International Journal for Traffic and Transport Engineering, 2019, 9(4): 456 - 470

DOI: http://dx.doi.org/10.7708/ijtte.2019.9(4).10 UDC: 656.125.073(497.5)

658.286(497.5)

VEHICLE ROUTING PROBLEM - CASE STUDY ON LOGISTICS

COMPANY IN CROATIA
Martina Jakara1, Jasmina Pašagić Škrinjar2, Nikolina Brnjac3
1 ,2, 3
University of Zagreb, Faculty of Transport and Traffic Sciences, Zagreb, Croatia
Received 6 June 2019; accepted 20 September 2019

Abstract: This paper defines the vehicle routing problem that logistic operators and other
traffic professionals face in organizing transportation and determining the optimal routes and
number of vehicles needed to provide a quality transport service. The purpose of the paper
is to outline the methods used to solve the problem and to compare their application. This
example shows the application of metaheuristic methods for solving vehicle routing problems
of a logistics company in the Republic of Croatia. The aim of this case study is to determine
the optimal solution using methods which represents such a distribution of available vehicles
that generates the lowest cost.

Keywords: transport logistics, transport networks, mathematical methods, optimization,

heuristics, metaheuristics, vehicle routing.

1. Introduction The problem of determining optimal routes

for a group of vehicles in the literature
Realization of technological processes is one has been described as a Vehicle Routing
of the most important elements of decision Problem (VRP). There are many different
making. Three levels of decision making mathematical models and algorithms for
can be analyzed in transportation: strategic, solving this problem.
tactical and operational level. At the strategic
level, a decision is made to determine the The purpose of this paper is to demonstrate
warehouse location. The choice of traffic the application of mathematical methods
mode and type of transport vehicle is made for solving vehicle routing problem. After
at the tactical level. At the operational analyzing the transport network of the
level, optimal routes are chosen for vehicles logistics company, a mathematical model
selected at the tactical level and based on the of the vehicle routing problem was optimized
location of the warehouse at the strategic using the Nearest Neighbor Heuristics and
level. The transport problem is the problem the VRP Spreadsheet Solver optimization
of determining the transport schedule of program, which combines different heuristic
a particular transport substrate from the methods to solve previously mentioned
source to the destinations whose substrate problem. The main goal of this paper is
demand is to be met, using the available to determine the optimal solution for
transport routes by the criterion of minimum distribution of available vehicles with the
transport cost or shortest transport route. lowest cost and to explain the results and the

1
Corresponding author: mjakara@fpz.hr

456
 Jakara M. et al. Vehicle Routing Problem - Case Study on Logistics Company in Croatia

difference in the application after comparing routes, lines, transport chains that enable
the applied methods. This paper presents the fast, safe and rational processes of production
determination of the transport plan in the of transport ser v ices. The purpose of
transport network of the observed enterprise. transport networks is to transport certain
The organization of vehicle routing will be goods, materials or passengers from one place
carried out in the city of Zagreb by example to another. The basic elements of a transport
of deliveries for Monday and Wednesday. network are nodes and links. A node is a place
or point of loading where loading/unloading,
2. Vehicle Routing Problem sorting, storage and change of transport
vehicle or mode of transport take place. Links
A key element in the logistics chain is the represent transport and transport activities
transportation system. As transport costs that connect nodes, creating a transport
account for the largest share of total logistics network. By connecting all source and
costs, improving transport efficiency has destination nodes with a sufficient number
a significant impact on increasing the of transshipment nodes, a transport network
overall performance of the logistics chain, is obtained (Woxenius, 1998; Woxenius,
thereby directly affecting the results of 1997). The transport network is a complex
logistics activities. Transportation costs system and it is therefore not surprising
vary depending on different industries that it has various problems. The resulting
(Pavlić Skender and Grčić Fabić, 2013). problems need to be optimized to maximize
They depend on fuel prices and often exceed the utilization of the transport network and
50% of total logistics costs. It is necessary to make transport processes more efficient.
to reduce the share of logistics costs in the The commitment to these issues is extremely
total costs of certain products or services, high, and with the development of logistics
which can be achieved by solv ing the and computer technology has become even
appointed transport problem. By analyzing greater. Therefore, various mathematical
the transport network and identif ying methods and programming tools are being
existing shortcomings and bottlenecks, it used today to optimize the problems
is possible to set up a transport problem and encountered on the transport network.
solve it with mathematical methods such as Some of the problems that arise on the
heuristics or mathematical models, which are transportation network are transportation
reducing the number of vehicles and costs of problems, the shortest route problem, the
distribution by optimizing other elements vehicle routing problem, location problems,
of the transport net work. Solv ing the the traveling salesman problem and many
transport problem determines the optimal others (Buntak et al., 2013).
transportation schedule, which implies a
plan of transportation of goods, which from One of the prominent problems in logistics
the point of origin to the destination has is the Vehicle Routing Problem (VRP). The
the minimum total transportation costs problem of routing vehicles requires finding
(Pašagić, 2003). the optimal route from the initial terminal
to all the designated destinations in such
The transport network can be defined as a way that demand at the destinations is
a system of interconnected and interest satisfied and the total distance traveled by
connected transport hubs, roads, corridors, the delivery vehicles is minimized (Dantzig

