The text presented here is not precisely as published by OUP, but modifications are minor. Illustrations are another matter. Where images used in the original book were not my copyright, I have in most cases been able to substitute links to coloured images on the web.

When this text was submitted as part of a PhD thesis in 1996, the Notes were greatly extended. Most readers may prefer to ignore them. They have been collected at the end of each chapter, with internal links leading to them and back to the text. They are a mixture of: simple page references; additional examples or quotations to justify generalisations; and some afterthoughts.

Note, 2003. This was an attempt to summarise the state of the art in 1984. Theory has probably changed since then (I haven't kept in touch) but the ideas are still of value. The next few chapters glanced at the question of creativity. Many readers protested "but, you can't teach creativity!" I understand their scepticism, but it depends on what you mean by "teach". It may be that our creativity level is fixed at birth, or by an early age, but it is certain that most people operate well below their personal best, often as a result of discouragement in childhood. I feel sure the techniques discussed in the next few chapters can enhance performance simply by raising awareness and by giving people 'permission' to try out new ideas and have fun doing so.

Introduction.

Previous chapters have placed the design of structures in the context of political and organizational factors, financial and functional planning, and the demands of architectural philosophy and aesthetics. Mention has been made of the technological context of safety, reliability, and performance criteria and the constraints imposed by the nature of the available site, the potential hazards, the imposed loadings, the available materials and the practical methods of fabrication and construction. In view of this enormous complexity, how does the designer come to grips with a problem and evolve a form of structure which takes into account every relevant factor?

The answer is, of course, that he does not. There exist in the profession certain recognized ways of dealing with the usual problems of structural design and the average engineer simply uses these. At any given time these recognized standard solutions may be out-of-step with trends in social needs or in economic conditions and may fail to take advantage of new developments in technology and theoretical understanding. However, on the whole, designs are produced which are satisfactory to society as far as it is aware of the possibilities.

On the other hand the gradual change which is seen to occur in the collective wisdom of the profession is inspired by those who produce original solutions which take account of changing conditions and opportunities. How does such 'creative' design take place?

Many designers feel that the process of creative design is inexplicable and that it must occur in the unconscious mind. They feel that even to try to discuss it will somehow stultify it. [Note 1.] Others are liable to take a more pedestrian view. When faced with a problem they normally look up a text book for a standard solution, or consult journals to find out how other people have tackled the problem in the past. To them design seems a straightforward process.

In between these two extremes come others who are willing to speak of a 'flash of inspiration'. They feel that after careful investigation of the problem, their subconscious mind begins to sift the factors involved and to produce possible solutions which they can then subject to conscious scrutiny. They thus think of design as a sort of interplay between conscious and subconscious processes.

In recent decades a great deal of effort has gone into analyzing and modelling conventional design methods and developing improved procedures to cope with the complex and open-ended problems found in fields such as aero-space, mechanical engineering and architecture.

The main characteristics of the design process.

Design in its widest sense is defined by the theorists as the translation of a perceived need into a solution to that need; bringing about a change of state; or moving from one situation to another. First, let us look at a thumb-nail sketch of the design process to identify its major features.

In an engineering context the perceived needs must be translated into engineering terms. Thus 'we need to get across the river' becomes 'we need a ferry' or 'we need a bridge'. Notice that immediately, in the definition of the problem, a broad solution has been proposed.

However, there are various types of bridge; suspension, arch, pier-and-beam. As soon as each one is proposed, new facts emerge about the need.

Fig. 13.1. In finding a structural solution to the problem of crossing a navigable river, the greater clearance and safety of a suspension bridge must be weighed against the greater cost. Topography and foundation conditions are other important factors. (a) Severn Bridge, Great Britain, 1966 (Engrs: Freeman Fox and Partners). (b) Medway Bridge, England, 1963 (Engrs: Freeman Fox and Partners).

'If we have a pier and beam bridge, we shall restrict the shipping channel; we need a wide channel and adequate height to clear the masts of ships'.

A suspension bridge seems clearly indicated, but suspension bridges are more expensive than other types and, if the terrain is flat, require long approach ramps. They also require suitable foundations for the high concentration of load at the two major supports.

'But our finances are limited and to make matters worse the approaches are heavily built up; it would be costly and politically difficult to acquire the necessary land. Let's have another think about it. Perhaps the ships will have to accept a restricted channel'.

Already we can see the major characteristics of the design process. It starts with the vague sensing of a need. This is followed by a definition of that need in engineering terms; the postulation of possible solutions; the examination of their merits and demerits and their cost; and, often, a return to the starting point in the light of this new knowledge to examine whether the initial definition of the problem should be modified.

If the problem is re-defined, a new line of investigation is opened parallel to the former. This further increases knowledge of the situation. It may lead to final resolution or to back-pressure for yet another re-definition.

As time progresses the design process appears to cycle. A number of cycles may occur before resolution.

The other major feature demonstrated in the sketch is the conflict which arises between the various requirements. A costly high-level bridge will place a greater burden on the tax-payer and cause more inconvenience to landowners and users on either side. A lower level bridge, rising to a narrow navigation span will inconvenience shipping and present a finite level of risk of collision. There is thus no one solution that will satisfy all the requirements. The corollary is that we must then choose between a number of possible solutions, each having advantages and disadvantages which must be weighed against each other on some subjectively established common scale of value.

The requirements, stated in engineering terms, define a 'design space'. As we have seen, each standard solution (e.g. suspension or girder bridge) will satisfy each requirement to a different degree and some of them not at all. A hypothetical 'perfect solution' would satisfy all requirements to the full. Such a situation would be defined as perfect 'fit' between the solution and the problem. In real design we can therefore talk about the 'degree of fit' achieved, a good solution achieving a higher degree of fit than a poorer one.

