Title: Design methods compared 2: Tactics
Pages: 46 - 50
Author: J. Christopher Jones
Text: Design methods compared 2: Tactics
by J. Christopher Jones
The expanding body of knowledge on systematic methods of designing can well create confusion and mistrust. In these two articles, the author describes a number of approaches and techniques which have been developed, and discusses the purposes for which they may be used. Last month he compared a variety of strategies-is, the sequence of stages which a designer decides to follow. In the following article, he goes on to talk about the tactics of designing-the techniques or tools which a designer uses at various stages within a design sequence.
The traditional techniques of industrial designers are the making of sketches, scale drawings, perspectives and models. Those of professional engineers are calculations, flow diagrams, cost analyses, tests and the like, which are intended to expose aspects of the problem that are not visible in a scale drawing. To this range of techniques one can now add a number of systematic methods of expressing the thoughts which designers traditionally keep to themselves.
The benefit of expressing design thinking systematically in terms of 'maps', or navigational aids', is to make the early stages of the design process accessible to many people instead of restricting it to an experienced few, .The main disadvantage of these techniques is that they introduce a lengthy and tedious stage before work can begin on the drawing board. However, this disadvantage of systematic methods can be regarded as the first step in overcoming the two major weaknesses of the scale drawing as a design tool:
1 A drawing can only be done by one person at a time; and
2 It takes a long time to do.
In applying systematic methods, we widen the range of experience that influences a design but add even more time to the process. However, these notations or languages are necessary steps to the transfer of the slow and tedious parts of designing to a computer, 3 and 4. Once this is achieved, the total design time - both thinking and drawing shrinks to a fraction of what it is at present.
Clearly we should be preparing ourselves for the time when, with the aid of computers, a team of design specialists will be able to design a product from inception to manufacturing instructions between breakfast and lunch!
Externalising and automating the design process Professor Moris Asimow of the University of California breaks down the process of engineering design into the four stages which are shown along the time axis. Professor Robert Mann of MlT has pointed out the large increase in the number of persons employed as a conventional design proceeds.
Both systematic designing and design automation tend to flatten this curve. The externalising of design thinking at the early stages widens the search for ideas and information and allows more people to be employed before the critical decisions are made. So there is more chance of spotting design errors at a stage when the cost of correcting them is low. The risk of expensive redsign at later stages is reduced.
Systematic methods applied at a later stage reduce the cost of subsequent work eg, value analysis, which cuts production cost, or reliability testing, which cuts maintenance cost. Alternatively, the systematising of later design stages permits the use of less expensive people, or eliminates human thinking altogether, by the invention of wholly determinate sequences which can be automated - eg, computers for drafting and for circuit design.
Widening the area of search Most of the systematic methods that have been published so far tend to widen the area of search for solutions to a design problem or for information that is relevant to it. Several of these methods are described below.
The checklist is perhaps the simplest systematic technique, and the most widely used.
A checklist is a set of prepared questions to which the designer seeks answers. The aim is to make it impossible to overlook one or more of a large number of requirements which a product may have to fulfil. A checklist inevitably restricts the designer to a range of alternatives that has been worked out in advance: but it offers him a valuable safeguard against repeating minor blunders that others have discovered by hard experience, s.
This unfortunate term refers to a simple method of expressing the ideas and experience of a group of persons who have been called together at the start of a design problem.
It is a means of collecting many uninhibited ideas in a short time. The essential point is that the leader enforces the rule of 'deferred evaluation'. No criticism is allowed until the meeting is over, and everyone is made to feel secure in mentioning the thoughts and ideas that they would normally reject as impractical or irrelevant. There is evidence that this method does not produce better ideas than solitary thought, but it is certainIy worth doing because it is so quick. Unfortunately, the widening of the area of search is so great that convergence at later stages becomes difficuIt.
A useful modification of the method is to ask each person to write down before the meeting everything that he thinks, or sees, or reads, that appears to be in any way relevant to the problem in hand. He writes each idea, fact or question on a separate card. At the meeting each person reads out a card in turn and others contribute any ideas which are suggested by what is read out.
It is usual to get about 150 items in about half an hour from a group of four to six people.
This term is used by William Gordon to describe a method of harnessing not only random thoughts but also large segments of memory, bodily function, brain, language and experience to a design problem.
The method is intended to uncover analogies that correspond closely to the problem and, at the same time, restructure it in a radical new way. A synectics session
2,3 and 4 Design automation and computer-aided design The time taken to make scale drawings is a large part of the time taken to develop and make a new product. Already there are computers which can make routine drawings without human intervention, 2. Here, 2a is computer-drawn; 2b is made by a human draughtsman. There are also computer programmes capable of selecting from a vast number of alternatives a single 'optimum' design which best suits a pre-selected list of requirements.
