SlideShare a Scribd company logo
1 of 5
Download to read offline
1
A feasible solutionalgorithmfor a
primitivevehicle routing problem
Cem Recai Çırak
31 December 2015
Abstract
This paper is written to give brief information about
a feasible solution algorithm offer for a primitively
defined vehicle routing problem. The paper broaches
to subject with a short review about on a vehicle
routing problem solution algorithms. After that;
definition of the problem at issue, choice of the type
of offered algorithm, brief explanation of the
algorithm, simulation results on pseudo-randomly
generated artificial problems and comparative
interpretation about the performance of the
algorithm are follows explicated.
1. Introduction
The vehicle routing problem (VRP) is a
combinatorial optimization and integer
programming problem seeking to service a
number of customers with a fleet of
vehicles. Proposed by Dantzig and Ramser
in 1959, VRP is an important problem in the
fields of transportation, distribution, and
logistics. [1] The Vehicle Routing Problem
(VRP) is a general name given to an entire
type of problems in which a set of routes for
a fleet of vehicles based at single or multiple
depots must be determined for a number of
physically dispersed customers. The
objective of the VRP is to deliver a set of
customers with known demands on
minimum-cost vehicle routes starting and
finishing at a depot.
A representation of a characteristic input
for a VRP problem and one of its possible
solution outputs can be seen below in
Figure 1.
Figure 1: An instance of a VRP and its solution
Approximately all of the most commonly
used techniques to solve Vehicle Routing
Problems are heuristics and metaheuristics
since there is no exact algorithm which
guarantees to find the optimal solution
within admissible computing time while the
number of customers are getting larger.
This is due to the issue that VRP is an NP-
Hard class problem and its exact algorithm
time complexity T(n)=O(cn). Classification of
solution techniques and some of the most
commonly used solution algorithms are
listed.
Exact Approaches: Exact approaches
compute every possible solution until the
best (optimal) solution is found. This
technique uses divide and conquer strategy
by applying a branch and bound or branch
and cut algorithms on a K-tree. Customers
are set as leaves of K-tree. Then problem is
divided into small sub-problems and solved.
2
Heuristics: Heuristic methods perform a
relatively limited exploration of the search
space and get a not too bad but much faster
solution by applying some experience-
based guesses. Heuristics are divided into
two parts as constructive methods and 2-
phase algorithms.
In constructive methods, a feasible solution
is provided by checking whether it satisfies
the cost requirement and do not improve
upon. Some reputed constructive
algorithms are below:
 Clark and Wright savings
 Matching-based savings
 Multi-route improvement heuristics
 Kinderwater and Savelsbergh
 Thomson and Psaraftis
 Van Breedam
In 2-phase algorithms, customers are set as
nodes and then divided into clusters and
the routes are determined in which there
are paths connecting the clusters
themselves as a feedback loop. 2-phase
algorithms are divided into two types:
 Cluster first, route second
 Taillard
 The Petal algorithms
 The Sweep algorithms
 Fisher and Jaikumar
 Route first, cluster second
In many versions of 2-phase algorithms, the
main improvements are included in the
applied clustering technique.
Meta-Heuristics: Meta-heuristics performs
a complete search in the most appropriate
regions of the solution area. Meta-
heuristics gets better quality solutions in
comparison with heuristics. Some reputed
meta-heuristics algorithms are below:
 Ant algorithms
 Constraint programming
 Genetic algorithms
 Tabu search
2. Problem Definition
A very simplified Single Depot Multiple
Route Vehicle Routing Problem is defined as
follows:
 There is a single depot at the origin.
 There are n customers with pseudo-
randomly generated coordinates and
demands with via computer
programming. Their coordinates are
defined with a radius between 1 and
100 and an angle between 0° and 359°
with respect to the depot at the origin
(0,0). Their demands are defined
between 1 and 100.
 All customers connected with a
complete weighted graph and only one
edge connects any two of customers.
Weights of edges are defined as directly
equal to the geometric distance
between two customers which are
consisting the edge.
 Unlimited number of trucks are
available with identical capacities.
However, a truck can be used if and only
if it is necessarily needed. In other
words, if total demand exceeds the
maximum truck capacity, then usage of
another truck is allowed.
 Each customer on the network must be
visited exactly one times. Therefore,
their demands must be satisfied at
once.
 Total cost is assumed equal to the sum
of the weights of used edges.
