Journal of Intelligent & Fuzzy Systems 40 (2021) 10417–10433 10417

DOI:10.3233/JIFS-200610
IOS Press

A system dynamics model for routing and

scheduling of cross-dock and transportation
in reverse logistics network of perishable
goods
Pezhman Abbasi-Tavallalia , Mohammad Reza Feylizadeha,b, * and Atefeh Amindousta
a Department of Industrial Engineering, Najafabad Branch, Islamic Azad University, Najafabad, Iran
b Department of Industrial Engineering, Shiraz Branch, Islamic Azad University, Shiraz, Iran

Abstract. Cross-dock is defined as the practice of unloading goods from incoming vehicles and loading them directly into
outbound vehicles. Cross-docking can simplify supply chains and help them to deliver goods to the market more swiftly and
efficiently by removing or minimizing warehousing costs, space requirements, and use of inventory. Regarding the lifetime
of perishable goods, their routing and scheduling in the cross-dock and transportation are of great importance. This study
aims to analyze the scheduling and routing of cross-dock and transportation by System Dynamics (SD) modeling to design
a reverse logistics network for the perishable goods. For this purpose, the relations between the selected variables are first
specified, followed by assessing and examining the proposed model. Finally, four scenarios are developed to determine the
optimal values of decision variables. The results indicate the most influencing factors on reaching the optimal status is the
minimum distance between the cross-dock and destination, rather than increasing the number of manufactories.

Keywords: Scheduling, routing, transportation, cross-dock, reverse logistics network, perishable goods

1. Introduction products are stored in a dock for a short time – and are
then directly delivered to the customers in at most 12
Inventory flow control is one of the crucial con- hours [3]. In other words, although the cross-docking
cepts of supply chain management. Cross-docking strategy removes the inventory operations of a tradi-
has been assumed as an efficient method to control tional warehouse, it allows goods to be classified and
inventory flow, which is essential to supply chain unloaded via the integration process and then loaded
management [1]. Cross-docking, as a warehousing in vehicles [4]. If not all the vehicles of the pick-up
strategy, is a suitable approach for reducing inventory navy can arrive at the cross-dock at the same time,
needed to fulfill the demands of customers, which can the integration process is postponed after collecting
help decrease inventory resources by simplifying the all goods; as a result, the waiting time and inventory
flow between manufactories and producers [2]. At level increase exponentially. Most studies on cross-
the cross-dock, goods are directly shipped from the docking have addressed the exponential concept, its
receiving dock-doors to shipping dock-doors – where physical design, and its location [5].
Almost one-third of the annual produced food in
∗ Corresponding author. Mohammad Reza Feylizadeh, E-mail: the world for human consumption (about 1.3 billion
feylizadeh@iaushiraz.ac.ir. tons) is lost or wasted. Food losses are about 680

ISSN 1064-1246/$35.00 © 2021 – IOS Press. All rights reserved.

10418 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation

and 310 billion $ in the industrialized and develop- Liao [9] presented a mathematical planning model
ing countries, respectively [6]. Even if just one-fourth for integrated vehicle routing and scheduling of
of the food currently wasted in the world could be multi-purpose cross-dock in post-distribution mode
saved, it would be sufficient to feed 870 million hun- involving different product types. Also, Gunawan
gry people in the world [7]. The time spent during et al. [10] studied the integration of the vehicle routing
the logistics operations and environmental conditions problem with a crossover dock called Vehicle Routing
during transportation and warehousing dramatically Problem with Cross-Docking (VRPCD). The purpose
affect this loss [8]. of this study is to find a set of routes to deliver prod-
Therefore, in this paper, the redesign of the reverse ucts from a set of suppliers to a set of customers
logistics network of perishable goods is examined at through a cross-dock so that operating and transporta-
the strategic level. So far, many studies have been tion costs are minimized without violating vehicle
done on this issue, most of which have been at capacity and time constraints.
the operational and tactical level using mathematical Gunawan et al. [10] studied the integration of the
planning. This study considered different scenarios at vehicle routing problem (VRP) with cross-docking.
the strategic level, which are dynamically applied to Accordingly, in this study, researchers studied the
the system over time. Finally, among these scenarios, issue of VRPCD, which is widely used because of
an optimal situation at the macro level, for routing and its ability to reduce the overall costs of a supply
scheduling cross-dock and transportation in reverse chain network. The purpose of VRPCD was to deter-
logistics network of perishable goods, is achieved. mine the number of vehicles used and vehicle-related
To investigate this issue and optimize it, we examine routes using the Adaptive Large Neighborhood
scheduling and routing of transportation and cross- Search (ALNS) algorithm. In another study, Birasnav
docking via the SD method. Since the SD method et al. [11] presented a Mix Integer Linear Program-
considers the interaction between effective factors, ming Model (MILPM). The MILPM presented in this
the obtained results are suitably valid. study for the Vehicle-Specific Routing problem was
In this study, the proposed model’s influential focused on the transfer of matchboxes from man-
variables are extracted by reviewing the literature, ufacturing companies to retailers via a cross-dock
followed by investigating the relations between them. (cross-linking center) using a service provider third-
The model is then analyzed and validated by spec- party logistics was centralized. Due to the importance
ifying cause-and-effect relationships and plotting of the problem, the researchers in this study presented
them in Vensim software. Finally, four scenar- a new solution based on the MILPM.
ios are defined to analyze different conditions to Jansen [11] analyzed the logistics network con-
reach the proposed model’s key variables’ optimal cerning the effects of a new warehouse. They
values. developed efficient routing and planning methods to
The remainder of this paper is organized as follows. use a complex logistics network. He & Li [12] intro-
Section 2 presents a literature review to extract effec- duced a Mixed-Integer Programming (MIP) model to
tive variables in scheduling and routing of cross-dock minimize distance in the routing process to describe
and transportation in the inverse logistics network the Dynamic Schedule Problem (DSP). Rahmandoust
of perishable goods. Section 3 presents the applied & Soltani [13] proposed a nonlinear multiproduct
methodology. Section 4 describes the case study Vehicle Routing Problem (VRP) with heterogeneous
and solves its problem by designing, validating, and vehicles to find the probable minimum number of
assessing the SD model. Finally, Section 5 represents cross dockings among the available sets of discrete
the concluding remarks. locations and minimize the total cost of establish-
ing cross-docking centers and vehicle transportation
costs (i.e., distribution and operation cost) as well.
Küçükoğlu & Öztürk [14] addressed the VRP and
2. Literature review packing problem with cross-docking and proposed a
Mixed-Integer Linear Programming (MILP) model.
To provide a literature review and cover all LIU & And [15] studied the problem of location
the above-stated subjects, an investigation was per- routing with the limitation of the payment time
formed on previous studies on scheduling and routing window. They also applied the fuzzy processing
cross-dock and transportation and redesigning the method to define customer satisfaction performance
logistics network of perishable goods. to reduce costs, improve customer satisfaction, and
 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation 10419

