Discrete Dynamics in Nature and Society

Discrete Dynamics in Nature and Society / 2019 / Article
Special Issue

Discrete Optimization for Dynamic Systems of Operations Management in Data-Driven Society

View this Special Issue

Research Article | Open Access

Volume 2019 |Article ID 2493638 | 11 pages | https://doi.org/10.1155/2019/2493638

Simulation Optimization of Discrete Logistics Processes: A Case Study on Logistics of an E-Commerce Enterprise in Shanghai

Academic Editor: Xinchang Wang
Received23 Nov 2018
Revised29 Dec 2018
Accepted03 Jan 2019
Published20 Jan 2019

Abstract

With the rapid development of online shopping in recent years, logistics distribution has received much attention from enterprises and online consumers. Logistics distribution involves many factors and complex processes; conventional qualitative methods are unable to provide an effective analysis. Thus, this paper sets a framework to solve the above problem. A case study of an E-commerce enterprise in Shanghai on logistics distribution is proposed to discretize the whole process and minimize the total costs. Then the AnyLogic software is used to simulate and optimize the system from three aspects, including routes selection, warehouses quantity, and warehouses layout. Finally, this paper analyzes the simulation results, which would provide some valuable references for practical logistics.

1. Introduction

In recent years, E-commerce industry has developed rapidly. Online shopping is almost necessary to everyone. Logistics distribution is the last link of online shopping whose importance is rising as society demand increases. Whether goods can be delivered to consumers in time affects the consumers satisfaction of this shopping directly. Especially in the annual “Double Eleven” shopping festival, parcel quantities in various regions have increased rapidly and many delivery points have exploded. As shown in Figure 1, the logistics orders of Double Eleven in 2018 have set a new record. Logistics industry has also transformed into the rapid development stage. It is estimated that distribution costs account for over 50% of the total operational costs. This problem needs to be solved urgently. Besides, logistics distribution as one of the three major contents (distribution, storage, and management) in the logistics field includes scheduling management, distribution tools, distribution routes, delivery time, natural environment, human resources, and so on. More are getting to the importance of logistics distribution. Therefore, in this paper, much attention is paid to optimize and analyze the logistics distribution, shorten the delivery time, improve the distribution efficiency, and reduce the distribution costs. The distribution tools are necessary in the whole process which occupy considerable resources under demands uncertainty. To lower the logistics costs dramatically, reducing this consumption is significant through optimization [1]. Hence this paper considers the vehicle as the breakthrough point to analyze the costs optimization in system.

Meanwhile, the research on vehicle is not a theoretical problem in logistics distribution process. Lots of factors are contained including vehicles quantity, distribution terminal, delivery time, unloading time, and demand changes. Conventional qualitative methods are not insufficient to solve it. In recent years, computational technology including hardware and software has developed rapidly. This technology characterizes reflecting on complicated processes or behaviors to solve problems through simulation. Simulation is a new subject that has gradually formed with the development of computer technology. It was firstly proposed in the early 20th century and was mainly utilized in water conservancy research. Simulation is the process of experimental research on the system by establishing and using the real system model. Similar to the application of algorithm on theoretical issues, simulation has significant effects on practical problems, particularly the complex and practical problems like logistics distribution. By discretizing and dividing the whole process into different parts, an integrated model is established and analyzed for every part to obtain the system data. According to the obtained data, optimal results are calculated [2]. The methods have more practical value than the theoretical algorithm. Better simulation methods have been presented in recent years as the computer technology develops. Simulation on the application of logistics distribution will have broader prospects.

This paper studies the discrete logistics processes which include many stochastic variables and factors. The method of mathematical modeling is not suitable. Therefore, the simulation is used to optimize the logistics distribution system and get the practical results.

