Home      

Home1966212567
 
Title: Design methods compared 1:Strategies

Pages: 32 - 35

            

Author: J. Christopher Jones

Text: Design methods compared

1: Strategies
by J. Christopher Jones
The study of systematic methods of designing has been gaining momentum in recent years. Different approaches have been adopted by different groups in different parts of the world. New techniques have been devised to serve the ultimate objective of solving design problems, if not more simply, then at least more successfully. One approach was explored in detail in Bruce Archer's series of articles Systematic Method for Designers'*. But it is not the only method. What other methods exist, and how do they compare? In this, the first of two articles, the author discusses a number of strategies that may be adopted. Next month he will write about tactics.

(table)
1 accept policy
2 define aim
3a modify
3b new design
4 process group
5 process version
6 number of beds
7 components
8 vessel cross-section
9 bed size
10 heater
11 controls
12 style
13 mechanical standards
14 works design
15 sales method

(caption)
1 A cyclic sequence
The cyclic nature of designing is illustrated by this retrospective record of the evolution of an automatic absorption drier for chemical man ufacture.
The decisions made at each stage are named in the blocks. The cycles, or iterations, appear in the three loops feeding back to earlier stages. The loop between the 'components' decision and the 'heater' decision was found in this case to be difficult to break out of. Each circuit of the loop resulted in an unacceptable rating of the heater if the other variables were kept within known safe limits.
This obstacle was overcome by an intuitive decision to set one variable outside these limits. Experimental work at a later stage showed this decision, or 'insight', to be correct. From The Chemical Engineer(1).

The literature on methods of designing is growing quickly, but it is not easy to read or to put to practical use. Much of it is both vague and dogmatic, and there is little reference to the work of practising designers. Having myself written about designing, I cannot give an unbiased view: but I may be able to save the time of those who do not want to read the million or so words that have been written so far.
Already there have been two conferences in this country: the Conference on Design Methods(5), which was published as a book in 1963, and the Symposium on Design Method of 1965, which is also to be published. There is Christopher Alexander's impressive book Notes on the Synthesis of Form(6), there are books by Osborn(7) and Gordon(8) on creative thinking, and by Asimow(9) on engineering design. Gosling(10), Hall(11) and others have written books on system engineering. Readers of DESIGN will know of L. Bruce Archer's series of articles. These substantial but not always very practical publications include references to a host of articles in journals, and a selection of them appears on page 35.
In attempting to make sense of a variety of systems and methods, I have found it useful to distinguish between the strategy and the tactics of designing. By 'strategy', I mean the sequence of stages which a designer decides to follow. These are dealt with in the present article. By 'tactics', I mean the techniques or tools which he uses at various stages within a design sequence. These will be dealt with in the second article next month.

Design sequences (or strategies)
One of the few points on which many design theorists are agreed is that there are three stages which are common to any design process. These stages are:
1 analysis - breaking the problem into a set of requirements;
2 synthesis - putting ideas together to form complete solutions; and
3 evaluation - estimating or measuring the degree to which solutions satisfy requirements.
When the results of evaluation are negative, the proposed solutions must be modified or rejected. Sometimes the evaluation shows up mistakes in analysis also. So the sequence, analysis - synthesis evaluation, may be repeated
several times before the designer finds out what he needs to know to put things right.
Cycling and re-cycling in this way over the whole problem, or over parts of it, eventually a DESIGN reprint, Council of Industrial Design, 1 5s

(chart)
2
PHASE I FEASIBILTY STUDY

Market info
Needs analysis
Valid
Desired inputs

Step 2 Design Problem
Tech info
System identification
Relevant
Engr. Statement of problem

Step 3 Synthesis
Creative Talent
Design Concepts
Plausible
Alternative solutions

Step 4 Physical realizability
Tech Know!
Physical analysis
Possible
Realizable solutions

