INTEGRATED AND FLEXIBLE INFORMATION SYSTEM FOR QUALITY CONTROL AT FOAM PRODUCTION AUTOMOTIVE INDUSTRY Valentim Realinho Escola Superior de Tecnologia e Gestão de Portalegre Lugar da Abadessa, Apartado 118, 7301-901 Portalegre, Portugal valentim.realinho@estg.pt Filipe Fidalgo Escola Superior de Tecnologia e Gestão de Castelo Branco Av. Empresário, 6200-767 Castelo Branco, Portugal ffidalgo@est.ipcb.pt Pedro Silva Escola Superior de Tecnologia e Gestão de Castelo Branco Av. Empresário, 6200-767 Castelo Branco, Portugal psilva@est.ipcb.pt ABSTRACT The detection of defects at the end of the manufacturing process, or even in the customer, is a costly and incompatible process with the actual quality demands and services. The quality control should be part of the entire productive process, through inspections and tests that allow the detection of failures and make the necessary corrections in anticipation, avoiding scrap production and diminishing costs of non-quality. The use of quality tools allows those detections, its characterization and the removal of root causes, in order to take actions that restore the process to normal conditions of operation. Acting directly in the process can do this. At this level, the statistical control process, supplies a set of effective tools for the prosecution of this goal, so they allow the detection of the special causes of variability. This paper aims to describe a conceptual model for process improvement, building an information system capable of, in real time, allowing control quality applied to scrap and rework. The system was tested and is now working in a foam production company for the automobile industry. KEYWORDS Statistical Process Control; Statistical Quality Control; Scrap; Rework; Foam; Information System. 1. INTRODUCTION The use of statistical process control in the industry seeking the efficiency, productivity and sales, was popularized, starting from Deming work, and is being frequently used today as a key tool for quality control. In order to use these tools, it becomes necessary large amounts of data that allow applying with rigidity the statistical methods that are in the theoretical base of these tools. The 100% inspection generates large amounts of data, speciality in plants with great volumes production as it is common in the automobile industry. Those data can be obtained in an automatic way, through the use of automata that do the necessary measure and put that information in a central database; or then it should be the operator, that introduces the data manually into the system as he does each inspection. Anyway, the next step is to analyze that data using the appropriate software that implements the most several tools generating alert whenever the process produce out of the expected. Of among the quality indicators that worry more the managers, we have the rework and scrap. According with Montgomery (2001), these problems, represent the largest source of productivity lack in the companies. These are the indicators that the model here proposed aims to control
through the conceptualization and development of an information system. The system collects data in real time in order to be analyzed by tools based on the theory of the statistical process control. 2. CONCEPTUAL MODEL The model is composed by several modules that allow an improvement of process performance through the use of statistical quality control tools. In each ending position, we have a computer with a touchscreen that picks up the data related with rework and scrap. It is the operator s responsibility to classify and register the defects detected. These units allow real time quality control from the plant floor, classifying the conformity of each produced part, and store the results in a central database. Figure 1. Conceptual system model Problem Detection In this stage, there are identified causes which are the origin of problems. The classical statistical control tools supply useful information for this job. In this model, the problem detection assumes a preponderant role when allowing an early identification of problems in the production line that can result in non conformity pieces. The classic tools of statistical quality control have an important contribute in the accomplishment of this task. For the effect, some of the classic tools as control charts, line graphs, histograms and operatingcharacteristics (OC) curves, are supplied. The quality characteristics to control, rework and scrap, suggest the use of p and np control charts. Alarms are used to generate alerts in real time, and make problems monitoring that are happen in the plant floor, facilitating a quickly intervention on the process. The model proposed for alarms is based on a table of defects classified by mould and defect code, allowing to inform the supervisor concerning the problem that is happening, whenever abnormal situations of behavior are detected. This model of alarms allows, in an easy way, the identification of two types of problems: 1. Process problem These are problems concerning the process variables that affect all moulds. Easily to identify when the same problem occurs for a considerable number of moulds. 2. Problem in moulds Identificable when a mould presents a high proportion of defects. It can be resulted of a process problem. Identifying Cause In this stage, there are identified causes which are the origin of problems. The classical statistical control tools supply useful information for this job. The knowledge base with causes for each problem type, and a
report of corrective actions implemented with the respective result, allows the creation of organizational memory, particularly relevant in environments where the tax of rotation of operators is elevated, as it is the case of the case study sector. Whenever a problem is detected, there should be identified the causes that are in the origin of it, and suggested by the system the corrective actions to be taken, for priority order, in agreement with the last knowledge in similar situations. Just as the alarms, these corrective actions are configured by the user, and can be changed as the knowledge on the process increases. Associated to each defect type, a journal is built, with the report of the whole relevant information on events in the production lines, and the way these influence the process, what can supply an added knowledge on the same situation. The departure tool for the identification of the cause root of the problem is the cause-effect diagram (ISHIKAWA, 1984). A form of using this diagram is through the method of the "5 Why". Mainly, this method consists of asking "why" systematically (they are usually made a maximum of five questions), until being discovered the root of the problem. Figure 2 illustrates the screen developed for the problem solving method. The cause-effect diagram of the defect that is to be studied, as well the use of the "5 Why" method, will allow to arrive to the cause root. Figure 2. Problem Solving and 5 Why method Corrective Action For each one of the root causes identified, it should be defined the corrective actions that allow to eliminate the problem. Those correctives actions, constitutes the action plan that contains the schedule and the responsible for each one. The methodology 5W1H can help in the development of action plans, and it consists of elaborating the action plan based on six questions that will define its structure. Those questions are: - WHAT: it defines what will be executed, containing the explanation of the action to be accomplished. Usually are used verbs in the infinitive, in a briefly way; - WHEN: it defines when the action will be executed. Beginning and end date of the action; - WHO: it defines the responsible for the action; - WHERE: it defines where the action will be executed; - WHY: it defines the reason for the action. It presents the immediate purpose of the action to be taken; - HOW : it defines the detail of as the action will be executed. This field is a complement for the first field (WHAT). Follow-up This stage is responsible for the verification of the efficiency of the corrective actions. It is composed by a group of tools that allow the analysis of the historical evolution of the process: control charts (p chart), line graphs, operation characteristic curves, pareto charts, histograms and data sheets.