For the simulation research of logistics distribution problems, the GPSS language (The General Purpose Systems Simulator) was firstly presented by American Geoffrey Gordon in 1961, which is a solution to discrete events, particularly the queuing phenomenon [3]. Considering the combination of continuous system and discrete system, simulation language of hybrid system occurred after the 1970s, for instance, SLAM language (Simulation Language for Alternative Modeling), which translates the program into FORTRAN language and then compiles it into machine codes with high execution rate. The language can be used in discrete systems, continuous systems, and hybrid systems consisting of both [4]. KV has established an interactive model to support the logistics planning on container operations, which can provide the evaluation of the ports performance, optimize the utilization rate of ports, and shorten the turnaround time of ships [5]. Ila et al. proposed an eight-step simulation model development process (SMDP) to the design, implementation, and evaluation of logistics and supply chain study by adopting discrete events [6]. Meng et al. modeled the problem of free delivery in E-commerce companies with uncertain demands and discussed the influence of uncertain factors on the optimal solution [7]. Geng et al. established a self-organized elastic supply chain model based on MAS and designed the local fitness function, neighbor structure, and community interaction rules with the enterprises as agent. The results indicate that the system has an aggregation effect and its evolution can be controlled by MAS parameters [8]. Cui et al. transformed one step of RUL estimate of simulation model in discrete events logistics system into two steps. An instance validated the effectiveness and testified the performance of the two-step RUL estimation which is better than the one-step estimation [9]. Li et al. established a nonlinear model for vehicle routing planning problems subject to time windows which considered the influence of road irregularities on fresh fruits and vegetables. Compared to the results with the conventional models, the new model is superior to the prior one [10]. Lin et al. constructed a method of using transit signals first in the case of emergency evacuation among a sudden disaster without a specialized-transit channel. The bus signal priority dominated by four factors which could evacuate personnel and lessen time as much as possible. Paramics is utilized to simulate this method which is better than the other methods [11]. Thies et al. studied the effects of resources sharing on potential savings in logistics industry through a model of discrete instances and the resource utilization in installation stage of offshore wind energy generator based on Agent. The simulation results show that weather has a significant influence on the installation time, use time, and resources utilization rate. Meanwhile, the resources sharing has a potential saving on the installation of offshore wind energy generator [12]. Li et al. adopted the method of dynamic traffic network analysis and discussed the optimization of regional traffic organization. Then they simulated it with self-developed software and proposed the optimized model and algorithm to prove the method on availability and feasibility [13]. Wei et al. identified that, in a discrete-time VMI supply chain system composing of one retailer and one manufacturer, production fluctuations can be interestingly stable even if the retailer subsystem is fluctuant. Simulation experiments are used to verify the theoretical results on inventory and production fluctuation [14]. Teodor Gabriel et al. proposed an innovative classification method for the use of simulation in a complicated system of multimodal transportation with multi-participant. This method identifies the main findings, trends, and future routes in multi-dimension of multimodal transportation [15]. Mandi et al. presented a grouping formula based on the branch pricing to study the opportunity-constrained vehicle routing problem with stochastic demands and enhanced the solution quality by simulation experiments and sensitivity analysis [16]. Hu et al. proposed an urban traffic model (AUTM) for predicting and avoiding the traffic congestion. This model is used to the simulation of large-scale practical cases in different cities under different congestion conditions and has satisfactory results [17]. In response to the increasingly complicated logistics systems studies on discrete or continuous process, the number of simulation software arises with the rapid development of computer technology in related fields [18], which has developed a tendency to visualization, modularity, and intelligence. Modeling speed is accelerated and simulation effect is improved through prefabrication of various components. Many simulation platforms in this area are commonly seen on the market, for instance, Arean, Witness, Flexsim, Promodel, Automod, and so on.

This paper considers adopting a simulation tool, Anylogic software, which is developed early in this century to visualize modeling with a wide application scope. Complicated logistics distribution problems are discretized and simulated from the perspective of different processes. Moreover, one E-commerce enterprise in Shanghai is studied as a case. This paper starts with continuous changes of warehouse quantities and demands and optimizes the route selection, warehouses quantity, and warehouses layout. Ultimately, some optimization suggestions are raised based on simulation results of the software.

2. Application of AnyLogic Simulation Platform in Logistics Distribution Field

AnyLogic, a commercial simulation software released in 2000 by the AnyLogic Company, is a powerful simulation platform which can be applied in a wide range of fields, including logistics simulation, supply chain simulation, virus pervasion, road traffic, pedestrian evacuation, military simulation, and so on. This platform can also be used in discrete events modeling, agent-based modeling, and dynamics system modeling. This paper combines the AnyLogic technology with a case of an E-commerce enterprise in Shanghai on logistics distribution to propose the optimization suggestions.

The detailed introduction on the distribution case of an E-commerce enterprise in Shanghai is shown as follows.

