Professional Documents
Culture Documents
Facility Location Selection: IDS 355 SUMMER 2020
Facility Location Selection: IDS 355 SUMMER 2020
Selection
IDS 355
SUMMER 2020
Learning Objectives
Why Locations?
Facility Selection Criteria
Facility Location Planning
A tool for Determining Location
Strategic Considerations
Strategic Location Concerns
Distance Modeling: Euclidean and Metropolitan Models
Geographic Information Systems
Locations
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Facility Selection Criteria
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Facility Location Planning
10-5
Regression Model for Motel Location
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Strategic Considerations
Competitive Clustering (Among Competitors)
(e.g. Auto Dealers, Motels)
Saturation Marketing (Same Firm)
(e.g. An Bon Pain, Ice Cream Vendors)
Marketing Intermediaries
(e.g. Credit Cards, HMO)
Substitute Communication for Travel
(e.g. telecommuting, e-Commerce)
Separation of Front from Back Office
(e.g. ATM, shoe repair)
Impact of the Internet on Service Location
(e.g. Amazon.com, eBay, FedEx)
10-8
Strategic Location Concerns
Access to customers
Complementary operations
Separation of facilities
Saturation marketing
Area coverage
Competitive clustering
Expansion strategies
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Access to Customers
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Complementary Operations
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Separation of Facilities
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Saturation Marketing
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Area Coverage
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Competitive Clustering
In competitive clustering, multiple competing stores are located near each other
and compete for the same customers
Motivating causes:
High customer traffic (smaller slice of much bigger pie)
Aggressive competitor location strategy
Prime location justifies high-risk, high-reward strategy
Access to suppliers or other business partners provides economies of scale to all
operations in the cluster
Sometimes, competitive clustering is sustainable and even creates a competitive
advantage for the businesses involved
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Expansion Strategies
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Distance Modeling
Many service operations need to be close to their customers
Delivery services
In-store purchases
Especially true for services handling many small orders
Distance raises operating costs or the real price to customers
Travel time
Fuel
Vehicle miles
Can these costs be measured?
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Distance Modeling (cont’d)
What should be measured is the actual travel cost:
For delivery services: possible but usually approximated:
Variable fuel and vehicle costs
Traffic conditions can change labor costs
Impractical to measure directly for customer visits:
Different travel modes: walking, bicycling, public transportation, driving
Time has a different value to different customers at different times
Instead, distance is often used to measure travel costs even though:
Distance models only approximate travel behavior
The direct route may not be the fastest or even available
Combining multiple stops in one trip changes cost functions
When selecting from a set of potential locations, the best one has the smallest sum of distances
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Euclidean (Straight-Line) Distance Model
Straight-line distance is accurate if travel actually is direct for customers or
delivery. Otherwise, it can be an acceptable approximation if:
Distances are long compared to any road constraints
Demand is aggregated over large zones
Formula:
Customer located at coordinates (xc, yc)
Service operation located at coordinates (xs, ys)
Distance =
Weighting may be applied as a multiplier to model multiple customers
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Example: Euclidean Distance
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Metropolitan Distance Model
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Example: Metropolitan Distance Model
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Geographic Information Systems
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
GIS Data
Annual Grocery
Spending per
# of Customers Household
Zone
A 12,000 $8,000
B 15,000 $6,000
C 8,000 $14,000
D 10,000 $11,000
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Huff Model
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Attraction
Each store j has an attraction A to zone i calculated as:
Where:
is the attractive force from store j to customers in zone i
is the size of store j
is the average travel time between store j and customers in zone i
indicates customer propensity to travel
Low values increase attraction, high values decrease it
In our examples, we’ll assume = 2
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Example: Attraction
Attraction is used to calculate the probability that a given customer will visit a
given store
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Visit Probability
The probability Pi, j that a random customer from zone i will visit store j is
determined by:
The attraction of j to zone i
Relative to the other stores attracting customers from zone i
Formula: , where:
Aij
is the attraction of store j to zone i Pij = n
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Example: Visit Probability
What is the probability that a random customer from zone A will visit store Z?
Using Excel, calculate visit probabilities from each store to each zone
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Store Sales
For any {zone, store} pair, store sales are calculated as a function of:
Number of customers
Spending per customer
Visit probability for that {zone, store} pair
Formula: , where:
= the expected sales at store j from customers in zone i
= the number of customers in zone i
= the relevant product budget of customers in zone i
= the visit probability between zone i and store j
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Example: Store Sales
Aggregate sales are found by adding the expected sales over all zones included in
the model
Comparing aggregate sales allows calculating market share over an entire set of
zones
Discuss: How to find market share if all customers are from one zone?
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Market Share
Market share Mj is the fraction of sales obtained by store j over all zones included
in the model
Formula: , where:
= expected sales at store j from all zones
= expected sales at all stores
Shortcut: Expected sales at all stores is the sum of all relevant spending over all
customers
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop
Example: Market Share
Although we have considered an existing set of stores, the Huff model offers a
framework for:
Estimating sales and market share of potential new locations
Estimating the impact of new market entrants
Extrapolating demographic trends to estimate future sales
5
201
st,
dqui
Lun
D.
and
sel
Ouk
A.
by
hted
yrig
Cop