- Email: [email protected]
Simulation-based rescheduling of the stacker–reclaimer operation
G Model
ARTICLE IN PRESS
JOCS-286; No. of Pages 6
Journal of Computational Science xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Journal of Computational Science journal homepage: www.elsevier.com/locate/jocs
Simulation-based rescheduling of the stacker–reclaimer operation Teus van Vianen ∗ , Jaap Ottjes, Gabriël Lodewijks Department of Marine and Transport Technology, Faculty 3ME, Delft University of Technology, The Netherlands
a r t i c l e
i n f o
Article history: Received 21 February 2014 Received in revised form 28 May 2014 Accepted 1 June 2014 Available online xxx Keywords: Stacker–reclaimer Rescheduling Discrete event simulation Dry bulk terminals
a b s t r a c t In this paper simulation is applied to reschedule the stacker–reclaimers operation to increase the dry bulk terminal’s performance by reducing the waiting time of cargo trains being loaded at the terminal. Stacker–reclaimers perform both the stacking and reclaiming of dry bulk materials. Due to the differences in loads between ships and cargo trains, the time needed for stacking and reclaiming varies considerably per job. The simulation tool developed can be used to support decisions when to interrupt ship servicing in favor of train loading based on the availability of transportation routes and expected disturbances. An experimental study demonstrated that ships and trains have to spend less time in the port when the stockyard lanes are accessible by two stacker–reclaimers due to the higher machines redundancy. Using the stacker–reclaimers rescheduling function the average port time of trains decreased without significantly affecting the port time of ships. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Dry bulk terminals are essential nodes in the major transportation links for coal and iron ore. These dry bulk materials are used for the worldwide production of electric energy and steel. A dry bulk terminal consists of a quayside where ships are loaded (export terminal) or unloaded (import terminal), a stockyard equipped with machines for the temporary storage of the dry bulk materials and a landside where transport modalities like cargo trains are serviced. Common machines at the stockyard are stacker–reclaimers. These machines combine the two functions of stacking and reclaiming into a single unit. Consequently, one of the two functions can be fulfilled at a time. During stacking, the material is transported through the machine to the end of the machine’s boom where the material is discharged and dumped on a pile. During reclaiming, the reclaiming device, which is in most cases a bucket wheel mounted at the end of the machine’s boom, digs the material out of the pile. The material is transported in reverse direction through the machine and dumped on a yard conveyor that conveys the material to its new destination. The terminal operation is complex because both ships and trains have to be served simultaneously and on time to prevent paying demurrage penalties to ship-owners or delivering an unacceptable service to rail operators [1]. Stacker–reclaimers have to handle the incoming as well as the outgoing flow of bulk materials.
∗ Corresponding author. Tel.: +31 152785935. E-mail address: [email protected] (T. van Vianen).
Consequently, their operation largely determines the terminal’s performance. The shiploads and trainloads vary considerably in volume and mass which causes a large variation in operation times. Large bulk ships can contain more than 250,000 t of dry bulk materials but the trainload is significantly smaller. In Western Europe, for example, the trainload is limited with 4000 t while in Australia and South Africa trains loaded with 35,000 t are not exceptional. Stacker–reclaimers are installed at import terminals as well as export terminals. In this paper the operation at import terminals will be discussed. When during ship unloading a train arrives that requests material that is stored in the reach of an active machine, the train has to wait before being serviced until the stacking operation is finished. The resulting waiting time can lead to an unsatisfactory service to train operators and cargo owners. Rescheduling the stacker–reclaimer’s operation by interrupting stacking and handling trains in between is a solution. However, ship unloading cannot be interrupted infinite times because terminal operators have limited time to unload ships. The paper’s objective is to apply simulation for the rescheduling of stacker–reclaimers to decrease the average time that cargo trains have to spend in the port while still guaranteeing the agreed ship port time. This paper is organized as follows. A literature review of stacker–reclaimer scheduling and the related job shop scheduling problem is given in Section 2. In Section 3, the stacker–reclaimers operation is explained and in Section 4, a simulation-based approach for the stacker–reclaimer rescheduling is introduced. An experimental study is included in Section 5 and finally, conclusions are given in Section 6.
http://dx.doi.org/10.1016/j.jocs.2014.06.004 1877-7503/© 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model JOCS-286; No. of Pages 6
ARTICLE IN PRESS T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
2
Fig. 1. Schematic representations of a dry bulk terminal with stockyard lanes (L1–L6), three stacker–reclaimers (SR1–SR3), two ship unloaders (UL1–UL2) and three railcar loaders (L1–L3).
2. Literature review During the scheduling of the operation at dry bulk terminals many decisions must be made. For example, where to berth the arriving ships, which quay crane must be assigned, where to store the material, which transportation route must be selected and which stacker–reclaimer must be used. Many researchers studied the terminal’s seaside operation; see for an extensive literature review [2]. A limited number of papers discussed the allocation of bulk materials at stockyards [3–9] and only two papers were found addressing the route scheduling problem [10,11]. The stacker–reclaimer scheduling problem was discussed by Hu and Yao [12]. The authors formulated the scheduling problem as a mixed integer programming model with the objective of minimizing the makespan (which is the total time between the start of the first operation and the end of the last operation) for a given set of handling operations. The approach developed was based on genetic algorithms using two types of chromosome representations. In the greedy assignment procedure, operations were assigned to machines based on their availability, minimized completion time and minimized setup times. Computational experiments were performed for a specific case for a planning horizon of 8 h. Hu and Yao assumed that the processing time per operation varies between the 60 and 150 min and that a stacker–reclaimer completes the operation without any interruption or shift. The stacker–reclaimer scheduling problem has similarities with the job shop scheduling problem. A stacking or reclaiming operation can then be defined as a job. Some authors extend the standard job shop scheduling problem by taken the machine breakdowns and the arriving of new priority jobs into account [13,14]. However, the assumption that a machine cannot be interrupted until the operation of the job is finished, prohibits the use of
the job shop scheduling problem for the rescheduling problem of stacker–reclaimers. 3. Stacker–reclaimers operation This section provides details about the stacker–reclaimers operation. Fig. 1 shows two typical stockyard layouts for import terminals where only the active belt conveyors are shown. In layout A (Fig. 1A) the stockyard contains four lanes. Both outer lanes (L1 and L4) are accessible by one stacker–reclaimer while the middle lanes (L2 and L3) can be reached by two stacker–reclaimers. In layout B (Fig. 1B) each stacker–reclaimer has exclusively access to two lanes. Usually, at import terminals the identity preserved storage policy is applied to assign piles to storage locations [6]. Piles are stored individually to prevent mixing and for realizing tracking and tracing of dry bulk materials. Consequently, a specific grade of material is then stored at one specific storage location and must be reclaimed by the correct stacker–reclaimer when requested to be loaded in rail cars. Fig. 1A and B shows an arbitrary situation where stacker–reclaimer SR3 reclaims material to be loaded in a cargo train by loader L1. At the same time, two ship unloaders (UL1 and UL2) unload a bulk ship (ship 1). Two transportation routes, formed by multiple belt conveyors in series, are used to transport the unloaded materials to the stacker–reclaimers SR1 and SR2. These machines stack the material into piles. Usually, there is limited storage area available for large piles. That’s why large shiploads (e.g., more than 100 kilotons [kt]) are split into multiple piles [15]. In Fig. 1, the load of ship 1 is divided across piles 1, 2 and 3. Fig. 1A shows that pile 2 is formed out of pile 2A and pile 2B and both stacker–reclaimers SR1 and SR2 are needed to create pile 2.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model JOCS-286; No. of Pages 6
ARTICLE IN PRESS T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
3
defined. In the simulation model all active elements act parallel in time, synchronized by the sequencing mechanism of the simulation software, in this case Delphi® , using the simulation application TOMAS [20]. 4.2. The simulation model
Fig. 2. An example of the several activities during the unloading of ship 1 (this ship was introduced in Fig. 1).