E-commerce enterprises usually distribute goods to customers in two steps: (1) deliver goods from large-scale warehouses to distribution stations; (2) deliver goods from distribution station to customers with numerous manpower and material resources. Thus, two-part costs occur. The second costs are much more than the first one due to the large number of involving personnel. Moreover, higher risks of traffic accidents and loss of goods are generated. Hence, to save costs and reduce risks, the E-commerce company in Shanghai has established commodity self-raising points in various regions. The company only dispatches vehicles transferring goods to the distribution points and customers pick up the goods themselves so that the costs of the second part can be completely saved. At the same time, the injury of the delivery personnel and the loss of the goods are dramatically reduced.

According to statistics, 51 self-raising points have been established in the main urban area of Puxi by the E-commerce company. The distribution map is obtained through the AnyLogic platform as follows in Figure 2.

Firstly, one large-scale warehouse is considered to construct in Northwest of Shanghai outside the main urban area, which is responsible for goods distribution to self-raising points in main urban area of Puxi. The location is shown in Figure 3.

The AnyLogic simulation is used to model and calculate the cost and the time requirement of completing 51 self-raising points on vehicles, which are salient criterion for assessing the solution.

Many practical factors need to be considered before starting the simulation. For instance, the time requirements for delivery beyond the limits are causing the compensation to the customer called tardiness cost which increases the total costs. Besides, the number of goods from every self-raising point affects the total delivery time and some time-sensitive delivery requirements for subsequent points.

Because of the different situations every day, the above two factors of every self-raising point can generate the orders quantity and the delivery time requirement through a random function as the fundamentals of the simulation. Thus, the following four items are included in every self-raising point: (1) name; (2) location; (3) goods quantity; (4) time limits.

While large-scale warehouses only involve delivery, only the following two items need to be considered: (1) name; (2) location.

Two solutions are considered in this paper for route planning: (1) the shortest route solution, which calculates the shortest route between two locations as the real path for vehicles travelling; (2) the time-limited precedence solution, which considers firstly to deliver the goods with time-sensitive requirements and then the shortest route solution is adopted. Solution one can reduce the vehicles travelling costs and increase the tardiness costs while solution two is completely the opposite. Ultimately, the simulation results are used to compare the two solutions on the total costs and the total time.

Setting up three agents for the simulation of this problem, they are as follows: (1) self-raising point agent; (2) warehouse agent, (3) distribution vehicle agent.

The distribution vehicle agent is the main activity target, including controlling the vehicle from the warehouse, searching for the closest self-raising point and unloading the goods, continuously searching for a new point, and returning to the warehouse until all goods are unloaded.

The time to accomplish the delivery of the self-raising points is calculated to compute the travelling costs. The time span consists of the vehicle travelling time from one point to another and the unloading time at the terminal.

The calculation formula of total costs is as follows:

where

Meanwhile, the total time of entire system is calculated to judge if the delivery solution meets the criterion.

The calculation formula is

where

The logical structure to implement the functions of the distribution vehicle agents is composed of four states, six transitions, and one selection structure, as shown in Figure 4.

The following operation interface can be obtained through the above analysis and modeling, which is shown in Figure 5.

Simulation to the self-raising points of the E-commerce enterprise vehicle distribution can be performed by selecting the route planning mode and clicking the running button. Relevant data is obtained.

3. Simulation Performance and Optimization Analysis

This paper considers the logistics distribution settings under single warehouse, double warehouses, and three warehouses. The raw data of these three settings are all in Table 1. In a sharp increase of orders for special events like “Double Eleven”, we have increased the order number by ten times on the basis of the original data, which makes the research more realistic. Finally, we compare the results of simulation and provide suggestions for improvements.