457
International Journal for Traffic and Transport Engineering, 2019, 9(4): 456 - 470

and Ramser, 1959). The classic problem of (Čolić, 2013). Depending on the constraints
routing a vehicle is to find a series of routes a route has, there are several types of routing
with minimal cost (finding the shortest problems.
route, minimizing the number of vehicles,
etc.) starting and ending routes at the source, According to (Laporte, 2007), every form
so that the demand of all destinations is met of VRP requires equal resolution, which
(Belfiore and Yoshizaki, 2009). VRP is a is to serve all users under the following
generic name for a series of problems that restrictions:
arise when distributing goods, where it is 1. each route star ts and ends at the
necessary to determine the route by which warehouse;
a vehicle will tour a certain number of cities 2. each user is visited exactly once;
at a minimum total cost and in the shortest 3. the total demand of each route does not
possible time. Vehicle Routing Problem exceed the capacity of the vehicle.
is defined as a number of vehicles with a
given capacity. Each customer order has a The optimal solution to a particular problem
specific location and size. There are default solving is usually attempted by mathematical
shipping costs between all locations. The methods. In problems where the input data
goal is to design an optimal itinerary so that are not constants but depend on time or
all customers are visited once and capacity some stochastic factor, it is not possible to
constraints are met. Troubleshooting construct a representative mathematical
involves finding the optimal sequence of model. Sometimes, due to the complexity
visiting locations, provided that only one of the data and the problem size, the
vehicle is served by each location. The mathematical model provides the optimal
limitation of the set condition is that the total solution but in a extremely long period of
demand of all users on a particular route does time. Heuristic methods help to arrive at
not exceed the capacity of the vehicle (Hasle a solution in a short time that may not be
and Kloster, 2007). VRP can be described as optimal but close to optimal. Heuristic
the problem of designing optimal delivery methods can be divided into special and
or collection routes from one or more general heuristics. Specials are designed
warehouses to a number of geographically for special types of optimization problems
dispersed entities (cities or users) that are while respecting their characteristics and
subject to additional restrictions (Toth and their special features. General heuristics,
Vigo, 2002). It is about the distribution of often referred as metaheuristics, are general-
goods over a given time, a certain number purpose methods that can be applied to
of users, a certain number of vehicles that any combinatorial optimization problem,
are housed in one or more warehouses. regardless of the specifics of its structure
The vehicles are moving along the default (Stanković and Pašagić Škrinjar, 2015).
transport network. Typically, the solution
to this problem is to find routes, where each The Nearest Neighbor Heuristics (NNH) is
route is driven by one vehicle starting from one of the first algorithms to solve vehicle
the original warehouse and returning to routing problems and is often used as an
it, all set limits must be met and the total initial solution for testing repair heuristics.
cost of the transportation (distance traveled It’s pretty fast, but the solutions are almost
by vehicles) must be kept to a minimum a lw ay s fa r f rom opt i ma l . T he N N H

458
 Jakara M. et al. Vehicle Routing Problem - Case Study on Logistics Company in Croatia

algorithm works by randomly selecting a 3. Optimal Solution of the Vehicle Routing