Step 5 Economic worth
Econ Factors
Economic analysis
Pay
Worthwhile Solutions

Step 6
Finance fcators
Financial analysis
Money
Set of Useful solutions

Symbology
Process
Info source
Outcome
Decision
Iterative or Feedback loop


PHASE II PRELIMINARY DESIGN

step 1
Experience & Techn
Seletion of design concecpt
Best
Tentative selection

Step 2
Engr science
Mathematical archetypes
Valid
Analytical formulation

Step 3
Math analysis
sensitivity analysis
Which?
Sensitive parameters

Step 4
Tech info
Compatability analysis
Fit?
Adjusted parameters

Step 5
Math & Test
Stability analysis
Stable
Adjusted for stability

Step 6
Math
Optimization
Best
Optimum

Step 7
Trends
Projection into the Future
Time
Behaviour over time

Step 8
Math & Test
Prediction of behaviour
Perform
expected performance

Step 9
Lab technic
Testing
Work?
Test results

Step 10
Experience
Simplification
Better
Accepted proposal

PHASE III DETAILED DESIGN

Step 1
Experience
Prepartion for design
Valid?
Budget & organization

Step 2
Techn know!
Description subsystems
Good?
Subsystem concepts

Step 3
Techn Know!
Description components
Good?
Component concepts

Step 4
Techn KNow!
Description of parts
Good?
Specifications and drawings

Step 5
Techn exper
Assembly drawings
Fit?
Complete descriptions

Step 6
shop
Experimental construction
Producible?
Prototype

Step 7
Lab
Product test programme
Does it work?
Test Data

Step 8
Math analysis
Analysis & prediction
Good?
Detected flaws

Step 9
Techn Know!
Redesign
Better?
improved design

3
training
brief
programming
experience
data collection
analysis
synthesis
development
communication
solution

(caption)
2 Pre-planned linear sequence
There is local feedback within each step, but not between stages. From Asimow2.

(caption)
3 and 4 Preplanned sequence wih feedback
The stages 0 to 8.2 are subdivided giving, in all, 229 sub-stages which are linked into a network of feedback loops. From Systematic Method for Designers 3.

reduces his initial uncertainty to the right level. This is the level at which he is willing to take the irreversible decisions he is obliged to issue as instructions to the manufacturers who are waiting to put his plans into effect, 1.

A fluid process
This description of the designer's thoughts and actions allows us to form only a very fluid picture of the design process.
It is a picture composed of cycles and subcycles which can be linked in many different ways according to the sequence of stages which the designer chooses. A useful visual image is that of a network marking the tracks which the designer follows as he explores the problem and his solutions to it.
Kinds of sequence
This idea of a network which a designer both creates and explores as he searches for a satisfactory design differs from the rather rigid sequences of pre-planned stages which are recommended by some design theorists.
There are, for instance, pre-planned linear sequences composed of a set number of stages, with no cycles or repeats between stages, 2. There are also linear sequences with one or two feedback loops nesting inside each other, 3 and 4. Computer programmes are usually of this type. Then
2 Pre-planned linear sequence
There is local feedback within each step, but not between stages. From Asimow 2.
3 and 4 Pre-planned sequence with feedback
The stages 0 to 8.2 are subdivided giving, in all, 229 sub-stages which are linked into a network of feedback loops. From Systematic Method for Designers 3.

(table)
4
0 Preliminaries
0.1 Receive enquiry
0.2 Evaluate enquiry
0.3 Estimate office workload
0.4 Prepare preliminary response

Phase 1 - "Receive brief, analyse problem, prepare detailed programme and estimate"

1 Briefing
1.0 Receive instructions
1.1 Define goals
1.2 Define constraints

2 Programming
2.1 Establish crucial issues
2.2 Propose a course of action
Phase 2 - "Collect data, prepare performance (or design) specification, reappraise proposed programme and estimate"

3 Data collection
3.1 Collect readily available information
3.2 Classify and store data

4 Analysis
4A Identify sub-problems
4.2 Analyse sub-problems about ends
4.3 Prepare performance specification
4.4 Reappraise programme and estimate

Phase 3 - "Prepare outline design proposal(s)"

