Simulation of EndofLife Computer Recovery Operations

1
Simulation of End-of-Life Computer Recovery
Operations

• Design Team
• Jordan Akselrad, John Marshall
• Mikayla Shorrock, Nestor Velilla
• Nicolas Yunis

Project Advisor Prof. James Benneyan
Project Sponsor Prof. Sagar Kamarthi
2
Background Information

• Research project by sponsor, Professor Kamarthi
• Sensors are being developed for computer
  components
• Sensor Embedded Computers (SEC)
• Sensors closely estimate remaining useful life
• Product Recovery Facilities (PRF) exist that
  refurbish computers
• Ongoing research to determine sensors impact on
  entire reclamation process

B A C K G R O U N D
3
Project Scope

• Determine the effect expected component life
  information has on a Product Recovery Facility

S C O P E
4
Project Goal

• Develop a simulation tool which models Product
  Recovery Facilities
• Comparative model analysis
• Apply optimization techniques across simulation
  model
• Determine if sensors improve the cost
  effectiveness of computer recovery operations

S C O P E
5
Refurbishing Process
S C O P E
6
Design Concepts Considered

• ARENA
• Complex logic needs to be implemented
• Excel Interface
• Amount of data is overwhelming to user
• Event Based Simulation
• Unnecessary due to lack of queuing

S I M U L A T I O N
7
Simulation Design

• Custom user interface
• C / .Net backend
• Serves as window into simulation
• Assists in debugging model
• Rapid development, run anywhere
• Human Factors Considerations
• Simple Interface with powerful capabilities
• Easy to run large scale experiments
• Data easily importable / exportable
• Built in graphing for real-time analysis

S I M U L A T I O N
8
Price Generation

• Arbitrary computer configurations
• Each price contributor given a weight to
  influence score
• Weights solved to maximize price vs. score
  correlation
• Generated equation used to price dynamically

S I M U L A T I O N
9
Simulation Demo
S I M U L A T I O N
10
Sensor Times Benefit

Minutes per Component
Sensors
No Sensors
A N A L Y S I S
11
Profit Contributors
A N A L Y S I S
12
Design of Experiments

• 2 level, 10 Factor Experiment
• 1024 Combinations, 15 Runs each
• Output for 3 performance objectives
• Profit, Waste, Reliability
• Minitab used for analysis
• Variable interactions examined
• Approximation equations developed
• Efficient set extracted

A N A L Y S I S
13
Interaction of Profit Factors

Purchasing Costs
Purchasing costs have the greatest effect on
profit
A N A L Y S I S
14
Reliability Analysis

Without sensors 23 failure rate
Failure rate increasing with sensor error
A N A L Y S I S
15
Estimating Life Without Sensors

• Dispose if probability component working in one
  year is less than tolerance

• Optimal tolerance 54

Profit vs Tolerance
O P T I M I Z A T I O N
16
Estimating Life With Sensors

• Expected life reported with mean at failure date
• Sensor error is in months of deviation from mean,
  default 6
• Sensor reading is corrected to prevent warranty
  failures

Profit vs Sensor Correction
Optimal profit at 1 deviation of correction
Percent Profit
Correction of Sensor
O P T I M I Z A T I O N
17
Multi-Criteria Optimization

• Maximize Profit
• Minimize Waste
• Maximize Reliability

• Surface is the efficient solution front
• Efficient implies non-dominated trade-off between
  values

O P T I M I Z A T I O N
18
Conclusions

• Fully developed simulation tool
• Easy to use
• Exceeds research needs
• Preliminary Analysis Performed
• Without sensors refurbishment is infeasible
• 23 failure rate
• With sensors
• 21 reduction in time spent per component
• 22 reduction in processing cost per component
• Sensors strongly recommended
• Overall profit increase 48
• Customer failure rate 3

C O N C L U S I O N S
19
Future Considerations

• Improve MTBF data accuracy
• Research shows MTBF specified by
• manufacturer is unreliable
• Ideas to enhance accuracy
• Facilities record component failure rates
• Sensors report failure time to manufacturer
• Integration into facility
• Simulator used as a prediction engine

C O N C L U S I O N S
20
Questions
Thank you
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
Waste Analysis

• Sensor Error vs Working Disposals

Working components disposed increases with
sensor error
Without sensors 11 of disposed components are
working
A N A L Y S I S
29
Sensor Cost Benefit

Dollars per Component
No Sensors
Sensors
A N A L Y S I S