You are on page 1of 15

A Coal Unloader: A Finite Queueing System with Breakdowns

Author(s): Kenneth Chelst, Andrea Zundell Tilles and J. S. Pipis
Source: Interfaces, Vol. 11, No. 5 (Oct., 1981), pp. 12-25
Published by: INFORMS
Stable URL: .
Accessed: 04/01/2015 14:57
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact


INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Interfaces.

This content downloaded from on Sun, 4 Jan 2015 14:57:41 PM
All use subject to JSTOR Terms and Conditions

TRANSPORTATION EQUIPMENT *This paper was refereed.25 October 5. These at Toledo and a demurrage cost is then incurred. No.000 to $60. the number of queued trains will then be held QUEUES?APPLICATIONS. coal Power Plant is a 3. Wayne Tilles Rock ville. 12 INTERFACES This content downloaded from 192. Maryland 20850 and J. Research. S. it is also one of the first of its size to utilize a unit train to supply its fuel. coal was to be brought into the plant entirely by rail. Management observed frequent queues of trains at the unloader system which they attributed to breakdowns of the unloader system. the plant increased its generating capacity. which has one unloader system to dump the as coal. There were difficulties in meeting the plant's coal needs with the existing rail transport system. 1981.245.INTERFACES Vol. factors are believed to have contributed to the rail shortfall. These delays were costly for a number of reasons: Several Trains waiting to be unloaded are not transporting coal. which results in a decreased throughput. Sciences A COAL UNLOADER: A FINITE QUEUEING SYSTEMWITH BREAKDOWNS* Kenneth Department of Industrial Engineering Detroit. More expensive backup vessels must then be used to satisfy coal requirements at an added cost of $30. However. 2000 Second Avenue. If the unloader is broken for long periods of time. the existing rail system became insuffi cient. The more recent plan has been to combine rail and vessel delivery of coal to The Monroe satisfy the plant's requirements.000-megawatt facility that requires approxi tons of coal annually. Plant. The the relationship between the number of trains. In addition to being one of the world's mately 6. trains may exceed local holding-track capacity.000 per equiva lent trainload. as requested by the first author. Trains were frequently attributed to unloader break queued at the unloader system. Pipis Detroit Edison. 11. Detroit. delays. Michigan Andrea Control Data Chelst and Operations Zundell Corporation. to explore the impact on the system of unloader break with adding a second unloader associated system. Michigan 48226 owns and operates the coal-fired Monroe Edison Company is generally in nearby states and brought by train to the plant from mines is unloaded by a single unloader system. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions October 1981 .72 on Sun.5 million largest coal-fired plants. The institute of Management Copyright ? 1981 0092-2102/81/1105/0012$01. The Detroit Abstract. At present there are between four and eight trains allocated to moving coal from the mines to the power plant. which management downs. Coal Power A queueing model was developed downs and the potential benefits was to study model also used and queueing throughput. 48202 State University. Originally.60. The major contributor is believed to be the designed single-car unloader.

L. The specific parameters the model was to explore were: the impact of a second unloader system on the coal throughput. which is to the coal throughput. decided to build an analytic model which focused on the unloader in order to isolate its effect. The earliest work in this area [White and Christie. and the primary statistic of concern is the average arrival rate of trains to the unloader system. significantly is considering To alleviate these problems. and changing the cycle time between mine and power plant (puchasing coal from different mine fields). the impact of other changes were studied. a limited number of trains). Because of the limited number of trains involved. the average time spent in the unloader system. In addition.245. Although entire rail coal movement it was system had already been built for other purposes.. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions . INTERFACES October 1981 13 This content downloaded from 192. FIGURE 1. Our problem involved a finite source of customers (i. reducing the frequency of unloader the winter months contribute It is thought that trains waiting to the problem of frozen coal and its associated added costs. and W. Detroit Edison the addition of a a simulation model of the second multimillion dollar unloader system. Literature Review The nature of the unloader system suggested a queueing model subject to service 1958] viewed the interruptions. which in this case is a loaded train. This statistic is not addressed in the priority equivalent literature. A schematic diagram of the unloader system is displayed in Figure 1. including increasing the number of trains. THE COAL SUPPLY SYSTEM.72 on Sun. we could queueing analyze this problem using only basic queueing theory at the level of Hillier and text on Operations Research Lieberman's [1979]. the average number of trains in queue.e. service interruption as a high priority class of customer which preempts the service of the primary customer. reducing the repair time.