5 Synthesis
5.1 Resolve problems about ends
5.2 Postulate means for reconciling divergent desiderata in specification
5.3 Develop solutions-in-principle to problems about means arising from specification
5.4 Postulate outline overall solution(s)

Phase 4 - "Develop prototype design(s)"

6 Development
6.1 Define design idea
6.2 Erect a key model
6.3 Develop sub-problem mutual solutions
6.4 Develop overall solution(s)
6.5 Validate hypotheses

Phase 5 - "Prepare (and execute) validation studies"

Phase 6 - "Prepare manufacturing documentation"

7 Communication
7.1 Define communication needs
7.2 Select communication medium
7.3 Prepare communication
7.4 Transmit information

8 Winding-up
8.1 Wind up project
8.2 Close records

there are branching sequences with several pre-planned activities going on in parallel and occasionally coming together for evaluation or review, 5 This is the characteristic shape of critical path networks of the design process. Some accounts of designing emphasise the random exploration of problems and solutions and the deliberate avoidance of a plan of action, 8. Traditionally, designers are accustomed to exploring only a small part of the problem at a time, keeping in the main to previous solutions that have proved their worth, 9. Another school of thought is that the designer is best left to decide what each stage is to be when he knows the outcome of the stage before it, 10.

Uncertainty and variety
There are two things to remember in assessing the relevance of this variety of design procedures to any particular situation:
1 The amount of exploration which a designer needs to carry out is related to his uncertainty at the start. In novel situations he needs to explore widely: in familiar situations he may be able to jump quickly to a good solution. A good strategy is one which generates information of just the right kind and quantity to reduce his present uncertainty to the level at which decisions can be taken. A poor strategy is one which either leaves the designer guessing or generates more information than he can use.
2 In all but the simplest design situations, there are likely to be several paths through the exploratory network leading to satisfactory solutions, or even to the same solution. The important point is to find any strategy that leads the designer quickly to his destination, and to avoid those paths which take him through a wilderness of information that may be irrelevant to the design aims.
We can see now that a designer should not use any sequence which occurs to him he should choose, or invent, a sequence to fit the problem, the budget, the deadline and his experience. He should look for a sequence that makes the best use of the available sources of information. These sources include his own memory, the memories and writings of others, and any surveys or experiments which he has the resources to undertake. His chosen sequence should oblige him to pose, and to answer, the questions which express the uncertainty with which he begins. (Questions

(caption)
5-7 Pre-planned branching sequence
'Value analysis' is the name of a specialised technique for making red uctions in the costs of engineering components.
5 shows the pre-planned sequence of stages of which value analysis is composed. 6 shows the resulting changes made in revising the design of a rocket motor component, and 7 shows the savings made and the cost of making them. The technique is intended to eliminate the inessential functions of a component and to satisfy the remainder with a smaller number of parts and manufacturing operations.
This is an example of supplementing the designer's thinking and experience by a fixed procedure which allows others (in this case the manufacturing, quality control and purchasing engineers) subsequently to apply their own knowledge and experience. From The Chartered Mechanical Engineer4.

(table)
Cost reduction on the pintle
Cost of raw form
Cost of labour
Total cost ($)
Original $2.31 $5.03 7.34
New $0.59 $1.63 2.22
Gross saving/piece 5.12
Cost of implementation 1100
Savings on 5000 pieces 24500

upon which other questions depend should, of course, be asked first.) He should avoid sequences that generate quantities of information which he cannot use, and he should keep to the cheapest and quickest methods of information gathering that he can get away with.
His measure, in balancing the unreliability of information at his disposal against the cost of verifying it, is the penalty which he expects to pay for making a design error. He may, for instance, decide that the cost of testing alternative designs is less than the penalty of finding that he has gone ahead with a single design that has to be abandoned.

