Optimal Network Flows in A Hinterland System with Dry Ports Considering Water Corridor

(1)

1

Optimal Network Flows in A Hinterland System

with Dry Ports Considering Water Corridor

Master’s Thesis, MSc Supply Chain Management University of Groningen, Faculty of Economics and Business

(2)

2

Abstract: The development of dry port logistic networks and the enhanced role of barge

transport has forced shippers to plan the distribution of imported freights with the consideration of the characteristics of this network. This paper focuses on the optimization of network flows through a dry-port based distribution network and investigates the usage of barge mode in the network planning. The analysis was conducted by formulating a network flow model. The model was tested and validated on a total of 162 problem instances through Mosel Xpress IVE. Furthermore, three factors including service time requirements, maximum delivery distance from dry ports to their served clients and capacity constraints of dry ports were tested separately to identify their effects on the usage of barge mode. The managerial insights are as follows: (1) tight service time requirements result in a lower use of barge mode; (2) the maximum delivery distance from dry ports to their served clients should be set up according to the characteristics of dry ports in terms of capacity and transportation mode; (3) the capacity constraints of dry ports should be adapted to demand volumes to ensure sufficient resources for barge and rail service.

(3)

3

& Sgalambro, 2013). In order to solve the problem, great efforts have been made for the enhancement of seaport capacities. On one hand, technical innovation has been initiated to improve operational efficiency at seaports. On the other hand, a lot of attention has been paid to the accessibility of hinterlands (Crainic et al.,2013).One of the solutions is to create inland terminals, such as dry ports (Henttu, Lättilä, & Hilmola, 2010). The dry port is “an inland

intermodal terminal directly connected to seaport(s) with high capacity transport mean(s), where customers can leave or pick up their standardized units as if directly to a seaport.”

(Roso, Woxenius, & Lumsden, 2009). The dry port itself serves as an intermodal terminal in the inland transportation system (Henttu et al., 2010). In an optimal scenario, the whole freight movement between seaports and dry ports is conducted by rail and then distributed by trucks from dry ports (Henttu et al., 2010).

The traditional concept of dry ports is based on a rail-truck intermodal network. However, driven by the continuous pursuance of sustainable logistics, recently barge transport has attracted more and more attention as a competitive alternative to rail transport due to its ability to provide cheap and large-capacity transport services (Konings,2007).The proportion of barge transport in the hinterland shipments has grown significantly since the mid-1980s (Pielage, Konings, Rijsenbrij, & Schuylenburg, 2007). Thanks to the favourable waterway conditions, container transport volumes by barge are developing especially steadily in Western Europe (Konings, Kreutzberger, & Maraš, 2013). In the ports of Rotterdam and Antwerp, container barging has established an important role in the hinterland transportation (Wiegmans & Konings, 2007). Besides, the Rotterdam Port Authority has already set a target for the modal split in 2020 of 41% barge, 17% rail, and 42% truck (Konings et al.,2013).

(5)

5 flows (Caris, Macharis, & Janssens, 2008). With the barge mode integrated in the dry port system, the shipment flows are distributed among three types of routes: direct shipment by truck, transshipment with barge mode in the first leg and truck in the second leg, and transshipment with rail mode in the first leg and truck in the second leg. The traditional aspect concerning decision-making in the network design is transportation costs. However, there are three other factors that need to be taken into considerations.

 Service time requirements

 Maximum delivery distance from dry ports to their served clients

 Capacity constraints of dry ports

First, the service time requirement refers to the maximum transit time from seaports to clients. This factor has direct impact on the selection of routes, because it specifies each route should satisfy the time requirement. Second, the maximum delivery distance from dry ports to the clients assigned to them is involved in the planning due to the consideration of the changes in service time requirements. A close delivery distances from dry ports to clients provides more options to deal with this issue, such as delivery by special trucks with reasonable prices or picking-up from dry ports arranged by clients. Finally, the capacity constraints of dry ports refers to capacity limitation and minimum utilization rates. These two factors directly influence the volumes of flows assigned to each route.

(6)

6 Accordingly, the research question is formulated as “How to make tactical plans in a dry-port based network design considering barge mode?” with two sub-questions:

 How to optimize the shipment flows in a dry port network with waterways and railways?

 How are the allocation of barge flows influenced by service time requirements, maximum delivery distance from dry ports to their served clients, and capacity constraints of dry ports?

The network flow problem was addressed by presenting an optimization model. The model is developed from a shipper’s perspective with an attempt to minimize the total transportation costs. It accommodates the characteristics of a dry-port based network with both waterway and railway corridors, transportation costs, service time requirements, maximum delivery distances from dry ports to clients, capacity limitation and minimum utilization rates of dry ports. The proposed model was tested and validated through Mosel Xpress IVE in the computational experiments with 162 randomly created instances. In addition, the effects of service time requirements, maximum delivery distance from dry ports to clients, capacity constraints of dry ports on the usage of barge mode were explored in the computational experiments.

The contributions of this paper are twofold: (1) it enriches the theory on tactical planning in the design of dry port networks; (2) it explores the usage of barge mode in network planning.

