SID: 311092926

SID: 420086635

Content
1.

Introduction.........................................................................................................................2
1.1 Background...................................................................................................................2
1.2 Objective.......................................................................................................................2

2.

Pre-screen of data................................................................................................................3

3.

Part1- Vehicle Routing Plans...............................................................................................4

3.1 Tour 1: Insertion heuristic combined sweeping method...............................................4
3.2 Tour 2: Nearest Neighbour Heuristic............................................................................6
3.3 Conclusion....................................................................................................................7

Part 2 - Distribution Centre allocation.................................................................................8

4.1 Cluster Selection...........................................................................................................8
4.2 COG model...................................................................................................................8
4.3 New Routing Plan.........................................................................................................9

5. Recommendation................................................................................................................10
6. Reference..............................................................................................................................12
Appendix 1...............................................................................................................................13
Appendix 2...............................................................................................................................14
Appendix 3...............................................................................................................................14
Appendix 4...............................................................................................................................15
Appendix 5...............................................................................................................................15
Appendix 6...............................................................................................................................16
Appendix 7...............................................................................................................................17

SID: 311092926
SID: 420086635

1.

Introduction

1.1 Background
BetterWay Manufacturing Pty Ltd is a manufacturing company dealing with single product
line. It centralises its operations and uses light trucks to ship products directly from the large
factory to its customers. Fortunately, the company has successfully grown and is leading its
customer base to jump at 25. The implication is that the company uses 5 fleets to provide
high level of customer service.
However, increasing the fleet is not the only solution. The company intends to design a new
routing plan which is apart from being efficient, eliminating non-value added cost of fleet.
BetterWay also has concern about price hike in fuel due to new policy of carbon tax. Under
this scenario, the top management prefer to take pro-active action. One of the alternatives is
to decentralise its operation by opening three distribution centres that are closer to its
customer. A successful facilities locating plan can lead to cost-effectiveness and sustainable
operation in the future.
1.2 Objective
The specific objectives of the report are:

Assess and evaluate given data to use appropriate vehicle routing methods.
Develop routing plans through three models, i.e. Nearest Neighbour, Insertion Heuristic

combined Sweeping method and transshipment model.

Clusters with COG model analysis and evaluation to determine location of distribution

centres.
Calculate the total distance required for each plan s in km (assumed unit).
Assess and compare model in terms of cost savings and total distance.
Recommendations to the better implementation of the optimised routing plan.

SID: 311092926
SID: 420086635

2.

Pre-screen of data

Data that provided by BetterWay include x and y coordinate of depot and customers, demand
of customers, default route, operational constraints, shipping cost of light and heavy trucks.
Two values need to be evaluated in the end of each routing plan, i.e. total cost of
transportation and total distance.
(1) TC= Distances travelled by light trucks*10 + Distances travelled by light trucks *1
(2) TD= Distances travelled by light trucks + Distances travelled by light trucks

Figure 2.1
The above figure shows the location of every customer, it is worth noticing a group of
customers (22, 20, 18, 19, and 16), which is far away from factory but have stronger demand.
This situation increases the difficulties when we design the routing plan. Likewise, an
extreme customer (No. 8) appears on the right bottom corner that might have impact for the
whole routing plan.
The total demand of 25 customers is 1340, which indicates that the average demand is 268.
Trucks constraint or capacity is 335, while all customers are supposed to be served exactly
once.

SID: 311092926
SID: 420086635

3.

Part1- Vehicle Routing Plans

3.1 Tour 1: Insertion heuristic combined sweeping method

Some of the knowledge of sweeping method is employed to determine the starting point of
five trucks. Research (Suthikamnarunai, 2008) has reviewed that sweep heuristic is one of the
most frequently used methods in CVRP problem, selecting the best route to deliver goods to
customers from/to desired depot.
Insertion heuristic which is quiet straight forward is originally used to solve TSP problems.
But it would be also employed in this tour since it can be applied to CVRPs and DVRPs
almost without modification (Laporte, 1991). And in order to apply Insertion, we have to
determine two nodes first, factory and beginning customer (Collins, 2013). In this report, we
will only use sweeping method to decide the starting point (customer seeds) of insertion
heuristic.
100
97.7

97.7

97.1
90.5

90

90.5
84.9

80

79.3
73.7

70

73.7

70.5

64.9

60

54.5

50

46.5
40.1

40

Customers
Factory

43.3

32.1

30

20.1
19.3

20
10

23.3
16.9

22.5
16.9
8.1 6.5
1.7

0
0

10

20

30

40

50

60

70

80

90

0.1
100

Figure 3.1
Based on the knowledge of sweeping method and the factory location, we divided customers
distribution into five areas equally by generating rays since five trucks are equipped to deliver
our goods. As shown in figure 3.1, customers are concentrated in area 3, while areas 4 and 5
just have a few. As a result, we decided that two customers seeds should be placed in area 3

