Design and Analysys of Ind. Experiments

Design and Analysis of Industrial Experiments
Statistica in azienda, Statistici in azienda Padova Complesso Santa Caterina 15 Giugno 2010

Agenda
Introduction Statistics Statistics DOE
of Tetra Pak
at Tetra Pak
support to PD process: V-model approach at Tetra Pak

Consumer satisfaction case study Process parameter optimization case study
Key
messages & Answer
Questions
External Tarantino-Leardi / Jun 2010

Tetra Pak Introduction
Tetra Pak is a systems supplier of

Processing solutions Packaging solutions Distribution solutions

Tetra Pak is global and works locally

Present
in more than 170 countries across 5 continents packaging material plants R&D units machine assembly plants
42 11 11

Development & Engineering

No.
of employees Sweden Italy Germany Switzerland 1031 466 19 26 4
Lund,
Modena, Stuttgart, Romont, Other
locations

Total world deliveries
2009
Carton packaging material, mio packs Distribution machines Packaging machines Processing units 145,030 1,113 351 1,699

Tetra Pak Group, January 2010

Machines in operation 51,859 processing units
9,048 packaging machines
16,641 distribution machines


Todays package portfolio

Agenda
of Tetra Pak
at Tetra Pak

Key
messages & Answer
Questions

Statistics at Tetra Pak
And ma n
y other
things

Agenda
of Tetra Pak
at Tetra Pak

Key
messages & Answer
Questions

Statistical support to PD process: DoE within V-model
State of the art? Market research & screening Commissioning SPC at Customercustomer site Root cause analysis Verify lines & continuous improvements consistency Screening
Confirm Requirements Confirmation runs Fulfilled
Combined with SPC Requirements validation Define Screening & system simulations System VVT Strategy Requirements
Validation Tests
Physical Validate the system Validation Optimization and Robustness verification Verification Robust Design Module Test
Integration Tests
Preliminary assessments Requirements Screening & Virtual verification Cascade VVT Plan & Robust Design Concept evaluation, trade studies Architecture Screening and confirmation Design Unit testing
Unit Test
Verify system requirements: optimization

Integration
Modules verification: Screening & Optimization

Detailed Design

Statistical support to PD process: Decision process
System Consistently Operating? Confirm Remaining areas Verify lines for improvement consistency & issues
State of the art?

Customer needs
Requirements Fulfilled
Define System Requirements
Physical Validation
Risk scenario What is in what is out?
Requirements Cascade
Reqs verifiable?
System Validated? System Verified?

Verification
Architecture Design
Integration
Detailed Design
Ready to Verify, Validate And test?


Agenda
of Tetra Pak
at Tetra Pak

Key
messages & Answer
Questions
Consumer Satisfaction of opening system

Case study

DOE at Tetra Pak Consumer Satisfaction case study The aim of this study was to identify the parameters that optimize the performance and the customer satisfaction Packaging types, dimensional, sensorial and sociological factors were studied.
External Tetra Pak Internal Tarantino-Leardi / Jun 2010
Tarantino/May 09

DoE at Tetra Pak Consumer Satisfaction case study This activity is part of the product test & consumer satisfaction activity during concept development phase.
1. A trade study furnishes the feasibility test cases that fit the targeted usage scenarios. Instrumented mock-ups are manufactured in order to exercise the alternative opening systems A representative set of consumers from the addressed population is selected. A short training set is proposed to every consumer and successive 5 randomized openings. Subjective satisfaction index and objective opening performance are registered and analyzed.
2.
3.
4.
5.

DoE at Tetra Pak Consumer Satisfaction case study
FACTORS UNDER STUDY Dimensional: height & Dimensional diameter
Sensorial: different grips
Competitors: Carton vs. Bottle
Sociological: age & gender


Due to subjective evaluations and low cost of the single test a mixed full factorial testing with multiple mid-points was planned and executed in order to eliminate risky confounding and assess single users biases.
Each consumer opened 5 consecutive randomized mock-ups after one or two training openings on the mid-points.

The two main responses evaluated are characterized by: Opening force: objective, continuous, normally distributed Consumer satisfaction: Subjective, semi-quantitative and comparative:
XX X XX
4 2 3 1
X
3
Min. 0
Max satisfaction
To determine the number of replicates in the experiment we used the power function on the base of historical information in order to optimizing the chances to identify at least one grade on the satisfaction scale. The sampling so determined was more than sufficient to characterize the opening force characteristics. External
Tarantino-Leardi / Jun 2010

Note: the runs are here not randomized but in reality they are. The responses are artificially changed for confidentiality reasons




Bottles Carton The green area inside the plot shows the range of diameter and height where the criteria: appraisal 2.5-5 and a reasonable torque are both satisfied. In the yellow one only one of the two responses is fits the criteria. This plot is used to find the best operating conditions for getting the desired dimensions of the cap.