Figure 2: Pseudo generated VRP scheme
3
n: Number of customers
m: Number of trucks used
i = 1, 2, …, n;
j = 1, 2, …, n;
k = 1, 2, …, m
C0: Depot
C1, C2, …, Ci, …, Cn: Customers
T1, T2, …, Tk, …, Tm: Trucks
M: Maximum truck capacity
Di: Demand of Ci,
where 0 < Di < 100, ∀i
Eij: Edge between Ci and Cj,
where i ≠ j, ∀i,j
Rk: Route of Tk,
which is a combination of Eijs
Wij: Weight of Eij,
where 0 < Wij < 200, ∀i,j
Wk: Total weight of Rk, the sum
of Wijs that belongs to Eijs
which are consisting Rk
C(Wk): Total cost function,
𝐶(𝑊𝑘) = ∑ 𝑊𝑘
𝑚
𝑘=1
Minimize total cost function:
min ∑ 𝑊𝑘
𝑚
𝑘=1
3. Solution Algorithm
A 2-phase heuristic algorithm is used for
solution. In first phase customers are
assumed as nodes and they are assigned
into a cluster. In second phase all clusters
are routed inside and minimized total
weights of each route are computed. Then
total cost function is calculated as sum of
total weights of all routes.
First phase algorithm works as:
1. Sort all customer according to their
angle with ascending order.
2. Then add sorted customers in order
into a cluster and subtract the
added customer’s demand from the
maximum truck capacity.
3. If remaining truck capacity is less
than the added customer’s demand,
remove back that last added
customer from the cluster.
4. Then create a new cluster and add
the removed customer into new
cluster and continue to add sorted
customers in order.
5. Check step 3 after each addition
operation and repeat steps 3 and 4
until all customers assigned into a
cluster.
Figure 3: Phase 1, clustered graph scheme
4
Second phase algorithm works as:
1. Start with first cluster. Add depot to
the cluster route as starting point.
Add all customers into an adjacent
list. Then sort all customers in the
cluster with respect to their
distances from the depot in
ascending order.
2. Select the closest customer and
assign it as current customer. Add
current customer to the cluster
route. Add its distance from the
depot to total weight of route. Then
mark current customer as visited.
3. Search in adjacent list except the
ones marked as visited. Find the
closest adjacent to current
customer and assign it as next
customer. Add next customer to the
cluster route. Add distance between
next customer and current
customer to total weight of route.
Then assign next customer as
current customer. Then repeat this
step until all adjacent list marked as
visited.
4. After all adjacent list is visited. Add
depot to the cluster route as
finishing point. Add depot distance
from the current customer to total
weight of route.
5. Apply first four steps to all clusters.
Then sum all total weight of routes
and find total cost function value.
Figure 4: Phase 2, routed graph scheme
4. Simulation Results
Average results are obtained over 1000
simulations for each different case, for
maximum truck capacity 500.
Table 1: Results for Maximum Truck Capacity 500
#*
Multiple Route Single Route
Cost Time #** Cost Time
10 532 1,6 6 496 1,9
20 1083 2,2 7 686 2,7
50 2438 4 8 1087 4,1
100 4954 7,8 8 1520 8,3
200 9996 13,1 9 2130 20
500 23811 30,2 9 3243 73,9
1000 40832 62,1 9 4523 235
10000 162200 654 9 12885 17515
Average results are obtained over 1000
simulations for each different case, for
maximum truck capacity 5000.
Table 2: Results for Maximum Truck Capacity 5000
#*
Multiple Route Single Route
Cost Time #** Cost Time
10 509 1,9 10 496 1,7
20 682 2,4 20 686 2,7
50 1070 4,4 50 1087 4,4
100 1692 8,3 65 1520 8,4
200 3168 16 83 2130 19,6
500 7907 39,1 85 3243 76,4
1000 15425 77,9 93 4523 225
10000 68479 817 97 12885 16330
* Number of customers
** Number of customers per route
Since the time complexity of overall
algorithm 𝑇(𝑛, 𝑚) = 𝑂(
𝑛2
𝑚
). This algorithm
performs approximately under linear time
complexity
5
5. Conclusion
Graph 1: Cost minimization comparison
Graph 2: Computation time comparison
As seen in Graph 1, while number of
customers is getting larger, multiple route
algorithm’s cost minimization performance
gets worse in comparison with single route
algorithm for maximum truck capacity 500.
However, if truck capacity increased (i.e. to
5000), multiple route algorithm’s
performance gets significantly much better.
As seen in Graph 2, while number of
customers is getting larger, multiple route
algorithm’s cost minimization performance
gets so much better in comparison with
single route algorithm for maximum truck
capacity 500. Also when truck capacity
increased, there is be no significant
increment in computation time of multiple
route algorithm.
Although, the offered algorithm has very
successful time complexity, its cost
minimization performance is not good
enough. Therefore, addition of 2-opt
algorithm to the end of this algorithm as
third step, may lead a significant
improvement in its cost minimization
performance also.
6. References
[1] Vehicle Routing Problem. (available on 31
December 2015). Retrieved from
http://neo.lcc.uma.es/vrp/

