See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330796754

A Sustainable Vehicle Routing Problem for Indian Agri-Food Supply Chain

Network Design

Conference Paper · December 2018

DOI: 10.1109/POMS.2018.8629450

CITATIONS READS

26 811

4 authors, including:

Rakesh Patidar Venkatesh Bellamkonda

PDPM Indian Institute of Information Technology, Design and Manufacturing Jaba… PDPM Indian Institute of Information Technology, Design and Manufacturing Jaba…
9 PUBLICATIONS 140 CITATIONS 1 PUBLICATION 25 CITATIONS

SEE PROFILE SEE PROFILE

Saurabh Pratap
Indian Institute of Technology (Banaras Hindu University) Varanasi
103 PUBLICATIONS 2,086 CITATIONS

SEE PROFILE

All content following this page was uploaded by Saurabh Pratap on 08 August 2019.

The user has requested enhancement of the downloaded file.

 A Sustainable Vehicle Routing Problem for Indian
Agri-Food Supply Chain Network Design

Rakesh Patidar Bellamkonda Venkatesh Saurabh Pratap

Mechanical Engineering dept. Computer Science Engg. dept. Mechanical Engineering dept.
PDPM IIITDM PDPM IIITDM PDPM IIITDM
Jabalpur, India Jabalpur, India Jabalpur, India
[email protected] [email protected] [email protected]

Yash Daultani
Operations Management.
ABV-IIITM
Gwalior, India
[email protected]

Abstract— The agri-food supply chain network complicated by the cogitation of the distinct product and
design plays a significant role in the economy of the process characteristics. The food supply chains,
country. The existing traditional agri-food supply necessitating production, and distribution, are increasingly
chain (AFSC) network in India is a vast supply chain dynamic and complex (Ahumada and Villalobos, 2009) [2].
which has been operating haphazardly in an The existing traditional agriculture supply chain is a vast
unstructured manner. The system is basically area which has been operating haphazardly in an
controlled by various intermediaries; this is leading unstructured manner since the inception.
unsustainable and inefficient supply chain in terms of
time and money. The most affecting factor in the agri- The system is basically controlled by various
food supply chain goes in credit of the unorganized intermediaries leading the system in unsustainable and
manner of transportation from the farmers to the inefficient in both time and money factors. The system has
nearest food hub (market) wherein, they sell their been always pushing the farmers to run down in a poor
products. This process has been a major bottleneck in profit or many a times in the loss. Apte and Vishwanathan
terms of profitability to the farmers wherein the (2000) [3] alluded that about 30% of the total costs are
transportation costs are the major contributor to the incurred in the distribution process. The most affecting
overall costs. In traditional Indian AFSC, factor in the supply chain goes in credit of the
geographically dispersed individual farmers bring unorganized manner of transportation of the supply from
their product into the market for selling. In this the farmers to the nearest hubs (market) wherein they can
phenomenon, they have to pay higher transportation exchange their products with the money. And this incurs a
cost. Therefore, we propose the vehicle routing for the lot many additional costs of holding the products,
collection of products from farmers to the hub. A additional transportation costs, and miscellaneous costs
single period vehicle routing model is developed to while making the products flow from farmers to the
form an optimum travel route incurring minimum nearest hubs and hardly leaving the farmers with minimal
costs and allowing a pool of products (from the profit (Patidar et al., 2018) [4]. The multi-depot vehicle
farmers) supply to reach the destination hub. Genetic routing problem (MDVRP) is a significant class of the
algorithm (GA) and particle swarm optimization vehicle routing problems which in turn can be used in the
(PSO) is used to solve the proposed mathematical complex supply chain scenario of Indian agriculture. So,
model and validated on a practical case scenario of the the authors of this paper have tried to solve the problem of
central part of India. increasing transportation costs at the end of farmers by
Keywords— Agri-Food supply chain, supply chain network setting up a vehicle routing from the hubs to the farmers
design, multi-depot vehicle routing in the village. The solution will be revamping the entire
traditional system and will be reducing the overall costs
I. INTRODUCTION incurred in the process. The model is proposed for single
According to Indian Economic Survey 2018 [1], period, single product and multiple vehicles with a
conducted by Ministry of Finance, Government of India, constant capacity. The paper focuses on the sustainable
agriculture sector is the domain in which more than 50 per development of the supply chain network of Indian
cent of the total workforce in India is employed which is agriculture by allowing the futuristic approaches for
around 17-18 percent to the country’s GDP is contributed transparent information flow in the supply chain.
by the same and also, estimated figures around 25.7 per cent
of the total agricultural workers would be dropped by 2050 II. LITERATURE REVIEW
from 58.2 per cent in 2001. The reduction of the workforce Generally, the food supply chain operations have been
in this massive sector has been observed since the inception into a lot of research because of the increasing demand for
of the complexion in the agro-food supply chain. The the quality of food items. Wastages are inexorable along
operations carried along the food supply chains are the agro-food supply chain; almost 33% of the global food

