=Paper=
{{Paper
|id=Vol-3013/20210121
|storemode=property
|title=An Agent-Based Simulation for Optimizing the Parameters of a Railway Transport System
|pdfUrl=https://ceur-ws.org/Vol-3013/20210121.pdf
|volume=Vol-3013
|authors=Viacheslav Matsiuk,Olga Galan,Andrii Prokhorchenko,Volodymyr Tverdomed
|dblpUrl=https://dblp.org/rec/conf/icteri/MatsiukGPT21
}}
==An Agent-Based Simulation for Optimizing the Parameters of a Railway Transport System==
https://ceur-ws.org/Vol-3013/20210121.pdf
An Agent-Based Simulation for Optimizing the Parameters of a
Railway Transport System
Viacheslav Matsiuk1, Olga Galan2, Andrii Prokhorchenko3 and Volodymyr Tverdomed2
1
National University of Life and Environmental Sciences of Ukraine, Heroiv Oborony Str.15 building 3, Kyiv,
Ukraine, 03041
2
State University of Infrastructure and Technology, Ivan Ogienko Str., 19, Kyiv, Ukraine, 03049
3
Ukrainian State University of Railway Transport, Feuerbach sq., 7, Kharkiv, Ukraine, 61001 4
Abstract
The article presents results of optimization of the railway transport system's parameters. The
researches were made by computer simulation. The simulation model is developed based on
real supply chains of iron ore concentrate from the Poltava Mining and Processing Plant
(Ukraine) to transition points within Ukraine (Pivdennyi seaport, Izmail seaport, Chop rail
station). The simulation model was developed and implemented in the AnyLogic and Java SE
environment and is based on discrete-event and agent-based principles. The simulation model
is the interaction by agents of the railway transport business processes, loading and unloading
points, and vehicles. Due to results of the optimization and sensitivity experiments was possible
to determine the optimal fleet of locomotives and cars; and to establish the basic transport-
technological indicators of the annual transport work.
Keywords 1
agent-based simulation, AnyLogiс, Java SE, railway transport technologies, parameter
optimization
1. Introduction
The main task of the National transport system is to fully meet the transportation needs of the
economy. In addition, there is a certain conflict of interest between transport companies, cargo owners
and passengers. On the one hand, customers demand the highest quality of the transportation services;
on the other hand, transport companies are limited in productive resources. Therefore, the key and
topical issue of transportation organization is the search for optimal technological parameters of
transport systems depending on the planned volume of transportation.
Establishing the optimal parameters of railway transport systems is one of the key points in their
design, construction and operation. While the criterion of optimality can be the maximum level of
reliability (reliability or fault tolerance) of processes [1] or safety in terms of the method of organization
[2] or maintenance of critical transport infrastructure [3]. These criteria are a systemic indicator of
effectiveness. Moreover, this issue is equally relevant both for global multimodal processes [4] and for
distributive city logistics [5].
The most common ways to optimize transport processes are analytical methods and models. Such
methods are actively used in the planning of transport processes [6], including the use of Big Data
technologies [7], the search for rational routes of distribution logistics [8].
However, the most effective tool for the study of complex and large technological systems, in terms
of reliability, completeness and convenience, is simulation. This tool is quite actively used in the study
of a wide range of applied transport issues, ranging from logistics of mining companies [9] and
optimization of ordering [10], ending with planning issues based on Big Data [11].
ICTERI-2021, Vol I: Main Conference, PhD Symposium, Posters and Demonstrations, September 28 – October 2, 2021, Kherson, Ukraine
EMAIL: vimatsiuk@gmail.com
ORCID: 0000-0003-2355-2564 (A. 1); 0000-0001-8350-6565 (A. 2); 0000-0003-3123-5024 (A. 3); 0000-0002-0695-1304 (A. 4)
©️ 2020 Copyright for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
2. Materials and methods
2.1. Theoretical substantiation
In accordance to the peculiarities of the organization of railway transport production, the main
technological parameters of railway transport systems have always been three components: the required
fleet of locomotives, the required fleet of cars and the available capacity of railway routes. Together,
these three elements directly proportionally determine the carrying capacity of the railway transport
system:
Ncc = f ( Nloc , Ncar , Ncap ) , (1)
where Nloc – regulatory fleet of locomotives;
Ncar – regulatory fleet of cars;
Ncap – regulatory capacity of the railway lines.
In expression (1), the parameter (Ncap) should be understood not as the maximum possible, given the
technical equipment of the railway transport infrastructure, the size of train traffic, but the set number
of trains that will ensure the planned volume of cargo Ncc. Then the parameters of the need for rolling
stock (Nloc, Ncar) will also depend on the planned size of the train traffic (Ncap).