starting city from a set of cities, adding it Problem
to a route and marking it as a visited. Then,
from a set of unvisited cities, the algorithm In this chapter, the real-world example
finds the closest city from the previously of routing a real logistics business is the
added one and adds it to the route which application of metaheuristics in solving
indicates that the city has been visited. This vehicle routing problems. The observed
step is repeated until all cities are visited company collects goods from a destination
(Carić, 2004). in the city of Zagreb and transports them
to its warehouse. This example shows the
Meta heu r ist ics-based sof t wa re tools organization and optimization of daily routes
provide quick solutions that are very close for Monday and Wednesday. For Monday,
to optimal solutions. Algorithms using exact ten destinations were defined, while eight
mathematical methods require a long-term destinations were defined on Wednesday.
process to solve, which is not appropriate Using the coordinates of the source location
in the field of logistics since it is necessary and all destinations, a matrix for the distances
to respond to customer requests as soon as between all locations was made for both days,
possible. Heuristic models are much more and the displayed matrices were asymmetric
applicable than exact mathematical models in (two-way). This way of making the matrix
solving problems with variables (eg. changes is more accurate for deliveries in the city
in infrastructure capacity over time, changes of Zagreb because of the large number of
in prices, etc.). The VRP Spreadsheet Solver one-way roads that significantly affect the
program was created using this approach, in difference in distance in two directions
which is possible to set up, solve and visualize between two points. Metaheuristic methods
the results of vehicle routing problem will determine the optimal number of
(V R P). The above program combines vehicles and the shortest routes for each
Excel, public GIS and metaheuristics. It vehicle to minimize the final distribution
can solve over 64 variants of VRP’s with cost. The Nearest Neighbor method and the
up to 200 users (Erdoğan, 2017). Among VRP Spreadsheet Solver program described
these are the problems mentioned earlier, in the previous section are used to resolve
including limitations on capacity, distance, vehicle routing problem. The type of goods
time, simultaneous delivery and collection, being transported is liquid cargo stored in
delivery and collection of returned goods. barrels. Each barrel occupies a cargo area of
The advantage of this tool is that it’s simple, 0,16 m2 (dimensions 40 x 40 cm). The weight
practical and easy to use. Because the of one barrel is 37 kilograms. The barrels
program is refined with a metaheuristic are constructed that they can be stacked on
approach, it has implemented an algorithm top of each other in height. In the example,
based on Large Neighborhood Search (LNS). a profit of 70 HRK (9,46 EUR) will be used
The strategy is to try capturing the widest for each barrel collected. Values in EUR are
possible solution spaces during the search obtained by dividing the HRK values by 7,4,
and to explore only those areas that promise which was the average of the HRK/EUR
to find quality solutions (Galić, 2018). ratio in September 2019.

459
International Journal for Traffic and Transport Engineering, 2019, 9(4): 456 - 470

Delivery vehicle used for city distribution is one level in the cargo area and given that
a small truck in third category of the Iveco they can be stacked in several rows, it is
brand, from which it can be concluded that concluded that the limiting factor in this
this is a homogeneous problem of routing case is mass.
the vehicle. The maximum load capacity
is 2,2 tons and the volume of cargo space 3.1. First Case of Delivery Optimization
is 16 m 3 . The vehicle has diesel powered – Monday
engine and uses 16 liters / 100 kilometers
for city driving. Diesel fuel price on the T he f irst example that w i l l show the
day 19 th of February, 2019 amounted 9,65 application of heuristics is the schedule of
HRK (1,3 EUR) per liter from which it can delivery destinations and delivery vehicles
be calculated that the cost of transport per in the city of Zagreb for Monday. Table 1
kilometer for this vehicle is 1,544 HRK(0,21 shows the locations of the source and the
EUR)/ km (9,65 HRK x 16/100). ten destinations to be visited on a given
day. Pickup amount of each destination is
With a maximum mass limit, a maximum of also shown in the table . Pickup amount is
55 barrels (2200 kg / 37 kg) can be loaded important in routing the vehicles because
into one vehicle. There are 36 barrels on of the limited capacity of the vehicle itself.

Table 1
Locations and Pickup Amount at Monday Destinations
Node Location ID Pickup amount
The source location I 0
Destination 1 D1 4
Destination 2 D2 6
Destination 3 D3 11
Destination 4 D4 13
Destination 5 D5 20
Destination 6 D6 17
Destination 7 D7 11
Destination 8 D8 9
Destination 9 D9 12
Destination 10 D10 8

The distance matrix in kilometers is shown in Table 2. The distances shown in the table
will be used to further calculate and resolve vehicle routing problem.