Cycle time we knew could not follow was an obvious minimum time for an un loaded train to go to the coal fields and return with a trainload of coal." we assumed an exponential probability density function for four time components of the system's operation: the time to unload a train the time to repair a broken unloader the time it takes for an unloaded train to go to the coal fields and return with a time Unloading time Repair Cycle time trainload of coal Failure time the time between breakdowns loader is operating Though the exponential assumption was breakdowns showed that the exponential for the repair time (Figure 2) and failure an exponential since there distribution. REPAIR TIME: A COMPARISONOF EMPIRICALDATA AND THE EXPONENTIALDENSITY. motivated 1977 data on 15 by tractability.60.MODEL AND ASSUMPTIONS we adopted an approach set forth by In beginning the model development. distribution was a reasonable approximation time. in Morris his Art article "The of To make the problem tractable [1967] Modeling. when the un continuously. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions .245.8 1977DATA 3 15 6 REPAIR TIMEINDAYS 14 INTERFACESOctober 1981 This content downloaded from 192.72 on Sun.8) N^l o 2. FIGURE 2. \ >i? CO as LU C3 \ \*^ EXP(-T/2.

In Figures 3 and 4 we present state transition rate diagrams for the one-unloader and two-unloader systems. transitions occur from the top row to the second row whenever the unloader breaks down. new trains arrive at a rate of (K? 2)X. When two unloaders are broken.) The queueing model we built.245. (Equivalently. respectively. If the top row of each figure were isolated.2). the system's state. we need to specify both i. When one unloader is broken. The probability of being in a particular state (i. when the single unloader is broken. TRANSITION RATE DIAGRAM OF SINGLE-UNLOADER QUEUEING SYSTEM. October 1981 INTERFACES at a since no train is unloaded 15 This content downloaded from 192. the number of trains queued at Monroe. 1979]. are made transitions to a higher state with the rate of (K?j)\j.Additional assumptions about system operation follow: Coal availability does not affect the train cycle rate. (The average repair time of the unloader is substan tially greater than the time necessary to reroute the train.j) is written as Pitj. we are in state (0.60. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions . No transitions to a lower state can occur. was amodified version of a server finite source queueing model which allowed for standard single and multiple server breakdowns. 1/X2 is the mean cycle time of the train. as mentioned earlier. arrival of a new train FIGURE 3. fjL2 is the rate at which an unloader is repaired (1/ju. All trains are the same size. The notation we use in describing our model is as follows: Xj is the arrival rate of an individual train when not in queue. Within the second row. Trains are unloaded at a rate of /?j for the single unloader system (Figure 3). we would simply have a standard finite source queueing model [Hillier and Lieber In order to describe for example. only one crew repairs it. two trains are in queue and man. k2 is the rate at which unloader breakdowns occur when an unloader system is operating.) pi1 is the rate at which a train is unloaded. the number of broken unloaders andy.72 on Sun. A train that is in a facility when one unloader breaks down is rerouted to the other unloader facility. which occurs at a rate of X2. A train can wait at the facility while an unloader is being repaired (this assumption only affects the demurrage cost). When. repair time).2 is the mean K is the total number of trains in the system. two crews work independently on them.2.

A standard program to invert a matrix and which is available on any computer could used to solve these equations. we need to add state (1. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions October 1981 . For the one-unloader model involving K trains. which allowed us to perform basic sensitivity analysis with little added cost.60. Breakdown transitions from row 2 to 3 occur at a rate of \2 >while repair transitions from row 3 to 2 occur at a rate of 2p2. Whenever occur a at a at rate downward and breakdowns of (i.0).Transitions within row 3 result from new train can be represented by systems of difference arrivals. In steady state these models equations which equate the rate into and out of each state (available upon request from the authors). the model results in 2^+1 simultaneous equations with 2K+ 1 unknowns. have been 16 INTERFACES This content downloaded from 192. The two-unloader model involves 3K+ 2 equations and an equal number of unknowns. which was not possible in the previous model.e. Any one of these equations is redun dant and must be replaced by an equation that sets the sum of all of the state probabilities equal to one. Our discussion of the transition rates within row 1 and between rows 1 and 2 two minor to Figure 4 of the two-unloader system with similarly applies service is modifications. In addition. there are two or more trains being unloaded. These equations were rewritten so that the right-hand side contained all zeroes except for the last equation. TRANSITION RATE DIAGRAM OF TWO-UNLOADER QUEUEING SYSTEM.72 on Sun. transitions) completed 2pl9 rate of 2X2 Row 2 ismodified to allow for completion of unloading a train at a rate of P! because coal can still be unloaded when only one unloader is broken. Lastly. the entire solution of these equations (AX one column of the inverse of the A matrix.245.. whose right-hand side was one. We therefore wrote our own simple program to solve this problem. since it is now possible to have one unloader broken and no trains waiting to be unloaded. we must now add another row of states to allow for two broken unloaders. Because of the special structure of the = b) is contained in only right-hand side.FIGURE 4.

