Supply Chain Analytics

BITS Pilani Dr Satyendra Kumar Sharma

Pilani | Dubai | Goa | Hyderabad

1
 BITS Pilani
Pilani | Dubai | Goa | Hyderabad

Network Flow Models

2
 Learning Objectives
After completing this chapter, students will be able to:

1. Connect all points of a network while minimizing

total distance using the minimal-spanning tree
technique
2. Determine the maximum flow through a network
using the maximal-flow technique
3. Find the shortest path through a network using the
shortest-route technique
4. Understand the important role of software in solving
network problems

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Figure 8.3 Elements of
manufacturing strategy formation
Source: Adapted from Service
Product
Mark by Andersen
Competitiveness
Consulting © 1989 Andersen
Consulting STRATEGIC
Supply Supply
Chain Chain
Design Strategy
STRUCTURAL

Plant
New Design
Product and Quality Supplier Logistics
Development Operations Management Management

FUNCTIONAL
Policies Facilities Organization
Information and and Purchasing and Change
Trans-
Technology Procedures portation Equipment Management

IMPLEMENTATION
Elements of Manufacturing Formation
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Transportation network

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 LP Formulation Special Cases
• Total supply exceeds total demand
• Total demand exceeds total supply
• The objective is maximizing profit or revenue
• Minimum shipping guarantee from i to j
• Maximum route capacity from i to j
• Unacceptable route

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Transshipment problems
• Transshipment problems are transportation problems in
which a shipment may move through intermediate nodes
(transshipment nodes)before reaching a particular destination
node.
• Transshipment problems can be converted to larger
transportation problems and solved by a special
transportation program.
• Transshipment problems can also be solved by general
purpose linear programming codes.
• The network representation for a transshipment problem with
two sources, three intermediate nodes, and two destinations

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Transshipment Problem

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

LP Formulation

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 100 45
Plant Warehouse

CHI STL DET CIN LOU IND MIL MIN

22 14 18 17 15 13 15 20

Figure Alpha transportation network

 100 45
Plant Warehouse

CHI STL DET CIN LOU IND MIL MIN

22 14 18 17 15 13 15 20

Figure Extended Alpha transportation network

 Set-Covering Models
• In a set-covering model, each member of a
given set (set 1) must be “covered” by an
acceptable number of another set (set 2).
– The objective in a set-covering problem is to
minimize the number of members in set 2
necessary to cover all the members in set 1.
– Set-covering models have been applied to areas as
diverse as airline crew scheduling, truck
dispatching, political redistricting, and capital
investment.
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
• Objective: To develop a binary model to find the minimum number of hub
locations that can cover all cities.
• Solution: Western Airlines wants to design a hub system that connects
flights to and from cities within 1000 miles of each hub.
• The table below lists the cities that are within 1000 miles of other cities.

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Hub location model
• The big picture model is shown below.
– There is a binary variable for each city to indicate whether a
hub is located there.
– The number of hubs that cover each city is constrained to be
at least one

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Excel Model

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Introduction
• Large scale problems may require hundreds or thousands of
iterations making efficient computer programs a necessity
• All types of networks use a common terminology
• The points on a network are called nodes and may be
represented as circles of squares
• The lines connecting the nodes are called arcs

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Minimal-Spanning Tree Technique
• The minimal-spanning tree technique involves connecting all
the points of a network together while minimizing the
distance between them
• The Lauderdale Construction Company is developing a
housing project
• They want to determine the least expensive way to provide
water and power to each house
• There are eight houses in the project and the distance
between them is shown in Figure 1

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Minimal-Spanning Tree Technique

• Steps for the minimal-spanning tree technique

1. Select any node in the network
2. Connect this node to the nearest node that minimizes the
total distance
3. Considering all the nodes that are now connected, find
and connect the nearest node that is not connected. If
there is a tie, select one arbitrarily. A tie suggests there
may be more than one optimal solution.
4. Repeat the third step until all nodes are connected