(7)

7

2. Literature Review

In this section, we first describe the concept of dry ports and a fully implemented dry port network. Then a review on previous literature concerning dry port networks and the network flow problems in intermodal system takes place.

2.1 The Dry Port Concept

The greatest feature which distinguishes dry ports from conventional inland terminals is that dry ports are used consciously to solve the problems triggered by the increasing container flows (Roso, 2007). Despite being an intermodal terminal, a dry port is capable of offering value-added services and administrative activities to facilitate seaport operations. Crainic et al.

(2013) specified three assistant roles of dry ports in the sea port system. The first is to provide additional logistic services which are not available or cost a lot at seaports, such as warehousing for different periods, storage of empty containers, container repair and maintenance (Henttu et al., 2010). The second is to smooth intermodal transport by streamlining transfer procedures. The third is to undertake some administrative functions of seaports, such as customs clearance and cargo inspections.

2.2 The Fully Implemented Dry Port Network

The fully implemented dry port logistic network includes three types of dry ports depending on their locations and functions in the network. We first introduce the three types of dry ports and then present the fully implemented network.

2.2.1 Distant Dry Ports

(8)

8 society’s perspective, negative environment impact is reduced due to the modal shift from rail to road (Woxenius et al., 2004).

2.2.2 Midrange Dry Ports

A midrange dry port is located within a certain distance from seaports. Its most distinguishable function is to assume freight consolidation (Woxenius et al., 2004).

Furthermore, technical facilities and equipment needed for customs inspections and security can be stored in the midrange dry ports, leaving more storage areas for seaports (Woxenius et al., 2004).

2.2.3 Close Dry Ports

A close dry port is very close to seaports. Despite serving as a consolidation point, the close dry ports offer more services to collaborate with the associated seaports. First, freights can be immediately distributed to close dry ports with the dedicated railway directly connecting inland (Woxenius et al., 2004). Second, close dry ports can serve as buffering points for container flows so that the shipment plans can be synchronized with vessel unloading schedules (Crainic et al., 2013).

2.2.4 The Fully Implemented Dry Port Network

(9)

9

Figure 1:Comparison between a conventional hinterland network and a fully implemented dry port concept (Source: Woxenius et al., (2004))

2.3 Literature on Dry-Port Based Networks and Network Flow Problems

The dry port concept and the dry-port based hinterland network have received much attention in the literature. Most studies focus on the definitions and functions of dry ports, for example, Roso et al. (2009), Jaržemskis & Vasiliauskas (2007), Cullinane & Wilmsmeier (2011), and

Roso & Lumsden (2009). Especially, some researchers explored the effects of dry port

implementation on the environment. Roso (2007) evaluated the dry port system from an

environmental perspective with the means of modelling and simulation. The findings showed the implementation of dry port concept lead to a decrease in CO2 emissions.

(10)

10

actors involved in the system. Caballini & Gattorna (2009) conducted a case study of the dry ports in Rivalta Scrivia and analysed their functions as the supplemental spaces to the Genoa port.

The benefits of dry port implementation have driven many researchers to focus on the dry port network design. For example, Monios & Wilmsmeier (2012) discussed three site development strategies to link seaports and hinterland terminals. Henttu, Lättilä, & Hilmola (2011) developed different gravitational models to find the optimal cost-efficient dry port network structure. Rosa & Roscelli (2009) investigated the feasibility of the project, which aimed at establishing an integrated seaport and dry-port transport network.

In the existing research on dry port network design, a lot of attention has been paid to the strategic issues, especially the location problem. For example, Ka (2011) combined both quantitative and qualitative methods for decision-making on dry port locations. Wang & Wei (2008) built a model to systematically evaluate the factors influencing locations and presented an index system to rank these factors. However, few papers focus on the tactical planning issues, such as the optimization of network flows.

Relevant studies regarding the network flow problem can be found in the literature on intermodal network design. Chang (2008) focused on the intermodal routing problem: how to choose the optimal shipment routes from an international carrier’s perspective. Meng & Wang (2011) studied the flow problem in an intermodal hub-and-spoke network with the considerations of multiple actors and container types. Boussedjra, Bloch, & El Moudni (2004) and Nassir, Khani, Hickman, & Noh (2012) aimed at optimizing the network flows by finding the shortest routes between origins and destinations.

(11)

11

 Not all the dry ports can accommodate barge service.

 In the shipment route passing through one dry port, the first leg of this shipment is conducted by barge or rail mode, and the second leg is transported by trucks.

Therefore, this paper explores the network flow problem by integrating the barge mode and the characteristics of dry port systems into the network design.

3. The Network Flow Problem Description

The network flow problem in this paper regards the distribution plans for importing freights. The shippers need to choose the routes for shipment flows and decide the flow volumes in order to minimize the total transportation costs.

(12)

12 With respect to transportation costs, shippers would prefer to consolidate shipment flows by barge mode first and then distribute them by truck. This is because the cost savings are realized by consolidating shipment flows (Ishfaq & Sox, 2010). Besides, barge mode is more cost-efficient than rail mode. However, it is necessary to take into account other factors, which will be elaborated as follows.

