Quality improvement of cans for canned food

Key words

Suppliers Selection, Process Capability, Statistical Process control, Sigma Level of Process, Defect per Million of Opportunities, DMAIC

Introduction

Onix is a factory that provides metallic components, some of them are cans for canned food. These cans are used by several clients that are big manufactures of conserve meat and fish. In the last weeks, Onix is receiving a lot of complaints about the quality of products supplied, most of them around the sealed parts and nicks on the body. For that reason the factory will employ the Six Sigma DMAIC procedure in order to find and eliminate the principal root cause of the problem.


Define

When the Onix management group receives the mentioned group of complaints, immediately activate its Quality Committee to investigate the main root causes of these problems. This team find out two possible and antagonistic root causes of this problem, as shown in the next chart.


Cause Type

Action

Possible Error

The fault is caused by a natural wear of the machinery

The workers can solve the problem fixing the machinery or replacing the affected parts

Error type 2: Consider the raw materials as the problem and do not adjust the machinery 

The cans defects are produced by defects with the raw materials, because the managers in the effort of reduce costs, buy tinplate with bad quality. This problem in unknown for the workers and the managers have no idea of damage produced.

There is nothing to do for the workers, only the managers can solve this problem, changing the shop policy.

Error type 1: when the work line produces a defective can, the natural reaction of workers is to fix the machinery; but the problem is on the material, so this fixing can aggravate the problems, meanwhile the real root cause persist undetected.


After this causes analyses, the work team has to answer the following question: What is the real root cause of this problem? To solve this problem, what we must do first is check the problems with suppliers, (the process stability and capability).

Measure

In this case we can’t check the process stability and capability before is discarded the problems with the raw materials, because problems with raw materials affects the process stability because it is a variation cause.

Specifically for this problem, the quality characteristic that we must take into account is the tinplate hardness; for that reason it was carried out specific controls for that characteristic, which have to be higher than 90 N/m2. The data sets obtained with the Vickers test are presented following:

Supplier 1

Supplier 2

Supplier 3

84,8949

97,697

85,7721

97,7484

87,119

89,4851

83,6844

100,368

89,3418

87,7264

94,259

88,3778

93,5217

89,360

88,0630

97,1654

98,981

89,6156

86,9147

99,737

89,5801

97,0011

98,848

87,6867

85,8637

103,940

89,7483

90,1963

76,806

91,1584

88,5920

108,281

88,8189

96,1181

95,032

91,2991

88,6737

84,369

89,8778

84,4503

103,696

88,5731

87,8797

84,386

90,4986

83,3948

112,514

87,1240

92,9444

91,732

88,4587

90,5980

100,175

90,8822

92,6927

103,243

88,3840

87,3310

93,394

89,3629


Analyze

In order to have a better interpretation of the hardness results, we assess the supplier capability, seeing that vendor 2 is the better, because have a good values of process capability, vendor 3 can accomplish the harness requirements but only with 3 sigma level and vendor 1 are not capable for our requirements.


Improve

After cheeked the quality of raw materials, is needed to verify the process statistical control and capability. For accomplish this goal the production is started using only the best quality raw materials, provided by supplier 2. During this production the work team verifies every lot of 10 000 units, with the objective of do not supply any lot with 0.5 % of defects; the obtained data is presented following.


Sample

Defects

1

21

2

24

3

16

4

12

5

15

6

5

7

28

8

20

9

31

10

25

11

20

12

24

13

16

14

19

15

10

16

17

17

13

18

22

19

18

20

39

21

30

22

24

23

16

24

19

25

17

26

15

Control

Parting on the obtained data for the lot defects, is executed a quality capability and process stability study, obtaining that in general the process is stable, despite it have two points out of control limits, besides the process is capable to accomplish the clients requirements . The 26 lots have 0.1984 % of defects; that means 1984 defects per million. See next graphic.

The Z metric converts the average percent of defects in a capability index similar to the other that can be calculated with continuous data. In most cases is desired a Z value higher that 4, in this case is obtained 2.8, for that reason we can assume that the process must be improved if we want to obtain world class capability indexes.

The tolerance limits show that is expected less of 95 % of lots with 27 defects.

Conclusions

 It was checked that the principal problem was the poor quality of raw materials, obtained from 3 different vendors that have significant differences between the hardness of tinplate provided. It is needed a rigorous control of raw materials to prevent this problem.

It is strongly recommended to maintain commercial relationships only with supplier 2 until the other suppliers can meet the quality requirements needed by Onix.