SID: 311092926
SID: 420086635

while only one seed would be placed in area 4, 5. Customers 11, 10, 18, 8, 17 have been
selected to be our customer seeds. In this sense, in light of the capacity constraints and
insertion heuristic, we constructed the routes below:

Figure 3.2
For instance, Truck 3 would go through customer 16, 19, 18, 20 and then return to factory.
The total distance and capacity for truck 3 are 193.937 kms and 290 respectively. From
appendix 1, it is shown that there are cross lines in each route (i.e. the route can be upgraded
by improvement method such as r-opt). In this report, swapping algorithm (4-opt) is
employed to improve the performance of the route (Collins, 2013). After improvement, route
information is shown below:

Figure 3.3
Appendix 2 showed that most of the routes have been optimized and the overall distance has
been reduced by 253.623 kms (1166.77-913.15). It is noted that total capacity for five trucks
is 1340, while the maximum capacity is 1675 (355*5).

SID: 311092926
SID: 420086635

3.2 Tour 2: Nearest Neighbour Heuristic

The idea of nearest neighbour heuristic is straight forward. Starting from one node, we add
nearest node to the tour and then repeat previous step (Collins, 2013). In this report, we stop
adding the nearest route while the capacity constraint is fulfilled. Then we would start a new
route for a new truck.
In order to calculate the shortest distance, we generated the distance matrix. It is worth
mentioning that node 26 represents the location of factory.

Figure 3.4
In this chapter, most of works have been done manually. It is relatively time-consuming to
calculate all the routes, which is one of the aspects that should be enhanced in the future. The
result is shown below:

SID: 311092926
SID: 420086635

Figure 3.5
Appendix 3 showed the map of trucks delivery. However, we can see that the outcome is not
efficient. Similar to Tour 1, we utilized swapping method to optimize the route. After the
optimization, we have the new result:

Figure 3.6
Comparing to the original Nearest Neighbour result, the total distance of optimal figure is
52.3 kilometres less than the original one, which is improved by the swapping algorithm. The
graphical map of five routes is shown in Appendix 4.
3.3 Conclusion

Figure 3.7
In this part, we used two alternative methods to provide two feasible vehicle routing plan for
BetterWay. We recommended that BetterWay should employ the insertion combined
sweeping method to plan their routing trip since it has less total distance, comparing to NN
method. The less distances trucks travelled indicate that company will save a lot on transport
cost. The distance difference of these two models is 58.55 kms.
It is worth noticing that there is an extreme customer #8 (95.5, 0.1) which would be the
reason that NN heuristic does not performed very well. The selection of customers seed
method is debatable, but it still delivers a better performance in vehicle routing plan selection.

SID: 311092926
SID: 420086635

Part 2 - Distribution Center allocation

Considering the impact of carbon taxes on the fuel price, BetterWay tends to develop three
distribution centers to alleviate its total transport cost in the future. Likewise, a successful
facility location can have significant impacts on inventory, transportation, and level of
customer satisfaction. Numerous considerations need to be made while deciding facility
location, such as, cost of land, cost of utilities, potential for expansion and etc (Collins,
2013). However, in order to simplify the case, we only consider the specific location of
distribution, cost optimization and the distance for BetterWay.
4.1 Cluster Selection
Before deciding the location, the first and foremost thing is to select the proper customers
cluster. With reference to Part 1, we employed some of the knowledge of sweeping method.
In this part, we could also determine the customers cluster with the help of sweeping
method. From Appendix 5, we can see the cluster we decided:

Figure 4.1

Figure 4.2

4.2 COG model

Centres of Gravity method (COG) is employed after establishing clusters. This method is
widely used for factories or warehouses based on transport costs alone (without the
consideration of fixed or other costs). Study (CFIRE, 2013) also states that logistics network
efficiency is highly influenced by transportation network model. With COG method, we are
able to determine the proper location for distribution centres.

SID: 311092926
SID: 420086635

As illustrated in figure 9, three basic variables are used in COG, which are volume flowing
from/to customer, transportation rate, and coordinate points. And bar X and Y represent the
coordinate points for distribution centre.
Appendix 5 showed some of the steps we established the COG model and calculated Dist.
Centre location. In the end, we have the location below:

Figure 4.3
4.3 New Routing Plan
Based on clusters and distribution centre location, we have a new route plan for BetterWay,
which is established with reference to insertion heuristic and the rules of avoiding overlapped
route (route lines overlapped).