More Related Content

What's hot

HOPX Crossover Operator for the Fixed Charge Logistic Model with Priority Bas...
HOPX Crossover Operator for the Fixed Charge Logistic Model with Priority Bas...HOPX Crossover Operator for the Fixed Charge Logistic Model with Priority Bas...
HOPX Crossover Operator for the Fixed Charge Logistic Model with Priority Bas...IJECEIAES
 
A bi objective minimum cost-time network flow problem
A bi objective minimum cost-time network flow problemA bi objective minimum cost-time network flow problem
A bi objective minimum cost-time network flow problemAmirhadi Zakeri
 
MC0079 SMU 2013 Falll Session
MC0079 SMU 2013 Falll SessionMC0079 SMU 2013 Falll Session
MC0079 SMU 2013 Falll SessionNarinder Kumar
 
Minimization of Assignment Problems
Minimization of Assignment ProblemsMinimization of Assignment Problems
Minimization of Assignment Problemsijtsrd
 
Singular Value Decomposition: Principles and Applications in Multiple Input M...
Singular Value Decomposition: Principles and Applications in Multiple Input M...Singular Value Decomposition: Principles and Applications in Multiple Input M...
Singular Value Decomposition: Principles and Applications in Multiple Input M...IJCNCJournal
 
DESIGN OF DELAY COMPUTATION METHOD FOR CYCLOTOMIC FAST FOURIER TRANSFORM
DESIGN OF DELAY COMPUTATION METHOD FOR CYCLOTOMIC FAST FOURIER TRANSFORMDESIGN OF DELAY COMPUTATION METHOD FOR CYCLOTOMIC FAST FOURIER TRANSFORM
DESIGN OF DELAY COMPUTATION METHOD FOR CYCLOTOMIC FAST FOURIER TRANSFORMsipij
 
Analysis of LDPC Codes under Wi-Max IEEE 802.16e
Analysis of LDPC Codes under Wi-Max IEEE 802.16eAnalysis of LDPC Codes under Wi-Max IEEE 802.16e
Analysis of LDPC Codes under Wi-Max IEEE 802.16eIJERD Editor
 
IRJET- Comparison for Max-Flow Min-Cut Algorithms for Optimal Assignment Problem
IRJET- Comparison for Max-Flow Min-Cut Algorithms for Optimal Assignment ProblemIRJET- Comparison for Max-Flow Min-Cut Algorithms for Optimal Assignment Problem
IRJET- Comparison for Max-Flow Min-Cut Algorithms for Optimal Assignment ProblemIRJET Journal
 
A Review: Compensation of Mismatches in Time Interleaved Analog to Digital Co...
A Review: Compensation of Mismatches in Time Interleaved Analog to Digital Co...A Review: Compensation of Mismatches in Time Interleaved Analog to Digital Co...
A Review: Compensation of Mismatches in Time Interleaved Analog to Digital Co...IJERA Editor
 
SMART Seminar Series: Improving Public Transport Accessibility via the Optimi...
SMART Seminar Series: Improving Public Transport Accessibility via the Optimi...SMART Seminar Series: Improving Public Transport Accessibility via the Optimi...
SMART Seminar Series: Improving Public Transport Accessibility via the Optimi...SMART Infrastructure Facility
 
19 avadhanam kartikeya sarma 177-185
19 avadhanam kartikeya sarma 177-18519 avadhanam kartikeya sarma 177-185
19 avadhanam kartikeya sarma 177-185Alexander Decker
 
Queuing theory and its applications
Queuing theory and its applicationsQueuing theory and its applications
Queuing theory and its applicationsDebasisMohanty37
 
Integrating fuzzy and ant colony system for
Integrating fuzzy and ant colony system forIntegrating fuzzy and ant colony system for
Integrating fuzzy and ant colony system forijcsa
 
Sampling-Based Planning Algorithms for Multi-Objective Missions
Sampling-Based Planning Algorithms for Multi-Objective MissionsSampling-Based Planning Algorithms for Multi-Objective Missions
Sampling-Based Planning Algorithms for Multi-Objective MissionsMd Mahbubur Rahman
 
Area efficient parallel LFSR for cyclic redundancy check
Area efficient parallel LFSR for cyclic redundancy check  Area efficient parallel LFSR for cyclic redundancy check
Area efficient parallel LFSR for cyclic redundancy check IJECEIAES
 
Channel and clipping level estimation for ofdm in io t –based networks a review
Channel and clipping level estimation for ofdm in io t –based networks a reviewChannel and clipping level estimation for ofdm in io t –based networks a review
Channel and clipping level estimation for ofdm in io t –based networks a reviewIJARIIT
 
Discrete-wavelet-transform recursive inverse algorithm using second-order est...
Discrete-wavelet-transform recursive inverse algorithm using second-order est...Discrete-wavelet-transform recursive inverse algorithm using second-order est...
Discrete-wavelet-transform recursive inverse algorithm using second-order est...TELKOMNIKA JOURNAL
 

What's hot (18)

HOPX Crossover Operator for the Fixed Charge Logistic Model with Priority Bas...
HOPX Crossover Operator for the Fixed Charge Logistic Model with Priority Bas...HOPX Crossover Operator for the Fixed Charge Logistic Model with Priority Bas...
HOPX Crossover Operator for the Fixed Charge Logistic Model with Priority Bas...
 
A bi objective minimum cost-time network flow problem
A bi objective minimum cost-time network flow problemA bi objective minimum cost-time network flow problem
A bi objective minimum cost-time network flow problem
 
Queueing theory
Queueing theoryQueueing theory
Queueing theory
 
MC0079 SMU 2013 Falll Session
MC0079 SMU 2013 Falll SessionMC0079 SMU 2013 Falll Session
MC0079 SMU 2013 Falll Session
 
Minimization of Assignment Problems
Minimization of Assignment ProblemsMinimization of Assignment Problems
Minimization of Assignment Problems
 
Singular Value Decomposition: Principles and Applications in Multiple Input M...
Singular Value Decomposition: Principles and Applications in Multiple Input M...Singular Value Decomposition: Principles and Applications in Multiple Input M...
Singular Value Decomposition: Principles and Applications in Multiple Input M...
 