The first factor is the service time requirement, which refers to the delivery time of each shipment flow from the seaport to the client. Given the increasing level of service quality required by clients, the network should be designed to satisfy service time requirements

Seaport

Rail dry port

Client

Figure 2: Shipment flows adapted from Feng, Zhang, Li, & Wang (2013) Barge dry port

Barge shipment flow

Rail shipment flow

(13)

13 (Rodriguez, Alvarez, & Barcos, 2007). This means the delivery time spent in each chosen route should be within the service time required by clients.

The second factor is the capacity constraints of dry ports, which are related to the capacity limitation and the minimum utilization rate. As for the capacity limitation, it is unrealistic to have unlimited capacity at each dry port. As for the minimum utilization rate, sufficient shipment flows have to be gained at dry ports by consolidating freights from different seaports ( Lin, Chiang, & Lin, 2014), because the economy of scale is the principle force behind using intermodal network (Ishfaq & Sox, 2011).

The third factor is the maximum delivery distances from dry ports to their served clients. A good network design should be capable of handling the uncertainty in clients’ requirements, such as changes in service times. The times spent on the shipments passing through dry ports is longer than direct truck shipments. In case of clients cutting short the service times later, a close distance between dry ports and clients provides more options for shippers to deal with the issues. For example, the shipper can arrange special trucks with a reasonable price or let the clients to pick up the cargo by themselves.

To sum up, the network flow problem in this paper refers to the optimal routes and volumes of shipment flows aiming at minimizing the total transport costs with considerations of transportation costs, service time requirements, maximum delivery distance from dry ports to their served clients, and capacity constraints of dry ports.

In order to clarify the problem, the assumptions regarding costs, time, capacity, and demand flows are made as follows:

Costs:

1. There are no storage fees

2. dry ports already exist so there are no opening costs. 3. No modal connectivity costs occur.

Time:

(14)

14 5. The service time requirement excludes the time in custom clearance and any other

administration activities. Capacity:

6. Barge, rail and truck have unlimited load capacity.

7. Each dry port has limited capacity which is associated with ingoing flows. 8. Each dry port has a minimum utilization rate.

9. The capacity levels of dry ports range from high to low.

10. Some dry ports are connected to the seaport by both barge and rail mode, while the other dry ports can only use either barge mode or rail mode.

Demand flows:

11. The demand flows can be split and routed on paths that pass through 0 or 1 dry port. 12. The demand is stable within a certain period.

13. Only imported shipment flows are considered. 14. There is no minimum quantity for split flows.

With respect to assumption 1,3,4 and 6, the storage fees, modal connectivity costs, waiting time and load capacity are associated with specific situations. Given the data collected for the computational experiments are not based on actual cases, we make the assumptions as above.

4. Mathematical Model

We address the problem of optimizing the network flows in terms of network flow routes and volumes. The objective is to minimize the total transportation costs.

(15)

15

4.1 Model Formulation

The dry-port based network is represented by a graph (S, D, C) in which S,D,C stand for the set of seaports, the set of dry ports, and the set of clients. These sets are indicated by i, k and j respectively. Some dry ports can only be connected to the seaport by barge or rail while the others can use both barge and rail modes to connect with seaports. B denotes the set of dry ports that can be reached by barge, and R denotes the set of dry ports that can be reached by rail in a way that D is equal to R∪B. For every i ∈ S, and every j ∈ C, Wj denotes the aggregated volumes of demand flows sent from seaport i to client j.

For each Wj, decision variables of fijtt, fikjb , fijk r represent the split flows of Wj. For example, considering the situation where Wj is split off to pass through three routes: (1) direct truck shipment (2) transhipment through barge dry port (3) transhipment through rail dry port, the values of f_ijtt, f_ikjb , f_ijkr show the volumes of flows travelling through each route mentioned above. The sum of the three values should be equal to the value of Wj. In addition, these decision variables also give information on the selection of dry ports and flow routes which are indicated by the indices of i,j and k.

The model is restricted by service time requirements, maximum delivery distance from dry ports to their served clients, capacity limitation and minimum utilization rate of dry ports. Each client j has a service time requirement T_j, so the shipment time for each route should be no longer than T_j. For the direct shipment, the shipment time from seaport i to client j is denoted by t_ijtt. For the dry port based shipment, the shipment time consists of the transit time t_ikb_{or t}

ik

r _{from seaport i to dry port k by barge or rail, and the transit time}_t kj

(16)

16 The purpose is to optimize shipment flows in a way that the total transportation costs are minimized. The costs include three parts: (1) the sum of direct shipment costs f_ijttc_ijtt from seaport i to client j; (2) the sum of barge dry port shipment costs f_ikjb (c_ikb + c_kjt ); (3) the sum of rail dry port based shipment costs f_ikjr (c_ikr + c_kjt ).

The sets, parameters and variables are present in Table 1,2, and 3.