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 What is Spanning Tree
A spanning tree T of an undirected graph G is a
subgraph that is a tree which includes all of the
vertices of G, with minimum possible number
of edges. In general, a graph may have several
spanning trees, but a graph that is not
connected will not contain a spanning tree (but
see Spanning forests below). If all of the edges
of G are also edges of a spanning tree T of G,
then G is a tree and is identical to T (that is, a
tree has a unique spanning tree and it is itself)

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Minimal-Spanning Tree Technique
• Network for Lauderdale Construction

3
2 5
4
3
5
3 7
7
1 2 2
3
3 8
5 1
2 6
6
4

Gulf
Figure .1

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Minimal-Spanning Tree Technique- Prim’s
Algorithm
• Start by arbitrarily selecting node 1
• The nearest node is node 3 at a distance of 2 (200 feet) and
we connect those nodes
• Considering nodes 1 and 3, we look for the next nearest node
• This is node 4, the closest to node 3
• We connect those nodes
• We now look for the nearest unconnected node to nodes 1, 3,
and 4
• This is either node 2 or node 6
• We pick node 2 and connect it to node 3

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Minimal-Spanning Tree Technique
• Following this same process we connect from node 2 to node
5
• We then connect node 3 to node 6
• Node 6 will connect to node 8
• The last connection to be made is node 8 to node 7
• The total distance is found by adding up the distances in the
arcs used in the spanning tree
2 + 2 + 3 + 3 + 3 + 1 + 2 = 16 (or 1,600 feet)

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Minimal-Spanning Tree Technique
• All iterations for Lauderdale Construction

3
2 5
4
3
5
3 7
7
1 2 2
3
3 8
5 1
2 6
6
4

Gulf
Figures .2 – 5

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Kruskal’s Alorithms
• Arrange all the weights in increasing order
with their connected arcs.
• Start with lowest and keep in mind no graph is
constructed.
• Stop the algorithms when n-1 arcs are used.

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Maximal-Flow Technique
• The maximal-flow technique allows us to determine the
maximum amount of a material that can flow through a
network
• Waukesha Wisconsin is in the process of developing a road
system for the downtown area
• They want to determine the maximum number of cars that
can flow through the town from west to east
• The numbers by the nodes indicate the number of cars that
can flow from the node

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Maximal-Flow Technique
• Four steps of the Maximal-Flow Technique
1. Pick any path from the start (source)
source to the finish
(sink)
sink with some flow. If no path with flow exists, then
the optimal solution has been found.
2. Find the arc on this path with the smallest flow
capacity available. Call this capacity C. This represents
the maximum additional capacity that can be
allocated to this route.

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Maximal-Flow Technique
• Four steps of the Maximal-Flow Technique
3. For each node on this path, decrease the flow
capacity in the direction of flow by the amount C. For
each node on the path, increase the flow capacity in
the reverse direction by the amount C.
4. Repeat these steps until an increase in flow is no
longer possible

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Maximum Flow Problem
• Given: Directed graph G=(V, E),
Supply (source) node O, demand (sink) node T
Capacity function u: E  R .
• Goal: Given the arc capacities,
send as much flow as possible
from supply node O to demand node T
through the network.
• Example: A
4

4
O B 6 D 5
5
4 4 T

C 5
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Capacitated Transshipment Problem

Max xk1 (k is sink node, 1 is source node)

s.t. xij - xji = 0 (conservation of flow)
i j

xij < cij (cij is capacity of ij arc)

xij > 0, for all i and j (non-negativity)

(xij represents the flow from node i to node j)

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Characteristics of a feasible
flow
• Let xij denote the flow through arc i  j .
• Capacity kij of arc i  j is the upper bound on the flow shipped
through arc i  j .
Thus, we have the following constraints:
0  xij  kij , for any arc i  j
• Every node i, except the source and the sink,
should satisfy the conservation-of-flow constraint, i.e.,
flow into node i = flow out of node i
In terms of xij the constraint is
 x ji   x ik , for any node i  O, T
arcs j i arcs i  k