The final configuration for the opening system design to achieve the target of this study is: Height: 20.35 mm Diameter: 39.5 mm Grip: G2 (not practically relevant) Grip is not practically relevant and so it was settled up to the state of the art without further developments.

Statistical support to PD process: DoE within V-model
Verify lines consistency
State of the art? MarketCustomer & screening research
Confirm Requirements Fulfilled
Validation Tests
Define System Requirements Physical Validation
Integration Tests
Requirements Cascade Module Test Architecture Design
Unit Test
Verification
Verify system requirements: Integration optimization
Detailed Design
Process parameter optimization

Case study

Process parameter optimization
The
aim of this study was to optimize the injection moulding process parameters to produce caps according to the dimensions identified in the previous study. particular, cap-lid diameter and cap total height were studied.
In
Internal Tetra Pak Internal Tarantino/Nov.09
Tarantino/May 09

DOE at Tetra Pak Process parameter optimization The injection moulding is a manufacturing process for producing parts from thermoplastic material.
1. 2. Granules of plastic powder are poured or fed into a hopper A heater heats up the tube and when it reaches a high temperature a screw thread starts turning. A motor turns a thread which pushes the granules along the heater section which melts then into a liquid. The liquid is forced into a mould where it cools into the desired shape (in this case a cap). The mould then opens and the unit is removed.
3.
4.
5.

DOE at Tetra Pak Process parameter optimization
FACTORS UNDER STUDY

INJECTION TIME: Time for the injection of the polymer into the mold cavity INJECTION TEMPERATURE: Temperature at which the heater heats up the tube HOLDING PRESSURE: Pressure applied by the screw to compensate the shrinkage of the plastic part HOLDING TIME: Time at which the screw applied the holding pressure COOLING TIME = Time to transform row plastic material into desired part

For each one of the factors, 2 levels were studied, a high level and a low level. We call these levels by 1 and +1 respectively (or just and +). Low level (-1) Injection time Injection temperature Cooling temperature Holding Time Holding pressure 0.2 sec 230 C 0.9 C 0.8 s 407 bar High level (+1) 0.4 sec 250 C 1.1 C 1.2 s 543 bar

With 5 factors and 16 runs we have a resolution V factorial design, i.e. main effects would be confounded with four-factor interactions, and two-factor interactions would be confounded with certain threefactor interactions.
3 additional midpoints were added. They are used to learn something about non-linear effect and to limit the effort of replications
-1 -1 0 1

To determine the number of replicates in the experiment we used the power function Power function is a function of the probability to reject a certain hypothesis Question: What is the size of difference in the response that we want to be able to detect (practical relevance)
Significance level: 0.05, risk to reject the hypothesis that the Effect is zero despite the fact that it is. (Type I risk)
1.0
Sample size needed.

Reps, C tr Pts Per Blk 1, 3
A ssumptions A lpha 0.05 S tD ev 0.002 # F actors 5 # C orner P ts 16 # Blocks none # Terms O mitted 0 Term Included In M odel C enter P oints Yes
Power Curve for 2-Level Factorial Design
0.8
0.6 Power 0.4
0.2
0.0 -0.02 -0.01 0.00 Effect 0.01 0.02
With such low variations in the experiment 1 replicate per run is good enough
Relevant difference: If the effect is 0.02 we want to detect it


RUN
Inj. Time 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 0.2 0.4 0.2 0.4 0.2 0.4 0.2 0.4 0.2 0.4 0.2 0.4 0.2 0.4 0.2 0.4 0.3 0.3 0.3 Inj. Temp 230 230 250 250 230 230 250 250 230 230 250 250 230 230 250 250 240 240 240
FACTORS
Cooling time 0.9 0.9 0.9 0.9 1.1 1.1 1.1 1.1 0.9 0.9 0.9 0.9 1.1 1.1 1.1 1.1 1 1 1 Holding time 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1 1 1 Holding pressure 543 407 407 543 407 543 543 407 407 543 543 407 543 407 407 543 475 475 475
RESPONSES
H 20.293 20.202 20.194 20.296 20.211 20.308 20.304 20.215 20.284 20.391 20.384 20.289 20.400 20.304 20.299 20.402 20.301 20.301 20.299 D 39.402 39.407 39.284 39.528 39.287 39.532 39.403 39.402 39.459 39.712 39.576 39.575 39.589 39.578 39.463 39.713 39.499 39.499
39.503
Note: the runs are here not randomized but in reality they are. The responses are artificially changed for confidentiality reasons

Randomizing the order of the runs is usually good. It is some kind of insurance that our conclusions will not be affected by uncontrolled variation of the test environment. but randomization is not always easy or even possible Drawbacks with randomization: Some factors are hard and time consuming to change The number of changes of factor levels might in itself be time consuming It might get difficult to keep track of the experiments.