Terminal operators and ship-owners make agreements about the maximum time a ship will spend in the port. When there is time left during ship unloading, this additional time allows terminal operators to perform either an activity in between or finishing the handling earlier than expected which will result in despatch. Fig. 2 shows an example of possible activities during the unloading of ship 1. If there is no berth or unloader available, the ship has to wait. In the case of Fig. 1, two piles can be stacked at the same time when two unloaders are available and two transportation routes can be formed from available belt conveyors. During operation, a machine or belt conveyor may break down unexpectedly that may result in an exceeding of the ship port time. The interruption of ship unloading is only useful when the entire shipload can be handled within the ship port time. However, terminal operators face difficulties by predicting if the ship port time can be achieved because the availabilities and break downs of equipment are not known on beforehand. The simulation tool provides suggestions if stacker–reclaimers rescheduling can be performed based on the transportation routes availabilities and breakdown predictions. 4. Simulation-based approach This section introduces the simulation-based approach for supporting decisions whether stacker–reclaimers can be rescheduled or not. Section 4.1 explains this approach, Section 4.2 details the simulation model and the verification of the simulation model is discussed in Section 4.3. 4.1. The followed approach Discrete-event simulation has already been applied by others in the bulk handling industry. For example, Lodewijks et al. [16] suggested that discrete-event simulation can be used as a design tool to study the effect of different control philosophies at dry bulk terminals. Meng et al. [17] stated that discrete-event simulation has proven to be a powerful decision support tool to be effective for capturing high degree of complexity and uncertainty inherent in bulk handling systems at coal mines carrying out what-if analysis. In this paper the process-interaction method, introduced by [18,19], was followed. The terminal’s layout was virtually broken down into relevant element classes each with their typical attributes resulting in an object oriented data structure of the system. For all active element classes process descriptions, which describe the functioning of each element as a function of time, were
Data of (historically served) ships can be used as input for ships with specific arrival times and shiploads. For each pile the storage time is drawn from a distribution. After finishing the pile’s stacking operation, a train generator is created that generates trains to be loaded. In this simulation model, it was assumed that these trains arrived with constant interarrival times during the pile’s storage time. In reality, the interarrival times of trains that must be loaded with material from a specific pile vary during the storage time. Each stacker–reclaimer contains several algorithms; an algorithm to select if a new arrived ship or train can be handled, an algorithm to select an available transportation route and an algorithm to allocate (a part of) the shipload to a storage location within the stacker–reclaimer’s reach that contains enough free area to store the load. The time needed to stack or reclaim the bulk materials is determined by dividing the job’s load by the machine’s capacity. When the rescheduling function is activated and a train arrives to pick up materials, an algorithm investigates if this train can be handled immediately. In this algorithm, the stacking operation will be interrupted in favor of reclaiming material that must be loaded into railcars when the requested material is stored in the machine’s reach and the following preconditions are met: 1. If a stacker–reclaimer is active with reclaiming then the reclaiming operation will not be rescheduled due to the limited reclaiming time. 2. A transportation route can be formed from idle belt conveyors to transport the requested material to the right railcar loader. 3. There is still time left within the agreed ship port time after serving the train in between plus two times repositioning of the stacker–reclaimer (repositioning requires normally between 15 and 30 min). 4. The maximum number of interruptions per ship is not reached. When requested material is stored at a stockyard lane that is accessible by two stacker–reclaimers and both machines are active with stacking, the stacker–reclaimer with the most available ship port time is selected to be rescheduled. The fourth precondition was introduced to prevent that ship unloading is interrupted frequently at the beginning of the unloading activity. This may result in an extension of the ship end time due to unexpected breakdowns during the remaining stacking operation. The question now is, how many times may the ship unloading be interrupted when the route availability and breakdowns are not known on beforehand? A fixed number of interruptions per ship will not be a useful parameter because the shipload can vary considerably. Therefore, it is proposed that the number of interruptions per ship depends on the shipload and will be defined by dividing the shipload with the input parameter ‘shipload distributor’. For example, a value for the shipload distributor of 20 kt will assign maximum four interruptions for a ship loaded with 86 kt and only one interruption if this ship was loaded with 30 kt. The transportation route that was used during stacking will also be claimed during the interrupted reclaiming operation. Consequently, the belt conveyors in this route cannot be assigned to other stacker–reclaimers and after finishing reclaiming, the stacker–reclaimer continues with the already started stacking operation.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model
ARTICLE IN PRESS
JOCS-286; No. of Pages 6
T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
4
Table 1 Input parameters for the experimental study.
2.0 Simulation results M/M/3
Wt [1/μ]
1.6
1.2
0.8
Input parameter
Value
Ships interarrival time distribution Shipload distribution Average shipload [kt] Average storage time [h] Storage time distribution Stacking and reclaiming capacity [kt/h] Stacker–reclaimers repositioning time [h] Technical availability Trainload [kt] Shipload distributor [kt]
NED Based on historical data 101 500 NED 2.5 ¼ 0.97 4 15
0.4
0.0 0.3
the ship has to wait until free area arises. Despite this variation the model can be considered as correct.