460
 Jakara M. et al. Vehicle Routing Problem - Case Study on Logistics Company in Croatia

Table 2
First case Distance Matrix - Monday
I D1 D2 D3 D4 D5 D6 D7 D8 D9 D10
I 0,00 17,70 16,11 20,22 16,88 19,46 18,88 18,14 16,96 17,21 18,49
D1 16,72 0,00 4,13 4,71 4,05 9,27 11,72 10,04 9,49 8,75 6,77
D2 14,95 3,81 0,00 2,31 1,66 6,87 8,65 6,97 6,42 5,68 3,70
D3 15,98 4,49 2,37 0,00 0,62 5,53 7,29 5,61 5,07 4,33 2,33
D4 17,27 4,97 2,85 1,29 0,00 5,23 7,00 5,32 4,77 4,03 2,87
D5 19,91 8,55 6,43 4,06 4,68 0,00 1,88 2,83 2,89 3,38 4,98
D6 18,76 12,30 9,57 7,64 8,26 1,77 0,00 1,41 1,96 2,23 5,90
D7 18,25 10,34 7,62 5,19 5,81 1,92 1,65 0,00 1,45 1,72 5,40
D8 18,80 12,34 9,61 7,68 8,30 2,77 2,19 1,45 0,00 2,27 5,94
D9 18,05 8,97 6,24 4,09 4,71 1,96 3,06 1,69 1,14 0,00 3,20
D10 17,00 6,39 3,67 1,74 2,36 4,57 6,34 4,66 4,11 3,37 0,00

3.1.1. Resolving the Vehicle Routing According to the matrix in Table 2, the
Problem for Monday with the Nearest minimum distance of 16,11 km is between
Neighbor Heuristics the source and Destination 2. Destination
2 has a pickup amount of 6 barrels, which
As noted above, the Nearest Neighbor increases the initial capacity of the vehicle
Heuristics starts from the source and from 0 to current 6. From Destination 2, the
searches for the nearest destination to set nearest unvisited neighbor is again searched
it as the first destination. After arriving for and that is Destination 4 which is 1,66
at the first destination, it re-searches km away with a pickup amount of 13 barrels.
for the closest un-visited destination in The capacity rate is increasing to 19. The
second iteration and sets it as the second next destination is Destination 3 which is
destination. T his process is repeated 1,29 km from Destination 4. Filled capacity
until all destinations are visited and the at Destination 3 is increasing to 30 barrels
vehicle is returned to its source location. from the maximum 55. Next location is
Given that there is also a capacity limit in Destination 10, 2,33 km away and the capacity
this case, it is necessary to set the pickup is being filled to 38 barrels. The next closest
amount 0 at source and add it to the pickup undisclosed location is Destination 9 at
amount of each destination visited. The last 3,37 km, which has a pickup amount of 12
destination before exceeding the permitted barrels. This fills the maximum capacity of
vehicle capacity is set as the last destination the delivery vehicle which returns to source
for that vehicle and returns to the source after Destination 9 by crossing 18,05 km. The
after visiting it. route of the first vehicle is shown in Figure 1.

461
International Journal for Traffic and Transport Engineering, 2019, 9(4): 456 - 470

Fig. 1.
Route of the First Vehicle by the Nearest Neighbor Heuristics - Monday

The second vehicle re-starts from the source greater pickup amount than can be loaded
and visits the nearest unvisited neighbor. into another vehicle, so the vehicle after
This is Destination 8 distant 16,96 km away Destination 6 returns to source location by
from the source. The pickup amount at passing 18,76 km.
Destination 8 is 9 barrels. From Destination
8, the closest unattended destination is There is only Destination 5 left to visit for
Destination 7 which is 1,45 km away with this heuristic, with a pickup amount of 20
11 barrels in pickup amount. Following is barrels. So the third vehicle from source to
Destination 6 in distance 1,65 km away Destination 5 will make 19,46 km, and 19,91
with a pickup amount of 17 barrels. The km back to source location. Table 3 shows
total capacity occupancy of the vehicle is the routes for all engaged vehicles according
37 and the remaining destinations are in to the specified method.

Table 3
Preview of the Routes of all Delivery Vehicles Hired on Monday According to the Nearest Neighbor Heuristics
First vehicle I – D2 – D4 – D3 – D10 – D9 – I
Second vehicle I – D8 – D7 – D6 – I
Third vehicle I – D5 – I