enhance efficiency by selecting cross-docking centers controlling the return of products for repair, reman-
and arrange routes properly. Baniamerian et al. [16] ufacturing, recycling, reuse, or burning/disposal. Yu
designed a Profitable Heterogeneous Vehicle Routing & Solvang [28] proposed a model to provide a set of
Problem with Cross-Docking (PHVRCD) to increase Pareto solutions between benefit and environmental
the total benefit of a cross-docking connected system. performance. They investigated the effect of system
Rahbari et al. [7] proposed a bi-objective MILP model flexibility on the sustainable reverse logistics system
for the problem of routing and programming vehi- design. Rahimi & Ghezavati [29] proposed a multi-
cles with cross-docking for perishable products under period multi-objective MILP model to design and
the uncertainty in the travel time. Nikolopoulou et al. program the network benefit, increase social effects,
[17] presented a new routing problem of the public and decrease environmental effects in a reverse logis-
vehicle with cross-docking using an adaptive mem- tics network. Trochu et al. [30] evaluated the design
ory programming method coupled with a Tabu Search problem of a reverse logistics network under envi-
algorithm. They designed a set of pick-up and deliv- ronmental policies to recycle wood waste in the
ery routes with the minimum travel distance. Nasiri et construction, renovation, and demolition industry.
al. [18] introduced a MILP model wherein the selec- Khodaparasti et al. [31] presented a modified alloca-
tion and allocation of the order were incorporated tion model to avoid unwanted defects in criteria. This
into the VRPCD to minimize total costs, including problem was expressed as a covering model involving
purchasing, inventory, transportation, cross-docking, the facility capacity and demand elasticity.
and early/tardy delivery penalty costs. Mancini [19] Many studies have been conducted on reverse
introduced and formulated the hybrid VRP, which is logistics of perishable goods, mostly investigating
a developed form of VRP. In their proposed model, specific objectives (i.e., cost decrease and satisfaction
Hiassat et al. [20] added location decision as a strate- increase in most cases) and limitations by mathemat-
gic decision to the model developed by Lee et al. ical programming. However, time plays a key role
[21], who discussed the warehousing and transfer in the supply chain of perishable goods and delay
of blood units from hospital to specialized center. (for any reason) in each stage of the supply chain
This study addressed a location-inventory-routing process can lead to perishing products and cause
model for perishable goods. Shuang et al. [22] intro- financial/environmental losses. Accordingly, regard-
duced a reverse logistics production routing model ing the objectives of reducing procurement time,
with greenhouse gas emissions control policy selec- decreasing costs, and enhancing customer satisfac-
tion. The researcher assumed a reverse supply chain tion, the present study was conducted to design and
with remanufacturing options under various green- analyze a pattern using SD to solve the problem when
house gas emissions control policies. Kuşakcı et al. it is possible to collect, recycle, and demolish prod-
[23] studied the optimization o a reverse logistics ucts. The incorporation of the three objectives into the
network under fuzzy demand to prevent the rapid supply chain configuration of this study contributes
consumption of natural resources and convert gen- to reaching the highest productivity. Although many
erated waste into value for the economy. Yavari researchers have addressed the subject of vehicle
& Geraeli [24] investigated the design of a green scheduling, they have mostly disregarded the con-
closed-loop supply chain network for degradable junction of scheduling and routing cross-dock and
products under uncertain conditions using a MILP transportation. To the best of our knowledge, the
model to minimize costs and environmental pollu- number of studies on this subject is limited. Besides,
tion. Zhang et al. [25] studied a stochastic reverse their assumptions are far from reality, such that they
production routing model with environmental consid- have been identified as impractical due to simplifying
erations. Furthermore, they incorporated the reverse the applied models. Moreover, in the present study,
supply chain model with the remanufacturing option the effect of delivery time and quality is investigated
to reduce greenhouse gas emissions. Gardas et al. [26] on the performance of the reverse logistics network;
examined the reverse logistics in the automobile ser- the effective processes in scheduling and routing of
vice sector to reduce exploration and production of cross-dock and transportation are identified in man-
oil using the multi-criteria decision-making method. ufacturing units to optimize the supply chain; the SD
Liao [27] proposed the reverse logistics network method is applied to specify parameters due to the
design for recycling products and remanufacturing. uncertain nature of the real world; their interactions
They developed a generic Mixed-Integer Nonlinear are assessed in addition to the effects of different
Programming (MINLP) to maximize total profit by factors on the concerned logistics system.
10420 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation

We can summarize the paper’s contributions as

follows:
– Taking into account practical approach by intro-
ducing a case study with a potential of different
dynamics systems in order to cover all aspects
of Routing and Scheduling of Cross-Dock and
Transportation in Reverse Logistics Network of
Perishable Goods
– Considering uncertainty in demand, distribution
and Scheduling of Cross-Dock
– We estimated customer satisfaction, cross-dock
costs and product storage time in cross-dock for
60 next period.
– We running four different scenarios and camper
the result together to find which variables in most
Fig. 1. Implementation steps of this study.
important to a problem statement.
– Learning from the above simulation results, the
major factors influencing the VRP and cross-
docking systems is its unreasonable structure.
Figure 1 shows the implementation steps of this
study.
This study was performed to schedule and route
3. Methodology
cross-dock and transportation of the reverse logis-
tics network of perishable goods. In this respect, a
The modeling process in this study is implemented
hybrid decision-making framework was introduced.
using the SD, which has approximate and uncertain
A brief explanation of the implementation steps of
nature. Therefore, the model simulation was car-
this study is as follows. In the first step, an SD model
ried out for scheduling and routing of cross-dock
was developed to simulate key parameters of schedul-
and transportation in the reverse logistics network of
ing and routing of cross-dock and transportation in the
perishable goods using a practical research methodol-
reverse logistics network. The main outputs of the SD
ogy. In this qualitative-quantitative study, the required
model include cross-dock cost, product lead time in
data were collected using interviews and question-
the cross-dock, customer satisfaction, inventory hold-
naires. The literature review was first used to identify
ing cost, and customer demand. The subsystems of
research variables and their relations. Authentic
the dynamic model are as follows: transportation cost,
papers and views of the elite were used to col-
the average loss of perishable goods, prime price of
lect data and specify effective factors in scheduling
cross-docking operation per product, an average of
and routing of cross-dock and transportation in the
variable warehousing costs, and customer demand. In
reverse logistics network of perishable goods. The
the second step, four scenarios were proposed to opti-
data were analyzed by the SD approach for answering
mize the decision variables of the model. Library data
the research questions. Also, this investigation was
and information of elites were used to identify the fac-
conducted regarding three factors: 1) cross-dock cost,
tors effective in scheduling and routing of cross-dock
2) product lead time in the cross-dock, and 3) cus-
and transportation in the reverse logistics network of
tomer satisfaction. This study was carried out through
perishable foods. In the next step, the interactions
the following steps:
between the effective factors were investigated. After
– Step 1: Performing a literature review and col- the elite confirmed the effects of the specified fac-
lecting views of elite tors, the obtained data were analyzed by the Vensim
– Step 2: Analyzing the collected factors to specify software. The obtained results can be generalized to
ultimate factors all of the cross-docks to identify the effective fac-
– Step 3: Modeling the SD considering these fac- tors of these systems in the perishable goods sector
tors and their interactions and optimize them. In this way, the system would
– Step 4: Determination of SD outputs be optimized, and the overall performance would be
– Step 5: Conclusion enhanced.
 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation 10421