NameLatitudeLongitudeNumber of ordersProduct categoryDelivery Time

s131.1431986121.42104071001100
s231.15728268121.4120836501150
s331.15546308121.43715711020
s431.16333719121.4147434901200
s531.1343072121.4481408301250
s631.16970695121.42529062020
s731.16707584121.44168963020
s831.18410185121.42661274020
s931.18935785121.42767374020
s1031.19520877121.41840941001300
s1131.1988346121.44611663020
s1231.18874702121.3715782020
s1331.20861033121.4197887501350
s1431.2231975121.39541633020
s1531.27327301121.43383364020
s1631.26069097121.44627383020
s1731.26820805121.45642571001400
s1831.28828157121.44294772020
s1931.29730166121.41107344020
s2031.24911288121.4890998601450
s2131.27418872121.46331774020
s2231.27321697121.46884533020
s2331.2796735121.4802926801500
s2431.30568695121.46691722020
s2531.23087848121.44370994020
s2631.28454415121.5018165901550
s2731.26859412121.53992822020
s2831.28451296121.51153453020
s2931.29881095121.4994631001600
s3031.30787625121.4993919901650
s3131.29473544121.54095583020
s3231.2911163121.5489509801700
s3331.33387484121.52881154020
s3431.31553869121.36996883020
s3531.31371912121.35057422020
s3631.35066483121.45383754020
s3731.1259537121.38661342020
s3831.18216029121.40562144020
s3931.22972111121.40526141001750
s4031.22416167121.37737223020
s4131.23514491121.40780124020
s4231.23383979121.43184263020
s4331.24339301121.4188242020
s4431.24539919121.41379983020
s4531.23813096121.3559084020
s4631.25328122121.38723423020
s4731.26148927121.41841972020
s4831.25603512121.37321773020
s4931.26155358121.35453723020
s5031.28576633121.34127854020
s5131.19722065121.46983444020

3.1. Comparison of Two Solutions in Single Warehouse Mode

Only one delivery at all distribution points is accomplished in this mode. Assuming the shortest route solution is selected to the next point, the following results are obtained by running the AnyLogic:

TotalRunCost:41657.899;

TotalDelayCost:1074320.791;

TotalCost:1115978.69.

Meanwhile, some critical data on the sequence of every distribution point, time nodes, and time length during distribution processes are shown in Table 2.


NameArrive TimeLeft TimeRun LengthUnload LengthdelayLengthrunCostDelay CostHouse Name

WareHouseA0.000.000.000.000.000.000.00WareHouseA
s508.0428.048.0420.000.00804.000.00WareHouseA
s3535.1845.187.1410.000.00714.000.00WareHouseA
s4859.1174.1113.9415.000.001394.000.00WareHouseA
s4676.4691.462.3415.000.00234.000.00WareHouseA
s4997.33112.335.8715.000.00587.000.00WareHouseA
s45119.04139.046.7120.000.00671.000.00WareHouseA
s40143.81158.814.7715.000.00477.000.00WareHouseA
s14163.05178.054.2415.000.00424.000.00WareHouseA
s41182.12202.124.0720.000.00407.000.00WareHouseA
s39204.91254.912.7950.000.00279.000.00WareHouseA
s43259.37269.374.4610.000.00446.000.00WareHouseA
s44271.22286.221.8515.000.00185.000.00WareHouseA
s42292.76307.766.5515.000.00655.000.00WareHouseA
s25311.13331.133.3620.000.00336.000.00WareHouseA
s13337.75362.756.6225.000.00662.000.00WareHouseA
s10367.06417.064.3150.0067.06431.0013412.WareHouseA
s9419.23439.232.1820.000.00218.000.00WareHouseA
s8441.26461.262.0320.000.00203.000.00WareHouseA
s38465.79485.794.5220.000.00452.000.00WareHouseA
s4490.17535.174.3845.00290.1438.0058032WareHouseA
s2536.73561.731.5725.00386.7157.0077346WareHouseA
s1566.45616.454.7150.00466.4471.0093290WareHouseA
s3621.10626.104.655.000.00465.000.00WareHouseA
s6633.17643.177.0710.000.00707.000.00WareHouseA
s7648.96663.965.7915.000.00579.000.00WareHouseA
s11671.47686.477.5215.000.00752.000.00WareHouseA
s51691.79711.795.3220.000.00532.000.00WareHouseA
s20724.95754.9513.1630.00274.91316.0054990WareHouseA
s23762.95802.958.0040.00262.9800.0052590WareHouseA
s22806.11821.113.1615.000.00316.000.00WareHouseA
s21823.64843.642.5320.000.00253.000.00WareHouseA
s17845.93895.932.2950.00445.9229.0089186WareHouseA
s16898.68913.682.7515.000.00275.000.00WareHouseA
s15917.43937.433.7520.000.00375.000.00WareHouseA
s18942.59952.595.1610.000.00516.000.00WareHouseA
s19959.43979.436.8420.000.00684.000.00WareHouseA
s47989.60999.6010.1710.000.001017.000.00WareHouseA
s241014.171024.1714.5710.000.001457.000.00WareHouseA
s301030.211075.216.0345.00380.2603.0076042WareHouseA
s291077.771127.772.5650.00477.7256.0095554WareHouseA
s261131.391176.393.6245.00581.3362.00116278WareHouseA
s281178.761193.762.3715.000.00237.000.00WareHouseA
s311200.601215.606.8415.000.00684.000.00WareHouseA
s321217.661257.662.0640.00517.6206.00103532WareHouseA
s271264.991274.997.3310.000.00733.000.00WareHouseA
s331288.871308.8713.8820.000.001388.000.00WareHouseA
s361325.591345.5916.7220.000.001672.000.00WareHouseA
s341371.661386.6626.0715.000.002607.000.00WareHouseA
s121419.801429.8033.1310.000.003313.000.00WareHouseA
s371446.561456.5616.7610.000.001676.000.00WareHouseA
s51470.341485.3413.7815.0012201378.00244068WareHouseA
WareHouseA1541.581541.5856.240.000.005624.000.00WareHouseA