462
 Jakara M. et al. Vehicle Routing Problem - Case Study on Logistics Company in Croatia

After the Nearest Neighbor Heuristics has which in this case is one. It is also shown that
determined all routes and the number of all vehicles must be returned to the source.
vehicles required to serve all destinations,
the cost of shipping costs will be shown After the first worksheet, a second worksheet
below. The total net profit of the deliveries is created which is called 1. Locations. The
made is calculated by reducing the transport locations with corresponding pickup amounts
costs from the profit of the total number of are shown in Table 1 at the beginning of the
freight collected. It can be seen in Table 1 description of the first case. In addition to
that the total number of barrels collected the information in Table 1, the coordinates
is 111, which multiplied by 70 HRK (9,46 of each location that were collected from
EUR) per barrel gives a profit of 7.770,00 the Google Maps page are written on the
HR K (1050 EUR). If you add up all the worksheet for the purpose of automatically
kilometers traveled, all three used vehicles calculating distances between locations. The
g ive you a number of 121 k m, wh ich following worksheet is 2. Distances, showing
according to the fuel price calculation at the the distance matrix between all locations
beginning of the chapter multiplies by 1,544 and the travel time between nodes. The
HRK (0,21 EUR). The total cost of delivery following worksheet sets out delivery truck
is 186,82 HRK (25,22 EUR) and when this information. In this example, a type of Iveco
is deducted from the profit obtained, it can truck with a total capacity of 2,2 tones was
be concluded that the Nearest Neighbor installed. The capacity of 55 barrels and a
Heuristics generated a net profit of 7.583,66 unit shipping cost of 1,544 HRK (0,21 EUR)
HRK (1.024,82 EUR). per kilometer are defined at the beginning
of the chapter. The required number of
3.1.2. Resolving the Vehicle Routing vehicles is set to 3 in the analysis because
Problem for Monday with the VRP a smaller number of vehicles does not meet
Spreadsheet Solver Optimization Program the capacity limit, while a larger number
of vehicles is not needed because the three
According to the Monday delivery data vehicles generate the lowest costs. After this
shown in Table 1, a problem has been worksheet, a solution worksheet is set up,
previewed in the VRP Spreadsheet Solver which is shown in Table 4. Table 4 shows the
optimization program, which solves the number of stops and the profit generated for
metaheuristic principle of vehicle routing and each vehicle, where each destination shows
offers an optimal solution. In this example, the time of arrival and departure, retention
there are one source and ten destinations. time, realized profit and the amount of cargo
The number of vehicle types is determined, collected.

463
International Journal for Traffic and Transport Engineering, 2019, 9(4): 456 - 470

Table 4
Preview the Vehicle Routing for Monday Obtained using the VRP Spreadsheet Solver
Net
Vehicle: V1 Stops: 5 2318,26 HRK (321,79 EUR)
profit:
Location Distance Driving Arrival Departure Working Profit collected
Load
name travelled time time time time HRK EUR
The source
0,00 0:00 08:00 0:00 0 0 0
location
Destination 1 17,70 0:32 08:32 08:46 0:46 280 37,84 4
Destination 2 21,23 0:44 08:58 09:14 1:14 700 94,59 10
Destination 4 22,89 0:50 09:20 09:43 1:43 1610 217,57 23
Destination 3 24,01 0:54 09:47 10:08 2:08 2380 321,62 34
The source
39,99 1:24 10:38 2:38 2380 321,62 34
location

Net
Vehicle: V2 Stops: 4 3297,72 HRK (445,64 EUR)
profit:
Location Distance Driving Arrival Departure Working Profit collected
Load
name travelled time time time time HRK EUR
The source
0,00 0:00 08:00 0:00 0 0 0
location
Destination 7 18,14 0:34 08:34 08:55 0:55 770 104,05 11
Destination 5 19,70 0:43 09:04 09:34 1:34 2170 293,24 31
Destination 6 21,58 0:55 09:46 10:13 2:13 3360 454,05 48
The source
40,34 1:25 10:43 2:43 3360 454,05 48
location

Net
Vehicle: V3 Stops: 4 1971,03 (266,36 EUR)
profit:
Location Distance Driving Arrival Departure Working Profit collected
Load
name travelled time time time time HRK EUR
The source
0,00 0:00 08:00 0:00 0 0 0
location
Destination 8 16,96 0:27 08:27 08:46 0:46 630 85,14 9
Destination 9 19,22 0:34 08:53 09:15 1:15 1470 198,65 21
Destination 10 22,43 0:46 09:27 09:45 1:45 2030 274,32 29
The source
38,19 1:15 10:14 2:14 2030 274,32 29
location

The total net profit gained by this method of routing is 7.587,01 HRK (1.025,27 EUR).
The routes obtained by the optimization program are displayed on the last worksheet.

464
 Jakara M. et al. Vehicle Routing Problem - Case Study on Logistics Company in Croatia

3.2. Second Case of Delivery Optimization delivery locations and delivery vehicles in the
- Wednesday Zagreb area for Wednesday. Table 5 shows the
source and the eight destinations to be visited
The second example that will show the on a given day and their pickup amount in the
application of metaheuristics is the schedule of number of barrels to be collected.

Table 5
Locations and Pickup Amount at Destinations for Wednesday
Node Location ID Pickup amount
The source location I 0
Destination 1 D1 30
Destination 2 D2 27
Destination 3 D3 21
Destination 4 D4 18
Destination 5 D5 24
Destination 6 D6 33
Destination 7 D7 23
Destination 8 D8 22

The distance matrix in kilometers is shown in Table 6. The distances shown in the table
will be used to further calculate and resolve the vehicle routing problem.