Figure 4.4
The capacity used is the same with the previous tour plan since it refers to the demand of
each customer. On the other hand, the total distance is far less than the previous route plan
due to three distribution centres.

SID: 311092926
SID: 420086635

5. Recommendation
This report provides three alternative ways for BetterWay to establish the optimal vehicle
routing plan. The purpose of the new routing plan is to reduce transport cost while
simultaneously satisfying customers demand.

Figure 5.1 Distance is in kms and Cost in dollars

The above table compare total distance and cost for each method we employed, which
implies that insertion heuristic is better than nearest neighbour method in this case, however,
the transshipment model shows better performance than these two methods in terms of longterm perspective (only transport cost is considered in this case). For the total cost,
transshipment model is roughly 3000 dollars less than NN and insertion model, while total
distance has almost the same result with slight decrease. This might be due to the fact that
distribution centre is now closer to the customers cluster and the company doesnt have bear
the cost of five trucks carrying comparatively small load for long distance from factory direct
to customers.. Thus, we would recommend BetterWay to conduct the new distribution centre
model plan which should be worth the investment in the long run .Again. there are several
factors effecting it such as construction cost, energy cost, running cost, customer demand in
future, recession etc. These need to be carefully assessed and evaluated by the management
before taking any further action
Limitations of this report did exist. When considering constraints, we only take transport fees
into account. Yet in reality, numerous influencers would have effects on vehicle routing plan
such as time window, labours cost, customer service level expected taxes, road policy, road
congestion etc. Therefore, these potential influencers are supposed to be considered in the
future.

SID: 311092926
SID: 420086635

Some other Recommendations

Reduce the delivery trips and increase the on-carriage to customers is crucial to save fuel
( kellner and Igl 2012).
We also recommend sharing transportation and distribution centers which are already being
running in the nearby customer location. As major cost in logistics is due to inventory and
transport, it needs to be cut. Using own expensive distribution centers gain causes centrally
operated supply chain which could result into more cost because of holding inventory,
storing cost, loading and unloading cost, labour cost, regular trucks coming back empty, and
producing higher emissions without producing any value added service. Sharing cargos can
cut transport cost as number of kilometres driven reduces with full capacity of trucks being
utilised ( Mallidisa et al. 2012)
Nowadays, eco-friendly freight/ green fleet transport is widely worldwide by big companies
like Walmart which could be considered as a good option to save both on cost and construct
environmental friendly supply chain.

SID: 311092926
SID: 420086635

6. Reference
Andrew Collins, 2013, Logistics System, Fundamentals of Computational Logistics, Sydney University Course
Resource, viewed on 7th June, 2013

N.Suthikamnarunai, 2008, A Sweep Algorithm for the Mix Fleet Vehicle

Routing Problem, Proceedings of the International MultiConference of Engineers and Computer Scientists, 2008
vol 2
CFIRE, Module Six: Logistics Network Design and Facility Location, viewed on 8 th Jun, 2013,

http://www.wistrans.org/cfire/logistics/P2/M6.pdf
Transshipment Models, viewed on 7th June 2013,
http://www.math.mrt.ac.lk/TMJAC/OPertional%20Research%20techniques
%201/OR%20Lecture%208.pdf
Kellner, F. And Igl,j 2012, Estimating the effect of changing retailing structures
on the greenshouse gas performance of FMCG distribution network, viewed 14 th
2013,
http://link.springer.com.ezproxy2.library.usyd.edu.au/content/pdf/10.1007%2Fs12
159-012-0063-3.pdf
Mallidisa, I., Dekkerb, R and Vlachosa, D. 2012, The impact of greening on
supply chain design and cost: a case for a developing region, Journal Of
Transport Geography, Volume 22, pp. 118-128, viewed 14 th June 2013,
http://www.sciencedirect.com.ezproxy2.library.usyd.edu.au/science/article/pii/S09
6669231200004X
Terreri, 2008,Eight Steps to a Greener Supply Chain, World Trade, viewed
14thJune 2013,
http://search.proquest.com.ezproxy2.library.usyd.edu.au/docview/228278244

Gilbert Laporte, 1991, The vehicle routing problem: An overview of exact and approximate
algorithms, European Journal of Operational research 59(1992) 345-358, North-Holland

SID: 311092926
SID: 420086635

Appendix 1

SID: 311092926
SID: 420086635

Appendix 2

Appendix 3

SID: 311092926
SID: 420086635

Appendix 4

Appendix 5

SID: 311092926
SID: 420086635

Appendix 6

SID: 311092926
SID: 420086635

Appendix 7