,

Prod,

Prod,

Prod.

Prod:

Prod:

Prodi

Prod:

# of Units

70

75

25

70

60

50

45

20

50

50

ShipmentWeight(in lbs)

1400

1500

500

1400

1200

1000

1125

500

1250

1250


Figure 1. Network of Customers and Suppliers

rate communication between supplier and cus-tomer. Tbe constant contact and on-line com- 4)munication available through EDI is suited tosuch an environment. EDi can enhance thebenefits of JIT through proper integration.

ILLUSTRATIVE EXAMPLE 5)

This section presents a numerical illustrationof LIS and its capabilities. The example is sim-ple in nature, to allow understanding ofthe var-ious aspects and nuances of the system. Theapplication is structured as a base problem withsub-problems designed to cover special cir-cumstances. Certain assumptions were made:1) The manufacturing environment is non-

JIT.2) The shipments to be processed span a two-

day interval.3) If more than one route is feasible, the one

having the earliest completion time is

6)

7)

chosen.There are two products - Prod, is a fin-ished good, weighing 20 pounds per unitand measuring 4'x4'x4', while Prod: is araw material, of weight 25 pounds per unitand size 4'x4'x4'.There is one truck available for use, mea-suring 9'x9'x40'. with a weight capacity oV48.000 pounds. Therefore, it can accom-modate 180 units of Prod, and Prod: col-lectively, assuming 2 units across. 2 unitshigh, and 9 units long.Tbe truck driver is allotted a one-hourlunch break.The vehicle leaves the plant at 9 a.m., hav-ing pre-loaded the goods to be deliveredin that trip.Tbe list of shipments has already beenprocessed tbrough Phases I and II. Thereforethe list is ready to be routed in Phase III.

1993 IN FORMA TION SYSTEM 13

Tahle lb. Dayi Shipment List

Customer/Supplier

C,

s.C:

S'

S:

c,


Day,

Day,

Day,

Day,

Dayt

Dayi

TimeWindow

9:00-10:15

9:00-11:00

9:00-12:00

11:00- 1:00

2:00- 3:30

2:30- 5:00

StopTime

(in mins)

51

30

52

32

20

30

Product

Prodi

Prod:

Prod,

Prod:

Prod:

Prodi

# of Units

70

45

75

50

20

25


1400

1125

1500

1250

500

500

Table lc, Day2 Shipment List

Customer/Supplier

C4

C

S-.


Day:

Day:

Day:

Day:

TimeWindow

9:00-11:00

10:30-12:30

11:00- 1:00

3:00- 5:00

StopTime

(in mins)

51

47

42

32

Product

Prod,

Prod,

Prod,

Prod:

# of Units

70

60

50

50


1400

1200

1000

1250

9) All sub-problems are independent unlessotherwise stated.

10) Prior to implementing LIS. consolidationof inbound and outbound shipments wereconsidered independently of one another.

11) Times for loading and unloading, by fork-lift, can be derived using Deming's-' rela-tionship between handling time and ship-ment weight. With times in minutes andweight in pounds, Demings's result isshipment-handling time = a + h weight'

wbere a.b. and c are regression coefficients.Values for these parameters are b = 0.66 forloading. b= ].Q for unloading, and a = 0.6 and

c = 0.54 in either case. Note that in the follow-ing calculations, times have been rounded tothe nearest minute.

Tablela shows the list of shipments to beconsidered for routing during the two-dayperiod; Tables lb and lc divide the list intoDay, and Day: shipments respectively. Thedelivery/pickup network is illustrated in FigureI. Numbers associated with the arcs representthe average time in minutes that it takes to trav-el from one stop to the other. Because adetailed network showing all paths would becomplicated, only tbose arcs used in the exam-pie are exhibited.