DESIGN OF DELAY COMPUTATION METHOD FOR CYCLOTOMIC FAST FOURIER TRANSFORM
DESIGN OF DELAY COMPUTATION METHOD FOR CYCLOTOMIC FAST FOURIER TRANSFORMDESIGN OF DELAY COMPUTATION METHOD FOR CYCLOTOMIC FAST FOURIER TRANSFORM
DESIGN OF DELAY COMPUTATION METHOD FOR CYCLOTOMIC FAST FOURIER TRANSFORM
 
Analysis of LDPC Codes under Wi-Max IEEE 802.16e
Analysis of LDPC Codes under Wi-Max IEEE 802.16eAnalysis of LDPC Codes under Wi-Max IEEE 802.16e
Analysis of LDPC Codes under Wi-Max IEEE 802.16e
 
IRJET- Comparison for Max-Flow Min-Cut Algorithms for Optimal Assignment Problem
IRJET- Comparison for Max-Flow Min-Cut Algorithms for Optimal Assignment ProblemIRJET- Comparison for Max-Flow Min-Cut Algorithms for Optimal Assignment Problem
IRJET- Comparison for Max-Flow Min-Cut Algorithms for Optimal Assignment Problem
 
A Review: Compensation of Mismatches in Time Interleaved Analog to Digital Co...
A Review: Compensation of Mismatches in Time Interleaved Analog to Digital Co...A Review: Compensation of Mismatches in Time Interleaved Analog to Digital Co...
A Review: Compensation of Mismatches in Time Interleaved Analog to Digital Co...
 
SMART Seminar Series: Improving Public Transport Accessibility via the Optimi...
SMART Seminar Series: Improving Public Transport Accessibility via the Optimi...SMART Seminar Series: Improving Public Transport Accessibility via the Optimi...
SMART Seminar Series: Improving Public Transport Accessibility via the Optimi...
 
19 avadhanam kartikeya sarma 177-185
19 avadhanam kartikeya sarma 177-18519 avadhanam kartikeya sarma 177-185
19 avadhanam kartikeya sarma 177-185
 
Queuing theory and its applications
Queuing theory and its applicationsQueuing theory and its applications
Queuing theory and its applications
 
Integrating fuzzy and ant colony system for
Integrating fuzzy and ant colony system forIntegrating fuzzy and ant colony system for
Integrating fuzzy and ant colony system for
 
Sampling-Based Planning Algorithms for Multi-Objective Missions
Sampling-Based Planning Algorithms for Multi-Objective MissionsSampling-Based Planning Algorithms for Multi-Objective Missions
Sampling-Based Planning Algorithms for Multi-Objective Missions
 
Area efficient parallel LFSR for cyclic redundancy check
Area efficient parallel LFSR for cyclic redundancy check  Area efficient parallel LFSR for cyclic redundancy check
Area efficient parallel LFSR for cyclic redundancy check
 
Channel and clipping level estimation for ofdm in io t –based networks a review
Channel and clipping level estimation for ofdm in io t –based networks a reviewChannel and clipping level estimation for ofdm in io t –based networks a review
Channel and clipping level estimation for ofdm in io t –based networks a review
 
Discrete-wavelet-transform recursive inverse algorithm using second-order est...
Discrete-wavelet-transform recursive inverse algorithm using second-order est...Discrete-wavelet-transform recursive inverse algorithm using second-order est...
Discrete-wavelet-transform recursive inverse algorithm using second-order est...
 

Viewers also liked

Maths Saves Money
Maths Saves MoneyMaths Saves Money
Maths Saves MoneyKartik918
 
Supply chain logistics : vehicle routing and scheduling
Supply chain logistics : vehicle  routing and  schedulingSupply chain logistics : vehicle  routing and  scheduling
Supply chain logistics : vehicle routing and schedulingRetigence Technologies
 
Leadership concepts and theories
Leadership concepts and theoriesLeadership concepts and theories
Leadership concepts and theoriesalaguraja76
 
Internet of Things
Internet of ThingsInternet of Things
Internet of ThingsVala Afshar
 

Viewers also liked (7)

Maths Saves Money
Maths Saves MoneyMaths Saves Money
Maths Saves Money
 
Routes optimization
Routes optimizationRoutes optimization
Routes optimization
 
VRP - Vehicle Routing Problem
VRP - Vehicle Routing ProblemVRP - Vehicle Routing Problem
VRP - Vehicle Routing Problem
 
Supply chain logistics : vehicle routing and scheduling
Supply chain logistics : vehicle  routing and  schedulingSupply chain logistics : vehicle  routing and  scheduling
Supply chain logistics : vehicle routing and scheduling
 
Transportation management
Transportation managementTransportation management
Transportation management
 
Leadership concepts and theories
Leadership concepts and theoriesLeadership concepts and theories
Leadership concepts and theories
 