Table 6
Distance Matrix for the Second Case - Wednesday
I D1 D2 D3 D4 D5 D6 D7 D8
I 0,00 17,77 16,93 16,99 22,82 17,48 22,09 21,96 19,63
D1 20,49 0,00 2,04 1,55 6,59 3,67 7,86 11,61 3,82
D2 18,28 2,26 0,00 0,78 4,29 1,47 5,62 5,53 3,17
D3 18,15 1,86 1,41 0,00 5,70 2,88 7,03 10,77 3,61
D4 21,52 7,19 4,89 5,71 0,00 3,73 2,74 4,35 10,02
D5 18,23 3,49 1,22 2,00 3,35 0,00 4,62 4,50 2,54
D6 20,95 6,30 4,41 5,18 1,76 3,15 0,00 1,52 4,49
D7 22,34 12,03 10,59 10,88 3,15 4,54 3,10 0,00 8,02
D8 20,50 4,02 2,71 3,00 5,11 2,74 8,57 7,16 0,00

3.2.1. Resolving the Vehicle Routing barrels, increasing the initial capacity
Problem for Wednesday with the Nearest occupancy of the vehicle from 0 to the
Neighbor Heuristics current 27. From Destination 2, the closest
unvisited neighbor is searched for and it can
According to the matrix in Table 6, the be seen that Destination 3 is 0,78 km away
shortest distance from the source is to with a pickup amount of 21 barrels. Capacity
Destination 2 in amount of 16,93 k m. is increased to 49, which re-routs this vehicle
Destination 2 has a pickup amount of 27 by returning it 18,15 km to its source.

465
International Journal for Traffic and Transport Engineering, 2019, 9(4): 456 - 470

Second vehicle re-starts from the source and amount of 18 barrels. The remaining capacity
visits the nearest unvisited neighbor. This is is not sufficient for further loading and the
Destination 5, which is 17,48 km away from third vehicle returns to the source distant
the source. Pickup amount at Destination 5 21,52 km.
is 24 barrels. From Destination 5, the closest
unvisited destination is Destination 8, 2,54 The remaining two destinations cannot be
km away with 22 barrels in pickup amount. combined into one vehicle because pickup
The total capacity of the vehicle is 46 and the amount exceeds the capacity of the vehicle,
remaining destinations have greater pickup so each will need to be delivered with one
amount than can be loaded into this vehicle, vehicle. Destination 7 has a pickup amount
so that the vehicle returns 20,50 km to its of 23 barrels. So the fourth vehicle will make
source after Destination 8. 21,96 km from Source location to Destination
7, and back 22,34 km to the source.
The next closest unvisited destination they
are looking at from the source is Destination The fifth vehicle is the last from the source
1, which is 17,77 km away and has a pickup travels 22,09 km to Destination 6 and 20,95
amount of 30 barrels. From Destination 1, km back to the source. Pickup amount at
out of the three remaining destinations, Destination 6 is 33 barrels. Table 7 shows the
Destination 4 is 6,59 km away with a pickup routes of all vehicles engaged for Wednesday.

Table 7
Preview of the Routes of all Delivery Vehicles Engaged on Wednesday Using the Nearest Neighbor Heuristics
First vehicle I – D2 – D3 – I
Second vehicle I – D5 – D8 – I
Third vehicle I – D4 – D1 – I
Fourth vehicle I – D7 – I
Fifth vehicle I – D6 – I

After the Nearest Neighbor Heuristics has of delivery is 323,62 HR K (43,73 EUR)
determined all the routes and the required and when this is deducted from the profit
number of vehicles, it is necessary to make a received, it can be concluded that application
calculation of the delivery costs. The total net of the Nearest Neighbor Heuristics in the
profit of the deliveries made is calculated by delivery for Wednesday resulted in a net
deducting the transport costs from the profit profit of 13.536,38 HRK (1.829,24 EUR).
of the total number of freight collected. It can
be seen from Table 4 that the total number 3.2.2. Resolving the Vehicle Routing
of barrels collected is 198, which multiplied Problem for Wednesday with the VRP
by 70 HRK (9,46 EUR) per barrel gives a Spreadsheet Solver Optimization Program
profit of 13.860,00 HRK (1872,97 EUR). If
all kilometers of all five vehicles are summed According to the Wednesday deliver y
up, the amount number is 209,6 km, which locations shown in Table 5, a problem has been
is multiplied by 1,544 HRK (0,21 EUR) at defined in VRP Spreadsheet Solver optimizer
the beginning of the chapter according to program. As in the Monday optimization
the fuel price calculation. The total cost example, the first worksheet specifies