This can be done automatically if the effects of varying each design dimension are known in detail. This condition is satisfied in such design problems as finding the best size for turbine blades and rotors, 3 (by permission of D. C. Boston). 4 shows Sketchpad, a computer controlled electronic drawing board at MIT which turns rough sketches into finished drawings, and which enables a designer to explore alternatives, make calculations and take decisions at much increased speed. This 'design automation' becomes decreasingly possible as we move from the certain ending of the design process to its uncertain beginning. Therefore design automation affects only the right hand side of '. 'Computer-aided design' (CAD) is potentially the means of extending the use of the computers to the left half of 1.
begins with a statement of the problem as given (PAG). This is analysed in discussion. Then there is a 'purge of obvious solutions', which clears the participants' minds of ideas that are easy to think up and which are unlikely to be more than permutations of what already exists. The points of real difficulty are then defined (PAU or 'problem as understood'). The chairman asks an 'evocative question' which calls for a solution in one of three main types of analogy.
The first of these is personal analogy, in which the designer identifies himself with the object or situation in question and imagines how he would react. The second kind is direct analogy, in which biological or other solutions to similar problems are examined. The third kind is symbolic analogy, in which the crucial or unclear parts of the problem are symbolised in any words, pictures or other images which suggest themselves. The group plays, in a leisurely and easy way, with each 'evocative question'. If a fruitful analogy is generated, its implications are examined in detail in relation to the problem as understood. Otherwise, the chairman re-directs discussion by posing another evocative question.
The virtue of synectics is that it submits a problem to re-structuring by unexplained processes of body, brain, language and perception that lie outside the scope of more deterministic methods of design.
5 A checklist The checklist is the best known, the simplest and perhaps the most practical aid to systematic designing. This example was prepared by the writer for the layout of displays and controls.
Checklist for the layout of displays and controls
1 What are the user's purposes or objectives ?
2 By what actions will the user attempt to achieve these objectives ?
3 Which actions are particularly important and which of little importance? (lmportance can be assessed by the
probability of great or small consequence of FAILURE to carry out each action.)
4 Which actions require continuous vision and which actions require only occasional glances ?
5 Which actions involve simultaneous vision of two or more items and which involve vision of only single items ?
6 What is the duration of each action?
7 What is the frequency of each action ?
8 Which body measurements are critical for each action ?
9 Which actions are likely to be fatiguing ?
10 Which actions call for the maximum forces the muscles concerned can exert, or for a particularly light touch ?
11 Which actions require an awkward grip or posture that greatly reduces the force which can be exerted or the distance that can be reached ?
12 Which actions are in awkward positions or directions in relation to the body ?
13 What degrees of body motion are compatible with carrying out the actions with the required efficiency and with a tolerable degree of comfort ?
14 Will the users be unfamiliar with the actions required, or highly skilled in carrying them out ?
15 Will a large proportion of users be encountering new equipment for the first time, and are they likely to compare it unfavourably with equipment with which they have become familiar through long experience ?
16 Will the users expect, or be very appreciative of great comfort and convenience, or will they be willing to tolerate considerable discomfort and inconvenience ?
6 Brainstorming Some of the ideas collected at a brainstorm session at which the problem was the layout of car controls.
6 Project: to systematically explore the possibility of improving the layout of car controls (assume hydraulic electrical and other non-mechanical linkages that are independent of control position). Assume existing car bodies, engines, transmissions and brakes, etc.
A random list of factors was compiled extending to all factors thought to influence the design, however obvious. This list also included additional factors arising from discussion of the original collection.
Little used controls to foot- wipers, start, etc
Visual display of neutral
Displays should give additional warning of critical conditions
All displays easily seen
Non-slip surfaces for controls
Controls should not interfere with vision
Steering wheel should not interfere with vision
Non-continuous steering rim
Have handlebar-type steering ?
Why a steering column ?
Have non-slip surface on steering device
Steering device should be adjustable for different people
Can left and right hand be given equal work ?
Steering should be easily possible with one hand
Depressable foot buttons to eliminate metal fatigue, etc
Auto emergency stop
Why have both foot and hand brake ?
Difference in foot size
All controls within easy reach
Pressing flexible floor for stop
Safety interlock on hand brake
Are foot controls separate enough for safety ?
Should hand brake be on floor or not ?
Displays should be capable of being read at a glance
Knock-off hand brakes
Foot operated hand brake
Pedal movement directions
Controls adaptable for left and right hand drive
Automatic driver -'George'the auto pilot
Controls grouped in logical order
One foot driving
No reversing gear-turn wheels back
Interchangeable dual control A-B/B-A
What type of car are we considering ?
Position of horn ?
Ashtray located on vertical control panel bad
Access for maintenance
Sufficient control range
Minimum number of controls
Easily adjustable to meet differing driver sizes
Lights and brake on steering device
No undue strength required to operate controls
Facing back with forward tv viewer (or mirror)
Aero engine and propeller
Existing classification of functions
Buying and Selling
Availability of goods
Power to engine
New classification of functions
Stiffness of edge
Control of angle
Control of Strokes
Maintenance of edge
Selection of goods
Payment for goods
Replacement of goods
Power to accelerate air
Power to expand air
Edges of clamp
Turbine in expanded air
7 Reclassification and innovation It appears that innovation depends upon a re-grouping of physical components and an underlying reclassification of functions. Three innovations, the safety razor, the supermarket and the jet engine, are compared in the table with their antecedents. In each case a new set of physical components has emerged. The functions assigned to each component are listed in the two inner columns.