Internet of Things
Internet of ThingsInternet of Things
Internet of Things
 

Similar to Feasible Solution Algorithm for Vehicle Routing Problem

An Enhanced Agglomerative Clustering Algorithm for Solving Vehicle Routing Pr...
An Enhanced Agglomerative Clustering Algorithm for Solving Vehicle Routing Pr...An Enhanced Agglomerative Clustering Algorithm for Solving Vehicle Routing Pr...
An Enhanced Agglomerative Clustering Algorithm for Solving Vehicle Routing Pr...ijtsrd
 
Dispatching taxi cabs with passenger pool
Dispatching taxi cabs with passenger poolDispatching taxi cabs with passenger pool
Dispatching taxi cabs with passenger poolBogusz Jelinski
 
The effective distribution of mtn vouchers in the kumasi metropolis
The effective distribution of mtn vouchers in the kumasi  metropolisThe effective distribution of mtn vouchers in the kumasi  metropolis
The effective distribution of mtn vouchers in the kumasi metropolisabdul rashid attah zakaria
 
Supply Chain Management - Optimization technology
Supply Chain Management - Optimization technologySupply Chain Management - Optimization technology
Supply Chain Management - Optimization technologyNurhazman Abdul Aziz
 
Parallel Artificial Bee Colony Algorithm
Parallel Artificial Bee Colony AlgorithmParallel Artificial Bee Colony Algorithm
Parallel Artificial Bee Colony AlgorithmSameer Raghuram
 
Mixed Integer Linear Programming Formulation for the Taxi Sharing Problem
Mixed Integer Linear Programming Formulation for the Taxi Sharing ProblemMixed Integer Linear Programming Formulation for the Taxi Sharing Problem
Mixed Integer Linear Programming Formulation for the Taxi Sharing Problemjfrchicanog
 
T HE PASSENGER VEHICLE ' S LOGISTICS TRANSPORTATION PLANNING PROBLEM BASED O...
T HE PASSENGER VEHICLE ' S LOGISTICS  TRANSPORTATION PLANNING PROBLEM BASED O...T HE PASSENGER VEHICLE ' S LOGISTICS  TRANSPORTATION PLANNING PROBLEM BASED O...
T HE PASSENGER VEHICLE ' S LOGISTICS TRANSPORTATION PLANNING PROBLEM BASED O...ijcsa
 
Optimal location of relief facility.pptx
Optimal location of relief facility.pptxOptimal location of relief facility.pptx
Optimal location of relief facility.pptxgyaneshtripathiirsme
 
Mb0048 operations research
Mb0048  operations researchMb0048  operations research
Mb0048 operations researchsmumbahelp
 
How to determine demand Centers-of-Gravity
How to determine demand Centers-of-Gravity How to determine demand Centers-of-Gravity
How to determine demand Centers-of-Gravity StellingConsulting
 
Mb0048 operations research
Mb0048   operations researchMb0048   operations research
Mb0048 operations researchsmumbahelp
 
IRJET- Real-Time Forecasting of EV Charging Station Scheduling for Smart Ener...
IRJET- Real-Time Forecasting of EV Charging Station Scheduling for Smart Ener...IRJET- Real-Time Forecasting of EV Charging Station Scheduling for Smart Ener...
IRJET- Real-Time Forecasting of EV Charging Station Scheduling for Smart Ener...IRJET Journal
 
Summary of “Efficient large-scale fleet management via multi-agent deep reinf...
Summary of “Efficient large-scale fleet management via multi-agent deep reinf...Summary of “Efficient large-scale fleet management via multi-agent deep reinf...
Summary of “Efficient large-scale fleet management via multi-agent deep reinf...MauroRubieri
 
Case Study for Plant Layout :: A modern analysis
Case Study for Plant Layout :: A modern analysisCase Study for Plant Layout :: A modern analysis
Case Study for Plant Layout :: A modern analysisSarang Bhutada
 

Similar to Feasible Solution Algorithm for Vehicle Routing Problem (20)

robotaxi2023.pdf
robotaxi2023.pdfrobotaxi2023.pdf
robotaxi2023.pdf
 
An Enhanced Agglomerative Clustering Algorithm for Solving Vehicle Routing Pr...
An Enhanced Agglomerative Clustering Algorithm for Solving Vehicle Routing Pr...An Enhanced Agglomerative Clustering Algorithm for Solving Vehicle Routing Pr...
An Enhanced Agglomerative Clustering Algorithm for Solving Vehicle Routing Pr...
 
Dispatching taxi cabs with passenger pool
Dispatching taxi cabs with passenger poolDispatching taxi cabs with passenger pool
Dispatching taxi cabs with passenger pool
 
MULTI-OBJECTIVE ANALYSIS OF INTEGRATED SUPPLY CHAIN PROBLEM
MULTI-OBJECTIVE ANALYSIS OF INTEGRATED SUPPLY CHAIN PROBLEMMULTI-OBJECTIVE ANALYSIS OF INTEGRATED SUPPLY CHAIN PROBLEM
MULTI-OBJECTIVE ANALYSIS OF INTEGRATED SUPPLY CHAIN PROBLEM
 
The effective distribution of mtn vouchers in the kumasi metropolis
The effective distribution of mtn vouchers in the kumasi  metropolisThe effective distribution of mtn vouchers in the kumasi  metropolis
The effective distribution of mtn vouchers in the kumasi metropolis
 