S = Set of seaports D = Set of dry ports C = Set of clients

B= Set of dry ports which can be reached by barge mode R=Set of dry ports which can be reached by rail mode D = B ∪ R

Table 1: Sets

Wj =The aggregated volumes of demand flows required by client j, ∀j ∈ C

c_ijtt_{= Unite direct truck transportation cost per flow from seaport i to client j,}_{∀i ∈ S; ∀j ∈ C} c_kjt _{= Unite truck transportation cost per flow from dry port k to client j, ∀k ∈ D; ∀j ∈ C} c_ikr _{= Unite rail transportation cost per flow from seaport i to dry port k, ∀i ∈ S; ∀k ∈ D} c_ikb_{= Unite barge transportation cost per flow from seaport i to dry port k, ∀i ∈ S; ∀k ∈ D} t_ijtt = Direct truck shipment time from seaport i to client j, ∀i ∈ S; ∀j ∈ C

t_kjt = Truck transit time from dry port k to client j, ∀k ∈ D; ∀j ∈ C t_ikb_{= Barge transit time from seaport i to dry port k,}_{∀i ∈ S; ∀k ∈ D} t_ikr _{= Rail transit time from seaport i to dry port k,}_{∀i ∈ S; ∀k ∈ D} T_j= Service time requirement by client j, j ∈ C

(17)

17 U_k= Minimum untilization rate at dry port k, k ∈ D

F_k= Total flows going into dry port k, k ∈ D

d_kj= The distance between dry port k and client j, ∀k ∈ D; ∀j ∈ C M = A large value

Table 2: Parameters

yk = {1, if dry port k is used_{0, otherwise} , k ∈ D

f_ijtt _{= The volumes of flows by direct truck shipment, ∀i ∈ S; ∀j ∈ C}

f_ikjb _{= The volumes of flows by barge dry port based shipment, ∀i ∈ S; ∀k ∈ D; ∀j ∈ C} f_ikjr = The volumes of flows by rail dry port based shipment, ∀i ∈ S; ∀k ∈ D; ∀j ∈ C

Table 3: Decision variables

We formulate the network flow model as follows:

M i n ∑_i∈S∑_k∈D∑_j∈Cf_ikjb (c_ikb + c_kjt )+ ∑_i∈S∑_k∈D∑_j∈Cf_ikjr (c_ikr + c_kjt ) +∑_i∈S∑_j∈Cf_ijttc_ijtt ( 1 ) Subject to:

(18)

18 F_k ≥ U_kCP_ky_k ∀k ∈ D ( 1 1 ) F_k ≤ CP_k ∀k ∈ D ( 1 2 ) f_ikjb = 0 ∀i ∈ S; ∀k ∈ R − B ∩ R; ∀j ∈ C ( 1 3 ) f_ikjr = 0 ∀i ∈ S; ∀k ∈ B − B ∩ R; ∀j ∈ C (14 ) f_ikjb ≥ 0 ∀i ∈ S; ∀k ∈ D; ∀j ∈ C ( 1 5 ) f_ikjr ≥ 0 ∀i ∈ S; ∀k ∈ D; ∀j ∈ C ( 1 6 ) f_ijtt ≥ 0 ∀i ∈ S; ∀j ∈ C ( 1 7 ) y_k ∈ {0,1} ∀k ∈ D ( 1 8 )

The objective function (1) minimizes the total transportation costs which include the barge transport costs, the rail transport costs and the truck transport costs. Constraint (2) stipulates that the sum of the flow volumes shipped to the same client should be equal to the total demand flows required by the client. Constraints (3)-(5) specify that the total time spent on each route should meet the service time requirement. Constraints (6) and (7) specify that the delivery distance between each dry port and the client assigned to it should be within K. Constraints (8) and (9) enforce that the dry port is selected only when there are flows going through it. Constraint (10) defines the total flow volumes going into one dry port. Constraint (11) specifies the minimum utilization of each dry port. Constraint (12) ensures the total flows going into each dry port do not exceed the capacity limitation. Constraint (13) stipulates that some dry ports can only be connected to seaports by rail. Constraint (14) stipulates that some dry ports can only be connected to seaports by barge. Constraints (15)-(17) ensure the value of flow volumes should not be negative. Constraint (18) enforces that the decision variable y_k is binary.

5. Computational Experiments

The purpose of the computational experiments is twofold:

(19)

19 maximum delivery distance from dry ports to their served clients as well as capacity limitations and minimum utilization rates of dry ports.

 To explore the effects of influential factors on the usage of barge mode to acquire managerial insights.

We first designed the problem instances in a two-stage way, where basic data was created in Stage One and the problem instances were designed in Stage Two. Then we tested and solved the instances by using Mosel Xpress IVE. The tests were run on an i3-4130T CPU with 2.90GHz clock and 8GB RAM.

5.1 Experimental Design

In Stage One, the basic data set, which is not critical to the usage of barge mode, was randomly created and kept as fixed numbers in all the problem instances generated in Stage Two. As long as the random data set is designed in line with reality and experimental objectives, this approach enables researchers to explore different aspects of problems (Ishfaq & Sox, 2011). The data was created in the following way. First, the horizontal and vertical coordinates of dry ports and clients were generated by randomly sampling from uniform distribution. The locations of the two seaports were set as fixed points. The coordinates were set to build a midrange dry port based network where the distance between dry ports and seaports are in the range of 50 km to 500 km (Woxenius et al., 2004). Then the Euclidean distances were calculated based on these data. Second, the transportation costs of a unit flow were obtained from the distance multiplying it by different cost-per-mile rates, depending on transportation modes. Finally, the transit times were the results of the distance divided by different transport speeds. Table 4 lists all the values of the parameters. The transport speeds referred to Pekin, Macharis, Meers, & Rietveld (2013), and the unite transportation costs came from Falzarano, Ketha, Hawker, Winebrake, Corbett, Korfmacher, Zilora, (2007).