The act of reclassifying and redistributing functions appears to be the essential step in innovation - and the least easily explained. The more widely one searches for information, the more difficult, the more personal becomes the achievement of a feasible reclassification. With external aids such as computers and systematic techniques, this stage is likely to become even more demanding of the human designer than it has been in the past. From Symposium on Design Method'.
Classification of design information
A major difficulty of a novel design problem is that of breaking it down into manageable and separate pieces.
A quick way of trying out alternative breakdowns or classifications is to sort the cards obtained by brainstorming into categories. Each person should do this by himself, choosing a category for each card as it comes or putting it into a category that he has already chosen. The group then meets again to select one of these classifications as a basis upon which to divide the work for the next stage.
Often there will be acute disagreement about the choice of classifications because of the personal element that is inseparable from the activity of perception (the imposing of a pattern on reality). An inventive designer can, atthis stage, introduce a novel breakdown of the problem that will lead eventually to a breakthrough, a. The important points are that few items appear in more than one category, and that each category corresponds to a physical component that is technically and economically feasible.
Matrices and nets
The 'interaction matrix' is a useful tool for checking the presence or absence of connections between a set of items, 8. The matrix shown here is used to investigate and record the need for any of three degrees of proximity between rooms. The main advantage of this method of 'thinking outside the brain' is that the chance of omitting any one of many possibilities is greatly reduced.
The interaction net, 9, is a means of displaying a picture of the problem that has been previously investigated and recorded in an interaction matrix. Each spot on the matrix is translated into a line of the net. Each item of the matrix is translated into a node or block. The resulting network gives the designer a clear picture of his problem, and makes it easy for him to eliminate design proposals which do not satisfy the requirements that are portrayed.
Design-by-parts (the 'in-out' or 'multiple concept' method) The principle here is to split the problem up into subfunctions or pieces, to find a range of acceptable solutions for each piece, and then to construct a new design out of the sub-solutions. The underlying assumption is that interactions between pieces of the problem can be safely disregarded or can be
8 Interaction matrix A useful tool for checking the presence or absence of connections between items.
9 Interaction net A means of portraying the pattern of interactions that are recorded in the matrix shown in a (the matrix link between the consulting room and office is omitted from the net). 8 and 9 are the work of Alan Murray and Derek Middleton of the CEGB architects' development group. These diagrams are reproduced by permission of the CEGB from its design guide for medical centres.
8.5. Morphological chart of means of obtaining thermal comfort
A Air temperature
B Mean radiant temperature
C Air movement
E Temperature gradients
Warm air from a central source
High temperature electrical heating
No special provision
By disposition of appliances
Air heated by a convector in room
High temperature from flame
Air heated by a convector radiator in room
Low temperature fluid carrying panels
Incidental from radiant source
Incidental from surfaces warmed by convection
Morphological chart A re-assessment of the functional elements of room heating on the left, and a set of alternative means of satisfying each function on the right. Taking one solution from each row gives a complete solution to the problem. The total number of solutions here is 4x5x4x2xl =160. J.K.Page calls this tactic "planned re-learning in the framework of a system that forces divergent thought rather than convergent thought". Chart reproduced by permission of Norris Brothers Ltd. consulting engineers, and B.P. Trading Ltd.
mutually reconciled at a later stage.
This assumption is a reasonable one when the pieces correspond to separate physical units such as pumps, heaters or buildings, and when the connections between pieces are all identifiable pathways for the flow of people, vehicles, materials or energy (eg, roads, pipes or cables). An elementary example of design-by-parts appears in to. This is a 'morphological chart' showing several ways of satisfying each of the functional pieces of a heating system. It was made in the course of a search for new methods of house heating. This kind of analysis is easier to manage if each function is independent. Sub-solutions A3, A4, B5 and E1 are examples of dependence between functions.
A more controlled and realistic way of designing by parts is to describe each functional piece by a performance specification and to carry out rough tests to determine the limits between which lie a range of acceptable solutions, 11-13.
Design-as-a-whole (the 'out-in' or 'single concept' method)
The alternative to design-by-parts is to operate on a single concept or model in which the interdependence of parts is clearly expressed.
The scale drawing is a model of this kind. It allows the design or to explore the geometrical interference between components much more accurately than he can do in his own mind, and to adjust repeatedly the shape and size of each component until a satisfactory whole is achieved.
This method of beginning with a single overall concept, which is adjusted in form as various constraints are relaxed or applied, is the basis of conventional design procedure. Our reason for wanting to get away from 'design-as-awhole' is that the model we use, a drawing or other representation of a complete solution, provides information about shape but not about function. It does not portray the pattern of functional interference which any particular geometry implies; and it provides no measure of the degree to which different functions are shared by the physical parts. This question is usually dealt with by a mixture of experience and mental juggling.
Christopher Alexander, in his article A City is not a Tree 3, has suggested the semi-lattice as a means of displaying the allocation of overlapping social activities to the parts of a city, 16. Possibly the use of semi-lattices as