On the other hand, the implementation of the planned transportation schedule will depend on the
available (sufficient) fleet of cars and locomotives. Therefore, the complex methodology for
determining the required parameters of the railway transport system to ensure the required carrying
capacity has certain difficulties due to the interdependence of variables and function (1). In addition, in
the conditions of branched railway networks, there are certain problems in estimating the required fleet
of locomotives and cars, as the required amount of transportation is a certain set within different
directions.
Given the peculiarity of the technological process of railways, the main element of their
technological equipment are locomotives. First, the locomotive fleet is the costliest element in the
structure of operating costs. In addition, the fleet of locomotives has the greatest impact on the
performance of the entire railway transport system. The delivery time (considering the expectation of a
free locomotive) should be the minimum possible. Therefore, it is advisable to form the following
optimization problem:
( )
Tdel = f Nloc , N car , N cc = f ( N cap ) → min,
р ( Nloc ) н , (2)
р ( N car ) н ,
де (Nloс), (Ncar) – utilization rate of locomotives and cars, respectively;
р – the limit of the rationality of the using parameters;
н – the limit of reliability (fault tolerance) of the using parameters.
Since the optimization model (2) is presented implicitly, its solution will be carried out by
simulation.
2.2 Development of a simulation model
Given the complexity and, at the same time, the discreteness of railway transport technologies, the
model will be based on discrete-event and agent-based principles. AnyLogic University Researcher
([12], License Serial Number # 03926) with a built-in compiler Java SE was chosen as the simulation
development environment. The simulation is carried out through the interaction of the following agents:
1. Main – the main agent through which the presentation and interaction of all other agents of
the model should be conduct.
2. Production – an agent of simulations of production processes of railway transport.
3. DestinationPoint (Ncc) – population of agents of delivery cargo points.
4. Locomotive (Nloс) – the population of agents of locomotives.
5. Car (Ncar) – population of car agents.
6. Order (Ncc) – the population of agents of applications for transportation.
The model is implemented on the example of supply of iron ore concentrate from Poltava Mining
and Processing Plant (Ukraine). Delivery is carried out by international traffic, all of which pass through
three exit points in Ukraine: the ports of Pivdenny and Izmail and the Chop railway station. Therefore,
the whole railway transport process is an interaction of the point of origin of the freight mass and the
three points of their conditional repayment. The operational work of the fleet of vehicles is executed
centrally within three railway directions. The Main agent by means of a real GIS-marking of railway
routes (fig. 1) carries out simulation of train movement by the network.
Figure 1: Presentation at Agent Main
The logic of the whole business process begins with the population of the DestinationPoint
agents (Fig. 2), which simulates the stochastic accumulation of cargo mass to the norm of composition.
Figure 2: Business process of the agents DestinationPoint
The source block exponentially generates events of a cargo consignment of cargo (Ncc) to the
point of accumulation - the sink block. At formation of necessary, for loading of one structure of weight
of cargo, the Java-code is realized:
«Order order = new Order( this );
send ( order, main.production);»
simulating the sending of an information request for readiness to send the appropriate consignment.
The main business process is modeled in the Production agent (Fig. 3, 4).
Figure 3: Business process of Production agent (train movement on the network)
The event of the information message about the readiness of the cargo consignment to be sent
by one of the three routes is sent to enter block (Fig. 3). Then the application is sent to the waitingCar
block, where at the entrance the condition of the required number of free cars available for loading is
checked:
«if (waitMainDepo.size() > 0 ){
waitingCar.stopDelay(waitingCar.get(0));
waitMainDepo.stopDelay(waitMainDepo.get(0));
}»
When the condition is met, the time delay for loading the corresponding consignment into the
selected cars (trainForming block) is simulated. After that, the loaded train is waiting for a free
locomotive (seize block). If there is a free locomotive (block locomPool, with the number of
locomotives equal to Nloc) block seize "captures" this resource and starts a subprocess that simulates the
processing of the train on departure and following it is by network to the destination: blocks start,
trainServiceAndWait, moveToDistPoint.
Block delay1 simulates the processing of the train on arrival at the destination; it is unloading
and processing before returning to the station of the next load. The moveTo block simulates the return
flight of the train to the loading station, where the production resources (locomotive and cars) are sent
to the sludge waiting for the next sending.
The second subprocess of the Production agent (Fig. 4) simulates the turnover of cars in the
train. This subprocess is almost completely controlled by the first subprocess (Fig. 3).
Figure 4: Business process of the Production agent (car turnover)
2.3 Implementation of the model and search for optimal parameters.
The model was implemented based on open data on the commercial activities of the Poltava Mining
and Processing Plant (Ukraine) for 2015 – 2017. Duration of the modeling time – one calendar year.