L. into The following statistics were calcu performance meaningful we lated for both unloader systems. The mean repair time. while the dumper is operating.6 trainloads per year.60. W L?kj.o> is broken is K Pu I The fraction of time the unloader is busy is K I= 7 of a queue of trains is 1 P0.5 days. the average equals arrival rate: *_1 r i = + + \t. it is a straightforward task to convert these more measures. 1961] to determine the average wait for a INPUT DATA AND RESULTS Data were collected from a record of unloader outages for the first 10months of 1977.6 days/breakdown. The average number of trains atMonroe. l/fil9 is 9. breakdowns 1977 data showed that the coal unloader system was out of service Similarly. it means each train can bring in a maximum of 74. was calculated as follows: There were 15 breakdowns during the time period. Thus l/\2 = (9.i + Put) which throughput.Once all of the Pu are calculated.n Pun) J KPOt0 This is then used in Little's = train.5-hour average unloading time. present the appropriate formulas only for the simpler one-unloader model. for a The time for a train to complete a trip from the unloader to the mines and back was 108 hours.72 on Sun. The mean time between breakdowns. In Figure 5 we have plotted the number October 1981 INTERFACES of trainloads throughput per year against 17 This content downloaded from 192. Approxi mately 400 trains were unloaded. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions the number of .1 i*"") <Po.245.5 hours per train. total of 42. The average unloading time. 1//a2> was = 42.o The probability The key Poj 1 statistic is the average coal (Po.8 days/repair. *1 Ln2.5 hours/train) x (400 trains)/15 = 253 hours/breakdown = 10. When this is coupled with a 9.5 1/jjl2 days/15 repairs = 2. formula [Little. \ j The fraction of time the unloader The fraction of time the unloader l i=0 is idle is P0. is given by K L= 1 = j I Pu.

RELATIONSHIPBETWEEN THE NUMBER OF TRAINS AND THE ANNUAL COAL THROUGHPUT: ONE- VS TWO-UNLOADER SYSTEMS. and 370 trainloads if both operate simultaneously. From Figure 5 we see that five trains and a single unloader can throughput 322 trainloads. one additional train an or a additional of two-thirds train requires approximately 18 INTERFACES This content downloaded from 192. curve curves were 3 1 models for and for described while the model earlier. FIGURE 5. Three plots were generated: (1) single unloader. queueing 2 requires only a slight modification of these models. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions October 1981 . The two unloaders can throughput 355 trainloads if only one operates at a time.72 on Sun. the single unloader throughput. 600r MAY TWO UNLOADERS-BOTH OPERATESIMULTANEOUSLY TWO UNLOADERS-ONE OPERATING AT A TIME ONE UNLOADER TRAINS INSYSTEM We can also look horizontally at the curves to determine how many additional trains are required for the single-unloader system to maintain a specified level of an To maintain annual 400 trainload throughput.trains in the system. operating one at a time.60. (3) two unloaders operating simultaneously.245. (2) two un The loaders.

72 on Sun. These differences may be trains.VS TWO-UNLOADER SYSTEMS. each train is potentially capable of carrying 74. eighth train only adds about 50 trainloads to the total throughput for the single unloader.operating two-thirds of the time (these two things are not equivalent) when compared to the two-unloader system.6 trainloads per year.60. after a coal strike. The average wait to begin unloading. adding a second to hours unloader reduces the wait from 6. With eight to hours 1.5 cause critical when conditions coal to freeze.1 (see Figure 6). 25 r 20 3 O X ONEUNLOADER 15 TWO UNL0ADERS-0NLY ONEOPERATING ATA TIME I ?< 10 oc L? > < 5h TWO UNL0ADERS-B0TH MAY OPERATESIMULTANEOUSLY 6 7 5 NUMBEROF TRAINS IN SYSTEM October 1981 INTERFACES 19 This content downloaded from 192.6 hours. Wq. while it adds 60 trainloads to the total throughput for the dual system. These differences become especially significant if. the addition of a second unloader.245. Remember. The single unloader may need as many as five or six additional trains to bring in the same amount of coal. Clearly. As the throughput required increases to 475 or 520 the trainloads. With FIGURE 6.7 hours 0. the wait is reduced from 20. for example. is even more dramatically affected by four trains in operation. single unloader requires one and one and a half additional trains. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions . the difference between the two systems grows as the demand respectively. abnormally high amounts of coal are needed in a relatively short time. return from each train sets inmore quickly for the single increases and diminishing unloader The system. THE RELATIONSHIPBETWEEN THE NUMBER OF TRAINS AND THE AVERAGE DELAY INQUEUE: ONE.