Consider Dayi scheduling first, referring toTable lb and Figure 1. Notice that there arethree stops with time-windows beginning at9:00. These must be routed so as to allowreaching all three within the allotted windows.Consider stopping at C: first. Expected deliverytime is 52 minutes; thus, there is not a suffi-cient interval to make it to Ci and unload, andhence C; cannot be the first stop. Now considerSi as the first stop. It is apparent that this pick-up cannot be accommodated until a delivery ismade; customer shipments make up 170 units,so there would be room for only part of the 45units from Si. That leaves Ci as the first stop.Using travel and pickup/delivery times asgiven, the resulting schedule is:

1) 9:10-10:01 at CI2) 10:06-10:36 at SI

Figure 2. Day 1 Routing

3) 10:41-11:33 at C:The next stop will be S?; this is the only intervalduring which that stop can be accommodated:

4) 12:13- 12:45 at S35) 12:45 - 1:45 driver takes lunch

All that remains is to sort C3 and S2. Considergoing to C next. This would require waiting 25minutes after arrival until the time-windowopens. Following drop-off of the goods anddeparture at 3:00, arrival at S: would occur toolate for the 3:30 close. Therefore, the conclud-ing schedule will be as follows:

6) 2:15-2:35 at S27) 3:20-3:50 at C>8) 4:00-5:14atplant(unloadingofthe 115

units of Prod: takes 74 minutes).Similarly, routing for Day2 (Table lc) can be

done to arrive at tbe following schedule:

1993 IN FORMA TION SYSTEM 15

Figure 3. Day 2 Routing

1) 9:15-10:06 at C42) 10:41-11:28 a t e .3) 11:43-12:25 at a4) 12:25 - 1:25 driver takes lunch5) idle time from 1:25 to 2:206) 3:00-3.32 at S47) 3:52 - 4:40 at plant (it takes 48 minutes

to unload the 50 units).Figures 2 and 3 show the routes for Day, andDay2, respectively.

There are other situations which could occurafter the routing is decided, requiring alter-ations to the schedule. A sample of ad hococcurrences will now be considered.

CASE 1: A Change in Delivery Date

Suppose that supplier S4 ealls, saying that theshipment of raw materials must be picked upon Dayi as opposed to Day., but in the same

time-window. There are two ways in which thischange of Dayi could be accommodated:a) 1) original routing up to and including S:

remains the same2) 3:05 - 3:37 at S43) 3:57-4:27 at G4) 4:37 - 6:07 at plant (unloading takes 90

minutes - have 165 units)b) I) original routing up to and including C

remains the same2) 4:10-4:42 at S43) 5:12 - 6:42 at plant (unloading again

takes 90 minutes).Option (a) would be the better choice, since

it requires 35 less minutes than option (b).However, this routing still requires an addiUon-al 53 minutes over the base case for Dayi. Theschedule for Day: would also be affected; thetruck could proceed to the plant after stopping


Figure 4a. Case 1: Day 1 Routing After Change

at Cf. and before the driver's lunch. Work forthe day would finish at 12:45 with no unload-ing required, approximately 4 hours earlier thanthe original Days schedule. (Note that thischange leaves the driver and vehicle availablefor several hours of additional work not includ-ed in this consolidation scheme.) Figures 4aand 4b show the altered routings for Dayi and

respectively.

CASE 2: Post-Analysis Evaluation ofEfficiency

Examination of the final routing in Figure 2reveals a backtracking from S. to S2. Crossingof routes such as this usually indicates a sub-optimal solution, but here the crossing is need-ed to accommodate the time-windows. Nowsuppose, only for a moment, that the respectivetime-windows constraints were eliminated.This would alter the schedule to:

1) ll:58-12:18atS22) 12:18-1:18 driver takes lunch3) 1:48-2:20 at S34) 2:40-3:10atC35) 3:20 - 4:34 at plant (unloading of the 115

units takes 74 minutes).This would result in a savings of 40 minutesover the original case. Therefore, negotiationswith these suppliers should be initiated in orderto change the time windows to include 11.58 -12:18 at S2 and 1:48 - 2:20 at S3. The optimalrouting, assuming these changes are possible, isgiven in Figure 5.

