@simul8 virtual user group, september: brian harrington, less is more

19
Successful Simulations using Less is More By: Brian Harrington 9/17/2013 MTN-SIM, LLC 1

Upload: simul8-corporation

Post on 21-May-2015

1.096 views

Category:

Technology


3 download

DESCRIPTION

@SIMUL8 Virtual User Group We know not everyone can take time out to attend conferences and user meetings, so we're making it easy for you to get involved with our series of virtual user groups. Learning Zone: Brian Harrington will call on his experience as a Six Sigma Black Belt Black Belt to share his thoughts on "A Six Sigma approach to building successful simulations.

TRANSCRIPT

Page 1: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Successful Simulations using Less is More

By: Brian Harrington

9/17/2013

MTN-SIM, LLC

1

Page 2: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Chaos of Change

We need to batch build the Bodyside

Outers.

The new laser welder looks like it

will be over 58 seconds.

How many carriers are on the FS

delivery system?

We will be running this 2-10 hr shifts.

How many units between Seg1 and

Seg2?

The Main Framing

conveyor runs at 45fpm.

There are 3 repair resources available.

We have 3 different Pallet types. What

is the build schedule?

Last week… we had a 3 hr breakdown.

Looks like we are 2 weeks behind!

3 seconds for the lift table to go up,

then another 3 seconds to drop.

2

Page 3: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Complex Systems and Shorten Development Time

This is a good equation for stress!

We’ll simulate it!

This just pushed the stress on the simulation team!

3

Page 4: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Less is More using 6-Sigma

DMAIC or DMADV steps: Define, Measure, Analyze, Improve, Control

Define, Measure, Analyze, Design, Verify

DES Steps: Objective, Assumptions, Data Collection, Build Model, Verify, Validate, Experimentation, Results

Very Similar steps!

4

Page 5: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Y=f(x’s) Transfer Function

6-Sigma focuses on Key Input Factors (x’s) to deliver your Response.

All of the x’s can be measured & controlled to increase accuracy & precision of hitting your Target (Y).

System/Process

Trivial Many (N’s)

Vital Few (X’s)

Inputs (N’s & X’s) Output (Y)

5

Page 6: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

The Art & Science of DES

Experience modelers tend to create a model with the least amount of objects (data) to meet the objective. A common modeling error is to add too much detail to the model; attempting to simulate every move or event within the system. The two key questions become: 1. What is the significant input? 2. How do we control it?

6

Page 7: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

The P-Diagram

The P-Diagram not only helps engineers to define the Key Parameters for a robust design, but also acts as an excellent communication tool for team reviews.

7

Page 8: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Basic Building Blocks

The 6 Basic Building Blocks: Start Point, Queue, Activity, Conveyor, Resource, and End Point.

8

Page 9: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

6 is all you Need

1. Work Item Types: Can represent parts, carriers, signals, phone calls, just about anything that requires a “Label Profile”.

2. Activities: Work Centers, machines, tasks, process steps, anything that requires a “Cycle Time”.

3. Storage Areas: Buffers, de-couplers, banks, magazines, anything that requires a finite space to occupy over time.

4. Conveyors: Moving parts from pt A to Pt B; Number of parts & Speed of conveyor.

5. Resources: Manpower, crews, forklifts, tugs; anything that require a certain resource to be present.

6. End Pt: Keep track of statistics and free memory!

9

Page 10: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Manufacturing Example

Problem Statement: 30% proposed increase in throughput of an overhead sequencing bank of finite capacity

Questions from management: Do we need to add capacity (additional lane), or can we maintain the current size and improve our sequencing routing logic?

The two key questions from Simulation Team: 1. What is the significant input? 2. How do we control it?

10

Page 11: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Sequence Bank (60 units)

Net JPH & Gross JPH into System Vehicle Mix Percent

# of Designated Lanes Capacity # Size of Deck

Overflow lanes Sequencing Logic

Repair Times Conveyor Cycle Times

# of vehicles on input transfers

The last goes on…

Throughput Net JPH

INPUT SYSTEM OUTPUT

Priority of Work # Resources

Build Schedule Routing Logic

Consecutive jobs

The list goes on…

CONTROLS

Human Decisions Internal Failures

Crew Factors Market Demand

NOISE

Objective: Do we need to add an additional Lane?

P-Diagram (Seq Bank)

11

Page 12: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Reduce Inputs & Get Answers

1. Examine the system assuming ideal (perfect) sequence logic.

2. Some of the N’s & X’s can be ignored.

3. Determine the breakpoint where the system fails; maybe it’s 23%.

Use the least amount of data to get initial answers to management ASAP!

12

Page 13: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Establish Credibility

1. The program team might have critical answers sooner; hence managers have additional time to secure funds.

2. The initial answers provided and backed by data will keep the team engaged, and eager to provide further details!

3. The simulation engineer learns more about the system, and can add additional details as necessary.

13

Page 14: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Graph your Data! One of the most basic steps in 6-Sigma; Exploit your data!

Stat-Fit for

SIMUL8

14

Page 15: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Leverage Statistical Distributions!

• Curve fit your data! Instead of using lengthy spreadsheets.

• Black-box; entire segments of the model can be collapsed using distributions.

• If using empirical datasets, drop them into a “S8 Probability Profile Distribution”

15

Page 16: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Use Known Distributions

The data collection phase of modeling can be the lengthiest and most time consuming.

i.e.)

Downtime (MTBF & MTTR); such as Exponential & Erlang respectively.

Cycle times often use a Fixed distribution; that is the “Design Cycle Time”.

16

Page 17: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Steady State

A common data collection error is to capture all data points, and attempt to force them into one distribution.

– Filter out the outliers; usually catastrophic points are outside the scope of the steady state system.

17

Page 18: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Tornado of Change

Utilizing Six-Sigma tools in conjunction with model building keeps the team informed on what is of most importance to get the program launched.

This is how several large manufacturing companies are able to shorten the overall launch time, and bring their new product to the market before the competition

If you find yourself caught up in the tornado of change; just remember “Less is More”!

18

Page 19: @SIMUL8 Virtual User Group, September: Brian Harrington, Less is More

Questions

19