• For any flow that satisfies the conservation-of-flow constraints,

flow out of the source = flow into the sink
This is the amount we want to maximize.
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Linear Program for the
Maximum Flow Problem
• Summarizing, we have the following linear program:

maximize x max
s.t. x
arcs j i
ji  xarcs i  k
ik , for any node i  O, T

x
arcs j T
Oj  x
arcs i  T
iT x max

0  xij  kij , for any arc i  j

This linear program can be solved by a Simplex method or Excel

Solver add-in.
 2
x12 x25
4 4
x13
1 3 x35
5 x56
5 6 6
x14 5
4 x34
4 x46
4 5

Objective function: maximize xmax

Capacity constraints:
x12 ≤ 4, x13 ≤ 5, x14 ≤ 4, x25 ≤ 4, x34 ≤ 4, x35 ≤ 6,
x45 ≤ 5, x56 ≤ 5
Conservation-of-flow constraint:
x12 = x25, x13 = x34+x35, x14+x34=x46, x24+x35=x56
Constraint for the sourse and sink node:
x12+x13+x14=x46+x56=xmax
Non-negativity constraints: xij≥0

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

• There is also a method, which can be applied
directly in the graph:
Augmenting Path Algorithm.

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Shortest-Route Technique
• The shortest-route technique finds how a person or item can
travel from one location to another while minimizing the total
distance traveled
• It finds the shortest route to a series of destinations
• Ray Design, Inc. transports beds, chairs, and other furniture
from the factory to the warehouse
• They would like to find the route with the shortest distance
• The road network is shown in Figure 3

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Shortest-Route Technique
• Roads from Ray’s plant to warehouse

200
2 4
Plant
0 10
10 0

10
50 0 150
1 6

20 0
0 10
40
3 5 Warehouse

Figure .3

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Shortest-Route Technique
• Steps of the shortest-route technique
1. Find the nearest node to the origin (plant). Put the
distance in a box by the node.
2. Find the next-nearest node to the origin and put the
distance in a box by the node. Several paths may have to
be checked to find the nearest node.
3. Repeat this process until you have gone through the
entire network. The last distance at the ending node will
be the distance of the shortest route.

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Shortest-Route Technique
• We can see that the nearest node to the plant is node 2
• We connect these two nodes
• After investigation, we find node 3 is the next nearest
node but there are two possible paths
• The shortest path is 1–2–3 with a distance of 150
• We repeat the process and find the next node is node 5 by
going through node 3
• The next nearest node is either 4 or 6 and 6 turns out to
be closer
• The shortest path is 1–2–3–5–6 with a distance of 290
miles

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Shortest-Route Technique
• First iteration for Ray Design
100
200
2 4
Plant
0 10
10 0

10
50 0 150
1 6

20 0
0 10
40
3 5 Warehouse

Figure .11

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Shortest-Route Technique
• Second iteration for Ray Design
100
200
2 4
Plant
0 10
10 0

10
50 0 150
1 6

20 0
0 10
40
3 5 Warehouse

150

Figure .12

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Shortest-Route Technique
• Third iteration for Ray Design
100
200
2 4
Plant
0 10
10 0

10
50 0 150
1 6

20 0
0 10
40
3 5 Warehouse

150 190

Figure .13

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Shortest-Route Technique
• Fourth and final iteration for Ray Design
100
200
2 4
Plant
0 10
10 0
290
10
50 0 150
1 6

20 0
0 10
40
3 5 Warehouse

150 190

Figure .14

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Solution Shortest route
Objective Function
Min 80x12 + 40x13 + 80x14 + 130x15 +
180x16 + 60x25
+ 100x26 + 30x34 + 90x35 + 120x36 +
30x43 + 50x45
+ 90x46 + 60x52 + 90x53 + 50x54 + 30x56
Node Flow-Conservation Constraints
x12 + x13 + x14 + x15 + x16 = 1 (origin)
– x12 + x25 + x26 – x52 = 0 (node 2)
– x13 + x34 + x35 + x36 – x43 – x53 = 0
(node 3)
– x14 – x34 + x43 + x45 + x46 – x54 = 0
(node 4)
– x15 – x25 – x35 – x45 + x52 + x53 + x54
+ x56 = 0 (node 5)
x16 + x26 + x36 + x46 + x56 = 1
(destination)