0.4
0.5
0.6
0.7
0.8
5. Simulation experimental results
ρ [-] Fig. 3. Verification of the simulation model (layout B) with an M/M/3-queuing model.
4.3. Verification Verification of simulation models is required to check the correct translation of the conceptual model into computer code and to determine if the simulation model performs as intended [21]. For the verification, the tracing function of TOMAS was used and simulation results were compared with analytical results. Analytical results were achieved using the multiple-server M/M/n queuing model, only for this queuing system analytical solutions exist [22]. An extra unloader was added to the layout (B) of Fig. 1B to enable the comparison with an M/M/3-queuing model. The following preconditions were used for the simulation model; the ship interarrival times and shiploads were represented by negative exponential distributions, the shipload was stored in one pile and there was no stochastic storage time. The average ship waiting time (Wt) as function of the inverse of the ship unloading rate (1/) was determined using simulation. Fig. 3 shows these results together with analytical results for the M/M/3-queuing model. The minor deviation for the increasing values for the average stacker–reclaimers utilization () can be explained by the fact that storage allocation is included in the simulation model. If there is no area available to store the new shipload,
Stacker–reclaimer rescheduling will be investigated for both layouts of Fig. 1 using the input parameters of Table 1. For both layouts it was assumed that the stockyard’s areas have sufficient storage capacity to prevent that ships cannot be unloaded because there was no storage area available. Previous research [23] has shown that the ships interarrival times can be modeled using standardized distributions depending on the terminal type. For this case, it was assumed that the terminal acts as a stevedoring import terminal that has to serve many clients and does hardly have any influence on the ships arrival times. The interarrival time distribution can then best be represented by the negative exponential distribution. Historical data of unloaded ships during three years of operation from a specific import terminal was used as input for the shiploads. In van Vianen et al. [23] this measured shipload distribution was listed. The transportation route consists of multiple belt conveyors in series together with a loader or an unloader and a stacker–reclaimer. If one machine or belt conveyor fails, the entire route fails. Tewari et al. [24] used negative exponential distributions for the failure and repair distributions to represent the technical availability of belt conveyors. Historical data of belt conveyors at an export terminal during one year of operation showed an average conveyor availability between 0.9 and 0.97. For this case, it was assumed that each (un)loader, stacker–reclaimer and belt conveyor had an identical technical availability of 0.97. ˙ in million tons per During the experimental study, the annual throughput (m) year [Mt/y] was varied and the average ship (Wship ) [h] and the average train port time (Wtrain ) [h] were measured. Fig. 4 shows for both layouts these measured port times. From this figure can be concluded that ships and trains spent less time in the port when layout A will be applied compared to layout B. The higher stacker–reclaimer redundancy on the middle lanes results in a better division of stacking and reclaiming operations across the stacker–reclaimers causing less waiting times for ships and trains. For both layouts the average ship and train port times were measured when the rescheduling function was activated. These results are shown in Fig. 5. From this figure can be concluded that rescheduling of stacker–reclaimers will decrease the average train port time. For layout A, the average ship port time remains more or less constant when the rescheduling function was activated but for annual throughput
(A)
(B) 150
150
Layout B Layout A
Layout B Layout A 120 Wtrain [h]
Wship [h]
120
90
90
60
60
30
30
0
0 10
12
14 ṁ
16
18
20
10
12
14 ṁ
16
18
20
Fig. 4. Measured average port times for ships (A) and for trains (B) versus the annual throughput for the layouts as shown in Fig. 1.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model
ARTICLE IN PRESS
JOCS-286; No. of Pages 6
T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
5
Fig. 5. Measured average port times for ships and trains versus the annual throughput for layout A (A) and layout B (B) of Fig. 1 as function of stacker–reclaimer rescheduling.
presented can be used for supporting decisions whether rescheduling of the stacker–reclaimers operation can decrease the average train port time while still guaranteeing the seaside performance. Results from an experimental study have shown that ships and trains are served quicker when wide stockyard lanes, which are accessible by two stacker–reclaimers, are used. From the experimental study can also be concluded that rescheduling of the stacker–reclaimers operation will decrease the average train port time. The reduction of the train port time caused a minor increase of the average ships port time when a terminal layout with dedicated stacker–reclaimers lanes was used. The average ships port time remains approximately constant for the wide-lane terminal layout.
3.5 3
ni [-]
2.5 2 1.5 1
0.5
Layout B Layout A
Acknowledgement
0 10
12
14
16
[Mt/y]
18
20
Fig. 6. Average number of interruptions per ship for both layouts versus the annual throughput.
values larger than 17 Mt/y, the average ship port time even decreased. The handling activities were better spread across the three stacker–reclaimers that resulted in a reduction of the average machine utilization. This reduction of the average stacker–reclaimer utilization lead to, especially for the relatively high values of the stacker–reclaimers utilization of 0.6 to realize an annual throughput of more than 17 Mt, a decrease of the average ship waiting time. The increase of the average ship port time for layout B when the rescheduling function was activated can be explained by the limited stacker–reclaimers redundancy. A temporary interruption of the stacking operation will cause an increase of the ship port time. The larger reduction of the average train port time for layout A (see Fig. 5A) compared to layout B (see Fig. 5B) can be explained by the average number of interruptions per ship (ni ). Fig. 6 shows these numbers versus the annual throughput for both layouts. For the same value of the shipload distributor, the ship unloading operation in layout B is more frequent interrupted than in layout A due to the lower stacker–reclaimers redundancy. Each interruption requires two times repositioning of the stacker–reclaimer, which takes half an hour, and introduces ineffective machines hours.