978-1-5386-9500-5/18/$31.00 ©2018 IEEE

production is squandered (Gustavsson et al., 2011) [5]. To villages as per the demand, and supply from the village,
reduce the losses incurred, factors like integration of food and various constraints for the splendor and smooth
production and distribution are required (Cook, 2002) [6]. endurance of the system.
According to Food Hubs (based in the USA as per NGFN  Quantity shipped from farmers to hub based on
report) [7] and New England Food Hub Networks [8], one demand of hub and availability of farmers
is the transportation of the food products and the other is  Number of vehicle required
product distribution. The overall costs incurred in the
 Vehicle route for each hub
complete supply chain increases by many times during the
phase of transportation. This is where the government and The model is assumed to have demand of single period,
many research associations are trying to build the bridge single product, multi depot (hub), and multiple vehicles
to reduce the cost. Zhang et al. (2003) [9] contrived a with constant capacity travelling for maximum limited
system of physical line of distribution to suffice the agro- distance with a constant transportation cost when the
food products standard essentialities that reduces the total vehicle is routed.
cost incurred in the shipment and storage. Lucas and
Chhajed (2004) [10] dealt with the locality complications The following is the mathematical modeling of our
and its execution in agricultural domain. The article defined problem.
establishes the solution methodology using various A. Set of Indices
operational research methods confronting with the
strategic production-distribution network design in f Set of farmers
agricultural sector. h Set of hubs
Blackburn and Scudder (2009) [11] bring forward a
v Set of vehicles,
cost minimization model for perishable product supply
chain design. The paper also speaks upon the deterioration
food products. Kopanos et al. (2012) [12] presented an i, j Any node in the proposed model
operations and logistics model for production-planning of
a multi-site, semi-continuous food processing industry
using Mixed Integer Programming (MIP). Jayaraman and
Ross [13] presented an efficient and practical
methodology for solving a multi-product multi-echelon
problem by the simulated annealing (SA) algorithm to
distribution network. Bachlaus et al. (2008) [14] had put
forward a nimble and an integrated multi-echelon supply
chain network. The problem was modeled as a
multifaceted mathematical programming model in order to
minimize the overall costs experienced and to maximize
the volume flexibility and plant flexibility and was solved
using HTPSO algorithm. An efficient improvisation in the
vehicle routing for the intake of the products from the
farmers (producers) based on the various constraints in the
Indian scenario, where the producers are scattered rather
than in an organized manner, wherein extensive work
focussing on the Indian scenario hasn’t been carried out.
The research has been restricted only to the cross-docking
and its operations happening inside a cross-dock has been
done. But the real-life application of the problem
especially in the agriculture domain hasn’t been done Fig. 1: Supply chain model for the collection of food items
extensively which can, in turn, reduce the total costs
incurred including fixed and variable costs which in turn B. Parameters
would help to maximize the profits of the farmers The harvest quantity available to supply from the
(producers). farmer f
III. MATHEMATICAL MODELING The quantity demand by hub h for

Distance from farmer i to farmer f ( )

The model is a two-echelon vehicle routing model for a
single period and a single product. There are multiple Distance from farmer f to hub h
fixed locations of hubs and we need to identify farmers
allocation to these hubs such total cost is minimized. The Transportation cost from farmer to farmer
total cost includes transportation cost from farmer to the Transportation cost from farmer to hub
hub, vehicle routing cost between farmer nodes and fixed
cost of vehicles. To develop a model for inbound product Constant capacity of the vehicles
collection system, we consider a set of geographically Quantity of food product flowing from node i to
dispersed farmers that produce one item. The items are node j
collected by a fleet of vehicles that are dispatched from
the hubs. The vehicle is routed from the hub to the nearest
C. Decision Variables Ensures and constraints the maximum capacity of the
vehicle is greater than the quantity supplied by the farmers
Quantity transported from farmer f to hub h

1 if the farmer f is assigned to hub h (9)

0 otherwise
The maximum load transported in between the node i and
node j is given by (Flow conservation)
1 if vehicle v is moving from node i to node j
0 otherwise (10)