466
 Jakara M. et al. Vehicle Routing Problem - Case Study on Logistics Company in Croatia

the number of sources and the number of of vehicles by analysis is set to four because
destinations to be served. In this example, a smaller number of vehicles does not meet
there is one source and eight destinations. the capacity limit. While a larger number of
The number of vehicle types is determined, vehicles is not needed as four vehicles generate
which in this case is one. It also specifies the the lowest costs. After this worksheet, a
condition that all vehicles must be returned solution worksheet is set up which is shown
to the source. The procedure for creating in Table 8. Using the VRP Spreadsheet Solver
worksheets is the same as in the previous case optimization software, a total net profit of
for Monday, except that the required number 13.601,17 HRK (1.838 EUR) was obtained.

Table 8
Preview of Vehicle Routing for Wednesday Obtained with VRP Spreadsheet Solver
Net
Vehicle: V1 Stops: 3 3575,01 HRK (483,11 EUR)
profit:

Location Distance Driving Arrival Departure Working Profit collected

Load
name travelled time time time time HRK EUR
The source
0,00 0:00 08:00 0:00 0 0 0
location
Destination 1 17,77 0:35 08:35 08:45 0:45 2100 283,78 30
Destination 8 21,59 0:46 08:56 09:06 1:06 3640 491,89 52
The source
42,09 1:32 09:52 1:52 3640 491,89 52
location

Net
Vehicle: V2 Stops: 3 3304,63 HRK ( 446,57 EUR)
profit:

Location Distance Driving Arrival Departure Working Profit collected

Load
name travelled time time time time HRK EUR
The source
0,00 0:00 08:00 0:00 0 0 0
location
Destination 2 16,93 0:34 08:34 08:44 0:44 1890 255,41 27
Destination 3 17,71 0:37 08:47 08:57 0:57 3360 454,05 48
The source
35,86 1:17 09:37 1:37 3360 454,05 48
location

467
International Journal for Traffic and Transport Engineering, 2019, 9(4): 456 - 470

Net
Vehicle: V3 Stops: 3 3221,59 HRK (435,35 EUR)
profit:

Location Distance Driving Arrival Departure Working Profit collected

Load
name travelled time time time time HRK EUR
The source
0,00 0:00 08:00 0:00 0 0 0
location
Destination 5 17,48 0:36 08:36 08:46 0:46 1680 227,03 24
Destination 7 21,98 0:52 09:02 09:12 1:12 3290 444,59 47
The source
44,31 1:44 10:04 2:04 3290 444,59 47
location

Net
Vehicle: V4 Stops: 3 3499,94 HRK (472,96 EUR)
profit:

Location Distance Driving Arrival Departure Working Profit collected

Load
name travelled time time time time HRK EUR
The source
0,00 0:00 08:00 0:00 0 0 0
location
Destination 6 22,09 0:48 08:48 08:58 0:58 2310 312,16 33
Destination 4 23,85 0:56 09:06 09:16 1:16 3570 482,43 51
The source
45,38 1:44 10:04 2:04 3570 482,43 51
location

3.3. Comparative Analysis of the Results to be more effective in finding the optimal
solution than the other. Table 9 shows a
By comparing the two used methods, the comparison of the results obtained for the
Nearest Neighbor Heuristics and the VRP Monday and Wednesday routing examples.
Spreadsheet Solver optimization program, it The net profits of the routes obtained and
can be concluded that one method has proven the number of vehicles engaged are shown.

Table 9
Comparison of the Results Obtained for the Methods Used
Nearest Neighbour Heuristics VRP Spreadsheet Solver
Net profit Net profit
Number of vehicles Number of vehicles
HRK EUR HRK EUR
Monday 7583,66 1024,82 3 7587,01 1025,27 3
Wednesday 13 536,38 1829,24 5 13 601,17 1838 4