Figure 4. Follow-up The information can be visualized in according to the following analysis dimensions: model, production shift, defect class (rework and scrap) or still for period of time. In the p charts, there are marked in red all the points out of the limits control. In yellow are the points that Hansen (1963) describes as indicators that require investigation (Violating runs). The operation characteristic curves (OC curves) allow to determine the probability of the process to present a certain tax of defects. The pareto chart present the most appealing defects for the specified conditions. 3. APPLICATION The system was tested in a foam production company for the automobile industry. The quality control is made by operators, through manual inspection of each part produced. The part is observed as they are produced and remove the ones that are not in agreement to the defined quality criteria. The non conformity is classified through touchscreen. During the first six months, was registered the production of 465.453 parts, with a rework of 22,43%. In this period the scrap was 1,94%. Table 1 presents production summary of rework and scrap by model during this period. Table 1. Production Summary by model Rework Scrap Model Description Production Nº % Nº % 5116022L ASSENTO TRÁS 1/3 B/C 84 121.331 45.741 37,70 2.795 2,30 5116026L ASSENTO TRÁS 2/3 B/C 84 117.293 24.770 21,12 1.768 1,51 5101335 ASSENTO FRENTE DIR. B 84 94.182 12.017 12,76 1.692 1,80 5101334 ASSENTO FRENTE ESQ. B 84 90.201 10.401 11,53 952 1,06 9279297 ASSENTO FRENTE DIR. C 84 12.543 3.463 27,61 449 3,58 9279296 ASSENTO FRENTE ESQ. C 84 11.425 3.596 31,47 516 4,52 5172592E ASSENTO TRÁS 2/3 B/C 84 COURO 5.825 1.222 20,98 374 6,42 5172588E ASSENTO TRÁS 1/3 B/C 84 COURO 4.201 1.224 29,14 161 3,83 5119478 ASSENTO TRÁS B/C 84 1/1 3.692 1.128 30,55 147 3,98 5170406 ASSENTO FRENTE ESQ. B 84 COURO 1.834 297 16,19 60 3,27 5170407 ASSENTO FRENTE DIR. B 84 COURO 1.688 306 18,13 51 3,02 5172248 ASSENTO FRENTE ESQ. C 84 COURO 649 141 21,73 27 4,16 5172249 ASSENTO FRENTE DIR. C 84 COURO 589 84 14,26 18 3,06 Total 465.453 104.390 22,43 9.010 1,94 The first four models represent 91% of total production, and they concern with the version of automobile model more sold. In the remaining models the rework and scrap is quite superior to the one of these, denoting a smaller concern with these parameters. The fact that the volume production is low, doesn't justify the costs to lower these parameters. Table 2 presents the same data grouped by month. In august, an accentuated decrease of the production was verified due to the annual stop of production lines for maintenance. Starting
from august, they started to work only with two shifts. This decrease translated herself in a quick break of the productive capacity. Table 2. Production Summary by month Rework Scrap Month Production Nº % Nº % Mai. 94.102 25.869 27,49 1.492 1,59 Jun. 90.341 18.507 20,49 1.308 1,45 Jul. 80.582 15.527 19,27 1.197 1,49 Ago. 44.778 11.002 24,57 929 2,07 Set. 75.165 15.667 20,84 2.144 2,85 Out. 80.485 17.818 22,14 1.940 2,41 Total 465.453 104.390 22,43 9.010 1,94 The evolution of rework and scrap can be observed in figure 5. In this figure the tendency line of rework was added, in order to be more easily verifiable a tendency for rework decrease. Figure 5. Follow-up Table 3. Rework by type and month (% of the monthly total) Defect May Jun Jul Aug Sep Oct Total by type Colapso aba 7,3 12,0 17,1 19,3 25,5 14,4 15,4 Bolha debaixo Velcro 12,3 15,4 13,7 19,0 11,4 13,1 13,8 Bolha junto moquette 3,1 7,3 10,1 9,3 18,9 19,3 11,2 Outro 5,9 7,2 10,4 7,1 9,0 8,6 8,0 Falta enchimento lateral 1/3 16,6 4,1 