Parameter Value

Barge speed (km/h) 12

Rail speed (km/h) 25

Truck speed (km/h) 60

(20)

20

Rail unite cost per km per flow 1,4

Truck unite cost per km per flow 1,5

Horizontal coordinate range of dry ports [100,500] Horizontal coordinate range of clients [400,800] Vertical coordinate range of dry ports [50,600] vertical coordinate range of clients [0,600]

Coordinates of seaport #1 (0,200)

Coordinates of seaport #2 (0,400)

Table 4:Parameter values

In Stage Two, the problem instances were created to simulate different scenarios. The general characteristics of all the scenarios are described as follows:

 The network has three layers including seaports, dry ports, and clients. The flows travel through direct shipments by truck or shipments passing through one dry port. The latter shipment consists of two legs: barge or rail mode is used to consolidate freights in the first leg, and truck mode is used to distribute freights from dry ports to clients in the second leg.

 The capacity levels for the dry ports are categories into two levels: high and low. The high-capacity dry port is connected to the seaports by both barge and rail, while the low-capacity dry port can only be reached by one mode. There are two high-capacity dry ports in each scenario.

 As for the low-capacity dry ports, half of them are barge dry ports and the others are rail dry ports in each scenario.

(21)

21 volumes, respectively. The service time requirements were set at three levels: 36, 48 and 72 hours. Here we assumed that all the clients imposed a uniform service time requirement in each instance. The maximum distance was set at three levels: 360, 480 and 600 km.

Code No. of

dry ports

No. of

clients Demand volumes

Service time requirements

(hours)

Maximum delivery distance from dry ports to

their served clients (km)

1 18 80 Large, medium, small 36,48,72 360,480,600

Table 5:Instance table

The capacity constraints of dry ports were set as fixed numbers in all the instances based on the scenario where there are 10 dry ports and 60 clients with medium scale of demand volumes. The figures as listed in Table 6 were designed in a way that the overall dry port capacities are capable of satisfying the total demand requirements. We used a uniform capacity limitation and minimum utilization rate for low-capacity dry ports and the same to high-capacity dry ports. In this way, the discrepancy between the overall available capacities and total demand flow volumes varied when the demand volumes increased from small scale to large scale.

Dry port type Capacity limitation Minimum utilization rate

Low-capacity dry port 2040 0.4

High-capacity dry port 8160 0.75

Table 6: Capacity constraints

5.2 Results and Discussions

(22)

22 service time requirements, the maximum delivery distance from dry ports to their served clients, and the capacity constraints of dry ports.

In order to explore the effects of the aforementioned factors, the results were analysed in a way that only one factor was changeable while others were constant. Since different scenarios show similar impact, we select one typical case for one factor to illustrate the analysis results.

5.2.1 The Effect of Service Time Requirements

Case description: 18 dry ports, 80 clients, large scale of demand volumes, 600 km maximum delivery distance from dry ports to their served clients, and 36, 48 and 72 hours service time requirements. The results are listed in Table 7.

72 hours 48 hours 36 hours

F(1)=8160 F(1)=8160 F(1)=8160 F(2)=8160 F(2)=6120 F(2)=0 F(3)=2040 F(3)=2040 F(3)=2040 F(4)=0 F(4)=816 F(4)=2040 F(5)=0 F(5)=2040 F(5)=2040 F(6)=0 F(6)=0 F(6)=2040 F(7)=0 F(7)=0 F(7)=0 F(8)=2040 F(8)=2040 F(8)=2040 F(9)=0 F(9)=0 F(9)=0 F(10)=0 F(10)=0 F(10)=0 F(11)=0 F(11)=0 F(11)=0 F(12)=2040 F(12)=2040 F(12)=2040 F(13)=816 F(13)=0 F(13)=0 F(14)=0 F(14)=0 F(14)=1166 F(15)=0 F(15)=0 F(15)=0 F(16)=2040 F(16)=2040 F(16)=0 F(17)=2040 F(17)=2040 F(17)=2040 F(18)=2040 F(18)=2040 F(18)=0

Total cost: 60460 Total cost: 60619.3 Total cost: 61147.3

No.of dry ports=9 No.of dry ports=10 No.of dry ports=9

Total flows=60060 Total flows=60060 Total flows=60060

Barge flows=25296 Barge flows=21774 Barge flows=13406

Rail flows=4080 Rail flows=7602 Rail flows=10200

Truck flows=30684 Truck flows=30684 Truck flows=36454

BFP=42.1179% BFP=36.2537% BFP=22.321%

(23)

23

TFP=51.0889% TFP=51.0889% TFP=60.696%

Rail flows(1)=0 Rail flows(1)=0 Rail flows(1)=0

Barge flows(1)=8160 Barge flows(1)=8160 Barge flows(1)=8160

Table 7:Results with different service time requirements

(Note: F means the volumes of flows going into the dry port. The number in the bracket represents the code of each dry port. The two high-capacity dry ports are coded with #1 and #2. BFP, RFP, and TFP represent the percentage of barge flows, rail flows, and direct truck flows in the total flow volumes. These symbols are also applied in Table 8 and 9.)

