Statistical Process Control 5 Part 10 ppsx

302 Process problem solving and improvement Steps in constructing a cause and effect diagram include identifying theeffect, establishing the goals, constructing a framework, recording al

302 Process problem solving and improvement

 Steps in constructing a cause and effect diagram include identifying theeffect, establishing the goals, constructing a framework, recording allsuggested causes, incubating the ideas prior to a more structured analysisleading to plans for action

 A variation on the technique is the cause and effect diagram with addition

of cards (CEDAC) Here the effect side of the diagram is quantified, with

an improvement target, and the causes show facts and improvement ideas.

 Scatter diagrams are simple tools used to show the relationship betweentwo factors – the independent (controlling) and the dependent (con-trolled) Choice of the factors and appropriate data recording are vitalsteps in their use

 Stratification is a sample selection method used when populations arecomprised of different characteristics It involves collecting or dividingdata into meaningful groups It may be used in conjunction with othertechniques to present differences between such groups

 The principles and techniques of problem solving and improvement may

be applied to any human activity regarded as a process Where barriers tothe use of these, perhaps in non-manufacturing areas, are found, training

in the basic approach of process data collection, analysis and ment action may be required

improve-References

Ishikawa, K (1986) Guide to Quality Control, Asian Productivity Association, Tokyo, Japan Lockyer, K.G., Muhlemann, A.P and Oakland, J.S (1992) Production and Operations

Management, 6th Edn, Pitman, London, UK.

Oakland, J.S (2000) Total Quality Management – text and cases, 2nd Edn,

Butterworth-Heinemann, Oxford, UK.

Pyzdek, T (1990) Pyzdek’s Guide to SPC – Vol One – Fundamentals, ASQC Quality Press,

Milwaukee WI, USA.

Sygiyama, T (1989) The Improvement Book – creating the problem-free workplace, Productivity

Press, Cambridge MA, USA.

Discussion questions

1 You are the Production Manager of a small engineering company and havejust received the following memo:

MEMORANDUMTo: Production Manager

From: Sales Manager

Subject: Order Number 2937/AZ

Joe Brown worked hard to get this order for us to manufacture 10 000 widgetsfor PQR Ltd He now tells me that they are about to return the first batch of

1000 because many will not fit into the valve assembly that they tell us they areintended for I must insist that you give rectification of this faulty batch numberone priority, and that you make sure that this does not recur As you know PQRLtd are a new customer, and they could put a lot of work our way

Incidentally I have heard that you have been sending a number of youroperators on a training course in the use of the microbang widget gauge foruse with that new machine of yours I cannot help thinking that you shouldhave spent the money on employing more finished product inspectors, ratherthan on training courses and high technology testing equipment

(a) Outline how you intend to investigate the causes of the ‘faulty’widgets

(b) Discuss the final paragraph in the memo

2 You have inherited, unexpectedly, a small engineering business which isboth profitable and enjoys a full order book You wish to be personallyinvolved in this activity where the only area of immediate concern is thehigh levels of scrap and rework – costing together a sum equivalent to about

15 per cent of the company’s total sales

Discuss your method of progressively picking up, analysing and solvingthis problem over a target period of 12 months Illustrate any of thetechniques you discuss

3 Discuss in detail the applications of Pareto analysis and cause and effectanalysis as aids in solving operations management problems Give at leasttwo illustrations

You are responsible for a biscuit production plant, and are concernedabout the output from the lines which make chocolate wholemeal biscuits.Output is consistently significantly below target You suspect that this isbecause the lines are frequently stopped, so you initiate an in-depthinvestigation over a typical two-week period The table below shows thecauses of the stoppages, number of occasions on which each occurred, andthe average amount of output lost on each occasion

304 Process problem solving and improvement

occurrences

Lost production (00s biscuits)

WRAPPING:

ENROBER:

How could stratification aid the analysis in this particular case?

4 A company manufactures a range of domestic electrical appliances.Particular concern is being expressed about the warranty claims on oneparticular product The customer service department provides the followingdata relating the claims to the unit/component part of the product whichcaused the claim:

Unit/component

part

Number of claims

Average cost of warranty work (per claim)

Discuss what criteria are of importance in identifying those unit/componentparts to examine initially Carry out a full analysis of the data to identifysuch unit/component parts.

5 The principal causes of accidents, their percentage of occurrence, and theestimated resulting loss of production per annum in the UK is given in thetable below:

all accidents

Estimated loss

of production (£million/annum)

(b) Give three other uses of this type of analysis in non-manufacturing andexplain briefly, in each case, how use of the technique aids improvement

6 The manufacturer of domestic electrical appliances has been examiningcauses of warranty claims Ten have been identified and the annual cost ofwarranty work resulting from these is as follows:

Cause Annual cost of warranty work (£)

ABCDEFGHIJ

1 090

2 130

30 690620

5 930970

49 980

1 060

4 980

3 020

306 Process problem solving and improvement

Carry out a Pareto analysis on the above data, and describe how the maincauses could be investigated

7 A mortgage company finds that some 18 per cent of application formsreceived from customers cannot be processed immediately, owing to theabsence of some of the information A sample of 500 incompleteapplication forms reveals the following data:

appro-8 A company which operates with a four-week accounting period isexperiencing difficulties in keeping up with the preparation and issue ofsales invoices during the last week of the accounting period Data collectedover two accounting periods are as follows:

Number of sales invoices issued 110 272 241 495

Number of sales invoices issued 232 207 315 270Examine any correlation between the week within the period and thedemands placed on the invoice department How would you initiate action

to improve this situation?