5,1 5,2 4,6 8,1 7,8 Colapso centro 2,6 8,8 8,9 7,2 6,4 5,6 6,4 Falta enchimento no canto 6,1 7,0 5,4 4,2 3,1 6,2 5,5 Colapso junto Velcro 4,7 6,0 6,3 7,5 4,8 2,4 5,1 Falta enchimento 2/3 8,1 3,1 3,3 2,1 2,2 5,3 4,3 Espuma rasgada sobre Velcro 10,4 8,1 1,6 1,3 0,4 0,7 4,1 Falta rebordo 3,4 2,3 3,2 3,8 2,4 8,2 3,9 Velcro descolado 6,4 3,9 4,2 2,7 3,1 2,1 3,9 Falta espuma vedação 2,0 2,6 2,7 3,1 1,8 2,8 2,5 Colapso centro/aba 2,7 2,9 2,2 2,0 0,8 1,1 2,0 Orelha rasgada 3,6 1,7 1,4 1,9 2,0 0,6 1,9 Colapso junto moquette 1,1 2,9 1,8 1,2 1,9 0,3 1,5 Falta espuma zona velcro 1,6 0,8 0,7 0,9 0,3 0,1 0,7 Quadrado imperfeito 0,3 0,7 0,6 1,0 1,1 0,4 0,7 Contracção espuma 0,2 1,1 0,8 0,6 0,1 0,4 0,5 Deficiente junção espuma 0,0 1,7 0,2 0,1 0,1 0,3 0,4 Falta enchimento 1/1 1,3 0,1 0,0 0,0 0,0 0,0 0,3 Colapso debaixo velcro 0,2 0,2 0,2 0,1 0,2 0,2 0,2 Table 3 presents rework by type and month (% of the monthly total). An important point of analysis of these tables is the fact that many defects exist classified as "Other". This happens whenever the operator doesn't get
to identify correctly the defect type. Globally, this classification doesn't introduce mistakes, but of the analysis point of view and identification of the problems leads to false tracks. We have minimized this subject, through specific training of operators, and with the use of visual helps in the defects register screens. 4. CONCLUSION In this article the applicability of information systems was demonstrated in the quality control of moulded foams production. For that, it was used the basic statistical process tools, based in a model that integrates them in a coherent way. The model allows the creation of a knowledge base to aid operators and engineers in the prevention and identification of special causes of variation of the process. The used approach and the applied techniques can substitute with advantage the manual techniques used usually in quality control, eliminating most of their deficiencies and inconveniences. The model is still sufficiently generic and flexible to be used in industrial processes of similar characteristics, namely where the need of controlling the quality of the piece exists, and where those controls should be made in real time and with high productive cadence. The gotten results in the company in study are globally satisfactory, allowing a rework reduction, eliminating manpower in the finishes productive cells. In the quantitative point of view, the use of the model, as allowed a decrease in rework in the order of the 7-8%. REFERENCES DEMING,E., 1986. Out of the Crisis. MIT Centre for Advanced Engineering Study, Cambridge DUNCAN,A,. 1986. Quality control and Industrial Statistics, 5th ed. Irwin, Illinois FEIGENBAUM,A., 1983. Total Quality Control. McGraw-Hill, New York HANSEN,B., 1963. Quality Control Theory and Application. Prentice-Hall, Englewood Cliffs, N.J. HO,S. and FUNG,C., 1994. Developing a TQM Excellence Model. The TQM Magazine, Vol. 6, Nº 6, MCB University Press ISHIKAWA,K., 1984. Quality Control Circles at Work. JUSE JURAN,J., 1992. Juran on Quality by Design: The New Steps for Planning Quality into Goods and Services. The Free Press, New York JURAN,J. and GRYNA,F., 1991. Controle da qualidade handbook: conceitos, políticas e filosofia da qualidade. McGraw-Hill e Makron Books do Brasil LAUNDON,K. and LAUDON,J., 2002. Management Information Systems: Managing the Digital Firm - Seventh edition. Prentice-Hall MONTGOMERY,D., 2001. Introduction Statistical Quality Control, Fourth Edition. John Wiley & Sons, New York WESTERN ELECTRIC COMPANY, 1956. Statistical Quality Control Handbook. Mack Printing Company, Easton, PA