According to Table 2, obviously the total time accomplishing the whole process (among vehicle departing and returning to the warehouse) is 1541.58 virtual time with single warehouse (the shortest route solution).

As shown in Table 2, some goods have time-restricted requirements that must be delivered before time limits, or tardiness costs occurred. The time-limited precedence solution is utilized to minimize the tardiness costs when the next station route needs to be selected. In other words, the goods with time requirements should be delivered first and then the shortest route solution is adopted to deliver goods without time requirements.

Through AnyLogic platform, the results of distribution costs in single warehouse (time-limited precedence solution) are as follows:

TotalRunCost:49796.137;

TotalDelayCost:59418.424;

TotalCost:109214.56.

Meanwhile, 1622.96 virtual times are required to accomplish the entire distribution procedure.

Comparisons of the results on the shortest route solution and the time-limited precedence solution in single warehouse mode are shown in Table 3.


ShortestPathLimitTimeCompare Result

TotalRunCost41657.89949796.137+19.54%
TotalDelayCost1074320.79159418.424-94.47%
TotalCost1115978.69109214.561-90.21%
FinishTime1541.5801622.960+5.28%

As presented in Table 3, the time-limited precedence solution causes the increments in total traveling costs by 19.54% and the abatements in total tardiness costs by 94.47%. The total costs are also reduced by 90.21%, because this solution can reduce the tardiness costs of the self-raising points with time limits effectively. Nevertheless, without adopting the shortest route solution, the vehicle travelling costs are raised and the total distribution time also increases by 5.28%. Because the amount of cost reduction is far larger than the increments, the total costs are reduced. In summary, under the circumstances of generating order number randomly, to the self-raising points with time-restricted requirements, the time-limited precedence solution has lower tardiness costs and total costs, but the traveling costs and distribution time increased merely.

Moreover, the total time span to complete the distribution is long whatever the two solutions are. In practice, reducing the delivery time to the customers significantly is an important problem the E-commerce companies face. This paper considers adding one warehouse, that is, double warehouses mode to solve this problem.

3.2. Simulation Comparison of Double Warehouses Mode and Single Warehouse Mode

The distribution in double warehouses mode is delivering goods to all the self-raising points simultaneously with two warehouses. According to the warehouse layout of the E-commerce enterprise, the distribution warehouse is added to the west of Shanghai interplaying with the original one. The distribution of double warehouses and self-raising points is shown in Figure 6.

The simulation consequences of double warehouses in the shortest route solution by AnyLogic are shown as follows:

TotalRunCost:42067.155;

TotalDelayCost:377939.599;

TotalCost:420006.754.

It takes 848.60 virtual time to accomplish the entire distribution process.

Similarly, following consequences of double warehouses can be obtained in the time-limited precedence solution:

TotalRunCost:51287.608;

TotalDelayCost:45760.399;

TotalCost:97048.007.

It takes 890.88 virtual time to accomplish.

The consequences of two solutions are compared in Table 4.