The direct impact of the changes in service time requirements is illustrated in the modal shift. As shown in Figure 3, a shorter service time requirement leads to a decrease in the barge flows and an increase in rail and direct truck flows. This is because the barge mode is cheap but takes the longer to transport cargo than rail and road modes. The effect is also reflected in a single dry port. For dry port #2, which is connected to the seaports by both rail and barge, the decrease in the service time requirements leads to a modal shift from barge to rail. Comparing Column #1 and #2, the barge flows drop from 8160 to 5454, while the rail flows rise from 0 to 666. In Column #3, no flows go into the dry port at all because it takes too much time to deliver goods even with rail mode in the first leg of the shipment.

Figure 3:The effect of service time requirements

0 5000 10000 15000 20000 25000 30000 35000 40000 30 40 50 60 70 80 TH e fl o w vo lu m e s

Service time requirement (hour)

(24)

24

Managerial insights: tight service time requirements result in a lower use of barge mode

because it is time-consuming compared to rail and truck mode. Hence, additional supporting services should be applied to compensate for this disadvantage in transport time, such as higher level of automation to improve cargo handling efficiency. Another option for shippers to promote barge transport is to differentiate logistical service levels by associating logistic prices with service time.

5.2.2 The Effect of Maximum Delivery Distance From Dry Ports to Their Served Clients

Case description: 18 dry port, 40 clients, large scale of demand volumes, 72 hours service time requirement, 360, 420 and 600 km maximum delivery distance from dry ports to their served clients. The result are listed in Table 8.

600 km 480 km 360 km F(1)=8160 F(1)=8160 F(1)=0 F(2)=6120 F(2)=6543 F(2)=6120 F(3)=2040 F(3)=1757 F(3)=2040 F(4)=0 F(4)=0 F(4)=0 F(5)=0 F(5)=0 F(5)=1500 F(6)=0 F(6)=0 F(6)=816 F(7)=0 F(7)=0 F(7)=0 F(8)=0 F(8)=0 F(8)=0 F(9)=0 F(9)=0 F(9)=0 F(10)=0 F(10)=0 F(10)=0 F(11)=0 F(11)=0 F(11)=0 F(12)=2040 F(12)=2040 F(12)=2040 F(13)=0 F(13)=0 F(13)=0 F(14)=0 F(14)=0 F(14)=0 F(15)=0 F(15)=0 F(15)=0 F(16)=2040 F(16)=2040 F(16)=2040 F(17)=0 F(17)=0 F(17)=0 F(18)=2040 F(18)=2040 F(18)=2040

Total costs: 29470.2 Total costs: 29498.6 Total costs: 29932.7

No. of dry ports=6 No. of dry ports=6 No. of dry ports=7

(25)

25

RFP=7.02358% RFP=6.04923% RFP=14.9974%

TFP=22.7406% TFP=22.2586% TFP=42.8611%

Table 8: Results with different maximum delivery distance

The decrease in the maximum delivery distance from dry ports and clients leads to an increase in the total transportation costs, but the changes in the flow volumes of the three transportation modes do not follow the same pattern as shown in the service time case. Figure 4 shows the barge flow volumes increase first and then decline as the maximum delivery distance drops from 600 km to 360 km. In contrast, the rail and direct truck flow volumes decline first and then rise. In order to explain this phenomenon, we first compare the results in Column #1 and #2. It is observed that a part of flows assigned to dry port #3 are transferred to dry port # 2, because some clients cannot be assigned to the distant dry port #3 but dry port #2 is close to them. The barge flows increase because the barge mode is used to reach dry port #2. Meanwhile, this flow shift relieves capacity in dry port #3, so a portion of direct truck shipments are replaced by dry port based shipments. This results in a decrease in the truck flow volumes. Then we compare the results in Column #2 and #3, when the distance constraint becomes highly strict, such as 360 km in Column #3, the dry port #1 which has a high capacity cannot be assigned to any clients. As a result, the flows that are originally allocated to this dry port are dispersed among some small capacity dry ports, leading to an increase in the number of selected dry ports and a flow shift from barge to rail and truck.

(26)

26

Figure 4:The effect of maximum delivery distance from dry ports to their served clients

Managerial insights: the maximum delivery distance from dry ports to their served clients

should be set up according to the characteristics of dry ports in terms of capacity and transportation mode. The shipper can relax the maximum distance restrictions to some extent in situations where some barge dry ports or high-capacity dry ports are not located in proximity to clients due to geographical conditions. In parallel with this plan, information technology should be utilized to offer cargo status to clients.

5.2.3 The Effect of Capacity Constraints of Dry Ports

Case description: 10 dry port, 60 clients, 72 hours service time requirement, 480 km maximum delivery distance from dry ports to their served clients, large, medium, and small scale of demand volumes. The results are listed in Table 9.