6. Conclusions In dry bulk import terminals stacker–reclaimers are occupied for a long time during stacking of materials from ships. When a train arrives at the terminal’s landside to pick up material that is stored in the reach of an active stacker–reclaimer, this train has to wait before getting serviced. Waiting of trains may result in an unacceptable service to rail operators. The simulation model
The authors acknowledge the terminal operators who provided operational data about the ships arrival process and storage process of bulk materials at their stockyards and gave valuable feedback during the research. References [1] R. Robinson, Regulating efficiency into port-oriented chain systems: export coal through the Dalrymple Bay Terminal, Australia, Marit. Policy Manag. 34 (2007) 89–106. [2] C. Bierwirth, F. Meisel, A survey of berth allocation and quay crane scheduling problems in container terminals, Eur. J. Oper. Res. 202 (2010) 615–627. [3] S. Dipsar, T. Altiok, Control policies for material flow in bulk-port marine terminals, in: Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics, Canada, 1998, pp. 3083–3088. [4] P. Molck, R. Goncalves, T. Caldas, J. Valentim, L. Lima, E. Newton, M. Franc¸a, R. Mendes, F. Gomide, Intelligent stockpile building in iron ore shipping yard, in: Proceedings of the Third International Conference on Intelligent Processing and Manufacturing of Materials (IPMM 2001), Canada, 2001, pp. 1–11. [5] M. Ago, T. Nishi, M. Konishi, Simultaneous optimization of storage allocation and routing problems for belt conveyor transportation, J. Adv. Mech. Des. Syst. Manuf. 1 (2007) 250–261. [6] J. Leech, Design of an efficient coal export terminal, in: Proceedings of the Queensland Mining & Engineering Exhibition, Australia, 2010. [7] J. Leech, Optimizing a bulk minerals export chain, Min. Mag. 11 (2012) 42–48. [8] N. Umang, M. Bierlaire, I. Vacca, Exact and heuristic methods to solve the berth allocation problem in bulk ports, Transp. Res. E 54 (2013) 14–31. [9] T. Robenek, N. Umang, M. Bierlaire, S. Ropke, A branch-and-price algorithm to solve the integrated berth allocation and yard assignment problem in bulk ports, Eur. J. Oper. Res. 235 (2) (2014) 399–411. [10] B. Kim, S.Y. Chang, J. Chang, Y. Han, J. Koo, K. Lim, J. Shin, S. Jeong, W. Kwak, Scheduling of raw-materials unloading from ships at a steelworks, Prod. Plan. Control 22 (4) (2011) 389–402. [11] T.A. van Vianen, D.L. Mooijman, J.A. Ottjes, R.R. Negenborn, G. Lodewijks, Simulation-based operational control of a dry bulk terminal, in: Proceedings
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model JOCS-286; No. of Pages 6
ARTICLE IN PRESS T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
6
[12] [13]
[14]
[15] [16] [17]
[18] [19] [20]
[21] [22] [23]
[24]
of the 2012 IEEE International Conference on Networking, Sensing and Control (ICNSC 2012), 2012, pp. 73–78. D. Hu, Z. Yao, Stacker–reclaimer scheduling in a dry bulk terminal, Int. J. Comput. Integr. Manuf. 25 (2012) 1047–1058. A.H. Nasr, T.Y. ElMekkawy, Robust and stable flexible job shop scheduling with random machine breakdowns using a hybrid genetic algorithm, Int. J. Prod. Econ. 132 (2011) 279–291. R. Sarker, M. Omar, S.M.K. Hasan, D. Essam, Hybrid evolutionary algorithm for job scheduling under machine maintenance, Appl. Soft Comput. 13 (2013) 1440–1447. T.A. van Vianen, J.A. Ottjes, G. Lodewijks, Stockyard dimensioning for dry bulk terminals, in: Proceedings of the 2012 Trail-Beta Congress, Rotterdam, 2012. G. Lodewijks, D.L. Schott, J.A. Ottjes, Modern dry bulk terminal design, Port Technol. Int. 43 (2009) 87–94. C. Meng, S.S. Nageshwaraniyer, A. Maghsoudi, Y.J. Son, S. Dessureault, Datadriven modeling and simulation framework for materials handling systems in coal mines, Comput. Ind. Eng. 64 (2013) 766–779. B.P. Zeigler, H. Praehofer, T.G. Kim, Theory of Modeling and Simulation, Academic Press, San Diego, 2000. G.S. Fishmann, Discrete Event Simulation–Modeling, Programming and Analysis, Springer-Verlag, New York, 2001. H.P.M. Veeke, J.A. Ottjes, Tomas tool for object-oriented modeling and simulation, in: Proceedings of the Business and Industry Simulation Symposium ASTC, United States, 1999. J.P.C. Kleijnen, Verification and validation of simulation models, Eur. J. Oper. Res. 82 (1995) 145–162. I. Adan, J. Resing, Queueing Theory – Lecture Notes, Eindhoven University of Technology, 2002. T.A. van Vianen, J.A. Ottjes, G. Lodewijks, Modelling arrival process at dry bulk terminals, in: Proceedings of the Bulk Ports, Terminals and Logistics 2012 Conference, Amsterdam, 2012. P.C. Tewari, I.P. Singh, M.K. Khare, Reliability analysis of a conveyor belt system, with only one server, subject to failures and idleness after repair, Microelectron. Reliab. 31 (1991) 823–826.
Teus van Vianen received his MSc degree in mechanical engineering at the Delft University of Technology in 2005. After his study, he worked for five years in the machinery industry. In June 2010, he returned to the Delft University and started his Ph.D. project to determine the fundamentals of dry bulk terminal design at the faculty of Mechanical, Marine and Materials Engineering (3ME). His main interests are dry bulk terminals, terminal design, queuing theory, routing & scheduling, storage allocation and automation of transport systems.