4. Case study investigated in the form of scheduling and routing

model of cross-dock and transportation in the logis-
This study investigated the use of the SD model tics network, as well as their effects on time, cost,
for Ramak Co., Iran. In this study, we used real data and customer satisfaction. Afterward, state variables
from Ramak Co. for five years, from 2013 to 2018. (i.e., time, cost, customer satisfaction) were defined
In the following, the executive steps of the research in the SD to design different strategies and scenarios.
will be explained in detail. Step 2: Identification of dynamic variables and
Step 1: Problem Statement hypotheses
In recent years, the study on the supply chain Table 1 lists the key criteria obtained from library
has received the attention of many researchers and research and views of elites, which were incorporated
industrialists. The supply chain is a set of methods into the model. The interactions of these criteria are
applied for the effective integration of suppliers, pro- depicted in Fig. 2.
ducers, distributors, and retailers to minimize system A dynamic hypothesis is a notion to decide which
costs, produce and distribute the correct number of structure can be formed in similar behavior in the
goods at the apt location and time, provide an accu- reference states. After making a hypothesis, some
rate distribution of resources, and meet customer diagrams are created to show the main driving mech-
demands [32, 33]. However, the supply chain for anisms of the dynamic behavior of the system. A
perishable goods has other conditions and features. model cannot be established without realizing feed-
Since the perishability of goods is a commonplace back loops itself. A good dynamic hypothesis and
phenomenon, products might lose abruptly or gradu- known fundamental mechanism mean that the system
ally their quality or value [34–36]. Fruits, vegetables, is provided with enough information to start calculat-
flowers, drugs, blood, dairy products, meat, and food ing surface and rate equations. Then, the individual
are examples of perishable goods. The perishability can formulate the next stage of the modeling process
of goods is the main concern to the supply chain [41]. Based on conversations with the elite and inves-
since the quality or value of most products decreases tigation of previous studies, it was revealed that some
during their lifetime. This perishability has a non- factors are more influential, and other factors were
linear function affected by many factors, including removed for the sake of simplification due to their
transportation and storage time [37, 38]. similar effects. In this study, four dynamic hypotheses
There are some critical factors in cross-dock which were proposed.
allow reaching the former objectives, namely appro-
priate decision-making for the allocation of trucks to – Dynamic hypothesis 1: Customer demand is in
doors, scheduling of trucks, and allocation of desti- direct proportion to cross-dock costs.
nations to trucks. After the departure of trucks from Cross-dock is a reliable method to reduce
cross-dock, crucial decisions need to be made, such inventory and meet customer demands. Since the
as routing of trucks and scheduling their path. It has total demand must be less than or equal to the
been intended to provide a more practical study and storage capacity, as the demand increases, the
decrease the defects of previous studies using the amount of goods in the warehouse to cover this
assumptions closer to reality. Therefore, this study demand increases, which leads to an increase in
aimed to apply an SD model in the form of schedul- the holding cost of the dock. As a result, the cost
ing and routing of vehicles in a logistics network to of cross-docking is increased.
reduce costs and time and increase customer satis- – Dynamic hypothesis 2: Product maintenance
faction level considering uncertainty in the logistics time is in direct proportion to inventory holding
network, particularly uncertainty in demand. Accord- cost in cross-dock.
ingly, different scenarios were defined to model According to Fallah et al. [42], the cost
factors affecting the scheduling and routing of cross- of holding, especially the cost of holding the
dock and transportation in the logistics network in a perishable goods, is not always the same and
cause-and-effect framework. Then the key objectives increases over time.
in this field were identified, based on the litera- – Dynamic hypothesis 3: Transportation cost is
ture review and interview with elites in the field of in reverse proportion to product storage time in
scheduling and routing of cross-dock and transporta- cross-dock.
tion in the logistics network. In the first step, effective According to the holding time of the prod-
indices for the conjunction of logistics networks were ucts, if the products need to be unloaded quickly
10422 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation

Table 1
Key variables affecting the scheduling and routing of cross-dock and transportation of the reverse logistics network of perishable goods
NO. Category Parameter Definition References
1 inventory holding cost Costs of holding each unit of inventory Rahmandoust & Soltani
Product storage time in

product in the warehouse [13]

2 Rate of incoming trucks Number of incoming trucks to the Rahbari et al. [7]
cross-dock per hour
cross-dock

3 Rate of outgoing trucks Number of outgoing trucks from the Rahbari et al. [7]
cross-dock per hour
4 Number of vehicles Number of vehicles available in the dock Rahmandoust & Soltani
[13]
5 Transportation cost Costs of available vehicles in the Rahmandoust & Soltani
cross-dock (e.g., fuel) [13]
6 Inventory rate of cross-dock Average cross-dock inventory per day Baniamerian et al. [16]
7 Rate of constant, required Average constant resources needed by Küçükoğlu & Öztürk [14]
resources the warehouse per day (e.g., electricity)
8 Reloading percentage Unloading incoming truck and loading Küçükoğlu & Öztürk [14]
outgoing truck
Cross-dock costs

9 Number of suppliers Number of suppliers cooperating with LIU & And [15]
the cross-dock
10 Number of doors and platforms Entrance and exit doors of the cross-dock LIU & And [15]
and loading/unloading platforms
11 Required space for storage of Required space to store products in the Nasiri et al. [18]
products cross-dock during the day (square meter)
12 Cost of required stuff to hold Pallet, container, etc. Baniamerian et al. [16]
inventory
13 Average costs of warehousing Costs depend on the number of products Küçükoğlu & Öztürk [14]
14 Time of activities Duration of activities from unlading Rahbari et al. [7]
process to loading and delivery process
15 Customers’ demand The average demand for the goods Nikolopoulou et al. [17]
Scheduling of transportation

managed by the cross-dock

16 Prime prices for the cross-dock Increase in the price of incoming good Baniamerian et al. [16]
operations per product due to the operations of cross-dock for
the delivery process
17 Rate of product delivery time The average time interval between order Pishvaee & Rabbani [39]
and delivery
18 Responsive percent of cross-dock Percentage of supply and Mogale et al. [34]
19 Time of shipment The average time interval between order Mogale et al. [34]
and shipment
20 Cost of each unit distance Costs of transportation for each unit path LIU & And [15]
21 The average distance of The average distance from manufactory Hiassat et al. [20]
manufactory from the destination to the customer (km)
Routing