Worked examples

1 Reactor Mooney off-spec results

A project team looking at improving reactor Mooney control (a measure ofviscosity) made a study over 14 production dates of results falling ±5 MLPoints outside the grade aim Details of the causes were listed (Table 11.6)

Table 11.6 Reactor Mooney off-spec results over 14 production days

Table 11.7 Reactor Mooney off-spec results over 14 production dates: Pareto analysis of reasons

Reasons for Mooney off-spec Tally Frequency % of total Cum %

Using a ranking method – Pareto analysis – the team were able to determinethe major areas on which to concentrate their efforts.

Steps in the analysis were as follows:

1 Collect data over 14 production days and tabulate (Table 11.6)

2 Calculate the totals of each cause and determine the order of frequency (i.e.which cause occurs most often)

3 Draw up a table in order of frequency of occurrence (Table 11.7)

4 Calculate the percentage of the total off-spec that each cause is responsiblefor

e.g Percentage due to reactor sticks = 32

80  100 = 40 per cent

Figure 11.14 Pareto analysis: reasons for off-spec reactor Mooney

310 Process problem solving and improvement

Figure 11.15 Income rank/contribution rank chart

5 Cumulate the frequency percentages

6 Plot a Pareto graph showing the percentage due to each cause and thecumulative percentage frequency of the causes from Table 11.7 (Figure11.14)

2 Ranking in managing product range

Some figures were produced by a small chemical company concerning thecompany’s products, their total volume ($), and direct costs These are given

in Table 11.8 The products were ranked in order of income and contributionfor the purpose of Pareto analysis, and the results are given in Table 11.9 Toconsider either income or contribution in the absence of the other could lead

to incorrect conclusions; for example, Product 013 which is ranked 9th inincome actually makes zero contribution

One way of handling this type of ranked data is to plot an income–contribution rank chart In this the abscissae are the income ranks, and the

Table 11.9 Income rank/contribution rank table

Total direct costs ($)

Total contribution ($)

312 Process problem solving and improvement

ordinates are the contribution ranks Thus product 010 has an income rank of

2 and a contribution rank of 7 Hence, product 010 is represented by the point(2,7) in Figure 11.15, on which all the points have been plotted in this way

3 Process capability in a bank

The process capability indices calculations in Chapter 10 showed that theprocess was not capable of meeting the requirements and management led aneffort to improve transaction efficiency This began with a flowcharting of theprocess as shown in Figure 11.16 In addition, a brainstorm session involvingthe cashiers was used to generate the cause and effect diagram of Figure11.17 A quality improvement team was formed, further data collected, andthe ‘vital’ areas of incompletely understood procedures and cashier trainingwere tackled This resulted over a period of six months, in a reduction inaverage transaction time and improvement in process capability

Figure 11.16 Flowchart for bank transactions

Figure 11.17 Cause and effect diagram for slow transaction times

12 Managing out-of-control

processes

Objectives

 To consider the most suitable approach to process trouble-shooting

 To outline a strategy for process improvement

 To examine the use of control charts for trouble-shooting and classify of-control processes

out- To consider some causes of out-of-control processes

In a manufacturing company, for example, the production people have theresponsibility for meeting production targets, which include those associatedwith the quality of the product It is unreasonable for them to acceptresponsibility for process output, efficiency, and cost while delegatingelsewhere responsibility for the quality of its output If problems of lowquantity arise during production, whether it be the number of tabletsproduced per day or the amount of herbicide obtained from a batch reactor,then these problems are tackled without question by production personnel.Why then should problems of – say – excessive process variation not fallunder the same umbrella?

Problems in process operations are rarely single-dimensional They have atleast four dimensions:

316 Managing out-of-control processes

 product or service, including inputs;

 plant, including equipment;

 programmes (timetables-schedules);

 people, including information.