COMPARISON TO PREVIOUS SYSTEM

It is important to compare the performanceof the joint consolidation schedule to that of theprior system. Consider the routings required forseparate consolidation:

1993 INFORMA TION SYSTEM 17

Dayi: Outbound1) 9:10-10:01 a t d2) 10:11-ll:03atCz3) 11:03 - 2:15 idle (with one hour lunch

for driver)4) 2:30-3:00 at Ci5) 3:10 back at plant (no unloading required)

Inbound1) 9:15--9:45 at S,2) 9:45-10:15 idle3) 11:00-l l :32atS.4) 11:32 - 1:30 idle (with one hour lunch

for driver)5) 2:00-2:20 at S26) 2:55 - 4:09 at plant (unloading again

takes 74 minutes)Day;: Outbound

1) 9:15-10:06 at C4

Figure 4b. Case 1: Day 2 Routing After Change

2) 10:41-11:28 a t e .3) 11:43-12:25 at C64) 12:45 back at plant (no unloading required)

Inbound1) 3:00-3:32atS42) 3:-52 - 4:40 at plant (unloading of the 50

units takes 48 minutes).

The primary difference between the routingsis the need for a second truck prior to imple-menting joint consolidation. Dayi requires twovehicles or a contract with atiother company tocover those shipments that cannot be accom-modated on the single truck. This necessitatesan additional capital expenditure for anothervehicle, and the salary of an extra driver (or anoutlay of cash if the shipment is contractedout).

( Plant


The second major difference is the extra idletime present before using LIS. Day, inboundand outbound schedules both show idle time fortbe truck and driver. Idle time is wasted timeand results in unnecessary expenses as the dri-ver must still be paid. Some idle time may bedesirable, but too much is detrimental to com-pany operations.

Both difficulties can be avoided throughsimultaneous inbound/outbound consolidation.as shown above. One vehicle can more effi-ciently bandle the shipment requirementsthrough application of LIS. Idle time can alsobe reduced, improving performance.

FUTURE ENHANCEMENTS

There are several ways in which LIS could bemodified in order to improve its current capabil-ities and hence expand its application. Certainassumptions could be relaxed as follows:1) Products with alternate storage requirements

could be considered.2) The maximum distance limit may be

relaxed to accommodate all shipments.3) Varying optimization criteria would be

available for selection.4) Altemative modes of transportation, sucb as

air and rail, could be considered.5) Allow trucks of different sizes or shapes.

They, too, would simply be treated as

Figure 5. Case 2: Day 1 Routing After Change

1993 INFORMATION SYSTEM 19

altemate modes.6) Extend application to different types of

businesses, such as merchandising firms orcontract-carrier operations. Use of LIS inthese areas should be encouraged.

7) Include provisions for multiple plants, eachwith its own inbound and outhoundshipments.

Modification to Phase I (the eliminationphase) may be required to accommodate thepreceding changes. Future versions of LIScould also be based on different prioritizationschemes from those in Phase II above (the sortprocedure).

The performance criteria and post-analysismodules could be expanded as well to includemeasures and capabilities other than those men-tioned. LIS could be designed to incorporateevaluation tools specific to the individual com-pany.

CONCLUSIONS

Simultaneous consolidation of inbound andoutbound shipments is a promising area for fur-ther research. As illustrated earlier, this method-ology could result in significant savings andimproved efficiency. Future endeavours couldbe directed to developing heuristics to handlesimultaneous consolidation. This may be moreefficient than using current versions of vehiclerouting software, which were not designed toaccommodate such a scenario. Algorithmswhich may serve as a base for this type of studycould include those mentioned by Casco,Golden, and WasiF* and by Min, Current, andSchilling.-^

It is true that introduction of a system such asLIS would require extensive coding and coordi-nation of various aspects of the company.However, for an organization considering ajoint inbound/outbound consolidation program,LIS will prove beneficial. The system willstreamline the consolidation process, andshould also result in improved efficiency andcost savings as well as major reductions in cap-ital expenditures.

ENDNOTES' Frank P. Buffa, "Inbound Logistics: Analysing

Inbound Consolidation Opportunities," InternationalJournal of Physical Distribution and MaterialsManagement, 16 (4), 1986. pp. 3-32.

