S. I. Nekrasov a) and V. I. Gornostaev b)
Currently, the solution of the traveling salesman problem has been introduced in almost every existing software package both in cargo transportation and logistics.[2] Unfortunately, the procurement of new application licenses for small enterprises may seem expensive. Additionally, most of the existing software provides solutions to trivial tasks, and the convenience of their use is mainly on the generation of reports and other accompanying documentation. It should also be noted that in most cases low-cost software systems do not have an analytical block to vary the parameters in order to further optimize them.
Thus, the study aims to develop a simulation model to improve the efficiency of the transportation process of an industrial enterprise, with the possibility of performing computer experiments with various system parameters. It is necessary to find the optimal path search, as well as to optimize the number of vehicles under desired conditions.
The following researchers have worked out the solutions for efficiency improvement of transport support: Shcherbakov, V.V., Shash, N.N., Mirotin, L.B., Kirkin, A.P., Gorev, A.E., George L. Michael.
Their studies show the theoretical foundations of efficiency improvement of transport support. In these references, the relevance of this issue is provided insight, as well as the main factors affecting the efficiency of the rolling stock [6].
Under these papers, the action mechanism of the transport support system was reviewed in detail, namely, its scope of application, various techniques for solving the set goals and objectives [9] These techniques are mainly intended for analyzing the transport support process of large enterprises.[1]
Today, many Russian enterprises, adapting foreign experience, are introducing various approaches to logistic support management based on MRP, MRPII, ERP, DRP, etc. systems into their production and economic activities [11]. Currently, one of the most attractive systems is the “just-in-time” (“just-in-time”, JIT-system). JIТ-systems are reflected in many business scopes, including in everyday life, for example, food delivery [3]. In the case of using JIT systems for industrial enterprises, one of the clear examples is the manufacturing and delivery process of bitumen. The process is marked by accurate delivery times and the production capacity of the enterprise. In this regard, under the current study, the model development will be performed and configured in accordance with the particularities of the manufacturing process and the implementation of bitumen delivery.
Currently, the solution of the traveling salesman problem has been introduced in almost every existing software package both in cargo transportation and logistics [8]. Unfortunately, the procurement of new application licenses for small enterprises may seem expensive. Additionally, most of the existing software provides solutions to trivial tasks, and the convenience of their use is mainly on the generation of reports and other accompanying documentation. Moreover, the abovementioned software packages do not have an analytical block to vary the parameters in order to further optimize them.
Having analyzed the software packages given in the table, we have concluded that to solve optimization problems, such as determining the rational number of vehicles for cargo transportation, depending on the enterprise's capacity and the number of incoming applications, it is better to develop a model on our own.
In our work, we used Anylogic. It is software – this is an object-oriented environment that has binding to GIS maps. It allows real-time modeling, as well as the design of unique objects in a software environment.
A distinguishing feature of AnyLogic is the ability to combine several modeling types at once under the solution of a single task. Totally, AnyLogic highlights three main types of simulation modeling: system dynamics, discrete event simulation and agent-based simulation.
Simulation of the transport support process. Let us consider a sample use of the combined approach. The simulated system is characterized equally as a sequence of actions, which is typical for the discrete event technique. Nevertheless, it consists of individual objects whose actions depend on their own particularities. The latter is already typical for agent-based modeling [4].
If we consider the agent-based approach in terms of practical application, then it can be accepted as a simulation modeling technique studying the actions of dispersed agents. Also, these actions define the behavior of the complete system. Agent is a link in the model that is capable of having a cache (memory), behavior, contacts, and so on. While applying, the main condition is the location and volume of the cargo to be delivered. After creating the first request, the model defines the optimal route and the number of necessary vehicles for delivery [7]. If a new request is added, it is recalculated. Considering the restrictions introduced by us (in our model, these restrictions are the shelf life of the cargo and the duration of the work shift). If the route derivation cannot be realized by the existing vehicles (for example, the load capacity or shift time does not allow), then a new one is added and other combinations are checked [10]. Figure 1 demonstrates a software job of work.
FIGURE 1. General model view.
Operation principle of the software:
Creating an optimal transportation route depending on the volume of the customer's order with the least number of vehicles used.
The order volume is controlled according to the schedule using discrete event simulation and a pseudo-random distribution function for dynamic route derivation.
The route is evaluated by a mathematical function using the solution of the Traveling Salesman problem: to determine the shortest path.
If the consumer's request was smaller than the possible volume of cargo transportation, then another order is chosen until the vehicle is fully filled [5]. If the order is received and the distance between consumers is below the critical one, then the shortest path is defined through two points. The algorithm for searching and processing the request is illustrated in figure 2.