Jaap Ottjes studied physics at Delft University of Technology and received the MSc degree in 1970. He obtained his Ph.D. degree in the ‘prediction of pressure losses in dilute phase pneumatic transport systems’ at the same university. After his Ph.D. research that he changed subject and specialized, partly as a consultant, on industrial logistics and simulation of transport systems. He was employed as associate professor in transport engineering and logistics at Delft University of Technology within the Department of Marine and Transport Technology at the faculty 3ME. His main research interests are in the field of industrial logistics and transport systems and in the modeling and simulation of these systems in particular. Professor Gabriel Lodewijks studied Mechanical Engineering at Twente University and Delft University of Technology from which he obtained a master degree (cum laude) in 1992. He specialized in transport technology, material engineering and dynamics. He obtained his Ph.D. degree at Delft University of Technology on the dynamics of belt conveyor systems in 1996. After 5 years of experience in the industry including three years in the USA, he was appointed as professor of transport engineering and logistics at the 3ME faculty of Delft University of Technology. He wrote over 250 papers and contributed to four books. He is frequently commissioned to act as an expert witness and he is further president of Conveyor Experts B.V.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
ARTICLE IN PRESS
JOCS-286; No. of Pages 6
Journal of Computational Science xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Journal of Computational Science journal homepage: www.elsevier.com/locate/jocs
Simulation-based rescheduling of the stacker–reclaimer operation Teus van Vianen ∗ , Jaap Ottjes, Gabriël Lodewijks Department of Marine and Transport Technology, Faculty 3ME, Delft University of Technology, The Netherlands
a r t i c l e
i n f o
Article history: Received 21 February 2014 Received in revised form 28 May 2014 Accepted 1 June 2014 Available online xxx Keywords: Stacker–reclaimer Rescheduling Discrete event simulation Dry bulk terminals
a b s t r a c t In this paper simulation is applied to reschedule the stacker–reclaimers operation to increase the dry bulk terminal’s performance by reducing the waiting time of cargo trains being loaded at the terminal. Stacker–reclaimers perform both the stacking and reclaiming of dry bulk materials. Due to the differences in loads between ships and cargo trains, the time needed for stacking and reclaiming varies considerably per job. The simulation tool developed can be used to support decisions when to interrupt ship servicing in favor of train loading based on the availability of transportation routes and expected disturbances. An experimental study demonstrated that ships and trains have to spend less time in the port when the stockyard lanes are accessible by two stacker–reclaimers due to the higher machines redundancy. Using the stacker–reclaimers rescheduling function the average port time of trains decreased without significantly affecting the port time of ships. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Dry bulk terminals are essential nodes in the major transportation links for coal and iron ore. These dry bulk materials are used for the worldwide production of electric energy and steel. A dry bulk terminal consists of a quayside where ships are loaded (export terminal) or unloaded (import terminal), a stockyard equipped with machines for the temporary storage of the dry bulk materials and a landside where transport modalities like cargo trains are serviced. Common machines at the stockyard are stacker–reclaimers. These machines combine the two functions of stacking and reclaiming into a single unit. Consequently, one of the two functions can be fulfilled at a time. During stacking, the material is transported through the machine to the end of the machine’s boom where the material is discharged and dumped on a pile. During reclaiming, the reclaiming device, which is in most cases a bucket wheel mounted at the end of the machine’s boom, digs the material out of the pile. The material is transported in reverse direction through the machine and dumped on a yard conveyor that conveys the material to its new destination. The terminal operation is complex because both ships and trains have to be served simultaneously and on time to prevent paying demurrage penalties to ship-owners or delivering an unacceptable service to rail operators [1]. Stacker–reclaimers have to handle the incoming as well as the outgoing flow of bulk materials.
∗ Corresponding author. Tel.: +31 152785935. E-mail address: [email protected] (T. van Vianen).
Consequently, their operation largely determines the terminal’s performance. The shiploads and trainloads vary considerably in volume and mass which causes a large variation in operation times. Large bulk ships can contain more than 250,000 t of dry bulk materials but the trainload is significantly smaller. In Western Europe, for example, the trainload is limited with 4000 t while in Australia and South Africa trains loaded with 35,000 t are not exceptional. Stacker–reclaimers are installed at import terminals as well as export terminals. In this paper the operation at import terminals will be discussed. When during ship unloading a train arrives that requests material that is stored in the reach of an active machine, the train has to wait before being serviced until the stacking operation is finished. The resulting waiting time can lead to an unsatisfactory service to train operators and cargo owners. Rescheduling the stacker–reclaimer’s operation by interrupting stacking and handling trains in between is a solution. However, ship unloading cannot be interrupted infinite times because terminal operators have limited time to unload ships. The paper’s objective is to apply simulation for the rescheduling of stacker–reclaimers to decrease the average time that cargo trains have to spend in the port while still guaranteeing the agreed ship port time. This paper is organized as follows. A literature review of stacker–reclaimer scheduling and the related job shop scheduling problem is given in Section 2. In Section 3, the stacker–reclaimers operation is explained and in Section 4, a simulation-based approach for the stacker–reclaimer rescheduling is introduced. An experimental study is included in Section 5 and finally, conclusions are given in Section 6.
http://dx.doi.org/10.1016/j.jocs.2014.06.004 1877-7503/© 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model JOCS-286; No. of Pages 6
ARTICLE IN PRESS T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
2
Fig. 1. Schematic representations of a dry bulk terminal with stockyard lanes (L1–L6), three stacker–reclaimers (SR1–SR3), two ship unloaders (UL1–UL2) and three railcar loaders (L1–L3).
2. Literature review During the scheduling of the operation at dry bulk terminals many decisions must be made. For example, where to berth the arriving ships, which quay crane must be assigned, where to store the material, which transportation route must be selected and which stacker–reclaimer must be used. Many researchers studied the terminal’s seaside operation; see for an extensive literature review [2]. A limited number of papers discussed the allocation of bulk materials at stockyards [3–9] and only two papers were found addressing the route scheduling problem [10,11]. The stacker–reclaimer scheduling problem was discussed by Hu and Yao [12]. The authors formulated the scheduling problem as a mixed integer programming model with the objective of minimizing the makespan (which is the total time between the start of the first operation and the end of the last operation) for a given set of handling operations. The approach developed was based on genetic algorithms using two types of chromosome representations. In the greedy assignment procedure, operations were assigned to machines based on their availability, minimized completion time and minimized setup times. Computational experiments were performed for a specific case for a planning horizon of 8 h. Hu and Yao assumed that the processing time per operation varies between the 60 and 150 min and that a stacker–reclaimer completes the operation without any interruption or shift. The stacker–reclaimer scheduling problem has similarities with the job shop scheduling problem. A stacking or reclaiming operation can then be defined as a job. Some authors extend the standard job shop scheduling problem by taken the machine breakdowns and the arriving of new priority jobs into account [13,14]. However, the assumption that a machine cannot be interrupted until the operation of the job is finished, prohibits the use of
the job shop scheduling problem for the rescheduling problem of stacker–reclaimers. 3. Stacker–reclaimers operation This section provides details about the stacker–reclaimers operation. Fig. 1 shows two typical stockyard layouts for import terminals where only the active belt conveyors are shown. In layout A (Fig. 1A) the stockyard contains four lanes. Both outer lanes (L1 and L4) are accessible by one stacker–reclaimer while the middle lanes (L2 and L3) can be reached by two stacker–reclaimers. In layout B (Fig. 1B) each stacker–reclaimer has exclusively access to two lanes. Usually, at import terminals the identity preserved storage policy is applied to assign piles to storage locations [6]. Piles are stored individually to prevent mixing and for realizing tracking and tracing of dry bulk materials. Consequently, a specific grade of material is then stored at one specific storage location and must be reclaimed by the correct stacker–reclaimer when requested to be loaded in rail cars. Fig. 1A and B shows an arbitrary situation where stacker–reclaimer SR3 reclaims material to be loaded in a cargo train by loader L1. At the same time, two ship unloaders (UL1 and UL2) unload a bulk ship (ship 1). Two transportation routes, formed by multiple belt conveyors in series, are used to transport the unloaded materials to the stacker–reclaimers SR1 and SR2. These machines stack the material into piles. Usually, there is limited storage area available for large piles. That’s why large shiploads (e.g., more than 100 kilotons [kt]) are split into multiple piles [15]. In Fig. 1, the load of ship 1 is divided across piles 1, 2 and 3. Fig. 1A shows that pile 2 is formed out of pile 2A and pile 2B and both stacker–reclaimers SR1 and SR2 are needed to create pile 2.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model JOCS-286; No. of Pages 6
ARTICLE IN PRESS T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
3
defined. In the simulation model all active elements act parallel in time, synchronized by the sequencing mechanism of the simulation software, in this case Delphi® , using the simulation application TOMAS [20]. 4.2. The simulation model
Fig. 2. An example of the several activities during the unloading of ship 1 (this ship was introduced in Fig. 1).