Flexibility and insight
It is equally important that the chosen design sequence fits the designer's inclinations and capabilities. It should place no restraints on his freedom to change his plans in the light of the new information and new insights that he gains as the work proceeds.
For this reason, it is useful to keep two streams of thought (and two sets of papers, the logical and the imaginative) going on in parallel, and to switch from one to the other whenever one feels the need. Occasionally, these streams can be brought together so that the best of each can be combined and the rest discarded, 11.
By now, it should be clear that systematic designing is not the exclusive use of this or that new technique. It is the rational choice by the designer of a strategy or sequence which he has good reason to believe is the best available method of posing and answering the questions that are relevant to his problem.
So designers who are attempting to apply either new or old techniques should continually attempt to justify their strategies, in these terms. They should declare their areas of uncertainty. They should be aware of their reasons for thinking that the sequence in which they are engaged is in fact likely to explore enough alternatives quickly enough and reliably enough to expose at least one demonstrably good solution to the problem.

(caption)
8 Random seqence
In choosing the next step, the designer deliberately ignores the brief, and his previous experience and information. This strategy may be useful in very uncertain design situations which need to be widely explored before taking up any aspect in detail.

9 Traditional sequence
The designer relies as far as he can on reliable information that has been generated in previous design sequences.

(diagram)
brief
re-assess an existing solution
explore a few minor modifications
adjust the existing solution to accommodate modifications
outcome

(caption)
10 Adaptive sequence
The designer exploits, as far as possible, the information which is generated at each stage as he proceeds from uncertainty to decision.

(diagram)
brief
decide what stage 1 is to be
carry out stage 1
outcome of stage 1
decide what stage 2 is to be
carry out stage 2
outcome of stage 2
decide what stage 3 is to be
carry out stage 3
outcome of stage 3
etc

(caption)
11 Logic and imagination
The thinking of designers cannot and should not be held to a rigid sequence of distinct stages: this would curtail the undoubted benefit of uninhibited jumps and leaps of the imagination. On the other hand, the human brain can deal with intricate detail only in small pieces. There can be little doubt that a good designer allows himself to operate both sequentially and by leaps of insight at different times.
The long periods of sequential and logical detail have been described by Professor Robert Mann of the Massachusetts Institute of Technology as "crank-turning" or "grunge", and the shorter imaginative periods as "creative leaps". A good designer keeps these two modes of design thinking apart by an effort of will so that neither inhibits the other. When design thinking is brought outside the brain as a systematic process, these two modes can be recorded on separate sets of paper work.
The results of leaps of insight are kept in one set until such time as enough factual and logical information has accumulated in the other set to test their relevance and value. The 'free-thinking) data are then compared with the 'sequential thinking' data. The next stage can be either the abandoning of insights, because they are seen not to fit the facts, or the abandoning of a logical sequence because it can now be seen to lead in an unfruitful direction. When things go well, both insights and sequential sequence point in the same direction.
ideas, insights or creative peaks that may be unrelated to the current strategy
track of the designer's attention

(diagram)
idea 1
idea 2
idea 3
strategy 1 (logical sequence)
idea 4
strategy 2 (logical sequence) originating in idea 2

(caption)
References

1 S. A. G regory, The Development of an Automatic Absorption Drier, Part 11: Report on the Principles Involved in
the Design, in The Chemical Engineer, No 184, December 1964
2 Moris Asimow, Introduction to Design, Prentice-Hall Inc. Englewood Cliffs, N.J., 1962
3 L. Bruce Archer, Systematic Method for Designers, ColD, 1965
4 H. Davies, Value Analysis as a Working Tool, The Chartered Mechanical Engineer, February 1966
5 J. Christopher Jones and D. G. Thornley (editors), Conference on Design Methods, Pergamon, 1963
6 Christopher Alexander, Notes on the Synthesis of Form, Harvard University Press, 1964
7 A. F. Osborn, Applied Imagination - Principles and Procedures of Creative Thinking, Charles Scribner's Sons, New
York, 1963
8 William Gordon, Synectics, Harper & Bros, New York, 1961
9 W. Gosling, The Design of Engineering Systems, Heywood, London 1962
10 D. Hall, A Methodology of Systems Engineering, D. Van Nostrand Corp, Princeton, 1962

 

 

about        contact        terms of use        image credits        © 2008