Data Collection and Analysis

1
Data Collection and Analysis
2
Introduction

• Data collection is the most challenging and
  time-consuming task in doing a simulation
• For new systems information is usually very
  sketchy and only rough estimates
• For existing systems information is seldom
  available in a form that is directly useable in
  building a model

3
Guidelines for Data Gathering

• Data gathering should not be performed
  haphazardly with the idea that you will sort
  through all the information once it is collected
• Data should be collected systematically seeking
  specific items of information that are needed for
  building the model
• Approach data collection with the end in mind

4
Guidelines for Data Gathering

• Identify triggering events
• Identify the causes or conditions that trigger
  the activities
• Example Gathering downtime data
• Useful to distinguish between downtimes that are
  due to failure, planned or scheduled maintenance
  or out of stock situations
• Questions to ask
• What triggers the movement of entities
• What triggers the use of a particular resource

5
Guidelines for Data Gathering

• Focus only on key impact factors
• Example Off shift activities
• Off shift activities such as maintenance should
  not be included in the model since it does not
  impact the model
• Ignore extremely rare downtimes, negligible move
  times, on-the-fly inspections, external set-ups,
  ect

6
Guidelines for Data Gathering

• Isolate actual activity times
• Example
• Do not include waiting time with activity time
• Look for common groupings
• When dealing with lots of of variety in a
  situation such as hundreds of part types of\r
  customer profiles it helps to look for common
  groupings or patterns
• Identify general categories to which all data can
  be assigned
• Calculate the percentage of data that fall in
  each category

7
Guidelines for Data Gathering

• Focus on the essence rather than substance
• A system definition for modeling purposes should
  capture the cause-and-effect relationships and
  ignore the meaningless details (called system
  abstraction)
• A system should be abstracted to the highest
  level possible while still preserving the essence
  of the system operation

8
Guidelines for Data Gathering

• Black box approach not concerned about the
  specific activities being performed (such as
  milling or inspection) but rather the impact that
  these activities have on resource utilization and
  entities flow

9
Guidelines for Data Gathering

• Separate input variables from response variables
• Input variables define how the system works
• Response variables describe how the system
  responds to a given set of input variables

10
Steps for Data Gathering

• Determine data requirements
• Identify data sources
• Collect data
• Make assumptions where necessary
• Analyze data
• Document and approve the data

11
Determine Data Requirements

• Structural Data Includes all the objects of the
  system to be modeled (Entities, Resources,
  Locations)
• Operational Data explain how the system
  operates (when, where and how events and
  activities take place)
• Logical or behavioral information about the
  system (routings, schedules, downtime, and
  resource allocation)
• Numerical Data quantitative information about
  the system (capacities, arrival rates, activity
  times, time between failures, etc)

12
Determine Data Requirements

• Use of a questionnaire
• To help focus data gathering efforts on the right
  information
• Ensure that meetings with others on whom you
  depend for model information are productive
• People respond better to specific questions
• Reduces the length of the interview

13
Identifying Data Sources

• Sources of data include
• Reports
• Personal interviews operators (work methods),
  maintenance personnel (repairs procedures),
  engineers (routings), managers (schedules and
  forecasts)
• Personal observations facility walk-through,
  time studies, work sampling
• Assumptions
• Historical records Production, sales, scrap
  rates, equipment reliability
• System documentation process plans, facility
  layouts, work procedures

14
Identifying Data Sources

• Sources of data include
• Comparisons with similar systems within the
  company, within the same industry, within any
  industry
• Vendor claims process times, reliability of new
  machines
• Design estimates process times, move times, and
  other for new system
• Research literature published research on
  learning curves, predetermined time studies
• If data if very difficult to access, assumptions
  must be made

15
Collecting the Data

• Go from general to specific
• Data collection continues through to the end of
  the project as objectives change and information
  that was unavailable at the beginning of the
  project now begins to materialize
• Data should be defined in the following sequence
• Define the overall entity flow
• Develop a description of operation
• Define incidental details and firm up data values
• Data will not come in this order
• The focus of attention should follow this order

16
Developing a Description of the System

• Develop a scripted flow chart
• Identify for each type of entity at each location
  in the system
• Time and resource requirements of the activity
• Where, when and what quantities entities get
  routed next
• Time and resource requirements for moving to the
  next location

17
Process Description Example
Patients enter Doctor Browns office and sign in
at the check in counter before taking a seat to
wait for the nurses call. Patients are summoned
to one of three examination rooms when one
becomes available. The nurse escorts patients to
the examination room, where they wait for Dr.
Brown to see them. After treatment, patients
return on their own to the checkout counter where
they make payment and possibly reschedule another
visit. This process is shown in the entity flow
diagram that follows.
18
Process Description ExampleEntity Flow Diagram
Check-in Counter
Waiting Room
Exam Room (3)
Check-out Counter
Patient
19
Process Description Example
20
Making Assumptions

• Assumptions are a part of every simulation model
• Never let the confidence you have in the results
  exceed the confidence you have in the assumptions
  in the model
• When comparing 2 models, we will exam relative
  performance rather than absolute performance
  thus as long as the assumptions are the same in
  both models, the comparison is valid
• All assumptions are not equal in impact on
  results
• A system with large processing times compared to
  move times with a move estimate off by 20 is not
  a problem
• A processing time with a 20 error could make a
  20 difference in throughput

21
Airplane ExampleEntity Flow Diagram
Station 1 Queue
Station 1
Station 2 Queue
Station 2
parts
Station 3 Queue
Station 3
Station 4 Queue
Station 4
Station 5 Queue
Station 5
Station 6 Queue
Station 6
22
Airplane Example

• We are assigned the task of improving the process
• What data will we need?
• How will we collect the data?
• Reports
• Personal interviews operators (work methods),
  maintenance personnel (repairs procedures),
  engineers (routings), managers (schedules and
  forecasts)
• Personal observations facility walk-through,
  time studies, work sampling
• Assumptions
• Historical records Production, sales, scrap
  rates, equipment reliability
• System documentation process plans, facility
  layouts, work procedures

23
Airplane Example

• When will we collect the data?
• How much data will we need?
• Why will we be required to have several data
  collection sessions?
• What types of information will be need?
• Structural Data, Operational Data, Numerical Data
  
• What should our approach be to collect the data?
• Start with the overall system at a high level and
  then work down to a smaller level of detail

24
Airplane Example Data Collection Brainstorm

• Waiting time
• Process time
• Arrival pattern
• batch size
• time B.A.
• Queue length
• Avg. cycle time
• Number of Resources
• People
• Equipment
• Product variety
• Distribution
• Rejection rate
• Resource utilization
• Shift schedule

• Total time to study system
• Failure distribution
• Travel time between stations
• Production rate/Throughput
• Reports
• - set-up time
• Failure duration distribution
• Failure procedure
• Quality checks
• Resource availability
• WIP
• Layout
• - travel distance
• Resource capacity
• Average time in queue
• Raw materials

• Inspection of raw materials
• Shortages
• Cross-training
• Level of detail
• Type of queue
• Bottleneck Occurrence