468
 Jakara M. et al. Vehicle Routing Problem - Case Study on Logistics Company in Croatia

As shown in Table 9, the route optimizer gave methods within the VRP Spreadsheet Solver
better results than the Nearest Neighbor’s software. The analysis of the results shows
handmade heuristic method. In the first that the software tool performs better than
example for determining routes on Monday, using the second method in the form of less
the results are about the same. The difference traveled distance, less engaged vehicles and
in cost is 3,35 HRK (0,45 EUR) and the ultimately lower delivery costs, which can
number of vehicles is the same, but in the lead to higher profit or decrease in the overall
example for Wednesday the profit is higher price of the service or product.
by 64,8 HRK (8,76 EUR) for metaheuristics
because it is by 41,96 km less and one vehicle The application of software tools for solving
less was used, which is an even greater saving vehicle routing problem greatly facilitates
because the budget did not use fixed vehicle decision making in the logistics sector. The
costs, as well as driver costs, which are higher role of mathematical methods and software
with more vehicles. tools will be increasing with the further
development of logistics and transportation
O f t he t wo met hod s u sed, t he V R P technology, which will greatly facilitate the
Spreadsheet Solver software has been shown work of logistics operators in their daily
to perform better than the Nearest Neighbor work in the transport system. The number
Heuristics, although the latter in some of logistic processes on the transport network
cases provides a satisfactory solution that, that directly affect the utilization of the
despite a different route and vehicle layout, transport network, will be reduced.
is qualitatively and quite close to optimal.
References
4. Conclusion
Belfiore, P.; Yoshizaki, H.T.Y. 2009. Scatter Search for a
In this paper, a mathematical model was set Real-Life Heterogeneous Fleet Vehicle Routing Problem
up on a real-life example of the business of a with Time Windows and Split Deliveries in Brazil,
reverse distribution company after defining European Journal of Operational Research 199(3): 750-758.
a transport problem where metaheuristic
methods were applied to solve the problem Buntak, K.; Grgurević, D.; Drožđek, I. 2012. Međusobni
of routing vehicles in the transport network. odnos logističkih i transportnih sustava [In English:
The purpose of applying mathematical Interrelation of logistics and transport systems]. Tehnički
methods is to reduce transportation costs glasnik 6(2): 228 –232.
by optimizing routes and the number of
vehicles. Minimizing transportation costs Carić, T. 2004. Optimizacija prometnih procesa [In
brings savings and increases profits for English: Optimization of Traffic Processes]. Faculty
businesses, so great attention must be taken of Transport and Traffic Sciences, Zagreb, Croatia.
in organizing transportation.
Čolić, P. 2013. Raspoređivanje vozila u stvarnom
For example, two metaheuristic methods vremenu uz pomoć genetskog programiranja [In English:
were used to optimize routes and solve Vehicle routing in real time with genetic programming].
vehicle routing problem - the Nearest Master Thesis, Faculty of Electrical Engineering and
Neighbor Heuristics method and several Computing, University of Zagreb, Croatia.

469
International Journal for Traffic and Transport Engineering, 2019, 9(4): 456 - 470

Da ntzig , G.B.; R a mser, J.H. 1959. T he Tr uck Pavlić Skender, H.; Grčić Fabić, M. 2013. Logistički
Dispatching Problem, Management Science 6(1): 80-91. špediter u fokusu prometnog i gospodarskog sustava
[In English: Logistics freight forwarder in the focus
Erdoğan, G. 2017. An open Source Spreadsheet Solver of the traffic and economic system]. Pomorski zbornik
for Vehicle Routing Problems, Computers & Operations 47-48: 95-107.
Research 84: 62-72.
Stanković, R.; Pašagić Škrinjar, J. 2015. Logistika i
Galić, A. 2018. Rješavanje problema usmjeravanja transportni modeli [In English: Logistics and transport
vozila hibridnim staničnim evolucijskim algoritmom models]. Faculty of Transport and Traffic Sciences,
[In English: Solving the Vehicle Routing Problem Using Zagreb, Croatia.
Hybrid Cellular Evolutionary Algorithm]. PhD thesis,
Faculty of Transport and Traffic Sciences, Zagreb, Toth, P.; Vigo, D. 2002. The Vehicle Routing Problem.
Croatia. Society for Industrial and Applied Mathematics,
Philadelphia, 29-51.
Hasle, G.; Kloster, O. 2007. Industrial Vehicle Routing.
In: Hasle G., Lie KA., Quak E. (eds) Geometric Modelling, Woxen ius, J. 1997. Ter m i na ls – A Ba r r ier For
Numerical Simulation, and Optimization. Springer, Berlin, Intermodality? In Proceedings of the Nordic Transport
Heidelberg, 397-435. Research’s conference on Intermodal Freight Transport. 22-
23 September Ebeltoft, Denmark, 1-16.
Laporte, G. 2007. What you should know about the
vehicle routing problem, Naval Research Logistics 54: Woxenius, J. 1998. Development of small-scale
811–819. intermodal freight transportation in a systems context.
PhD thesis, Department of Logistics Chalmers,
Pašagić, H. 2003. Matematičke metode u prometu [In Göteborg.
English: Mathematical methods in traffic]. Faculty of
Transport and Traffic Sciences, Zagreb, Croatia.

470