Terminal operators and ship-owners make agreements about the maximum time a ship will spend in the port. When there is time left during ship unloading, this additional time allows terminal operators to perform either an activity in between or finishing the handling earlier than expected which will result in despatch. Fig. 2 shows an example of possible activities during the unloading of ship 1. If there is no berth or unloader available, the ship has to wait. In the case of Fig. 1, two piles can be stacked at the same time when two unloaders are available and two transportation routes can be formed from available belt conveyors. During operation, a machine or belt conveyor may break down unexpectedly that may result in an exceeding of the ship port time. The interruption of ship unloading is only useful when the entire shipload can be handled within the ship port time. However, terminal operators face difficulties by predicting if the ship port time can be achieved because the availabilities and break downs of equipment are not known on beforehand. The simulation tool provides suggestions if stacker–reclaimers rescheduling can be performed based on the transportation routes availabilities and breakdown predictions. 4. Simulation-based approach This section introduces the simulation-based approach for supporting decisions whether stacker–reclaimers can be rescheduled or not. Section 4.1 explains this approach, Section 4.2 details the simulation model and the verification of the simulation model is discussed in Section 4.3. 4.1. The followed approach Discrete-event simulation has already been applied by others in the bulk handling industry. For example, Lodewijks et al. [16] suggested that discrete-event simulation can be used as a design tool to study the effect of different control philosophies at dry bulk terminals. Meng et al. [17] stated that discrete-event simulation has proven to be a powerful decision support tool to be effective for capturing high degree of complexity and uncertainty inherent in bulk handling systems at coal mines carrying out what-if analysis. In this paper the process-interaction method, introduced by [18,19], was followed. The terminal’s layout was virtually broken down into relevant element classes each with their typical attributes resulting in an object oriented data structure of the system. For all active element classes process descriptions, which describe the functioning of each element as a function of time, were
Data of (historically served) ships can be used as input for ships with specific arrival times and shiploads. For each pile the storage time is drawn from a distribution. After finishing the pile’s stacking operation, a train generator is created that generates trains to be loaded. In this simulation model, it was assumed that these trains arrived with constant interarrival times during the pile’s storage time. In reality, the interarrival times of trains that must be loaded with material from a specific pile vary during the storage time. Each stacker–reclaimer contains several algorithms; an algorithm to select if a new arrived ship or train can be handled, an algorithm to select an available transportation route and an algorithm to allocate (a part of) the shipload to a storage location within the stacker–reclaimer’s reach that contains enough free area to store the load. The time needed to stack or reclaim the bulk materials is determined by dividing the job’s load by the machine’s capacity. When the rescheduling function is activated and a train arrives to pick up materials, an algorithm investigates if this train can be handled immediately. In this algorithm, the stacking operation will be interrupted in favor of reclaiming material that must be loaded into railcars when the requested material is stored in the machine’s reach and the following preconditions are met: 1. If a stacker–reclaimer is active with reclaiming then the reclaiming operation will not be rescheduled due to the limited reclaiming time. 2. A transportation route can be formed from idle belt conveyors to transport the requested material to the right railcar loader. 3. There is still time left within the agreed ship port time after serving the train in between plus two times repositioning of the stacker–reclaimer (repositioning requires normally between 15 and 30 min). 4. The maximum number of interruptions per ship is not reached. When requested material is stored at a stockyard lane that is accessible by two stacker–reclaimers and both machines are active with stacking, the stacker–reclaimer with the most available ship port time is selected to be rescheduled. The fourth precondition was introduced to prevent that ship unloading is interrupted frequently at the beginning of the unloading activity. This may result in an extension of the ship end time due to unexpected breakdowns during the remaining stacking operation. The question now is, how many times may the ship unloading be interrupted when the route availability and breakdowns are not known on beforehand? A fixed number of interruptions per ship will not be a useful parameter because the shipload can vary considerably. Therefore, it is proposed that the number of interruptions per ship depends on the shipload and will be defined by dividing the shipload with the input parameter ‘shipload distributor’. For example, a value for the shipload distributor of 20 kt will assign maximum four interruptions for a ship loaded with 86 kt and only one interruption if this ship was loaded with 30 kt. The transportation route that was used during stacking will also be claimed during the interrupted reclaiming operation. Consequently, the belt conveyors in this route cannot be assigned to other stacker–reclaimers and after finishing reclaiming, the stacker–reclaimer continues with the already started stacking operation.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model
ARTICLE IN PRESS
JOCS-286; No. of Pages 6
T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
4
Table 1 Input parameters for the experimental study.
2.0 Simulation results M/M/3
Wt [1/μ]
1.6
1.2
0.8
Input parameter
Value
Ships interarrival time distribution Shipload distribution Average shipload [kt] Average storage time [h] Storage time distribution Stacking and reclaiming capacity [kt/h] Stacker–reclaimers repositioning time [h] Technical availability Trainload [kt] Shipload distributor [kt]
NED Based on historical data 101 500 NED 2.5 ¼ 0.97 4 15
0.4
0.0 0.3
the ship has to wait until free area arises. Despite this variation the model can be considered as correct.