The initial data are indicated in the table. 1, the simulation results are summarized in the table. 2, 3 and
fig. 5 – 8.
Table 1
Initial data on the supply of iron ore concentrate in Ukraine from the Poltava Mining and Processing
Plant (Ukraine) for 2015 – 2017
Parameter Value
Annual quantity of sent cargo, by route (in trains), Ncc:
- South port 1375
- the port of Izmail 625
- Chop station 900
Number of cars in the train 56
Train weight, net tons 3920
Route speed of the train by network, km / h 40
The required number of locomotives, Nloc calculated
The required number of cars, Ncar calculated
Table 2
Optimal calculated values
Estimated parameter Value
Estimated number of locomotives, Nloc 9
Estimated number of cars, Ncar 616
Locomotives’ utilization factor, (Nloс) 0.66
Cars’ utilization factor, (Ncar) 0.64
Table 3
The model results (one model year)
Indicators of the time (hours)
Average batch delivery time 15.6
Turnover of cars, of them: 33.3
- in a usage (movement and conducting operations) 20.2
- waiting for cargo 12.0
- waiting for locomotives 1.1
Locomotive turnover 27.3
Mileage indicators, km
Weighted average route length 796
Dimensions of trains:
- in average per the day 7.9
- in average per the year 2895
Car mileage:
- in average per the day 574
- in average per the year 209493
Locomotive mileage:
- in average per the day 701
- in average per the year 256047
Indicators of efficiency of transportation work (per year)
Thousand tons per car 18
Mln. tons-km per car 3859
Thousand tons per locomotive 1261
Mln. tons-km per locomotive 322858
The need of rolling stock for the organization of supply 1 mln. tons per year:
- the needs in locomotives, Nloc: 0.793
- the needs in сars, Ncar: 54.3
Thanks to optimization experiments, the dependence of the average delivery time on the
calculated fleet of locomotives and cars are obtained.
50 ,1.0
40 ,0.8
Delivery time, hours
Utilization factor,
(Nloс), (Ncar)
30 ,0.6
20 ,0.4
10 ,0.2
0 ,0.0
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Delivery time Utilization factor of cars Utilization factor of locomotives
Figure 5: The dependence of delivery time and coefficients of use of cars and locomotives depending
on the estimated fleet of locomotives (fleet and volume of traffic are constant, correspond to the
values of tables 1 and 2).
0.14
0.12
0.1
Probability
0.08
0.06
0.04
0.02
0
,23.2
,36.8
,11.2
,12.0
,12.8
,13.6
,14.4
,15.2
,16.0
,16.8
,17.6
,18.4
,19.2
,20.0
,20.8
,21.6
,22.4
,24.0
,24.8
,25.6
,26.4
,27.2
,28.0
,28.8
,29.6
,30.4
,31.2
,32.0
,32.8
,33.6
,34.4
,35.2
,36.0
,37.6
,38.4
The middle of the infraval grouping, hours
Figure 6: Density of time distribution f consignment of cargo
,0.30
,0.25
,0.20
Probability
,0.15
,0.10
,0.05
,0.00
,10.6
,2.2
,0.2
,0.6
,1.0
,1.4
,1.8
,2.6
,3.0
,3.4
,3.8
,4.2
,4.6
,5.0
,5.4
,5.8
,6.2
,6.6
,7.0
,7.4
,7.8
,8.2
,8.6
,9.0
,9.4
,9.8
,10.2
,11.0
,11.4
,11.8
,12.2
The middle of the infraval grouping, hours
Figure 7: Density of distribution of time of waiting of a locomotive
9
8
7
Parties of cargo, in trains
6
5
4
3
2
1
0
47
93
1013
1059
1105
1151
1197
1243
1289
1335
1381
1427
1473
1519
1565
1611
1657
1703
1749
1795
1
139
185
231
277
323
369
415
461
507
553
599
645
691
737
783
829
875
921
967
The end of the model time, for five years
Figure 8: The formation of a queue of ready-to-ship cargo waiting for cars (one unit corresponds to
the rate of loading into the warehouse, i.e., 3920 tons)
The density of the locomotive waiting time distribution (Fig. 7) is exponential, which confirms
the typical process of failure formation. The obtained "typicality" indicates a completely natural
modeling process and the adequacy of the simulation model itself.
The nature of the formation of the queue of ready-to-ship cargo indicates sufficient fault tolerance
of the delivery process (Fig. 8): the queue is formed, but the technological system has enough internal
reserves to get out of the state of temporary failure.