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Models for Facility Location and Capacity Allocation

• Maximize the overall profitability of the supply chain network while

providing customers with the appropriate responsiveness

• Many trade-offs during network design

• Network design models used to decide on locations and capacities and to

assign current demand to facilities

Dr S K Sharma, BITS Pilani 46

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Models for Facility Location and Capacity Allocation

• Important information
Location of supply sources and markets
Location of potential facility sites
Demand forecast by market
Facility, labor, and material costs by site
Transportation costs between each pair of sites
Inventory costs by site and as a function of quantity
Sale price of product in different regions
Taxes and tariffs
Desired response time and other service factors
Dr S K Sharma, BITS Pilani 47
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Phase II: Network Optimization Models

Dr S K Sharma, BITS Pilani 48

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Capacitated Plant Location Model

= number of potential plant locations/capacity

= number of markets or demand points = 1 if plant i is open, 0 otherwise
= annual demand from market j = quantity shipped from plant i
to market j
= potential capacity of plant i

= annualized fixed cost of keeping plant i open

= cost of producing and shipping one unit from plant i to market j (cost includes
production, inventory, transportation, and tariffs)

subject to

Dr S K Sharma, BITS Pilani 49

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Capacitated Plant Location Model

Dr S K Sharma, BITS Pilani 50

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Capacitated Plant Location Model

Dr S K Sharma, BITS Pilani 51

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Capacitated Plant Location Model

Dr S K Sharma, BITS Pilani 52

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Phase III: Gravity Location Models

Dr S K Sharma, BITS Pilani 53

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Gravity Location Model

xn, yn:Coordinate location of either a market or supply source n

Fn:Cost of shipping one unit for one mile between the facility and
either market or supply source n

Dn:Quantity to be shipped between facility and market or supply

source n

(x, y) is the location selected for the facility, the distance dn between the
facility at location (x, y) and the supply source or market n is given
by

Dr S K Sharma, BITS Pilani 54

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Gravity Location Model

Dr S K Sharma, BITS Pilani 55

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Gravity Location Model

Coordinates
Transportation Cost Quantity in Tons
Sources/Markets $/Ton Mile (Fn) (Dn) xn yn
Supply sources
Buffalo 0.90 500 700 1,200
Memphis 0.95 300 250 600
St. Louis 0.85 700 225 825
Markets
Atlanta 1.50 225 600 500
Boston 1.50 150 1,050 1,200
Jacksonville 1.50 250 800 300
Philadelphia 1.50 175 925 975
New York 1.50 300 1,000 1,080
k
Total transportation cost TC  d n Dn Fn
n 1

Dr S K Sharma, BITS Pilani 56

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Gravity Location Model

Dr S K Sharma, BITS Pilani 57

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Gravity Location Model

Dr S K Sharma, BITS Pilani 58

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Gravity Location Model

• For each supply source or market n, evaluate dn

• Obtain a new location (x’, y’) for the facility, where

• If the new location (x’, y’) is almost the same as

(x, y) stop. Otherwise, set (x, y) = (x’, y’) and go to step 1

Dr S K Sharma, BITS Pilani 59

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Phase IV: Network Optimization Models

Demand City
Production and Transportation Cost
per Thousand Units (Thousand $)
Monthly Monthly
Capacity Fixed Cost
(Thousand (Thousand $)
Supply City Atlanta Boston Chicago Denver Omaha Portland Units) K f
Baltimore 1,675 400 985 1,630 1,160 2,800 18 7,650
Cheyenne 1,460 1,940 970 100 495 1,200 24 3,500
Salt Lake 1,925 2,400 1,450 500 950 800 27 5,000
City
Memphis 380 1,355 543 1,045 665 2,321 22 4,100
Wichita 922 1,646 700 508 311 1,797 31 2,200
Monthly 10 8 14 6 7 11
demand
(thousand
units) Dj