0.4
0.5
0.6
0.7
0.8
5. Simulation experimental results
ρ [-] Fig. 3. Verification of the simulation model (layout B) with an M/M/3-queuing model.
4.3. Verification Verification of simulation models is required to check the correct translation of the conceptual model into computer code and to determine if the simulation model performs as intended [21]. For the verification, the tracing function of TOMAS was used and simulation results were compared with analytical results. Analytical results were achieved using the multiple-server M/M/n queuing model, only for this queuing system analytical solutions exist [22]. An extra unloader was added to the layout (B) of Fig. 1B to enable the comparison with an M/M/3-queuing model. The following preconditions were used for the simulation model; the ship interarrival times and shiploads were represented by negative exponential distributions, the shipload was stored in one pile and there was no stochastic storage time. The average ship waiting time (Wt) as function of the inverse of the ship unloading rate (1/) was determined using simulation. Fig. 3 shows these results together with analytical results for the M/M/3-queuing model. The minor deviation for the increasing values for the average stacker–reclaimers utilization () can be explained by the fact that storage allocation is included in the simulation model. If there is no area available to store the new shipload,
Stacker–reclaimer rescheduling will be investigated for both layouts of Fig. 1 using the input parameters of Table 1. For both layouts it was assumed that the stockyard’s areas have sufficient storage capacity to prevent that ships cannot be unloaded because there was no storage area available. Previous research [23] has shown that the ships interarrival times can be modeled using standardized distributions depending on the terminal type. For this case, it was assumed that the terminal acts as a stevedoring import terminal that has to serve many clients and does hardly have any influence on the ships arrival times. The interarrival time distribution can then best be represented by the negative exponential distribution. Historical data of unloaded ships during three years of operation from a specific import terminal was used as input for the shiploads. In van Vianen et al. [23] this measured shipload distribution was listed. The transportation route consists of multiple belt conveyors in series together with a loader or an unloader and a stacker–reclaimer. If one machine or belt conveyor fails, the entire route fails. Tewari et al. [24] used negative exponential distributions for the failure and repair distributions to represent the technical availability of belt conveyors. Historical data of belt conveyors at an export terminal during one year of operation showed an average conveyor availability between 0.9 and 0.97. For this case, it was assumed that each (un)loader, stacker–reclaimer and belt conveyor had an identical technical availability of 0.97. ˙ in million tons per During the experimental study, the annual throughput (m) year [Mt/y] was varied and the average ship (Wship ) [h] and the average train port time (Wtrain ) [h] were measured. Fig. 4 shows for both layouts these measured port times. From this figure can be concluded that ships and trains spent less time in the port when layout A will be applied compared to layout B. The higher stacker–reclaimer redundancy on the middle lanes results in a better division of stacking and reclaiming operations across the stacker–reclaimers causing less waiting times for ships and trains. For both layouts the average ship and train port times were measured when the rescheduling function was activated. These results are shown in Fig. 5. From this figure can be concluded that rescheduling of stacker–reclaimers will decrease the average train port time. For layout A, the average ship port time remains more or less constant when the rescheduling function was activated but for annual throughput
(A)
(B) 150
150
Layout B Layout A
Layout B Layout A 120 Wtrain [h]
Wship [h]
120
90
90
60
60
30
30
0
0 10
12
14 ṁ
16
18
20
10
12
14 ṁ
16
18
20
Fig. 4. Measured average port times for ships (A) and for trains (B) versus the annual throughput for the layouts as shown in Fig. 1.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model
ARTICLE IN PRESS
JOCS-286; No. of Pages 6
T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
5
Fig. 5. Measured average port times for ships and trains versus the annual throughput for layout A (A) and layout B (B) of Fig. 1 as function of stacker–reclaimer rescheduling.
presented can be used for supporting decisions whether rescheduling of the stacker–reclaimers operation can decrease the average train port time while still guaranteeing the seaside performance. Results from an experimental study have shown that ships and trains are served quicker when wide stockyard lanes, which are accessible by two stacker–reclaimers, are used. From the experimental study can also be concluded that rescheduling of the stacker–reclaimers operation will decrease the average train port time. The reduction of the train port time caused a minor increase of the average ships port time when a terminal layout with dedicated stacker–reclaimers lanes was used. The average ships port time remains approximately constant for the wide-lane terminal layout.
3.5 3
ni [-]
2.5 2 1.5 1
0.5
Layout B Layout A
Acknowledgement
0 10
12
14
16
[Mt/y]
18
20
Fig. 6. Average number of interruptions per ship for both layouts versus the annual throughput.
values larger than 17 Mt/y, the average ship port time even decreased. The handling activities were better spread across the three stacker–reclaimers that resulted in a reduction of the average machine utilization. This reduction of the average stacker–reclaimer utilization lead to, especially for the relatively high values of the stacker–reclaimers utilization of 0.6 to realize an annual throughput of more than 17 Mt, a decrease of the average ship waiting time. The increase of the average ship port time for layout B when the rescheduling function was activated can be explained by the limited stacker–reclaimers redundancy. A temporary interruption of the stacking operation will cause an increase of the ship port time. The larger reduction of the average train port time for layout A (see Fig. 5A) compared to layout B (see Fig. 5B) can be explained by the average number of interruptions per ship (ni ). Fig. 6 shows these numbers versus the annual throughput for both layouts. For the same value of the shipload distributor, the ship unloading operation in layout B is more frequent interrupted than in layout A due to the lower stacker–reclaimers redundancy. Each interruption requires two times repositioning of the stacker–reclaimer, which takes half an hour, and introduces ineffective machines hours.