B04 05 1116
B04 05 1116B04 05 1116
B04 05 1116
 
Operations Research Project
Operations Research ProjectOperations Research Project
Operations Research Project
 
Supply Chain Management - Optimization technology
Supply Chain Management - Optimization technologySupply Chain Management - Optimization technology
Supply Chain Management - Optimization technology
 
Parallel Artificial Bee Colony Algorithm
Parallel Artificial Bee Colony AlgorithmParallel Artificial Bee Colony Algorithm
Parallel Artificial Bee Colony Algorithm
 
Mixed Integer Linear Programming Formulation for the Taxi Sharing Problem
Mixed Integer Linear Programming Formulation for the Taxi Sharing ProblemMixed Integer Linear Programming Formulation for the Taxi Sharing Problem
Mixed Integer Linear Programming Formulation for the Taxi Sharing Problem
 
Bg4102418422
Bg4102418422Bg4102418422
Bg4102418422
 
T HE PASSENGER VEHICLE ' S LOGISTICS TRANSPORTATION PLANNING PROBLEM BASED O...
T HE PASSENGER VEHICLE ' S LOGISTICS  TRANSPORTATION PLANNING PROBLEM BASED O...T HE PASSENGER VEHICLE ' S LOGISTICS  TRANSPORTATION PLANNING PROBLEM BASED O...
T HE PASSENGER VEHICLE ' S LOGISTICS TRANSPORTATION PLANNING PROBLEM BASED O...
 
Optimal location of relief facility.pptx
Optimal location of relief facility.pptxOptimal location of relief facility.pptx
Optimal location of relief facility.pptx
 
Mb0048 operations research
Mb0048  operations researchMb0048  operations research
Mb0048 operations research
 
How to determine demand Centers-of-Gravity
How to determine demand Centers-of-Gravity How to determine demand Centers-of-Gravity
How to determine demand Centers-of-Gravity
 
Mb0048 operations research
Mb0048   operations researchMb0048   operations research
Mb0048 operations research
 
IRJET- Real-Time Forecasting of EV Charging Station Scheduling for Smart Ener...
IRJET- Real-Time Forecasting of EV Charging Station Scheduling for Smart Ener...IRJET- Real-Time Forecasting of EV Charging Station Scheduling for Smart Ener...
IRJET- Real-Time Forecasting of EV Charging Station Scheduling for Smart Ener...
 
Aaai00 120
Aaai00 120Aaai00 120
Aaai00 120
 
Summary of “Efficient large-scale fleet management via multi-agent deep reinf...
Summary of “Efficient large-scale fleet management via multi-agent deep reinf...Summary of “Efficient large-scale fleet management via multi-agent deep reinf...
Summary of “Efficient large-scale fleet management via multi-agent deep reinf...
 
Case Study for Plant Layout :: A modern analysis
Case Study for Plant Layout :: A modern analysisCase Study for Plant Layout :: A modern analysis
Case Study for Plant Layout :: A modern analysis
 

More from Cem Recai Çırak

AM Radio Receiver with Automatic Gain Control Unit
AM Radio Receiver with Automatic Gain Control UnitAM Radio Receiver with Automatic Gain Control Unit
AM Radio Receiver with Automatic Gain Control UnitCem Recai Çırak
 
OKIDA Electronics Internship Report
OKIDA Electronics Internship ReportOKIDA Electronics Internship Report
OKIDA Electronics Internship ReportCem Recai Çırak
 
Arçelik A.Ş. Dishwasher Plant Internship Report
Arçelik A.Ş. Dishwasher Plant Internship ReportArçelik A.Ş. Dishwasher Plant Internship Report
Arçelik A.Ş. Dishwasher Plant Internship ReportCem Recai Çırak
 
Arçelik A.Ş. Dishwasher Plant Internship Project Presentation
Arçelik A.Ş. Dishwasher Plant Internship Project PresentationArçelik A.Ş. Dishwasher Plant Internship Project Presentation
Arçelik A.Ş. Dishwasher Plant Internship Project PresentationCem Recai Çırak
 
2 Level Guitar Hero Final Report
2 Level Guitar Hero Final Report2 Level Guitar Hero Final Report
2 Level Guitar Hero Final ReportCem Recai Çırak
 
JCN Electronics The Servant Project Report
JCN Electronics The Servant Project ReportJCN Electronics The Servant Project Report
JCN Electronics The Servant Project ReportCem Recai Çırak
 
Symmetric TSP with Euclidean Distances for Settlement Places in Turkey
Symmetric TSP with Euclidean Distances for Settlement Places in TurkeySymmetric TSP with Euclidean Distances for Settlement Places in Turkey
Symmetric TSP with Euclidean Distances for Settlement Places in TurkeyCem Recai Çırak
 

More from Cem Recai Çırak (9)

AM Radio Receiver with Automatic Gain Control Unit
AM Radio Receiver with Automatic Gain Control UnitAM Radio Receiver with Automatic Gain Control Unit
AM Radio Receiver with Automatic Gain Control Unit
 
OKIDA Electronics Internship Report
OKIDA Electronics Internship ReportOKIDA Electronics Internship Report
OKIDA Electronics Internship Report
 