3. Discussion and conclusion
Centralized management of the fleet of locomotives and cars can significantly increase the efficiency
of their use, as applications for freight are formed in one turn. This is especially true of branched
networks. The presented simulation model allows estimating the real need for locomotives and cars in
the conditions of organization of supply chains on the branched transport network. The agent principle
of model building and formalization of business process logic allowed modeling real supply chains in
the conditions of formation of a single queue for transportation: one point of departure and several
delivery points. It is under such conditions that it is theoretically possible to achieve significant
productivity and efficiency of transportation organization. Given that the presented business process
can be considered close to the typical in the realities of Ukrainian railways, it can be argued that for the
railway transport network of Ukraine, in theory, it is quite possible to achieve results in other segments
of freight. To organize the transportation of one million tons per year of consignments, the need for
locomotives and cars will be approximately 0.793 and 54.3, respectively.
Another important result of this study is the confirmation of the greatest impact on the stability and
efficiency of the organization of rail freight transportation of the locomotive fleet. Reducing the number
of locomotives below the critical value leads to significant delays in the movement of goods and
disruption of the stability of most units of the railway, including the cars fleet. The increase in the fleet
of locomotives has almost no effect on the acceleration of delivery and reduction of the load of the
transport system, and only leads to a decrease in the load factor of the locomotives themselves.
4. References
[1] V. Matsiuk et al., “Improvement of efficiency in the organization of transfer trains at
developed railway nodes by implementing a ‘flexible model,’” Eastern-European J.
Enterp. Technol., 2019, doi: 10.15587/1729-4061.2019.162143.
[2] V. Samsonkin, O. Goretskyi, V. Matsiuk, V. Myronenko, A. Boinik, and V. Merkulov,
“Development of an approach for operative control over railway transport technological
safety based on the identification of risks in the indicators of its operation,” Eastern-
European J. Enterp. Technol., vol. 6, no. 3–102, pp. 6–14, 2019, doi: 10.15587/1729-
4061.2019.184162.
[3] M. Katsman, V. Myronenko, V. Matsiuk, and P. Lapin, “Approach to Determining the
Parameters of Physical Security Units for a Critical Infrastructure Facility,” Reliab.
Theory Appl., vol. 16, no. 1(61), pp. 71–80, 2021, doi: 10.24412/1932-2321-2021-161-
71-80.
[4] A. Mazaraki, V. Matsiuk, N. Ilchenko, O. Kavun-Moshkovska, and T. Grygorenko,
“Development of a multimodal (railroad-water) chain of grain supply by the agent-based
simulation method,” Eastern-European J. Enterp. Technol., vol. 6, no. 3 (108), pp. 14–
22, Dec. 2020, doi: 10.15587/1729-4061.2020.220214.
[5] N. Shramenko, D. Muzylyov, and V. Shramenko, “Methodology of costs assessment for
customer transportation service of small perishable cargoes,” Int. J. Bus. Perform. Manag.,
vol. 21, no. 2, pp. 132–148, 2020, doi: 10.1504/IJBPM.2020.106113.
[6] R. Lovelace, J. Parkin, and T. Cohen, “Open access transport models: A leverage point in
sustainable transport planning,” Transp. Policy, vol. 97, pp. 47–54, Oct. 2020, doi:
10.1016/j.tranpol.2020.06.015.
[7] O. Iliashenko, V. Iliashenko, and E. Lukyanchenko, “Big Data in Transport Modelling
and Planning,” in Transportation Research Procedia, Jan. 2021, vol. 54, pp. 900–908, doi:
10.1016/j.trpro.2021.02.145.
[8] V. Sakharov, S. Chernyi, S. Saburov, and A. Chertkov, “Automatization Search for the
Shortest Routes in the Transport Network Using the Floyd-warshell Algorithm,” in
Transportation Research Procedia, Jan. 2021, vol. 54, pp. 1–11, doi:
10.1016/j.trpro.2021.02.041.
[9] O. V. Gorbova, “Formalization of the technological processes at railway stations based on
the step-by-step modeling,” Sci. Transp. Progress. Bull. Dnipropetr. Natl. Univ. Railw.
Transp., vol. 0, no. 5(83), pp. 71–80, Oct. 2019, doi: 10.15802/stp2019/181850.
[10] M. Bučková, M. Krajčovič, and M. Edl, “Computer Simulation and Optimization of
Transport Distances of Order Picking Processes,” in Procedia Engineering, Jan. 2017, vol.
192, pp. 69–74, doi: 10.1016/j.proeng.2017.06.012.
[11] D. Ziemke, B. Charlton, S. Horl, and K. Nagel, “An efficient approach to create agent-
based transport simulation scenarios based on ubiquitous Big Data and a new, aspatial
activity-scheduling model,” in Transportation Research Procedia, Jan. 2021, vol. 52, pp.
613–620, doi: 10.1016/j.trpro.2021.01.073.
[12] https://anylogic.help/anylogic/index.html