ShortestPathLimitTimeCompare Result

TotalRunCost42067.15551287.608+21.92%
TotalDelayCost377939.59945760.399-87.89%
TotalCost420006.75497048.007-76.89%
FinishTime848.600890.880+4.98%

As shown in the Table 4, in double warehouses mode with the time-limited precedence solution, the total traveling costs and delivery time increase by 21.92% and 4.98%, the total tardiness costs and the total costs reduce by 87.89% and 76.89%. The reasons are the same as the single warehouse mode that without adopting the shortest route solution results in an increase in the travelling costs and entire delivery time. Nevertheless, the tardiness costs are significantly reduced so that the total costs are reduced.

In order to comprehend the influence of the warehouse quantity on the distribution costs and the delivery time, above results are compared to gain Tables 5 and 6.


SingleHouseDoubleHouseCompare Result

TotalRunCost41657.89942067.155+1%
TotalDelayCost1074320.791377939.599-64.82%
TotalCost1115978.690420006.754-62.36%
FinishTime1541.580848.600-44.95%


Single HouseDouble HouseCompare Result

TotalRunCost49796.13751287.608+3%
TotalDelayCost59418.42445760.399-22.99%
TotalCost109214.56197048.007-11.14%
FinishTime1622.960890.880-45.11%

By comparing Tables 5 and 6, the double warehouses model can significantly reduce the total time compared to the single warehouse model in the two solutions. Particularly under the circumstances of the shortest route solution, the tardiness costs and the total costs are reduced dramatically, because the two warehouses distribute goods simultaneously to different self-raising points which reduces the time. In the shortest route solution, the delay time is so long that reducing delivery time can have great effects, while, in the time-limited precedence solution, the delay time is tiny without evident efficiency. In conclusion, the double warehouses mode has more effects than the single one in tardiness costs and total costs in the shortest route solution.

Moreover, in special shopping festivals, for instance, 11.11, 618, and so on, goods quantities increase dramatically on every self-raising point of the E-commerce companies. Therefore, this article will enlarge ten times of the order number for each point to study this problem with AnyLogic.

Simulation results of double warehouses in the shortest route solution are shown (the order number is magnified by ten times):

TotalRunCost:42067.155;

TotalDelayCost:8148314.73;

TotalCost:8190381.884.

Meanwhile, the overall distribution time is 6428.60 virtual time.

Simulation time of double warehouses in the time-limited precedence solution is as follows (the order number is magnified by ten times):

TotalRunCost:51287.608;

TotalDelayCost:2758085.139;

TotalCost:2809372.747.

Meanwhile, the overall distribution time is 6470.88 virtual time.

Compared with the normal order number, two tables are obtained as in Tables 7 and 8.


NomalOrders10timesOrdersCompare Result

TotalRunCost42067.15542067.155+0%
TotalDelayCost377939.5998148314.73+2055.98%
TotalCost420006.7548190381.884+1850.06%
FinishTime848.6006428.600+657.55%


NomalOrders10timesOrdersCompare Result

TotalRunCost51287.60851287.608+0%
TotalDelayCost45760.3992758085.139+5927.23%
TotalCost97048.0072809372.747+2794.83%
FinishTime890.8806870.880+671.25%

As shown in Tables 7 and 8, the tardiness costs, the total costs, and the time requirements are many times larger than the normal one during the special shopping festival. The double warehouses model is completely incapable of meeting the practical demands. Therefore, this paper considers adding another warehouse to reduce the time and costs in three warehouses mode.

3.3. Three-Warehouse Mode and the Influence of Distribution Area Adjustment on the Simulation Results

Three warehouses are utilized to be distributed simultaneously in three-warehouse mode which is also the fundament for multiple warehouse distribution research. In this paper, the warehouse location of the E-commerce enterprise in Shanghai is taken as example. We expand the double one to three-warehouse model. The distribution of warehouses and self-raising points are shown in Figure 7.

The simulation consequences of three warehouses in the shortest route solution can be obtained by AnyLogic as follows:

TotalRunCost:45398.807;

TotalDelayCost:132487.919;

TotalCost:177886.726.

Meanwhile, the overall distribution time is 641.76 virtual time.

Similarly, the simulation consequences of three warehouses in the time-limited precedence solution are obtained as follows:

TotalRunCost:52468.394;

TotalDelayCost:15795.939;

TotalCost:68264.333.

Meanwhile, the overall distribution time is 676.48 virtual time.