Large Medium Small

F(1)=8160 F(1)=8097 F(1)=7295 F(2)=8160 F(2)=6120 F(2)=0 F(3)=2040 F(3)=2040 F(3)=2040 F(4)=2040 F(4)=0 F(4)=0 F(5)=2040 F(5)=1112 F(5)=816 F(6)=2040 F(6)=2040 F(6)=2040 F(7)=2040 F(7)=0 F(7)=0 F(8)=816 F(8)=0 F(8)=1338 F(9)=2040 F(9)=869 F(9)=816 0 5000 10000 15000 20000 25000 300 350 400 450 500 550 600 650 Th e fl o w vo lu m e s

Maximum delivery distance from dry ports to their served clients (km)

(27)

27

F(10)=0 F(10)=0 F(10)=0

Total costs: 46334.2 Total costs: 27733.8 Total costs: 18558.8

No.of dry ports=9 No.of dry ports=6 No.of dry ports=6

BFP=46.937% BFP=55.4612% BFP=51.9147%

RFP=18.0527% RFP=19.0875% RFP=26.8996%

TFP=35.0103% TFP=25.4513% TFP=21.1856%

Table 9: Results of different capacity constraints

The effort of capacity constraints should be studied together with demand volumes, because it is the discrepancy between the overall available capacities and total demand flow volumes influencing the distribution of flows.

Since the capacity limitation and the minimum utilization rate were set as fixed numbers based on the scenario with medium scale of demand volumes , when the demand volumes increase from small scale to large scale, the three columns in table 9 represent overcapacity, moderate capacity, and undercapacity situations from left to right. To be specific, in this case, overcapacity refers to the situation where the overall capacity of dry ports is nearly 1.8 times more than the total demand flows; moderate capacity means the capacity match with the demand flows; undercapacity means the capacity is nearly 72% of the demand flows.

(28)

28 capacity in Column #3, the percentage decreases due to insufficient capacity of barge service. The reason why the rail flow percentage already declines in Column #2 is that the demand volumes almost reach the capacity limitation of rail service in this situation.

Managerial insights: the capacity limitation and the minimum utilization rate should be

adapted to demand volumes to ensure sufficient resources for barge and rail service. However, capacity constraints of dry ports are usually predetermined. Given the dynamic market where the demands fluctuate, one possible option is to find subcontractors for the expansion of dry port capacity.

6. Conclusions

This research explored the tactical planning issue in a network design. We focused on how to optimize the network flows in a dry-port based network with both waterways and railways. The optimization decisions include the selection of network flow routes and the determination of flow volumes. In addition, we investigated the usage of barge mode in the network design by analysing the effects of influential factors on the distribution of flows. The contributions of this paper are: (1) it enriches the theory on tactical planning in the design of dry port networks; (2) it explores the usage of barge mode in network planning.

We addressed the problem by presenting a network flow model. The model is developed from a shipper’s perspective with decision variables regarding the routes and volumes of flows. It accommodates the characteristics of dry-port based network with barge and rail mode, transportation costs, service time requirements, maximum delivery distance from dry ports to their service clients, and capacity constraints of dry ports. The purpose is to minimize the total transportation costs. The novelty of the proposed model is twofold: (1) it accommodates the characteristics of the dry-port network with waterways and railways (2) it takes into account the maximum distances from dry ports to their served clients.

(29)

29 constraints of dry ports, were analysed separately to identify their effects on the usage of barge mode. The managerial insights are present as follows: (1) tight service time requirements result in a lower use of barge mode, so additional supporting services such as higher level of automation should be applied to compensate for this disadvantage. Another option is to differentiate logistic service levels by associating logistic prices with service time; (2) the maximum delivery distance from dry ports to their served clients should be set up according to the characteristics of dry ports in terms of capacity and transportation mode. The maximum distance restrictions can be relaxed to some extent for distant barge dry ports or high-capacity dry ports. In parallel with this plan, information technology should be utilized to offer cargo status to clients; (3) the capacity limitation and the minimum utilization rate should be adapted to demand volumes to ensure sufficient resources for barge and rail service. One possible option is to find subcontractors for the expansion of dry port capacity.

(30)

30

Reference

Boussedjra, M., Bloch, C., & El Moudni, A. (2004). An exact method to find the intermodal shortest path (ISP). In Networking, Sensing and Control, 2004 IEEE International Conference

on (Vol. 2, pp. 1075-1080). IEEE.

Chang, T. S. (2008). Best routes selection in international intermodal networks. Computers &

operations research, 35(9), 2877-2891.

Crainic, T., Dell’Olmo, P., Ricciardi, N., & Sgalambro, A. (2013). Optimizing Dry-Port-Based

Freight Distribution Planning.

Caris, A., Macharis, C., & Janssens, G. K. (2008). Planning problems in intermodal freight transport: accomplishments and prospects. Transportation Planning and Technology, 31(3), 277-302.

Caballini, C., & Gattorna, E. (2009). The expansion of the port of Genoa: the Rivalta Scrivia dry port. Development of Dry Ports, 73.

Cullinane, K., & Wilmsmeier, G. (2011). The contribution of the dry port concept to the extension of port life cycles. Handbook of Terminal Planning, 359-379.

Feng, X., Zhang, Y., Li, Y., & Wang, W. (2013). A Location-Allocation Model for Seaport-Dry Port System Optimization. Discrete Dynamics in Nature and Society, 2013.