The indiscriminate involvement of research/technology people in shooting tends to polarize attention towards the technical aspects, with thecorresponding relegation of other vital parameters In many cases the human,managerial, and even financial dimensions have a significant bearing on theoverall problem and its solution They should not be ignored by taking aproblem out of its natural environment and placing it in a ‘laboratory’.The emphasis of any ‘trouble-shooting’ effort should be directed towardsproblem prevention with priorities in the areas of:

trouble-(i) maintaining quality of current output;

(ii) process improvement;

(iii) product development

Quality assurance, for example, must not be a department to be ignored wheneverything is running well, yet saddled with the responsibility for solvingquality problems when they arise Associated with this practice are thedangers of such people being used as scapegoats when explanations to seniormanagers are required, or being offered as sacrificial lambs when customercomplaints are being dealt with The responsibility for quality must always liewith operators of the process and the role of QA or any other support function

is clearly to assist in the meeting of this responsibility It should not beacceptable for any group within an organization to approach another groupwith the question, ‘We have got a problem, what are you going to do about it?’Expert advice may, of course, frequently be necessary to tackle particularprocess problems

Having described Utopia, we must accept that the real world is inevitablyless than perfect The major problem is the one of whether a process has thenecessary capabilities required to meet the requirements It is against thisbackground that the methods in this chapter are presented

12.2 Process improvement strategy

Process improvement is neither a pure science nor an art Procedures may bepresented but these will nearly always benefit from ingenuity It is traditional

to study cause and effect relationships However, when faced with a

multiplicity of potential causes of problems, all of which involve imperfectdata, it is frequently advantageous to begin with studies which identify onlyblocks or groups as the source of the trouble The groups may, for example,

be a complete filling line or a whole area of a service operation Thus, thepinpointing of specific causes and effects is postponed

An important principle to be emphasized at the outset is that initial studiesshould not aim to discover everything straight away This is particularlyimportant in situations where more data is obtainable quite easily

It is impossible to set down everything which should be observed incarrying out a process improvement exercise One of the most important rules

to observe is to be present when data are being collected, at least initially Thisprovides the opportunity to observe possible sources of error in the acquisition

of data It may be found that improvements are necessary in the data collectionmethod or the type of measuring equipment itself Direct observation of datacollection may also suggest assignable causes which may be examined at thetime This includes the different effects due to equipment changes, varioussuppliers, shifts, people skills, etc

In trouble-shooting and process improvement studies, the planning of dataacquisition programmes should assist in detecting the effects of importantchanges The opportunity to note possible relationships comes much morereadily to the investigator who observes the data collection than the one whosits comfortably in an office chair The further away the observer is locatedfrom the action, the less the information (s)he obtains and the greater thedoubt about the value of the information

Effective methods of planning process investigations have been developedover the past quarter of a century Many of these began in the chemical,electrical and mechanical engineering industries The principles and practicesare, however, universally applicable Generally two approaches are available,

as discussed in the next two subsections

Effects of single factors

The effects of many single variables (e.g temperature, voltage, time, speed,concentration) may have been shown to have been important in other,similar studies The procedure of altering one variable at a time is oftensuccessful, particularly in well-equipped ‘laboratories’ and pilot plants.Frequently, however, the factors which are expected to allow predictionsabout a new process are found to be grossly inadequate This is especiallycommon when a process is transferred from the laboratory or pilot plant tofull-scale operation Predicted results may be obtained on some occasionsbut not on others, even though no known changes have been introduced Inthese cases the control chart methods of Shewhart are useful to check onprocess stability

318 Managing out-of-control processes

Group factors

A trouble-shooting project or any process improvement may begin by anexamination of the possible differences in output quality of different people,different equipment, different product or other variables If differences areestablished within such a group, experience has shown that careful study ofthe sources of the variation in performance will often provide importantcauses of those differences Hence, the key to making adjustments andimprovements is in knowing that actual differences do exist, and being able topinpoint the sources of the differences

It is often argued that any change in a product, service, process or plant will

be evident to the experienced manager This is not always the case It isaccepted that many important changes are recognized without resort toanalytical studies, but the presence, and certainly the identity, of manyeconomically important factors cannot be recognized without them Processesare invariably managed by people who combine theory, practical experienceand ingenuity An experienced manager will often recognize a recurringmalfunctioning process by characteristic symptoms As problems becomemore complex, however, many important changes, particularly gradual ones,cannot be recognized by simple observation and intuition no matter howcompetent a person may be as an engineer, scientist, or psychologist Noprocess is so simple that data from it will not give added insight into itsbehaviour Indeed many processes have unrecognized complex behaviourwhich can be thoroughly understood only by studying data on the productproduced or service provided The manager or supervisor who accepts andlearns methods of statistically based investigation to support ‘technical’knowledge will be an exceptionally able person in his area

Discussion of any trouble-shooting investigation between the appropriatepeople is essential at a very early stage Properly planned procedures willprevent wastage of time, effort and materials and will avoid embarrassment tothose involved It will also ensure support for implementation of the results ofthe study (See also Chapter 14.)

12.3 Use of control charts for trouble-shooting

In some studies, the purpose of the data collection is to provide information

on the relationships between variables In other cases, the purpose is just tofind ways to eliminate a serious problem – the data themselves, or a formalanalysis of them, are of little or no consequence The application of controlcharts to data can be developed in a great variety of situations and provides asimple yet powerful method of presenting and studying results By this means,sources of assignable causes are often indicated by patterns or trends The use

of control charts always leads to systematic programmes of sampling and