Comparing the mentioned consequences with the two warehouses in two solutions, a significant improvement has occurred. However, by observing the three warehouses and distribution sites in Figure 7, the new warehouse WareHouseC is close to the WareHouseB. Therefore, the location of WareHouseC is adjusted to the southwest of Shanghai so that the nearby self-raising points have a relatively close warehouse to save the delivery time and total costs. The distribution of the adjusted warehouse and self-raising points are shown in Figure 8.

As shown in Figure 8, the WareHouseC is located in the southwest of Shanghai and these three warehouses are distributed evenly after adjustment of the location.

With the AnyLogic, the simulation consequences of three warehouses (adjusted location) in the shortest route solution can be obtained:

TotalRunCost:42057.032;

TotalDelayCost:49689.094;

TotalCost:91746.126.

Meanwhile, the overall distribution time is 697.08 virtual time.

Similarly, the simulation consequences of three warehouses (adjusted location) in the time-limited precedence solution can be obtained:

TotalRunCost:49418.701;

TotalDelayCost:0;

TotalCost:49418.701.

Meanwhile, the overall distribution time is 747.00 virtual time.

Comparison of the simulation results in three warehouses mode without adjusting, Tables 9 and 10, is obtained.


OriginalChangedCompare Result

TotalRunCost45398.80742057.032-7.36%
TotalDelayCost132487.91949689.094-62.5%
TotalCost177886.72691746.126-48.42%
FinishTime641.760697.080+8.62%


OriginalChangedCompare Result

TotalRunCost52468.39449418.701-5.81%
TotalDelayCost15795.9390-100%
TotalCost68264.33349418.701-27.61%
FinishTime676.480747.000+10.42%

As shown in Tables 9 and 10, after the adjustment of warehouse location, the traveling costs, the tardiness costs, and the total costs are reduced, particularly in the time-limited precedence solution. All the time requirements in initialization data are met as the warehouse location adjusts and the tardiness costs are not generated. In this paper, the overall distribution time increases, which is primarily due to comprehensive conditions of self-raising points. If differences are shown, then the overall distribution time may be decreased.

4. Conclusions

This paper uses AnyLogic simulation software to model and simulate the vehicle distribution process. Then the results are analyzed and optimized on three factors including routes selection, warehouses quantity, and warehouses layout. As shown in the simulation consequences, the time-limited precedence solution can dramatically reduce the tardiness costs and the total costs; increasing the warehouses quantity can significantly lessen the overall delivery time; vehicle travelling costs, tardiness costs, and total costs can also be reduced by distribution of warehouse locations reasonably, which also have an influence on the overall delivery time. What can also be observed from the results is that the method studying logistics distribution by AnyLogic is feasible, which can visualize complicated problems and improve operability effectively. More optimal methods and algorithms like heuristic can be used in future research. The optimization module also can be contained.

Data Availability

All the data used to support the findings of this study are included in our manuscript and are available from the corresponding author upon request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