6. Conclusions In dry bulk import terminals stacker–reclaimers are occupied for a long time during stacking of materials from ships. When a train arrives at the terminal’s landside to pick up material that is stored in the reach of an active stacker–reclaimer, this train has to wait before getting serviced. Waiting of trains may result in an unacceptable service to rail operators. The simulation model
The authors acknowledge the terminal operators who provided operational data about the ships arrival process and storage process of bulk materials at their stockyards and gave valuable feedback during the research. References [1] R. Robinson, Regulating efficiency into port-oriented chain systems: export coal through the Dalrymple Bay Terminal, Australia, Marit. Policy Manag. 34 (2007) 89–106. [2] C. Bierwirth, F. Meisel, A survey of berth allocation and quay crane scheduling problems in container terminals, Eur. J. Oper. Res. 202 (2010) 615–627. [3] S. Dipsar, T. Altiok, Control policies for material flow in bulk-port marine terminals, in: Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics, Canada, 1998, pp. 3083–3088. [4] P. Molck, R. Goncalves, T. Caldas, J. Valentim, L. Lima, E. Newton, M. Franc¸a, R. Mendes, F. Gomide, Intelligent stockpile building in iron ore shipping yard, in: Proceedings of the Third International Conference on Intelligent Processing and Manufacturing of Materials (IPMM 2001), Canada, 2001, pp. 1–11. [5] M. Ago, T. Nishi, M. Konishi, Simultaneous optimization of storage allocation and routing problems for belt conveyor transportation, J. Adv. Mech. Des. Syst. Manuf. 1 (2007) 250–261. [6] J. Leech, Design of an efficient coal export terminal, in: Proceedings of the Queensland Mining & Engineering Exhibition, Australia, 2010. [7] J. Leech, Optimizing a bulk minerals export chain, Min. Mag. 11 (2012) 42–48. [8] N. Umang, M. Bierlaire, I. Vacca, Exact and heuristic methods to solve the berth allocation problem in bulk ports, Transp. Res. E 54 (2013) 14–31. [9] T. Robenek, N. Umang, M. Bierlaire, S. Ropke, A branch-and-price algorithm to solve the integrated berth allocation and yard assignment problem in bulk ports, Eur. J. Oper. Res. 235 (2) (2014) 399–411. [10] B. Kim, S.Y. Chang, J. Chang, Y. Han, J. Koo, K. Lim, J. Shin, S. Jeong, W. Kwak, Scheduling of raw-materials unloading from ships at a steelworks, Prod. Plan. Control 22 (4) (2011) 389–402. [11] T.A. van Vianen, D.L. Mooijman, J.A. Ottjes, R.R. Negenborn, G. Lodewijks, Simulation-based operational control of a dry bulk terminal, in: Proceedings
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004
G Model JOCS-286; No. of Pages 6
ARTICLE IN PRESS T. van Vianen et al. / Journal of Computational Science xxx (2014) xxx–xxx
6
[12] [13]
[14]
[15] [16] [17]
[18] [19] [20]
[21] [22] [23]
[24]
of the 2012 IEEE International Conference on Networking, Sensing and Control (ICNSC 2012), 2012, pp. 73–78. D. Hu, Z. Yao, Stacker–reclaimer scheduling in a dry bulk terminal, Int. J. Comput. Integr. Manuf. 25 (2012) 1047–1058. A.H. Nasr, T.Y. ElMekkawy, Robust and stable flexible job shop scheduling with random machine breakdowns using a hybrid genetic algorithm, Int. J. Prod. Econ. 132 (2011) 279–291. R. Sarker, M. Omar, S.M.K. Hasan, D. Essam, Hybrid evolutionary algorithm for job scheduling under machine maintenance, Appl. Soft Comput. 13 (2013) 1440–1447. T.A. van Vianen, J.A. Ottjes, G. Lodewijks, Stockyard dimensioning for dry bulk terminals, in: Proceedings of the 2012 Trail-Beta Congress, Rotterdam, 2012. G. Lodewijks, D.L. Schott, J.A. Ottjes, Modern dry bulk terminal design, Port Technol. Int. 43 (2009) 87–94. C. Meng, S.S. Nageshwaraniyer, A. Maghsoudi, Y.J. Son, S. Dessureault, Datadriven modeling and simulation framework for materials handling systems in coal mines, Comput. Ind. Eng. 64 (2013) 766–779. B.P. Zeigler, H. Praehofer, T.G. Kim, Theory of Modeling and Simulation, Academic Press, San Diego, 2000. G.S. Fishmann, Discrete Event Simulation–Modeling, Programming and Analysis, Springer-Verlag, New York, 2001. H.P.M. Veeke, J.A. Ottjes, Tomas tool for object-oriented modeling and simulation, in: Proceedings of the Business and Industry Simulation Symposium ASTC, United States, 1999. J.P.C. Kleijnen, Verification and validation of simulation models, Eur. J. Oper. Res. 82 (1995) 145–162. I. Adan, J. Resing, Queueing Theory – Lecture Notes, Eindhoven University of Technology, 2002. T.A. van Vianen, J.A. Ottjes, G. Lodewijks, Modelling arrival process at dry bulk terminals, in: Proceedings of the Bulk Ports, Terminals and Logistics 2012 Conference, Amsterdam, 2012. P.C. Tewari, I.P. Singh, M.K. Khare, Reliability analysis of a conveyor belt system, with only one server, subject to failures and idleness after repair, Microelectron. Reliab. 31 (1991) 823–826.
Teus van Vianen received his MSc degree in mechanical engineering at the Delft University of Technology in 2005. After his study, he worked for five years in the machinery industry. In June 2010, he returned to the Delft University and started his Ph.D. project to determine the fundamentals of dry bulk terminal design at the faculty of Mechanical, Marine and Materials Engineering (3ME). His main interests are dry bulk terminals, terminal design, queuing theory, routing & scheduling, storage allocation and automation of transport systems.
Jaap Ottjes studied physics at Delft University of Technology and received the MSc degree in 1970. He obtained his Ph.D. degree in the ‘prediction of pressure losses in dilute phase pneumatic transport systems’ at the same university. After his Ph.D. research that he changed subject and specialized, partly as a consultant, on industrial logistics and simulation of transport systems. He was employed as associate professor in transport engineering and logistics at Delft University of Technology within the Department of Marine and Transport Technology at the faculty 3ME. His main research interests are in the field of industrial logistics and transport systems and in the modeling and simulation of these systems in particular. Professor Gabriel Lodewijks studied Mechanical Engineering at Twente University and Delft University of Technology from which he obtained a master degree (cum laude) in 1992. He specialized in transport technology, material engineering and dynamics. He obtained his Ph.D. degree at Delft University of Technology on the dynamics of belt conveyor systems in 1996. After 5 years of experience in the industry including three years in the USA, he was appointed as professor of transport engineering and logistics at the 3ME faculty of Delft University of Technology. He wrote over 250 papers and contributed to four books. He is frequently commissioned to act as an expert witness and he is further president of Conveyor Experts B.V.
Please cite this article in press as: T. van Vianen, et al., Simulation-based rescheduling of the stacker–reclaimer operation, J. Comput. Sci. (2014), http://dx.doi.org/10.1016/j.jocs.2014.06.004