Falzarano, A. M., Ketha, S., Hawker, J. S., Winebrake J., Corbett, J., Korfmacher., K., Zilora, S. (2007). Development of an intermodal network for freight transportation analysis. Rochester

Institute of Technology.

Henttu, V., Lättilä, L., & Hilmola, O. P. (2010). Financial and environmental impacts of a dry port to support two major Finnish seaports. Lappeenranta University of Technology,

Department of Industrial Management, Research Report, 224, 141.

Henttu, V., Lättilä, L., & Hilmola, O. P. (2011). Optimization of relative transport costs of a hypothetical dry port structure. Transport and Telecommunication, 12(2), 12-19.

Ishfaq, R., & Sox, C. R. (2011). Hub location–allocation in intermodal logistic networks.

European Journal of Operational Research, 210(2), 213-230.

Ishfaq, R., & Sox, C. R. (2012). Design of intermodal logistics networks with hub delays.

(31)

31 Ishfaq, R., & Sox, C. R. (2010). Intermodal logistics: the interplay of financial, operational and service issues. Transportation Research Part E: Logistics and Transportation Review, 46(6), 926-949.

Jaržemskis, A., & Vasiliauskas, A. V. (2007). Research on dry port concept as intermodal node.

Transport, 22(3), 207-213.

Konings, R., Kreutzberger, E., & Maraš, V. (2013). Major considerations in developing a hub-and-spoke network to improve the cost performance of container barge transport in the hinterland: the case of the port of Rotterdam. Journal of Transport Geography, 29, 63-73.

Konings, R. (2007). Opportunities to improve container barge handling in the port of Rotterdam from a transport network perspective. Journal of Transport Geography, 15(6), 443-454.

Ka, B. (2011). Application of fuzzy AHP and ELECTRE to China dry port location selection.

The asian journal of shipping and logistics, 27(2), 331-353.

Lin, C. C., Chiang, Y. I., & Lin, S. W. (2014). Efficient model and heuristic for the intermodal terminal location problem. Computers & Operations Research, 51, 41-51.

Monios, J., & Wilmsmeier, G. (2012). Port-centric logistics, dry ports and offshore logistics hubs: strategies to overcome double peripherality?. Maritime Policy & Management, 39(2), 207-226.

Meng, Q., & Wang, X. (2011). Intermodal hub-and-spoke network design: incorporating multiple stakeholders and multi-type containers. Transportation Research Part B:

Methodological, 45(4), 724-742.

Nassir, N., Khani, A., Hickman, M., & Noh, H. (2012). Algorithm for intermodal optimal multidestination tour with dynamic travel times. Transportation Research Record: Journal of

the Transportation Research Board, (2283), 57-66.

Pekin, E., Macharis, C., Meers, D., & Rietveld, P. (2013). Location Analysis Model for Belgian Intermodal Terminals: Importance of the value of time in the intermodal transport chain.

Computers in Industry, 64(2), 113-120.

Pielage, B. J., Konings, R., Rijsenbrij, J., & Schuylenburg, M. V. (2007). Barge hub terminals: a perspective for more efficient hinterland container transport for the Port Rotterdam. In 48th

Annual Transportation Research Forum, Boston University, Boston, MA (pp. 19-27).

(32)

32 Roso, V. (2007). Evaluation of the dry port concept from an environmental perspective: A note.

Transportation Research Part D: Transport and Environment, 12(7), 523-527.

Roso, V. (2008). Factors influencing implementation of a dry port. International Journal of

Physical Distribution & Logistics Management, 38(10), 782-798.

Roso, V., & Lumsden, K. (2009). The dry port concept: moving seaport activities inland.

UNESCAP, Transport and Communications Bulletin for Asia and the Pacific, 5(78), 87-102.

Roso, V., Woxenius, J., & Olandersson, G. (2006). Organisation of Swedish dry port terminals.

Division of Logistics and Transportation, Chalmers University of Technology, Gothenburg.

Rosa, A., & Roscelli, R. (2009). Innovative ideas and design of an integrated dry port and seaport system. Development of Dry Ports, 57.

Rodriguez, V., Alvarez, M. J., & Barcos, L. (2007). Hub location under capacity constraints.

Transportation Research Part E: Logistics and Transportation Review, 43(5), 495-505.

Van Riessen, B. (2013). Service network design for an intermodal container network with

flexible due dates/times and the possibility of using subcontracted transport (Doctoral

dissertation, Delft University of Technology).

Woxenius, J., Roso, V., & Lumsden, K. (2004). The dry port concept–connecting seaports with their hinterland by rail. ICLSP, Dalian, 22-26.

Wiegmans, B. W., & Konings, R. (2007). Strategies and innovations to improve the performance of barge transport. European journal of transport and infrastructure research

EJTIR, 7 (2).

Wang, C., & Wei, J. (2008, December). Research on the dry port location of Tianjin port based on analytic network process. In Business and Information Management, 2008. ISBIM'08.

Optimal Network Flows in A Hinterland System with Dry Ports Considering Water Corridor

Optimal Network Flows in A Hinterland System

with Dry Ports Considering Water Corridor

Contents