Main Effects Plot for H

Data Means
Inj. Time 20.350 20.325 20.300 20.275 Inj. Temp C ooling time
Main Effects Plot for D

Data Means
Point Ty pe C orner C enter
39.60 39.55 39.50 39.45 Inj. Time Inj. Temp C ooling time
Point Type C orner C enter
Mean
Mean
20.250 0.2 20.350 20.325 20.300 20.275 20.250 0.8 1.0 1.2 407 475 543 0.3 Holding time 0.4 230 240 250 0.9 1.0 1.1
39.40 0.2 39.60 39.55 39.50 39.45 39.40 0.8 1.0 1.2 407 475 543 0.3 Holding time 0.4 230 240 H olding pressure 250 0.9 1.0 1.1
Holding pressure

Interaction Plot for H
Data Means
230 240 250 0.9 1.0 1.1 0.8 1.0 1.2 407 475 543 20.4
Inj. Time
20.3
20.2 20.4
Inj. Time 0.2 0.3 0.4 Inj. Temp 230 240 250
Point Type Corner Center Corner Point Type Corner Center Corner Point Type Corner Center Corner Point Type Corner Center Corner
Inj. Temp
20.3
20.2 20.4
Cooling time
20.3
20.2 20.4
Cooling time 0.9 1.0 1.1 Holding time 0.8 1.0 1.2
Holding time
20.3
20.2
Holding pressure

Interaction Plot for D
Data Means
230 240 250 0.9 1.0 1.1 0.8 1.0 1.2 407 475 543 39.6 Inj. Time 39.5 39.4 39.6 Inj. Temp 39.5 39.4 39.6 Cooling time 39.5 39.4 39.6 Holding time 39.5 39.4
Inj. Time 0.2 0.3 0.4 Inj. Temp 230 240 250
Point Type Corner Center Corner Point Type Corner Center Corner Point Type Corner Center Corner Point Type Corner Center Corner
Cooling time 0.9 1.0 1.1 Holding time 0.8 1.0 1.2
Holding pressure

Normal Plot of the Standardized Effects
(response is H, Alpha = 0.05)
99
99
Normal Plot of the Standardized Effects

(response is D, Alpha = 0.05)
Effect Type Not Significant Significant
F actor A B C D E N ame Inj. Time Inj. Temp C ooling time H olding time H olding pressure
95 90 80 C A DE D
95 90 80 AE E A
Effect Ty pe Not Significant Significant

Percent
Percent
70 60 50 40 30 20 10 5
70 60 50 40 30 20 10 5
50
100 150 Standardized Effect
200
20
40
60 80 100 120 Standardized Effect
140
160
180
Pareto Chart of the Standardized Effects

(response is H, Alpha = 0.05) 4.3
E D C A DE B CE AC AE CD AB BC BE AD BD
Pareto Chart of the Standardized Effects

(response is D, Alpha = 0.05) 4.3
D E A AE DE C B CD AC CE BC BE AD BD AB
F actor A B C D E N ame Inj. Time Inj. Temp C ooling time Holding time Holding pressure
Term
Term
50
100 150 Standardized Effect
200
20
40
60 80 100 120 Standardized Effect
140
160
180


Contour Plot of H vs Holding pressure, Holding time

540 520 500 480 460 440 420 0.8 0.9 1.0 Holding time 1.1 1.2
20.20 20.24 20.28 20.32 H < > 20.20 20.24 20.28 20.32 20.36 20.36
Contour Plot of D vs Holding pressure, Holding time

540 520 500 480 460 440 420 0.8 0.9 1.0 Holding time 1.1 1.2
39.30 39.35 39.40 39.45 39.50 D < > 39.30 39.35 39.40 39.45 39.50 39.55 39.55
Holding pressure
Hold Values Inj. Time 0.2 Cooling time 0.9
Holding pressure
Hold Values Inj. Time 0.2 C ooling time 0.9

Contour Plot of H
540 520
H 20.3 20.38 Hold Values Inj. Time 0.2 Cooling time 0.9
The white area inside the plots shows the range of holding time and holding pressure where the criteria for both response variables are satisfied. This plot is used to find the best operating conditions for getting the right height and the right diameter of the caps
Holding pressure
500 480 460 440 420 0.8 0.9 1.0 Holding time 1.1 1.2
Contour Plot of D
540 520 500 480 460 440 420 0.8 0.9 1.0 Holding time 1.1
D 39.3 39.55 Hold Values Inj. Time 0.2 Cooling time 0.9
Holding pressure
External 1.2 Tarantino-Leardi / Jun 2010