Arçelik A.Ş. Dishwasher Plant Internship Report
Arçelik A.Ş. Dishwasher Plant Internship ReportArçelik A.Ş. Dishwasher Plant Internship Report
Arçelik A.Ş. Dishwasher Plant Internship Report
 
Arçelik A.Ş. Dishwasher Plant Internship Project Presentation
Arçelik A.Ş. Dishwasher Plant Internship Project PresentationArçelik A.Ş. Dishwasher Plant Internship Project Presentation
Arçelik A.Ş. Dishwasher Plant Internship Project Presentation
 
Wald-Wolfowitz Runs Test
Wald-Wolfowitz Runs TestWald-Wolfowitz Runs Test
Wald-Wolfowitz Runs Test
 
Capacity Planning
Capacity PlanningCapacity Planning
Capacity Planning
 
2 Level Guitar Hero Final Report
2 Level Guitar Hero Final Report2 Level Guitar Hero Final Report
2 Level Guitar Hero Final Report
 
JCN Electronics The Servant Project Report
JCN Electronics The Servant Project ReportJCN Electronics The Servant Project Report
JCN Electronics The Servant Project Report
 
Symmetric TSP with Euclidean Distances for Settlement Places in Turkey
Symmetric TSP with Euclidean Distances for Settlement Places in TurkeySymmetric TSP with Euclidean Distances for Settlement Places in Turkey
Symmetric TSP with Euclidean Distances for Settlement Places in Turkey
 

