Decision Support Systems

1
Decision Support Systems

• Jerry Banks

2
Decision Support Systems

• Supply chain management problems are not well
  defined enough to trust them to the computer
• Humans need to use their intuition and wisdom
• The first S in DSS indicates that they only
  support the decision maker

3
Decision Support Systems

• DSS range from spreadsheets to expert systems
• Spreadsheets
• Users perform their own analysis
• Expert systems
• Capture knowledge in various areas and suggest
  alternatives

4
Decision Support Systems

• Three major components of a DSS
• Input data
• Analytical tools
• Visualization techniques

5
Input data for logistics network design

• Manufacturer data
• Location
• Production capacity
• Production cost
• Transportation costs to warehouses

6
Input data for logistics network design

• Warehouse data
• Location
• Fixed costs
• Variable costs
• Inventory turnover
• Transportation costs (to retailer)

7
Input data for logistics network design

• Retailer
• Location
• Annual demand (by product)
• Product
• Volume
• Weight
• Holding cost

8
Input data for logistics network design

• Consider
• The type of data in the last 3 slides may not be
  readily available in the company database
• Even if accessible, they may not be in the right
  format
• Collecting, tabulating, and verifying data can
  take lots of time

9
Analytical tools

• Queries
• Vast quantities of data make manual analysis
  difficult
• How many clients do we service in Florida?
• How many customers in Mexico purchased more than
  10,000?

10
Analytical tools

• Statistical analysis
• Trends and patterns
• What was the average inventory in the Memphis
  warehouse?
• What is the average number of stops on the
  Northeast Georgia route?
• What is the variance in the demand for customers
  in California?

11
Analytical tools

• Data mining
• Search for hidden patterns, trends, and
  relationships
• Men purchase diapers and beer on Friday afternoon

12
Analytical tools

• Online analytical processing (OLAP) tools
• More sophisticated than spreadsheets, easier to
  use than database models
• Let users navigate through the hierarchies and
  dimensions of business data by drilling down, up,
  or across levels
• Sophisticated tools to analyze the data
• Tools to present results

13
Analytical tools

• Simulation
• Imitating reality in the computer
• Capturing randomness
• After the model has been verified and validated
• Simulation output can be used to analyze the real
  system
• What is the average time in the production
  system?
• What fraction of production lots were late?

14
Analytical tools

• Artificial intelligence
• Rule based systems
• Troubleshooting complex problems
• Intelligent agents
• Planning and executing elements in the supply
  chain
• Example An agent that communicates with other
  agents at warehouses to locate missing items

15
Analytical tools

• Operations research
• Exact algorithms
• The best possible solutionoptimal
• Can take a long time to execute
• Optimality is not always neededapproximations
  suffice since input data is often approximated or
  aggregated

16
Analytical tools

• Operations research
• Heuristics
• Good, but not necessarily optimal, solutions
• Run much faster than exact algorithms
• Good overview
• Optimizing your supply chain A model approach,
  Inbound Logistics, April, 2003

17
10 rules of logistics optimization

• Suggestions by Don Ratliff
• Founder of CAPS and Velant
• Strategic level decisions
• Facility location
• Execution level decisions
• Loads, routes, schedules for trucks that deliver
  products

18
10 rules of logistics optimization

• Objectives must be quantifiable and measurable
• If you cant (dont) measure it, how do you know
  when it is accomplished?