The final configuration for the injection moulding process to achieve the target of this study is: Injection time: 0.21 sec Cooling time: 1.10 sec Holding time: 1 sec Holding pressure: 407 The injection temperature is unimportant and so it was settled up at 230 C

Agenda
of Tetra Pak
at Tetra Pak

Key
messages & Answer
Questions

Key messages
DoE increases value of the overall system lifecycle planned activity Careful preliminary tests maximise the success as a part of proper planning. DoE design is easy but proper planning, randomization, preparation and execution is another game. DoE is not the Panacea to clarify all the uncertanties characteristics of the system during the development. DoE complements very well with the majority of the other statistical techniques.

Agenda
of Tetra Pak
at Tetra Pak

Key
messages & Answer
Questions

Question & Answer
Who should you contact

To know more and keep updated
Pietro Tarantino Expert Advisor
D&E - Packaging Technology Engineering Excellence Systems Engineering Methodology Pietro.tarantino@tetrapak.com +39 059898389
www.tetrapak.com
Carlo Leardi Expert Advisor

D&E - Carton Value Systems Engineering Systems Engineering Validation Carlo.Leardi@tetrapak.com +39 059898389
Thank you for attention!
Turning used cartons into an asset

Separating paperboard from plastic and aluminium
Pulp Products
Collected cartons
Repulping
Poly/Al
Products
External Tarantino-Leardi / Jun 2010 TP1137, JH/200903
Recycled cartons a valuable asset

Raw material for a wide range of new products
Cardboard Trays Dry food boxes Household tissue
Egg cartons
Envelopes
Paper cores
Plasterboard liner
Frozen food boxes
Industrial tissue
Office paper
Paper bags
Recycling a growing industry

Ensuring efficient re-use of valuable resources 33% of beverage cartons recycled in EU (2008) 18.7% Tetra Pak cartons recycled world-wide (2009) We actively support increased recycling and consumer awareness

Design and Analysys of Ind. Experiments

Uploaded by

Copyright:

Available Formats

You might also like

Design and Analysys of Ind. Experiments

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Design and Analysys of Ind. Experiments

Uploaded by

Copyright:

Available Formats

Design and Analysis of Industrial Experiments

Design and Analysis of Industrial Experiments

support to PD process: V-model approach at Tetra Pak

messages & Answer

External Tarantino-Leardi / Jun 2010

Design and Analysis of Industrial Experiments

Tetra Pak is a systems supplier of

External Tarantino-Leardi / Jun 2010

Design and Analysis of Industrial Experiments

Tetra Pak is global and works locally

External Tarantino-Leardi / Jun 2010

Design and Analysis of Industrial Experiments

Development & Engineering

of employees Sweden Italy Germany Switzerland 1031 466 19 26 4

Modena, Stuttgart, Romont, Other

External Tarantino-Leardi / Jun 2010

Design and Analysis of Industrial Experiments

Total world deliveries

Design and Analysis of Industrial Experiments

Tetra Pak Group, January 2010

9,048 packaging machines

16,641 distribution machines

Design and Analysis of Industrial Experiments

Todays package portfolio

External Tarantino-Leardi / Jun 2010

Design and Analysis of Industrial Experiments

support to PD process: V-model approach at Tetra Pak

messages & Answer

External Tarantino-Leardi / Jun 2010

Design and Analysis of Industrial Experiments

External Tarantino-Leardi / Jun 2010

Design and Analysis of Industrial Experiments

support to PD process: V-model approach at Tetra Pak

messages & Answer

External Tarantino-Leardi / Jun 2010

Design and Analysis of Industrial Experiments

Verify system requirements: optimization

Modules verification: Screening & Optimization

Design and Analysis of Industrial Experiments

State of the art?

Define System Requirements

Risk scenario What is in what is out?

System Validated? System Verified?

Ready to Verify, Validate And test?

Design and Analysis of Industrial Experiments

support to PD process: V-model approach at Tetra Pak

messages & Answer

External Tarantino-Leardi / Jun 2010

Consumer Satisfaction of opening system

Design and Analysis of Industrial Experiments

External Tetra Pak Internal Tarantino-Leardi / Jun 2010

Design and Analysis of Industrial Experiments

Design and Analysis of Industrial Experiments

Sensorial: different grips

Competitors: Carton vs. Bottle

Sociological: age & gender

Design and Analysis of Industrial Experiments

External Tarantino-Leardi / Jun 2010