To unload 21. A 20% decrease in the cycle time produces a throughput for the single unloader which is generally greater than what the throughput would be for the dual system under present cycle time Additional inmodel 20 INTERFACES This content downloaded from 192.245.VS TWO-UNLOADER SYSTEMS. number of trains. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions October 1981 . THE RELATIONSHIPBETWEENAVERAGEWAITING TIME AND THE COAL f HROUGHPUT:ONE.8 hours. -TWO UNLOADERS '(STRAINS) ONE UNLOADER (STRAINS) 24 HRSWAITING analyses were carried out to determine the sensitivity of the results to parameters. and the average delay will be 17. The single unloader requires seven trains to match the throughput.These same two sets of performance measures can be better perceived by com bining them into a single figure.6 hours. and average waiting time. the dual system requires six trains and the average delay will be 0. We present an analysis of cycle time. FIGURE 7. A decrease in the cycle time increases the throughput but also increases the average delay. In Figure 7 we have graphed tons per day.72 on Sun.60.5 x 105 tons/day. Cycle time could changes be changed by purchasing coal from different mines.

4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions . 600 r TWO UNLOADERS 990 ONE UNLOADER 900 490 < UJ > w 400 z < 390 h 300 290 ONE UNLOADER TWO UNLOADERS 200 4 _L 3 _L 6 J_ 7 8 NUMBER OF TRAINS INSYSTEM The number of trainloads was found not to be very sensitive to 20% changes in the mean time between unloader failures or to the failure duration.(Figure 8). The single unloader will feed through 470 trainloads with six trains. 21 This content downloaded from 192. with an 86. FIGURE 8. the impact on waiting time of a 20% change is not as significant. Delay in unloading for the single-unloader system was the parameter most affected by these changes.4 hour cycle time.72 on Sun. The dual system will always have significantly shorter waiting periods (Figure 9). two unloaders will feed through 445 trainloads with six trains if the cycle time remains at 108 hours. SENSITIVITYOF COAL THROUGHPUTTO CHANGES IN TRAIN CYCLE TIME: ONE- VS TWO-UNLOADER SYSTEMS.60. In contrast. As would be expected. A 20% increase INTERFACES (decrease) October 1981 in cycle time reduced (increased) the delay by about 12%. the single unloader was more affected by the changes.245.

Results from runs of the analytic model historical data for the one-unloader were checked against simulation runs and the results were close enough to use the analytic model for planning purposes. validated against actual data instead. SENSITIVITYOF AVERAGE WAITING TIME TO CHANGES IN THE TRAIN CYCLE TIME:ONE. 22 October 1981 INTERFACES This content downloaded from 192. In Table 1we present an illustrative compari son for a six-train configuration.VS TWO-UNLOADER SYSTEMS.FIGURE 9. The simulation model made no assump it used tions about the probability density function of the various time elements. A more complete and complex simulation model written in GPSS/V was available to Detroit Edison.60. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions . The simulation model had been previously case.245.72 on Sun. respectively. /^(+20%) ONE UNLOADER (-20%) (+20%) TWO UNLOADERS COMPARISON WITH A SIMULATION MODEL The analytic models required the solution of (2K+ 1) and (3K+2) simultaneous equations.