22 The distance of manufactories The distance of manufactories from Hiassat et al. [20]
from cross-dock cross-dock (km)
23 The average time interval The time interval between receipt of LIU & And [15]
between unloading and goods and their distribution
distribution of products
24 Health rate of the delivered The ratio of Intact goods delivered to the Kovačić et al. [40]
Perishable good

product customer to total goods

25 The average lifetime of Expiration period Yavari & Geraeli [24]
perishable products
26 The average loss of perishable Non-delivered goods until their Yavari & Geraeli [24]
products expiration
27 Cost of warehousing and waste Costs of warehousing and effacement of Rahbari et al. [7]
effacement perished goods in the distribution process
28 Average return cost Costs of product return by the customer Rahimi & Ghezavati [29]
logistics

to dock
Reverse

29 Rate of returned products The ratio of returned goods to total goods Shuang et al. [22]
30 Price reduction of the returned Product price reduction after re-pricing Shuang et al. [22]
product
 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation 10423

Fig. 2. The proposed model.

from the dock, vehicles with higher capacity or et al. [45] stated that a 3 to 5-year period is rea-
more numbers are needed. So, the storage time sonable for investigating the results. Thus, 36
of the products decreases, the transportation cost months was selected in the present study based
increases. on a large number of variables, their many cal-
– Dynamic hypothesis 4: The average prime price culation results, views of the elite, and literature
of cross-dock operations for each product is in review so that sufficient time would be available
inverse proportion to customer satisfaction. for the operation of feedbacks.
Price is one of the four major factors involved – Model boundary: Processes, information feed-
in the marketing mix and is of strategic impor- back, delay in policy-making, and time are
tance because it is closely related to product the important elements of SD modeling. Since
positioning. Also, the issue of price and pricing endogenous and exogenous variables are spec-
has a significant impact on other marketing mix ified based on boundary alone, the boundary
factors, including product features and product adjustment is similarly important in SD model-
distribution and promotion decisions. Using a ing. Boundaries need to be determined regarding
marketing mix can ultimately increase customer the problem that is supposed to be solved. The
satisfaction. If the other three items, except the SD is a field wherein the modeling must be
price, are assumed to be constant, the lower the problem-based. Analysis of the problem and its
price, the higher the customer satisfaction. Then, signs help the researcher in this field [15]. Table 1
the cost price affects the final price. presents the key variables in problem modeling.
– Research loops: In this study, the cause-and-
Step 3: Plot of SD model Figure 2 introduces effect method was applied to generate the flow
the entire state, rate, and auxiliary variables listed diagram for modeling and show interactions
in Table 1, as well as their interactions. between effective factors. As already stated,
the factors listed in Table 1 were investigated,
– The time horizon of modeling: An appropriate and the cause-and-effect relations were speci-
time horizon plays a key role in the output results fied. Finally, flow diagrams of state, auxiliary,
of the SD and leads to more real results consider- and rate variables were obtained. Figures. 3,
ing the apt time [43]. Under normal conditions, 4, and 5 present the product storage time in
although the effects of learning and feedback cross-dock, cross-dock costs, and customer sat-
loops are not short-term, the time horizon of isfaction, respectively.
cause-and-effect loops is short. Accordingly, it
is necessary to make a balance between these Boloori et al. [46] showed that several variables
two sections. Sapiri et al. [44] and Andersen could affect the product storage time in cross-dock.
10424 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation

Fig. 3. Flow diagram of state variable of product storage time in cross-dock.

Fig. 4. Flow diagram of cross-dock costs.

The volume of goods being loaded in the dock is For example, the consumed fuel is higher for more
raised by increasing the rate of incoming trucks to number of vehicles and, in turn, transportation costs
cross-dock (since the number of vehicles is height- increase. Also, transportation costs are dependent on
ened). Accordingly, the time of activities for those some other variables, including each unit distance,
goods is prolonged, leading to an increase in the prod- percentage of reloading, and distance of manufac-
uct storage time in the cross-dock. Similarly, a low tories from the warehouse. Hence, a higher cost of
rate of outgoing trucks means an increase in the load- unit distance, more massive depreciation costs of the
ing time of goods in trucks and/or an imbalance of vehicle per km, and a higher percentage of reload-
goods and the number of vehicles, which prolongs the ing (due to more number of trucks or their loading
product storage time in the cross-dock. Therefore, the efforts) leads to an increase in transportation costs.
variable effects of vehicle numbers are definite on the The distance of manufactories from the warehouse
rate of incoming/outgoing trucks, and the only other can also have an impact on this variable because a
variable affected by the number of vehicles is trans- longer distance entails further movement of vehicles
portation cost, whose relationship is not complex. resulting in huger costs. This variable also has an
 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation 10425

Fig. 5. Flow diagram of customer satisfaction.

impact on the rate of incoming trucks. For different inventory. Also, the average distance of manufactory
distances between manufactories and warehouse, the from customers affects the product shipment time;
number of incoming trucks to the cross-dock is more because with an increase in the distance, the product
divergent in unit time. is delivered to the customer with delay. Moreover,
The inventory rate of the cross-dock is known as an increase in these two variables yields a reduc-
one of the outputs of product storage time in the cross- tion in the cross-dock responsiveness in the unit time
dock (a state variable). The cross-dock inventory in since the products are delivered to the customer later.
unit time is affected by this state variable. Generally, Figure 4 indicates the flow diagram of cross-dock
as the product storage time is prolonged, the cross- costs.
dock inventory increases in unit time due to the lack of Based on Fig. 4 and the study of Baniamerian et al.
shipment of goods or products from the dock. More- [16], variables with interactions with the state vari-
over, at greater cross-dock inventory rates, a larger able of cross-dock costs include the rate of constant,
space is needed to hold them, i.e., the space required required resources, the average of variable warehous-
for storing products is enlarged. Furthermore, with ing costs, and cross-dock inventory rate. As the rate
an increase in the inventory, the variable holding of constantly required resources and the average of
costs are escalated due to the need for further space, variable warehousing costs increase, cross-dock costs
equipment, and workforce. Also, inventory affects are elevated. Also, with an increase in the inven-
the rate of outgoing trucks. For a larger inventory, tory rate, inventory holding and cross-dock costs are
a further shipment of products and goods from the escalated as well. The average of warehousing costs
cross-dock is required due to the nature of cross- is also dependent on other variables. In this regard,
dock (i.e., the smallest surface). Therefore, as can for a shorter product lifetime, more expensive equip-
be observed, there are interactions between variables. ment is required for keeping them, which leads to
Inventory rate is also dependent on customer demand higher variable costs. Customer demand affects both
so that a higher volume of goods is needed for a variable cost and inventory rate. More products are
greater demand to be responsive, which increases the needed to meet the larger customer demand, which
10426 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation

increases inventory volume and consequent inventory

holding and warehousing costs. Figure 5 shows the
flow diagram of customer satisfaction.
Customer satisfaction is another state variable of
the present study. According to Baniamerian et al.
[16] delivery time is one of the critical variables influ-
encing customer satisfaction. Customer satisfaction
is enhanced by shortening delivery time due to lower
waiting times. Also, an increase in the intact deliv-
ered goods improves customer satisfaction since a
customer expects to receive an intact product. This
variable is manifested in the rate of returned prod-
ucts, i.e., customer satisfaction is lower at a higher
rate of returned products. The rate of delivery time Fig. 6. Effect of removing the factor of equipment cost required
is also affected by the distance of manufactory from for holding products on the cross-dock cost.
customer and responsiveness percentage of the cross-
dock such that a longer distance lengthens delivery
time and more powerful responsiveness shortens it
since more number of products are delivered in a
shorter time. With an increase in the rate of returned
products, its costs are elevated and the reduction of
the prime price of cross-dock operations per prod-
uct is higher. Warehousing and effacement costs are
affected by the average loss of total perishable goods
and affect the average prime price of cross-dock oper-
ations per product. For larger losses, the effacement
costs are raised, leading to an increase in the average
prime price. Besides, losses are escalated by decreas-
ing the average lifetime of perishable goods since if
a product is not delivered to the customer on time,
it will perish. The average time interval between the
unloading process and distribution has an impact on Fig. 7. Effect of removing the number of vehicles on the product
storage time in the cross-dock.
the losses, i.e., a longer time interval yields further
loss.
Step 4: Model test of this variable leads to a reduction of cross-
Jeng & An [47] proposed a set of tests for model dock costs. Figure 6 demonstrates the effect of
validation, which were performed in the present removing the factor of equipment cost required
study. for holding products on the cross-dock cost.
Figure 7 shows the effect of removing the number
• Boundary adequacy test: Given that key factors of vehicles on the product storage time in the cross-
affecting a model have been investigated in pre- dock.
vious studies and verified by elites, they have The removal of this factor causes a significant
already proved their effects. In the next steps, increase in the product storage time in the cross-dock
their effects on the entire model are indicated as well. Figure 8 indicates the effect of removing the
by removing each of the factors. Figure 6 shows factor of the average lifetime of perishable products
the effect of removing the factor of equipment on customer satisfaction. As can be observed, the
cost required for holding products, which has removal of this variable decrease’s customer reduc-
an impact on cross-dock cost. The removal of tion in general.
this factor implies neglecting its effects in the
simulations (rather than the absence of this vari- • Structure assessment test: Structure validation
able in the real word). As can be seen, these suggests that the applied relationships in the
variables are highly correlated, i.e., the removal model sufficiently consider real relationships in
 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation 10427

Fig. 10. Model behavior when the average of variable warehousing

Fig. 8. Effect of removing the factor of the average lifetime of costs approach infinity.
perishable products on customer satisfaction.

Fig. 9. Validation of equations’ structure in Vensim softwareTM .

line with the project objective. In this study,

Fig. 11. Model behavior under boundary conditions of outgoing
since the model equations were written in Ven- truck rate (approaching zero).
sim software, their accuracy was verified by the
software.
– Situation 2: A decrease in the rate of outgoing
• Boundary conditions test: This test is performed
trucks to zero (Fig. 11)
to investigate the model behavior when model
inputs are under boundary conditions, i.e., when As shown in Fig. 11, if the rate of outgoing trucks
they reach their highest or lowest levels. In other approaches small values, the product storage time
words, this test evaluates whether the model is in the cross-dock is dramatically prolonged. More
sustainable under these conditions. In the bound- specifically, when trucks do not depart from the
ary adequacy test, the status of variables under cross-dock, holding time increases as a natural con-
boundary conditions (the minimum and maxi- sequence.
mum values) is investigated as follows.
– Situation 3: The health rate of the delivered prod-
– Situation 1: An increase in the average of ware- uct is at the lowest level (Fig. 12)
housing costs toward infinity (Fig. 10)
As can be seen in Fig. 12, if the health rate of the
If the average of variable warehousing costs delivered products approaches its lowest limit, the
dramatically increases, the only output is the aug- level of customer satisfaction is drastically reduced
mentation of cross-dock cost, which is contrary to since customers expect to receive intact products.
the objectives of this study. In other words, since the Thus, the decrease in customer satisfaction is directly
average of variable warehousing costs is a part of the proportional to the health rate of the delivered prod-
total cross-dock cost, they are directly proportional. ucts.
10428 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation

Fig. 12. Model behavior under boundary conditions of the health Fig. 14. Enhancement of customer satisfaction.
rate of the delivered products.

Fig. 15. Product storage time in the cross-dock.

Fig. 13. Model outputs in 36- and 60-month periods.

• Integration error test: This test shows the sen- identifying the criteria. As shown in Fig. 14,
sitivity of the model results in selecting the customer satisfaction can be enhanced by con-
temporal period. To perform this test, the tem- trolling and reducing the average time interval
poral period was changed from 36 to 60 months. between the unloading process and distribu-
In Fig. 13 (36- and 60-month period diagrams), tion of products (duration between receiving
no variation is observed in the model behavior a product by the cross-dock and its distribu-
by changing the temporal period. Also, effective tion). As can be observed in Figs.15 and 16,
factors will enhance the performance provided with an increase in the number of manufactories,
that they are controlled. the product storage time in the cross-dock and
• Behavior reproduction test: The behavior repro- cross-dock costs are prolonged. Accordingly, the
duction test is conducted to determine whether product storage time in the cross-dock and cross-
this model can reproduce and replicate the dock cost can be diminished by reducing an
system behavior in real conditions. Regarding optimal number of manufactories. However, per-
the wide investigations of previous studies, the formance enhancement is a function of many
present study addresses the factors affecting factors whose coordination takes a longer time.
product storage time in the cross-dock, cross-
dock cost, and customer satisfaction. Therefore, Step 5: Design of system alternatives and their
it is possible to predict the system behavior after assessment
 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation 10429

Fig. 16. Cross-dock cost.