References

  1. L. Zhen, “Tactical berth allocation under uncertainty,” European Journal of Operational Research, vol. 247, no. 3, pp. 928–944, 2015. View at: Publisher Site | Google Scholar | MathSciNet
  2. L. Zhen, E. P. Chew, and L. H. Lee, “An integrated model for berth template and yard template planning in transshipment hubs,” Transportation Science, vol. 45, no. 4, pp. 483–504, 2011. View at: Publisher Site | Google Scholar
  3. P. F. I. Casas and J. Casanovas, “JGPSS, an open source GPSS framework to teach simulation,” in Proceedings of the 2009 Winter Simulation Conference, WSC 2009, vol. 1-4, pp. 256–267, IEEE, Austin, Tex, USA, December 2009. View at: Google Scholar
  4. C. D. Pegden, A. Alan, and B. Pritsker, “SLAM: Simulation language for alternative modeling,” Simulation, vol. 33, no. 5, pp. 145–157, 1979. View at: Publisher Site | Google Scholar
  5. K. V. Ramani, “An Interactive Simulation Model for the Logistics Planning of Container Operations in Seaports,” Simulation, vol. 66, no. 5, pp. 291–300, 1996. View at: Publisher Site | Google Scholar
  6. I. Manuj, J. T. Mentzer, and M. R. Bowers, “Improving the rigor of discrete-event simulation in logistics and supply chain research,” International Journal of Physical Distribution & Logistics Management, vol. 39, no. 3, pp. 172–201, 2009. View at: Publisher Site | Google Scholar
  7. Q.-C. Meng, T. Zhang, M. Li, and X.-X. Rong, “Optimal order strategy in uncertain demands with free shipping option,” Discrete Dynamics in Nature and Society, Art. ID 578280, 6 pages, 2014. View at: Publisher Site | Google Scholar | MathSciNet
  8. L. Geng, R. B. Xiao, and X. Xu, “Research on MAS-based supply chain resilience and its self-organized criticality,” Discrete Dynamics in Nature and Society, vol. 2014, Article ID 621341, 14 pages, 2014. View at: Publisher Site | Google Scholar
  9. Y. Cui, J. Shi, and Z. Wang, “Discrete Event Logistics Systems (DELS) simulation modeling incorporating two-step Remaining Useful Life (RUL) estimation,” Computers in Industry, vol. 72, pp. 68–81, 2015. View at: Publisher Site | Google Scholar
  10. P. Q. Li, J. He, D. Y. Zheng, Y. S. Huang, and C. H. Fan, “Vehicle routing problem with soft time windows based on improved genetic algorithm for fruits and vegetables distribution,” Discrete Dynamics in Nature and Society, vol. 2015, Article ID 483830, 8 pages, 2015. View at: Publisher Site | Google Scholar
  11. C. Y. Lin and B. W. Gong, “Transit-based emergency evacuation with transit signal priority in sudden-onset disaster,” Discrete Dynamics in Nature and Society, vol. 2016, Article ID 3625342, 13 pages, 2016. View at: Publisher Site | Google Scholar
  12. T. Beinke, A. Ait Alla, and M. Freitag, “Resource Sharing in the Logistics of the Offshore Wind Farm Installation Process based on a Simulation Study,” International Journal of e-Navigation and Maritime Economy, vol. 7, pp. 42–54, 2017. View at: Publisher Site | Google Scholar
  13. S. B. Li, G. M. Wang, T. Wang, and H. L. Ren, “Research on the Method of Traffic Organization and Optimization Based on Dynamic Traffic Flow Model,” Discrete Dynamics in Nature and Society, vol. 2017, Article ID 5292616, 9 pages, 2017. View at: Publisher Site | Google Scholar
  14. Y. Wei, F. Chen, and H. Wang, “Inventory and production dynamics in a discrete-time vendor-managed inventory supply chain system,” Discrete Dynamics in Nature and Society, vol. 2018, Article ID 6091946, 15 pages, 2018. View at: Publisher Site | Google Scholar | MathSciNet
  15. T. G. Crainic, G. Perboli, and M. Rosano, “Simulation of intermodal freight transportation systems: a taxonomy,” European Journal of Operational Research, vol. 270, no. 2, pp. 401–418, 2018. View at: Publisher Site | Google Scholar | MathSciNet
  16. M. Noorizadegan and B. Chen, “Vehicle routing with probabilistic capacity constraints,” European Journal of Operational Research, vol. 270, no. 2, pp. 544–555, 2018. View at: Publisher Site | Google Scholar | MathSciNet
  17. W. B. Hu, H. Wang, Z. Y. Qiu, L. P. Yan, C. Nie, and B. Du, “An urban traffic simulation model for traffic congestion predicting and avoiding,” Neural Computing Applications, vol. 30, no. 6, pp. 1769–1781, 2018. View at: Publisher Site | Google Scholar
  18. L. Zhen, “Modeling of yard congestion and optimization of yard template in container ports,” Transportation Research Part B: Methodological, vol. 90, pp. 83–104, 2016. View at: Publisher Site | Google Scholar

Copyright © 2019 Xiang Xu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


More related articles

1149 Views | 514 Downloads | 0 Citations
 PDF  Download Citation  Citation
 Download other formatsMore
 Order printed copiesOrder

Related articles

We are committed to sharing findings related to COVID-19 as quickly and safely as possible. Any author submitting a COVID-19 paper should notify us at help@hindawi.com to ensure their research is fast-tracked and made available on a preprint server as soon as possible. We will be providing unlimited waivers of publication charges for accepted articles related to COVID-19. Sign up here as a reviewer to help fast-track new submissions.