Feasible Solution Algorithm for Vehicle Routing Problem

  • 1. 1 A feasible solutionalgorithmfor a primitivevehicle routing problem Cem Recai Çırak 31 December 2015 Abstract This paper is written to give brief information about a feasible solution algorithm offer for a primitively defined vehicle routing problem. The paper broaches to subject with a short review about on a vehicle routing problem solution algorithms. After that; definition of the problem at issue, choice of the type of offered algorithm, brief explanation of the algorithm, simulation results on pseudo-randomly generated artificial problems and comparative interpretation about the performance of the algorithm are follows explicated. 1. Introduction The vehicle routing problem (VRP) is a combinatorial optimization and integer programming problem seeking to service a number of customers with a fleet of vehicles. Proposed by Dantzig and Ramser in 1959, VRP is an important problem in the fields of transportation, distribution, and logistics. [1] The Vehicle Routing Problem (VRP) is a general name given to an entire type of problems in which a set of routes for a fleet of vehicles based at single or multiple depots must be determined for a number of physically dispersed customers. The objective of the VRP is to deliver a set of customers with known demands on minimum-cost vehicle routes starting and finishing at a depot. A representation of a characteristic input for a VRP problem and one of its possible solution outputs can be seen below in Figure 1. Figure 1: An instance of a VRP and its solution Approximately all of the most commonly used techniques to solve Vehicle Routing Problems are heuristics and metaheuristics since there is no exact algorithm which guarantees to find the optimal solution within admissible computing time while the number of customers are getting larger. This is due to the issue that VRP is an NP- Hard class problem and its exact algorithm time complexity T(n)=O(cn). Classification of solution techniques and some of the most commonly used solution algorithms are listed. Exact Approaches: Exact approaches compute every possible solution until the best (optimal) solution is found. This technique uses divide and conquer strategy by applying a branch and bound or branch and cut algorithms on a K-tree. Customers are set as leaves of K-tree. Then problem is divided into small sub-problems and solved.
  • 2. 2 Heuristics: Heuristic methods perform a relatively limited exploration of the search space and get a not too bad but much faster solution by applying some experience- based guesses. Heuristics are divided into two parts as constructive methods and 2- phase algorithms. In constructive methods, a feasible solution is provided by checking whether it satisfies the cost requirement and do not improve upon. Some reputed constructive algorithms are below:  Clark and Wright savings  Matching-based savings  Multi-route improvement heuristics  Kinderwater and Savelsbergh  Thomson and Psaraftis  Van Breedam In 2-phase algorithms, customers are set as nodes and then divided into clusters and the routes are determined in which there are paths connecting the clusters themselves as a feedback loop. 2-phase algorithms are divided into two types:  Cluster first, route second  Taillard  The Petal algorithms  The Sweep algorithms  Fisher and Jaikumar  Route first, cluster second In many versions of 2-phase algorithms, the main improvements are included in the applied clustering technique. Meta-Heuristics: Meta-heuristics performs a complete search in the most appropriate regions of the solution area. Meta- heuristics gets better quality solutions in comparison with heuristics. Some reputed meta-heuristics algorithms are below:  Ant algorithms  Constraint programming  Genetic algorithms  Tabu search 2. Problem Definition A very simplified Single Depot Multiple Route Vehicle Routing Problem is defined as follows:  There is a single depot at the origin.  There are n customers with pseudo- randomly generated coordinates and demands with via computer programming. Their coordinates are defined with a radius between 1 and 100 and an angle between 0° and 359° with respect to the depot at the origin (0,0). Their demands are defined between 1 and 100.  All customers connected with a complete weighted graph and only one edge connects any two of customers. Weights of edges are defined as directly equal to the geometric distance between two customers which are consisting the edge.  Unlimited number of trucks are available with identical capacities. However, a truck can be used if and only if it is necessarily needed. In other words, if total demand exceeds the maximum truck capacity, then usage of another truck is allowed.  Each customer on the network must be visited exactly one times. Therefore, their demands must be satisfied at once.  Total cost is assumed equal to the sum of the weights of used edges. Figure 2: Pseudo generated VRP scheme
  • 3. 3 n: Number of customers m: Number of trucks used i = 1, 2, …, n; j = 1, 2, …, n; k = 1, 2, …, m C0: Depot C1, C2, …, Ci, …, Cn: Customers T1, T2, …, Tk, …, Tm: Trucks M: Maximum truck capacity Di: Demand of Ci, where 0 < Di < 100, ∀i Eij: Edge between Ci and Cj, where i ≠ j, ∀i,j Rk: Route of Tk, which is a combination of Eijs Wij: Weight of Eij, where 0 < Wij < 200, ∀i,j Wk: Total weight of Rk, the sum of Wijs that belongs to Eijs which are consisting Rk C(Wk): Total cost function, 𝐶(𝑊𝑘) = ∑ 𝑊𝑘 𝑚 𝑘=1 Minimize total cost function: min ∑ 𝑊𝑘 𝑚 𝑘=1 3. Solution Algorithm A 2-phase heuristic algorithm is used for solution. In first phase customers are assumed as nodes and they are assigned into a cluster. In second phase all clusters are routed inside and minimized total weights of each route are computed. Then total cost function is calculated as sum of total weights of all routes. First phase algorithm works as: 1. Sort all customer according to their angle with ascending order. 2. Then add sorted customers in order into a cluster and subtract the added customer’s demand from the maximum truck capacity. 3. If remaining truck capacity is less than the added customer’s demand, remove back that last added customer from the cluster. 4. Then create a new cluster and add the removed customer into new cluster and continue to add sorted customers in order. 5. Check step 3 after each addition operation and repeat steps 3 and 4 until all customers assigned into a cluster. Figure 3: Phase 1, clustered graph scheme
  • 4. 4 Second phase algorithm works as: 1. Start with first cluster. Add depot to the cluster route as starting point. Add all customers into an adjacent list. Then sort all customers in the cluster with respect to their distances from the depot in ascending order. 2. Select the closest customer and assign it as current customer. Add current customer to the cluster route. Add its distance from the depot to total weight of route. Then mark current customer as visited. 3. Search in adjacent list except the ones marked as visited. Find the closest adjacent to current customer and assign it as next customer. Add next customer to the cluster route. Add distance between next customer and current customer to total weight of route. Then assign next customer as current customer. Then repeat this step until all adjacent list marked as visited. 4. After all adjacent list is visited. Add depot to the cluster route as finishing point. Add depot distance from the current customer to total weight of route. 5. Apply first four steps to all clusters. Then sum all total weight of routes and find total cost function value. Figure 4: Phase 2, routed graph scheme 4. Simulation Results Average results are obtained over 1000 simulations for each different case, for maximum truck capacity 500. Table 1: Results for Maximum Truck Capacity 500 #* Multiple Route Single Route Cost Time #** Cost Time 10 532 1,6 6 496 1,9 20 1083 2,2 7 686 2,7 50 2438 4 8 1087 4,1 100 4954 7,8 8 1520 8,3 200 9996 13,1 9 2130 20 500 23811 30,2 9 3243 73,9 1000 40832 62,1 9 4523 235 10000 162200 654 9 12885 17515 Average results are obtained over 1000 simulations for each different case, for maximum truck capacity 5000. Table 2: Results for Maximum Truck Capacity 5000 #* Multiple Route Single Route Cost Time #** Cost Time 10 509 1,9 10 496 1,7 20 682 2,4 20 686 2,7 50 1070 4,4 50 1087 4,4 100 1692 8,3 65 1520 8,4 200 3168 16 83 2130 19,6 500 7907 39,1 85 3243 76,4 1000 15425 77,9 93 4523 225 10000 68479 817 97 12885 16330 * Number of customers ** Number of customers per route Since the time complexity of overall algorithm 𝑇(𝑛, 𝑚) = 𝑂( 𝑛2 𝑚 ). This algorithm performs approximately under linear time complexity
  • 5. 5 5. Conclusion Graph 1: Cost minimization comparison Graph 2: Computation time comparison As seen in Graph 1, while number of customers is getting larger, multiple route algorithm’s cost minimization performance gets worse in comparison with single route algorithm for maximum truck capacity 500. However, if truck capacity increased (i.e. to 5000), multiple route algorithm’s performance gets significantly much better. As seen in Graph 2, while number of customers is getting larger, multiple route algorithm’s cost minimization performance gets so much better in comparison with single route algorithm for maximum truck capacity 500. Also when truck capacity increased, there is be no significant increment in computation time of multiple route algorithm. Although, the offered algorithm has very successful time complexity, its cost minimization performance is not good enough. Therefore, addition of 2-opt algorithm to the end of this algorithm as third step, may lead a significant improvement in its cost minimization performance also. 6. References [1] Vehicle Routing Problem. (available on 31 December 2015). Retrieved from http://neo.lcc.uma.es/vrp/