Fig. 17. Different scenarios for the state variable of product storage
time in the cross-dock.
In the final step, after analyzing the key factors
and their effects, it is necessary to define different
scenarios to reach practical strategies by specifying
various values of effective indices. In this study, as
per the views of the elite and some of the authentic
papers, different values were selected for variables in
both controllable and non-controllable fields for the
organization. These variables were accessible in var-
ious conditions, whether by the organization or the
environment. Hence, four scenarios were proposed
based on different values of these indices (Table 2
and Figs. 17–21). These four scenarios were writ-
ten considering the variables that could be changed
by the system, variables that were not decided by
the system but rather by the external environment,
and a combination of applicable values announced by Fig. 18. Different scenarios for the state variable of cross-dock
decision-makers. Table 2 lists the value sets of each cost.
scenario and their increase or decrease rate (based on
the nature of a variable).
routing strategies, which can have an impact
– Scenario 1: The number of manufactories on the transportation cost and rate of incom-
increased by 5, which increases the cross-dock ing trucks. The average distance of manufactory
inventory rate and the number of doors and from the destination declined by 5%, which
wharves, leading to an increase in the cross-dock affects the responsiveness percentage of cross-
cost. Besides, the distance of manufactories dock (percentage of supply and delivery) and
from the warehouse declined by 8% through product delivery time/rate. On the other hand,

Table 2
Specifications of the four scenarios
Variable Scenario 1 Scenario 2 Scenario 3 Scenario 4
Number of manufactories 5 –2 3 –3
Distance of manufactories from the ware house –8% –3% –5% –6%
Average distance of manufactory from the destination –5% –4% –2% –6%
Costs of required equipment for product storage 10% 15% 20% 25%
Percentage of reloading –5% –7% –8% –10%
Number of vehicles 10 15 20 25
Average time interval between unloading process and distribution of products –5% –7% –8% –10%
Average lifetime of perishable products 1% 2% 3% 4%
10430 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation

average of variable warehousing costs and the

rate of constantly required resources. More-
over, the percentage of reloading with a 5%
reduction leads to a decrease in transportation
costs. The number of vehicles increased by 10,
which affects the number of required doors and
platforms, cross-dock inventory rate, and rate
of incoming trucks. The average time interval
between the unloading process and distribution
of products decreased by 5%, leading to a reduc-
tion in the average loss of perishable products
and an increase in the health rate of the delivered
goods. Finally, the average lifetime of perishable
goods increased by 1% via applying storage poli-
Fig. 19. Different scenarios for the state variable of customer sat- cies, which affects the average loss of perishable
isfaction. goods, the health rate of delivered prod-
ucts, and the average of variable warehousing
costs.
– Scenario 2: The number of manufactories
declined by 2, the distance of manufactories
from the warehouse declined by 3%, average
distance of manufactory from the destination
diminished by 4%, cost of required equipment
for product storage elevated by 15%, reloading
percentage decreased by 7%, the number of vehi-
cles increased by 15, the average time interval
between the unloading process and distribution
of products reduced by 7%, and the average life-
time of perishable products increased by 2%.
– Scenario 3: The number of manufactories
Fig. 20. Different scenarios for maintenance cost.
increased by 3, the distance of manufactories
from the declined by 5%, average distance of
manufactory from the destination decreased by
2%, cost of required equipment for product
storage increased by 20%, reloading percent-
age diminished by 8%, the number of vehicles
increased by 20, the average time interval
between the unloading process and distribution
of products decreased by 8%, and the average
lifetime of perishable products prolonged by 3%.
– Scenario 4: The number of manufactories
decreased by 3, the distance of manufactories
from the warehouse decreased by 6%, average
distance of manufactory from the destination
diminished by 6%, cost of required equip-
ment for product storage increased by 25%,
reloading percentage decreased by 10%, the
Fig. 21. Different scenarios for customer demand. number of vehicles increased by 25, the aver-
age time interval between the unloading process
the cost of required equipment for product stor- and distribution of products declined by 10%,
age declined by 10% to reduce product losses, and the average lifetime of perishable products
whose impacts have also been observed on the increased by 4%.
 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation 10431

Based on these scenarios, the model is re- – Learning from the above simulation results, the
implemented. The changes of state variables (product major factors influencing the VRP and cross-
storage time in the cross-dock, cross-dock cost, and docking systems is their unreasonable structure.
customer satisfaction) and main variables (mainte- – The results indicate the most influencing factor
nance cost and customer demand) are depicted in on reaching the optimal status is the minimum
Figs. 17–21. distance between the cross-dock and destination,
According to Figs. 17–21, scenario 4 yielded the rather than increasing the number of manufac-
best results in terms of product storage time in tories.
the cross-dock, customer satisfaction, and customer
demand. So, it can be stated that when the number of The results of the scenarios show that the vari-
manufactories decreased by 3, the distance of man- ables of product holding time in cross-dock, customer
ufactories from the warehouse is decreased by 6%, satisfaction, consumer demand, cross-dock cost, and
average distance of manufactory from the destina- holding cost can not be optimized simultaneously and
tion is diminished by 6%, cost of required equipment should be determined according to the company’s
for product storage is increased by 25%, reloading goals. Also, based on the company’s priorities, the
percentage is decreased by 10%, the number of vehi- strategy will be selected. On the other hand, as multi-
cles is increased by 25, the average time interval objective problems, the best method to achieve the
between the unloading process and distribution of desired state can be found among all these goals.
products is reduced by 10%, and the average life-
time of perishable products is increased by 4%. As
to the cross-dock, scenario 3 outperformed other sce- 5. Conclusion
narios. Therefore, when the number of manufactories
was increased by 3, the distance of manufactories Over the past decade, cross-dock has been recog-
from the warehouse reduced by 5%, average distance nized as a leading technology for the transportation
of manufactory from the destination decreased by of goods. The known variety of VRP and VRPCD is
4%, cost of required equipment for product storage created in several procurement-programming fields.
increased by 20%, reloading percentage diminished Cross-dock is an almost new warehousing strategy in
by 8%, the number of vehicles increased by 20, the the procurement field, which is defined as integrat-
average time interval between the unloading process ing orders from receiving cargo to the classification
and distribution of products decreased by 7%, and the of outgoing goods in a distribution center. In the
average lifetime of perishable products prolonged by present study, an SD model was used to analyze these
3%. Finally, in terms of the maintenance cost, sce- components’ relationships. After designing four sce-
nario 1 generated the best results. According to this narios and approaching the specified values listed
scenario, the number of manufactories increased by in Table 2, it was observed that scenario 4 outper-
5, the distance of manufactories from the warehouse formed other scenarios in terms of product storage
decreased by 8%, average distance of manufac- time in the cross-dock, customer satisfaction, and
tory from the destination decreased by 5%, cost of customer demand. Also, scenarios 3 and 1 generated
required equipment for product storage decreased by the best results regarding the cross-dock and main-
10%, percentage of reloading reduced by 5%, the tenance costs, respectively. Based on the obtained
number of vehicles increased by 10, the average time results, it is recommended that:
interval between the unloading process and distribu-
tion of products decreased by 5%, the average lifetime – The number of manufactories should be reduced
of perishable goods increased by 1% via applying to decrease the cross-dock inventory rate and the
storage policies. By the way, the novel findings are as number of doors and wharves, i.e., a decrease in
follow: the cross-dock cost.
– Manufactories with the minimum distance
– We estimated customer satisfaction, cross-dock from the cross-dock and destination should be
costs, and product storage time in cross-dock for selected to shorten product delivery time.
60 next period. – Product storage machinery should be more
– We ran four different scenarios and camper the equipped to increase product lifetime.
result to find which variables most important to – Various strategies should be designed to reduce
a problem statement. the reloading rate.
10432 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation

– The average lifetime of perishable goods should [5] S.M. Mousavi and B. Vahdani, A robust approach to
be prolonged by applying product storage poli- multiple vehicle location-routing problems with time win-
dows for optimization of cross-docking under uncertainty,
cies. J Intell Fuzzy Syst 32(1) (2017), 49–62. doi: 10.3233/JIFS-
151050.
In summary, there are usually two techniques for [6] M. Rabbani, H. Farrokhi-Asl and H. Rafiei, A hybrid
increasing profitability: increasing sales (revenue) genetic algorithm for waste collection problem by het-
erogeneous fleet of vehicles with multiple separated
and reducing costs. According to the organization’s
compartments, J Intell Fuzzy Syst 30(3) (2016), 1817–1830.
policies, these results show that when the first method doi: 10.3233/IFS-151893.
is to increase sales, increase customer satisfaction [7] A. Rahbari, M.M. Nasiri, F. Werner, M.M. Musavi and F.
and thus increase consumer demand and ultimately Jolai, The vehicle routing and scheduling problem with
reduce product storage time at the Cross-dock to cross-docking for perishable products under uncertainty:
Two robust bi-objective models, Appl Math Model 70
improve delivery time. The organization can consider (2019), 605–625. doi: 10.1016/j.apm.2019.01.047.
reducing losses due to the destruction of goods (due [8] A. Hendalianpour, M. Hamzehlou, M.R. Feylizadeh, N. Xie
to long-term storage of goods). For this purpose, it and M.H. Shakerizadeh, Coordination and competition in
is suggested to achieve the fourth scenario’s optimal two-echelon supply chain using grey revenue-sharing con-
tracts, Grey Syst Theory Appl, 2020.
values using appropriate policies. On the other hand, [9] T.W. Liao, Integrated Outbound Vehicle Routing and
for organizations that are looking to reduce costs Scheduling Problem at a Multi-Door Cross-Dock Terminal,
to increase their profitability and reduce the cost of IEEE Trans Intell Transp Syst, 2020.
cross-dock in their plan, it is recommended to achieve [10] A. Gunawan, A.T. Widjaja, P. Vansteenwegen and F.Y.
Vincent, Adaptive large neighborhood search for vehi-
the optimal values set in the third scenario, by imple- cle routing problem with cross-docking, in 2020 IEEE
menting appropriate policies. Since any business’s Congress on Evolutionary Computation (CEC), 2020, pp.
ultimate objective is selected based on its policies, 1–8.
the best choice among these scenarios is determined [11] W. Jansen, Efficient Routing and Planning within the Com-
plex Logistical Network: Based on the Integration of a New
based on preset objectives and obtained results. In Warehouse, AGV Transports and Increased Transportation
this regard, strategies need to be proposed to help Rates, 2019.
an organization reach these values for the concerned [12] P. He and J. Li, The two-echelon multi-trip vehicle rout-
variables. ing problem with dynamic satellites for crop harvesting and
transportation, Appl Soft Comput J 77 (2019), 387–398. doi:
In future studies, researchers can develop some 10.1016/j.asoc.2019.01.040.
approaches to meeting these values for each of [13] A. Rahmandoust and R. Soltani, Designing a location-
the variables and address managerial strategies to routing model for cross docking in green supply chain,
Uncertain Supply Chain Manag 7(1) (2019), 1–16. doi:
implement this model. On the other hand, crisis man-
10.5267/j.uscm.2018.7.001.
agement strategies can be examined when imposed [14] İ. Küçükoğlu and N. Öztürk, A hybrid meta-heuristic
variables (non-controllable by the organization) are algorithm for vehicle routing and packing problem with
varied. cross-docking, J Intell Manuf 30(8) (2019), 2927–2943. doi:
10.1007/s10845-015-1156-z.
[15] H. LIU and C.L. And, Optimization for Multi-objective
Location-routing Problem of Cross-docking with Fuzzy
Time Windows, J Univ Electron Sci, 2019.
References [16] A. Baniamerian, M. Bashiri and R. Tavakkoli-Moghaddam,
Modified variable neighborhood search and genetic algo-
[1] J. Nalepa and M. Blocho, Adaptive guided ejection search rithm for profitable heterogeneous vehicle routing problem
for pickup and delivery with time windows, in Journal of with cross-docking, Appl Soft Comput J 75 (2019), 441–460.
Intelligent and Fuzzy Systems 32(2) (2017), 1547–1559. doi: doi: 10.1016/j.asoc.2018.11.029.
10.3233/JIFS-169149. [17] A.I. Nikolopoulou, P.P. Repoussis, C.D. Tarantilis and E.E.
[2] X. Wang, F. Yang and D. Lu, Multi-objective location- Zachariadis, Adaptive memory programming for the many-
routing problem with simultaneous pickup and delivery to-many vehicle routing problem with cross-docking, Oper
for urban distribution, J Intell Fuzzy Syst 35(4) (2018), Res 19(1) (2019), 1–38. doi: 10.1007/s12351-016-0278-1.
3987–4000. doi: 10.3233/JIFS-169721. [18] M.M. Nasiri, A. Rahbari, F. Werner and R. Karimi, Incor-
[3] P. Liu, A. Hendalianpour, J. Razmi and M.S. Sangari, porating supplier selection and order allocation into the
A solution algorithm for integrated production-inventory- vehicle routing and multi-cross-dock scheduling prob-
routing of perishable goods with transshipment and lem, Int J Prod Res 56(19) (2018), 6527–6552. doi:
uncertain demand, Complex Intell Syst, pp. 1–17, 2021. 10.1080/00207543.2018.1471241.
[4] S. Gelareh, F. Glover, O. Guemri, S. Hanafi, P. Nduwayo [19] S. Mancini, The Hybrid Vehicle Routing Problem, Transp
and R. Todosijević, A comparative study of formulations Res Part C Emerg Technol 78 (2017), 1–12. doi:
for a cross-dock door assignment problem, Omega (United 10.1016/j.trc.2017.02.004.
Kingdom) 91 (2020), 102015. doi: 10.1016/j.omega. [20] A. Hiassat, A. Diabat and I. Rahwan, A genetic algo-
2018.12.004. rithm approach for location-inventory-routing problem with
 P. Abbasi-Tavallali et al. / A system dynamics model for routing and scheduling of cross-dock and transportation 10433