Dr S K Sharma, BITS Pilani 60

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Network Optimization Models

• Allocating demand to production facilities

= Number of factory locations
= Number of markets or demand points xij = Quantity shipped from factory i to market
= Annual demand from market j j
= Capacity of factory i
= Cost of producing and shipping one unit from factory i to market j

n m
Min cij xij
i 1 j 1 subject to
n

x
i 1
ij  D j for j 1,..., m
m

x
j 1
ij  K i for i 1,..., n

Dr S K Sharma, BITS Pilani 61

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Network Optimization Models

• Optimal demand allocation

Atlanta Boston Chicago Denver Omaha Portland

TelecomOne Baltimore 0 8 2

Memphis 10 0 12

Wichita 0 0 0

HighOptic Salt Lake 0 0 11

Cheyenne 6 7 0

Dr S K Sharma, BITS Pilani 62

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Capacitated Plant Location Model

• Merge the companies

• Solve using location-specific costs
yi = 1 if factory i is open, 0 otherwise
xij = quantity shipped from factory i to market j

Dr S K Sharma, BITS Pilani 63

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Capacitated Plant Location Model

Dr S K Sharma, BITS Pilani 64

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Capacitated Plant Location Model

Dr S K Sharma, BITS Pilani 65

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Capacitated Plant Location Model

Dr S K Sharma, BITS Pilani 66

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Capacitated Plant Location Model

Dr S K Sharma, BITS Pilani 67

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Capacitated Model with Single Sourcing

• Market supplied by only one factory

• Modify decision variables
yi = 1 if factory i is open, 0 otherwise
xij = 1 if market j is supplied by factory i, 0 otherwise

subject to

Dr S K Sharma, BITS Pilani 68

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Capacitated Model with Single Sourcing

Dr S K Sharma, BITS Pilani 69

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Capacitated Model with Single Sourcing

• Optimal network configuration with single sourcing

Open/
Closed Atlanta Boston Chicago Denver Omaha Portland
Baltimore Closed 0 0 0 0 0 0

Cheyenne Closed 0 0 0 0 0 0

Salt Lake Open 0 0 0 6 0 11

Memphis Open 10 8 0 0 0 0

Wichita Open 0 0 14 0 7 0

Dr S K Sharma, BITS Pilani 70

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Locating Plants and Warehouses Simultaneously

Dr S K Sharma, BITS Pilani 71

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Locating Plants and Warehouses Simultaneously

• Model inputs
m = Number of markets or demand points
n = Number of potential factory locations
l = Number of suppliers
t = Number of potential warehouse locations
Dj = Annual demand from customer j
Ki = Potential capacity of factory at site i
Sh = Supply capacity at supplier h
We = Potential warehouse capacity at site e
Fi = Fixed cost of locating a plant at site i
fe = Fixed cost of locating a warehouse at site e
chi = Cost of shipping one unit from supply source h to factory i
cie = Cost of producing and shipping one unit from factory i to
warehouse e
cej = Cost of shipping oneDrunit from
S K Sharma, BITSwarehouse
Pilani e to customer j 72
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Locating Plants and Warehouses Simultaneously

• Goal is to identify plant and warehouse locations and quantities shipped

that minimize the total fixed and variable costs
Yi = 1 if factory is located at site i, 0 otherwise
Ye = 1 if warehouse is located at site e, 0 otherwise
xej = Quantity shipped from warehouse e to market j
xie = Quantity shipped from factory at site i to warehouse e
xhi = Quantity shipped from supplier h to factory at site i

Dr S K Sharma, BITS Pilani 73

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
Locating Plants and Warehouses Simultaneously

subject to

Dr S K Sharma, BITS Pilani 74

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
 Accounting for Taxes, Tariffs, and Customer
Requirements

• A supply chain network should maximize profits after tariffs and taxes while meeting customer service
requirements

• Modified objective and constraint

Dr S K Sharma, BITS Pilani 75

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956
THANK YOU

76
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956