a fixed number of trains). Perhaps of greater to raise is that the models allowed. the analytic model. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions . To the surprise of management. the ease and low cost of gathering information from the analytic model allowed certain trends to be recognized.. A single run of the simulation model on an IBM 370/158 used an average of 2 and 50 seconds CPU time and cost $23. within IMPLEMENTATION The results of the study were presented to upper management and are being used to analyze whether or not to purchase the second unloader system. provides ongoing information about changing in order tomeet system constraints (such as coal throughput the system configuration and wait time). the answers were available an hour. creasing Speeding up recognized.70 flat one for To obtain reliable estimates handling rate. This was raised a question that could be answered with one run exemplified when management statistics of the analytic model. This enables management and the people who run the system to tune to seasonal the system consider and exceptional needs (such as recovery from a coal use miners' Little's definition the strike). It enables "what if" questions to be answered in a shorter period of time. Our analytic models provided management with tools to answer rapidly some very specific opera Our models tional questions: trains should be run to bring in X trainloads of coal? trains can be run without the average delay exceeding time? How many How many waiting a specific In summary. The savings in computer costs and run time is more than just a financial factor. models are being used regularly to "update the intuition of decision makers.TABLE 1. whereas the simulation model would have taken at least one day. In addition. a single run of the analytic model could provide for a range of train configurations.) just configuration (e. = Analytic Model 400 trainloads 18 = 469 System x = 4 s. a has effect than different repairs reducing the breakdown frequency. Two-Unloader 450 The major advantages of the analytic models were the computer costs and run time. management significance and explore new questions: What What is the effect if only one of the unloaders can run at a time? is the relationship between unloader down time and train waiting time? that they must differentiate between ways of in showed management a unloader concept not previously availability.d. Simulation Model = 428 trainloads System x One-Unloader s. To [Little. and continue to enable.245.94. COAL THROUGHPUTCAPACITY (x) COMPARED FOR SIX TRAINS MOVING COAL. 1970] of implementation.g. which was originally only to developed the addition of an unloader.60." evaluate INTERFACES October 1981 23 This content downloaded from 192. at least 10 runs were typically made with different seed values and at the above cost.d. In contrast.72 on Sun. minutes (This included a $2.

16. Morris. 1958. 1979. D. and Lieberman. S. J. pp. F. with Vol. 383-387. Management Vol. B466-B484. San Francisco. No. "The Art of Modeling." Priorities 24 INTERFACESOctober 1981 This content downloaded from 192. pp. Formula: The Concept 1967.245." Vol. "Models Science Vol. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions .. S. L = \W. and Managers: Little J. 1970. Research. Little. or with Breakdowns. No. 9. H. 6.. W. Inc. 8. "A Proof for the Queueing 3. Holden-Day.. J.60. B707-B717.AC KNOWLEDGEMENT I would like to thank Professor Kolesar comments whose significantly improved the style of the paper. 1961. "Queueing and Christie. 12. C. D. C. No. References Hillier." Management L. Operations 3rd Edition. pp. 79-95. 13. Research 1. C. G.." Operations of a Decision Science Preemptive Research Calculus. Operations T. pp. No. White.72 on Sun.

discrete simulation model of the total system (mines. The attached improvements in the existing transportation paper was used in the initial the current of development stages system and is presented for your use and general information. power plants. emissions. Operations Fuel Supply Department (313) 237-7885 RLHrjmg attachment INTERFACES October 1981 25 This content downloaded from 192. Gene Woolsey Research for Operations Institute Colorado School of Mines 80401 Golden. vessel to assist us sites) system.72 on Sun. long it appears that any new long term coal purchases will be Generally. The Fuel Supply Department is responsible for negotiating contracts for the procurement of all fuel to be consumed for electrical of low sulphur coal will be generation. limited by the available coal transportation alternatives rather than or mining constraints. Sincerely. 4 Jan 2015 14:57:41 PM All use subject to JSTOR Terms and Conditions .Detroit 2000SecondAvenue . Hall Division Director. coal reserve we are developing a Presently. This model in addition power generation areas useful in other such its designed purpose has been very for increasing as (a) identifying several other alternatives coal through put at this power plant and (b) identifying sev eral ways of reducing waiting with severe delays associated Michigan winters.. however some additional term contracts will be required. 1979 Dr. Woolsey: in the attached The Analytical paper queuing model discussed was designed whether a second coal to assist in evaluating the coal would handling capability unloading facility improve at the Monroe Power Plant. Detroit Edison's electric largest to fulfilling source. VJV! (313)237-8000 November 8. The success of this modeling effort has resulted in an increased use and recognition of analytical modeling as a system evaluation tool at Detroit Edison and in the Fuel Supply Department. Large quantities to comply with the changing regulations required regarding sulphur dioxide Much of this coal is already under contract. and proposed transloader delivery in selecting coal purchase strategies and identifying possible network.245. Colorado Dear Dr. 48226 Michigan V":-^r\ v Detroit. R.60. L.