19
10 rules of logistics optimization

• Models must faithfully represent the actual
  logistic processes
• Weight and volume of products may be what is
  needed, not just weight

20
10 rules of logistics optimization

• Data must be accurate, timely, and comprehensive
• Theres a tendency to use old data even after it
  has run its course

21
10 rules of logistics optimization

• Integration must supply fully automated data
  transfer
• Manual data hampers accuracy and timeliness

22
10 rules of logistics optimization

• Optimized plans must be delivered in a form that
  facilitates execution, management and control
• Get the solutions to the people that will use
  them in a practical way

23
10 rules of logistics optimization

• Algorithms must intelligently exploit individual
  problem structure
• Say that there are 40 shipments on a truck
• There are 40! ways the deliveries can be made
• Its not possible to calculate all of the
  possibilities
• Algorithms take advantage of the special
  structure to reduce the complexity to a
  manageable size

24
10 rules of logistics optimization

• Computing platforms must have sufficient power to
  produce optimum plans in the time required
• Too much data, too little time

25
10 rules of logistics optimization

• People responsible for the technology must have
  the domain and technology expertise required to
  support the models, data, and optimization
  engines
• Optimization engines are very complex so you need
  a rocket scientist

26
10 rules of logistics optimization

• Business processes must support optimization and
  have the ability to continuously improve

27
10 rules of logistics optimization

• Return on investment must be provable,
  considering the total cost of technology, people,
  and operations

28
Applications and tools
29
Presentation tools

• The ordinary ones
• Reports, charts, spreadsheets, CAD layouts
• Animations
• Simulation of a material handling system
• Simulation of a distribution center
• Geographical information systems (GIS)

30
GIS

• Capabilities include
• Mapping
• Database
• Query
• Geographic data manipulation
• Spatial data analysis
• Geocoding
• Others

31
GIS

• Advantage
• Combines all of the capabilities mentioned above
• Also, distance calculations can be made (USA)
• Some features are in Excel

32
GIS

• Applications
• Network analysis
• Transportation
• Telecom
• Site selection
• Routing
• SCM

33
www.streetdirectory.com
Street maps for several countries in the Far
East especially detailed for Singapore.
34
www.mapquest.com
Provides driving distances and maps for any
location in the USA.
35
Map-based logistics support for mobile commerce
Janny Leung, City University of Hong Kong
36
www.caliper.com
Solution to the traveling salesman problem.
37
www.caliper.com
Solution to the vehicle routing problem.
38
www.caliper.com
Distribution analysis warehouse to retail.
39
www.caliper.com
Site location analysis.
40
The problem with supply chain software, Sodhi,
ORMS Today, 6/03

• One study says that the major SCM vendors are in
  financial trouble
• lt20 believe that their software has shown a
  favorable ROI
• The software is over-hyped
• The managers at the application level dont know
  how to use the software

41
The problem with supply chain software, Sodhi,
ORMS Today, 6/03

• A second study is more positive
• 75 of users feel that their vendor either met or
  exceeded expectations
• Only 24 said the integrator met expectations

42
The problem with supply chain software, Sodhi,
ORMS Today, 6/03

• Possible causes of problems with SCM software
• Many SCM companies are internet companies only
• Customers overestimate vendors technological
  capability to deliver

43
The problem with supply chain software, Sodhi,
ORMS Today, 6/03

• Possible causes of problems with SCM software
• Customer and vendor fail to understand
  integration other than hook one database to
  another
• Customers lack understanding of the software
• Training is inadequate

44
2003 Software Survey SCM, Askoy/Debrez, ORMS
Today, 6/03

• Using the web, they identified 160 software
  companies that provide SCM support
• Eliminated 2/3rds of the companies that supply
  execution support only
• Of the 52 remaining, 30 responded to the survey

45
2003 Software Survey SCM, Askoy/Debrez, ORMS
Today, 6/03

• For each of the 30 companies they show
• Year product introduced
• Platforms supported
• Recommended hardware
• Price information
• Industry targeted and company size
• ERP software specifications
• Methodology used
• Number of firms using the software
• Significant installations

46
2003 Software Survey SCM, Askoy/Debrez, ORMS
Today, 6/03

• Less than ½ the SCM software purchased two years
  ago is up and running
• Problems include
• Problem complexity
• Inadequate management and technical support, both
  internal and external
• Installation lead times and flaws
• Poor project management
• Unrealistic customer expectations

47
End