perishable products, J Manuf Syst 42 (2017), 93–103. doi: [34] D. Mogale, M. Kumar, K. Kumar and M. Tiwari, Title Grain
10.1016/j.jmsy.2016.10.004. silo location-allocation problem with dwell time for opti-
[21] L.K. Lee, P.C.Y. Chen, K.K. Lee and J. Kaur, Menstrua- mization of food grain supply chain network Submission
tion among adolescent girls in Malaysia: A cross-sectional Files Included in this PDF, 2018.
school survey, Singapore Medical Journal 47(10) (2006), [35] A. Hendalianpour, Optimal lot-size and Price of Perishable
869–874. Goods: A novel Game-Theoretic Model using Double Inter-
[22] Y. Shuang, A. Diabat and Y. Liao, A stochastic reverse val Grey Numbers, Comput Ind Eng 149 (2020), 106780.
logistics production routing model with emissions control [36] A. Hendalianpour and J. Razmi, Customer satisfaction mea-
policy selection, Int J Prod Econ 213 (2019), 201–216. doi: surement using fuzzy neural network, Decis Sci Lett 6(2)
10.1016/j.ijpe.2019.03.006. (2017), 193–206. doi: 10.5267/j.dsl.2016.8.006.
[23] A.O. Kuşakcı, B. Ayvaz, E. Cin and N. Aydın, Optimiza- [37] Z. Sazvar, M. Sepehri and A. Baboli, A Multi-
tion of reverse logistics network of End of Life Vehicles objective Multi-Supplier Sustainable Supply Chain with
under fuzzy supply: A case study for Istanbul Metropoli- Deteriorating Products, Case of Cut Flowers, IFAC-
tan Area, J Clean Prod 215 (2019), 1036–1051. doi: PapersOnLine 49(12) (2016), 1638–1643. doi: 10.1016/j.
10.1016/j.jclepro.2019.01.090. ifacol.2016.07.815.
[24] M. Yavari and M. Geraeli, Heuristic method for robust [38] L. Bertazzi and J.W. Ohlmann, Direct k-routing versus
optimization model for green closed-loop supply chain net- cross-docking: worst-case results, Optim Lett, pp. 1–8,
work design of perishable goods, J Clean Prod 226 (2019), 2020.
282–305. doi: 10.1016/j.jclepro.2019.03.279. [39] M.S. Pishvaee and M. Rabbani, A graph theoretic-based
[25] Y. Zhang, H. Alshraideh and A. Diabat, A stochastic reverse heuristic algorithm for responsive supply chain net-
logistics production routing model with environmental con- work design with direct and indirect shipment, Adv Eng
siderations, Ann Oper Res 271(2) (2018), 1023–1044. doi: Softw 42(3) (2011), 57–63. doi: 10.1016/j.advengsoft.2010.
10.1007/s10479-018-3045-2. 11.001.
[26] B.B. Gardas, R.D. Raut and B. Narkhede, Reducing the [40] D. Kovačić, E. Hontoria, L. Ros-McDonnell and M.
exploration and production of oil: Reverse logistics in the Bogataj, Location and lead-time perturbations in multi-level
automobile service sector, Sustain Prod Consum 16 (2018), assembly systems of perishable goods in Spanish baby food
141–153. doi: 10.1016/j.spc.2018.07.005. logistics, Cent Eur J Oper Res 23(3) (2015), 607–623. doi:
[27] T.Y. Liao, Reverse logistics network design for product 10.1007/s10100-014-0372-5.
recovery and remanufacturing, Appl Math Model 60 (2018), [41] E. Pruyt, Small System Dynamics Models for Big Issues:
145–163. doi: 10.1016/j.apm.2018.03.003. Hop, Step and Jump towards Real-World Dynamic Com-
[28] H. Yu and W.D. Solvang, Incorporating flexible capacity in plexity, 2013.
the planning of a multi-product multi-echelon sustainable [42] M. Fallah, R. Tavakkoli-Moghaddam, M. Alinaghian and
reverse logistics network under uncertainty, J Clean Prod A. Salamatbakhsh-Varjovi, A robust approach for a green
198 (2018), 285–303. doi: 10.1016/j.jclepro.2018.07.019. periodic competitive VRP under uncertainty: DE and PSO
[29] M. Rahimi and V. Ghezavati, Sustainable multi-period algorithms, J Intell Fuzzy Syst 36(6) (2019), 5213–5225.
reverse logistics network design and planning under uncer- [43] L. An and J.J. Jeng, On developing system dynamics model
tainty utilizing conditional value at risk (CVaR) for recy- for business process simulation, in Proceedings – Win-
cling construction and demolition waste, J Clean Prod 172 ter Simulation Conference 2005 (2005), 2068–2077. doi:
(2018), 1567–1581. doi: 10.1016/j.jclepro.2017.10.240. 10.1109/WSC.2005.1574489.
[30] J. Trochu, A. Chaabane and M. Ouhimmou, Reverse logis- [44] H. Sapiri, J. Zulkepli and N. Ahmad, Introduction to System
tics network redesign under uncertainty for wood waste Dynamic Modelling and Vensim Software, 2017.
in the CRD industry, Resour Conserv Recycl 128 (2018), [45] D.F. Andersen, E. Rich and R. Macdonald, System Dynam-
32–47. doi: 10.1016/j.resconrec.2017.09.011. ics Applications to Public Policy, in System Dynamics, New
[31] S. Khodaparasti, M.E. Bruni, P. Beraldi, H.R. Maleki York, NY: Springer US, 2020, pp. 253–271.
and S. Jahedi, A multi-period location-allocation model [46] A. Boloori, M. Zandieh, A.R. Boloori Arabani, S.M.T.
for nursing home network planning under uncertainty, Fatemi Ghomi and M. Zandieh, A multi-criteria cross-
Oper Res Heal Care 18 (2018), 4–15. doi: 10.1016/j.orhc. docking scheduling with just-in-time approach Multi-
2018.01.005. Follower Bi-level Programming Problem View project
[32] A. Hendalianpour et al., Hybrid Model of IVFRN-BWM Mixed-model assembly line sequencing problem View
and Robust Goal Programming in Agile and Flexible project A multi-criteria cross-docking scheduling with just-
Supply Chain, a Case Study: Automobile Industry, IEEE in-time approach, Artic Int J Adv Manuf Technol 49(5–8)
Access 7 (2019), 71481–71492. doi: 10.1109/ACCESS. (2010), 741–756. doi: 10.1007/s00170-009-2429-5.
2019.2915309. [47] J.J. Jeng and L. An, System dynamics modeling for
[33] A. Hendalianpour, M. Fakhrabadi, M.S. Sangari and SOA project management, in Proceedings – IEEE Inter-
J. Razmi, A Combined Benders Decomposition and national Conference on Service-Oriented Computing and
Lagrangian Relaxation Algorithm for Optimizing a Multi- Applications, SOCA 2007, 2007, pp. 286–291. doi:
Product, Multi-Level Omni-Channel Distribution System, 10.1109/SOCA.2007.45.
Sci Iran, 2020.