D. The model The number of vehicles required for the routing should be
less than the available number of vehicles (K)
Objective Function
(11)
Minimize
IV. SOLUTION APPROACH
Z=
The proposed problem is formulated as mixed integer
non-linear programming (MINLP), we have used the two
meta-heuristic approach (Genetic Algorithm) and Particle
+ (1) Swarm Optimization to minimize the total transportation
cost associated with agro-food supply chain. The proposed
Objective function (1) minimizes total cost which includes vehicle routing problem is NP hard problem [15].
transportation cost from farmer to the hub, vehicle
movement cost between farmer nodes and fixed cost of Genetic Algorithm
used vehicles. The Genetic Algorithm is a nature inspired algorithm,
developed by Goldberg & Deb (1991) [16]. In this
Subjected to: approach, we have developed the feasible chromosomes
that will represent the scheduling of vehicles with respect
Quantity shipped from farmers to hub must be greater to destination nodes and demand capacity [17]. The
than demand of that hub completed process of Genetic Algorithm has been
described below and flowchart of GA has been shown in
(2) figure 2.

Quantity shipped from farmer to hub must be less than Initial Population Chromosome
available at farmer The scheduling of vehicles is generated through the
generation of random numbers. For each block of
(3) chromosomes of vehicle scheduling, on the basis of
generated random numbers, the destination hubs are
One farmer is assigned to only one hub assigned.

Crossover
(4)
From the generated feasible chromosomes, two
chromosomes have to select from the given population on
Vehicle arrival at any node j from any i is only once
the basis of assign random numbers ( ) and
(5) crossover probability (Pc). If the , then there will
be the single point crossover between the selected
Vehicle departure from any node j to any node i is only chromosomes and generate the new (child) chromosome.
once
Mutation
(6)
The Each block of child chromosome is mutated with a
mutation probability pm=0.06. The mutation changes the
Ensure a consecutive movement at node i-r-j
schedule of vehicle from one farmer to other farmer and
one hub to another hub.
(7)
Considers if a food collection vehicle arrives and departs Fitness Evaluation
the hub in the pickup process of product from farmers
The fitness of the chromosomes is evaluated on the basis
of proposed objective function. The fitness function must
(8)
satisfy all the constraints simultaneously. If not, then Figure 3: Updation of Particle Position
follow the again same steps. Updation of Memory
In this process, we update the particle best position local
best pbest and global best position gbest (as shown in
equation c).

(c)

Stoppin
g
Criteri
a
If, we
Figure 2: Flowchart of Genetic Algorithm get best
possibl
Particle Swarm Optimization e
The PSO is a robust stochastic optimization technique solutio
based on the movement and intelligence of swarms [16]. It n
has been applied successfully to a wide variety of search f(pbest)
and optimization problems and built by abstracting the and
working mechanism of natural phenomenon. The steps of f(gbest)
particle swarm optimization are as follows: then
stop the solution. Either repeats the step of updation of
Initialization velocity and position.
In first step, we have to generate the random initial Figure 4: Flowchart of Particle Swarm Optimization
population of particles position and on the basis of
population, generate the velocities for certain dimensional The figure 4 describes the process flowchart of Particle
problem. The lower and upper bound decides the solution Swarm Optimization.
space of proposed problem.
V. RESULT & DISCUSSION
Updation of velocity
In each and every iteration, there will be velocity updation
on the basis of eq. (a) In this problem, we have tested the model on two case
scenarios for the one-month data set. In first case, there
are 50 farmers, 10 hubs and 10 vehicles. The Genetic
(a) Algorithm and the Particle Swarm Optimization are used
to determine the minimal total transportation cost. The
where, V(t+1) and xt represents the particle velocity and GA operating parameters are crossover probability
particle position respectively. The pbest represents the pc=0.8, mutation probability pm=0.06 and the number of
individual local best position and gbest represents the population is 10. The PSO, operating parameters are
global best solution at that iteration. The W is the inertia acceleration constant: c1 = 2.5, c2 = 0.5, inertia weight w =
weight factor that affects the dynamic fly of the particles. 0.6.
C1 and C2 are the acceleration constant. Figure 4 (a & b) shows the result obtained by Genetic
Algorithm and Particle Swarm Optimization. For the GA,
Position Updation solution converges at 140 generations, total transportation
In next step, the position of particle will be updated on the cost is 0.81 million Rupees and for the PSO, the solution
basis of following equation (b) converges at 135 generations, total transportation cost is
0.79 million Rupees.
(b)

where, xt+1 is new position of particle and xt is the initial

position and the new updated velocity of particle is Vt+1.