= Thomas E. Vollmann. William L. Berry and D. Clay

Whybark, Manufacturing Planning and Control Systems.Homewood, Illinois: Richard D. Irwin Inc., 3rd Ed., 1992(chapter 18).

' John E. Tyworth, Joseph L. Cavinato and C. JohnLangley, Jr., Traffic Management: Planning. Operationsand Control. Reading, Massachusetts: Addison-WesleyPublishing Company, 1987 (chapter 10).

' Carlos F. Daganzo. "On the Coordination of Inboundand Outbound Schedules at a Transportation Terminal."Proceedings of llth International Symposium on Trans-portation and Traffic Theory. Yokohama. Japan, 1990.

' Robert C. Bushnell, James T. Low and James B.Wiley, "Transportation Network Models: Past Problemsand Prospects for the 198O's." International Journal ofPhysical Distribution and Materials Management, 11 (8)1981, pp. 38-45.

" Craig M. Gustin, "Trends in Computer Application inTransportation and Distribution Management." Inter-national Journal of Physical Distribution and MaterialsManagement, 14(1). 1984. pp. 52-60,

' Roger J. Calantone and Michael H. Morris, "TheUtilisation of Computer-Based Decision Support Systemsin Transportation," International Journal of PhysicalDistribution and Materials Management. 15 (7), 1985 p17.

" John E. Tyworth, Joseph L. Cavinato and C. JohnLangley, Jr., op. cit., (chapter 17).

" John C, Langley Jr., "Information-Based DecisionMaking in Logistics Management," International Journalof Physical Distribution and Materials Management 15(7), 1985, pp. 41-55.

'" George T. Haley and R. Krishnan, "It's Time forCALM: Computer-Aided Logistics Management,"International Journat of Physical Distribution andMaterials Management, 15(7), 1985, pp. 19-31.

" Andre Langevin and Yves Saint-Mleux, "A DecisionSupport Sy.stem for Physical Distribution Planning,"accepted for publication in Decision Support Systems1991.

'' Daniel O. Casco, Bruce L. Golden and Edward A.Wasil, 'Vehicle Routing With Backhauls: Models,Algorithms and Case Studies," Vehicle Routing: Methodsand Studies. Amsterdam, The Netherlands: ElsevierScience Publishers. B.V.. 1988, pp. 127-147.

" C. John Langley. op. cit.. p. 43.

'" Alan Rushton and John Oxley, Handbook of Logisticsand Distribution Management. London, England: KoganPage Ltd., 1989. pp. 24-30.

" John E. Tyworth, Joseph L.Cavinato and C. JohnLangley, Jr., op. cit., pp. 444-446, 456-458.

" D. Firth, J. Apple. RR. Denham. J. Hall, P. Inglis andA.L. Saipe, Profitable Logistics Management. Toronto.Ontario: McGraw-Hill Ryerson. 1988, pp. 251-256.

Richard Schonberger, Japanese ManufacturingTechniques, New York: The Free Press, 1983. p. 16.

Yasuhiro Monden. Toyota Production System,Norcross, Georgia: Institute of Industrial Engineers, 1983.p.2.

" James H. Bookbinder and David M. Dilts. "LogisticsInformation Systems in a Just-In-Time Environment."Journal of Business Logistics, 10(1). 1989, pp. 50-67.

'" Douglas M. Lambert and James R. Stock, StrategicLogistics Management. Homewood, Illinois: Richard DIrwin, Inc., 3rd Ed., 1993, pp. 535-536.


-' Alan J. Stenger "Information Systems in LogisticsManagement: Past, Present and Future," TransportationJournal, 29 (3), 1986. pp. 65-82.

- C. John Langley, op. cit., pp. 45.- W. E. Deming, "On a Rational Relationship for

Certain Costs of Handling Motor Freight: Over The" Transportation Journal, 17(4). 1978, pp. 5-11,

•' Daniel O. Casco. Bruce L. Golden and Edward A.Wasil, op. cit.

" Hokey Min, John Current and David Schilling, 'TheMultiple Depot Vehicle Routing Problem With Baek-hauling." Journat of Business Logistics, 13 (I) , 1992, pp.259-288,

James h Bookbinder

Documents