Figure 4 (a) Results Figure 4 (a) Results

obtained (GA) obtained (PSO)
 semicontinuous food industries,” Omega, vol. 40, no. 5, pp.
634–650, 2012.
[13] V. Jayaraman and A. Ross, “A simulated annealing
Table 1 represents the three case instances of Agro-food methodology to distribution network design and management,”
supply chain network design Eur. J. Oper. Res., vol. 144, no. 3, pp. 629–645, 2003.
[14] M. Bachlaus, M. K. Pandey, C. Mahajan, R. Shankar, and M.
K. Tiwari, “Designing an integrated multi-echelon agile supply
Case f, h, v GA PSO chain network: A hybrid taguchi-particle swarm optimization
TTC* TTC* approach,” J. Intell. Manuf., vol. 19, no. 6, pp. 747–761, 2008.
(Rs.) (Rs.) [15] J. K. Lenstra, “Complexity of Vehicle Routing and Scheduling
Instance 1 30, 5, 5 4,61,004 4,08,473 Problems,” vol. 11, pp. 221–227, 1981.
[16] D. E. Goldberg and K. Deb, “Analysis of Selection A
Comparative Used in Genetic Algorithms Schemes.”
Instance 2 50, 10, 10 8,12,005 7,98,520 [17] S. Pratap, M. K. B, N. Cheikhrouhou, and M. K. Tiwari, “The
Robust Quay Crane Allocation for a Discrete Bulk Material
Instance 3 60, 12, 10 14,08,68 13,87,97 Handling Port,” pp. 1174–1178, 2015.
7 5

TTC*: Total Transportation Cost; GA: Genetic Algorithm; PSO:

Particle Swarm Optimization

VI. CONCLUSIONS
The mathematical model developed for the sustainable
supply chain network of the Indian food supply chain
is expected to revamp the traditional process and
letting the farmers receive a better margin of profits
than the former traditional ways. This model has been
developed to attain maximum efficiency in terms of
agro-food products flow and rationalizing the amount
the farmers will be receiving. The further development
or enhancement of the model could include the
information supply chain and higher terms of
efficiency and transparency in the system could be
attained. This mathematical model if implemented by
enhancing the supply chain of the information flow
will definitely help in attaining the Government of
India’s visionary agenda of doubling the farmers’
income by 2022 by increasing the profits of the
farmers.
REFERENCES
[1] G. of I. Ministry of Finance, “Economic Survey 2017-18,” vol.
I, 2018.
[2] O. Ahumada and J. R. Villalobos, “Application of planning
models in the agri-food supply chain: A review,” Eur. J. Oper.
Res., vol. 196, no. 1, pp. 1–20, 2009.
[3] UDAY M. APTE & S. VISWANATHAN, “International
Journal of Applications: A Leading Journal of Supply Chain
Effective Cross Docking for Improving Distribution
Efficiencies,” no. December 2012, pp. 37–41, 2010.
[4] R. Patidar, S. Agrawal, and S. Pratap, “Development of novel
strategies for designing sustainable Indian agri- fresh food
supply chain,” Sdhan, vol. 0123456789, 2018.
[5] C. & C. U. S. Jenny Gustavsson, Global food losses and food
waste - Extent, causes and prevention. 2011.
[6] R. L. Cook, “The U.S Fresh Produce Industry: An Industry in
Transition.,” pp. 5–30.
[7] R. Sturm and C. R. Gresenz, “National Survey,” vol. 324, no.
January, pp. 1–5, 2002.
[8] N. E. with funding from U. A. L. F. P. P. F. 2017 P. G. Karen
Karp & Partners For Farm To Institution, “EXPLORING
OPTIONS TO ENHANCE FOOD HUB COLLABORATION
&,” 2017.
[9] G. Zhang, W. Habenicht, W. Ernst, and L. Spieß, “Improving
the structure of deep frozen and chilled food chain with tabu
search procedure,” vol. 60, pp. 67–79, 2003.
[10] M. T. Lucas and D. Chhajed, “Applications of location
analysis in agriculture: A survey,” J. Oper. Res. Soc., vol. 55,
no. 6, pp. 561–578, 2004.
[11] J. Blackburn and G. Scudder, “Supply chain strategies for
perishable products: The case of fresh produce,” Prod. Oper.
Manag., vol. 18, no. 2, pp. 129–137, 2009.
[12] G. M. Kopanos, L. Puigjaner, and M. C. Georgiadis,
“Simultaneous production and logistics operations planning in

View publication stats