In traditional, non-industrialized societies, whose technology and needs change very slowly, a very high degree of fit is considered to be achieved in areas such as shelter and transport. Jones cites the traditional farm wagon as a prime example of this harmony in which every design feature fulfils several objectives at once, without interfering with those of other features (Jones 1970, pp.17-20). Such harmony is achieved slowly over many years by a process akin to biological evolution; a series of small alterations occurring regularly and being rejected or surviving according to the test of practical application. A significant and under-rated factor in this largely unconscious process is probably the gradual redefinition of needs and expectations to accord with what is possible in terms of practicable economical solutions.

As suggested in the introduction, what actually goes on in most design offices at the present time is very similar to this process, though the rate of change is greater and the process is more self-conscious.

The variety of practical situations.

Before discussing conventional design in more detail it is necessary to emphasize the variety of the phenomena that for convenience we are going to treat as an entity.

In the average small, harried consulting firm, designing factory sheds and two-storey commercial buildings, it may be possible for a single engineer to handle one or more jobs on his own. Although complexity in analysis and design is not directly proportional to the size of project, the problems of communication will be much less than on a large project. The need to deliver reliable designers quickly and to maintain a satisfactory profit margin will encourage the use of standard solutions for what will be perceived as standard problems.

When a large government instrumentality or consulting firm sets about the design of something the size of a power station the situation is quite different. Internal organization and communication becomes as large a problem as external relationships, with the organization divided, perhaps, into a planning section, an architectural section, a mechanical and electrical section and a structural section. Each of these may be further sub-divided. The structural design will be split between several engineers, each responsible for a part of the structure: the machine hall superstructure and overhead crane runways; the generator supports; the inlets and outlets for water; the control building, etc.

In such a case the chief engineer has to exercise what might be described as a form of generalship. Several alternatives may be examined to a fair depth and carefully assessed for cost and other attributes before a choice is made. On the other hand, the very size of the organization and its inevitably hierarchical nature tend to make it inflexible as problems are encountered in the detailed design stages.

A similar difference exists between the innovative project such as the Sydney Opera House roof in which applied research and model investigations may be required, and the design of routine factory sheds and medium-sized office blocks.

Despite these differences it is necessary for the sake of brevity to speak of 'design' as if it were a single entity.

So far the term 'client' has been used to signify the person or organization which owns or controls the project and normally briefs the architect. In the ensuing discussion of the design process in structural engineering the term 'client unit' will be used to refer to the person or administrative unit which briefs the structural designer, and to which he must respond with design proposals. As well as the owner, this could equally be the architect, or the planning section of a government instrumentality passing on a project for detailed structural design, or simply a senior structural engineer briefing one of his assistants. The individual or team responding to the briefing will be called the 'design unit'.

The strategy of design.

The first question the designer must ask himself when confronting a problem is 'where do I start?'. The next is 'where do I go from here?' and the final one is 'when do I stop?'. The answers constitute his strategy in tackling the problem. Strategy may be seen as consisting of two basic elements, 'Direction' and 'Balance'.

In small, routine projects, the choice of starting point for optimization within the technological sub-cycle is probably not very critical. Real difficulties arise when the larger, interdisciplinary design cycle is considered.

Since it is impossible to consider, or even be aware of, every relevant factor at the outset, the designer must start by selecting a manageable group of factors which he considers to be the most significant and produce an initial response to these. He must then check whether this proposal, or some modification of it, will satisfy the remaining requirements.

The initial selection will depend heavily on the interests, training and philosophy of the individual. An engineer approached directly by a client to design a factory would, perhaps, accept without question the client's specifications concerning the requirements of function, would fit around this a form which was ideal for carrying the loads, would think next about the cladding and internal environment, and lastly see if he could make it look presentable. An architect would be more likely to view the situation firstly in terms of visual aesthetics, secondly in terms of function and lastly in terms of structure. In neither of these cases is a global optimum likely to be attained and the attempts of each expert to achieve an optimum in his own area are fairly meaningless.

Lawson (1980) sees such biases as natural and inevitable. However, the best strategist is obviously one who maintains a wide vision and, from experience or knowledge, senses which are likely to be the dominant factors in a design so that a start can be made by working from them to the lesser factors.

The process of 'direction' in design strategy continues with the making of similar decisions at intermediate stages as the design progresses. Sometimes the designer must choose between two or more lines of attack which appear equally plausible. In this case he may, as we have seen, pursue them simultaneously for a certain distance, so that he can make his decision in the light of improved knowledge. At some point, however, he must decide to concentrate on a single one.

If a particular approach proves unrewarding or demands more effort than appears justified he must decide whether to press on or abandon it. When an impasse is reached he must decide which needs are important and which can be sacrificed or only partially met in the new approach. This is characterized as a choice between 'Evolution' and 'Revolution'. As usual the designer must make these decisions on the basis of limited data.

As the problems associated with the form of the Sydney Opera House roof became apparent, the client could have rejected it or Arup could have declared it impossible to achieve. There could have been a retreat to a 'safe' conventional, well-proven solution or another attempt at an equally bold one. The course adopted was intermediate between the two; a radical change from the proposed structural concept which permitted retention of an approximation to the original geometry.

These considerations introduce the question of Balance in design strategy. They occur in factories as much as in opera houses. Should trial designs be made in both steel and concrete, or is one 'obviously' more suited than the other? Is it worth considering timber or brick? Will excessive cost 'definitely' rule out a client's desire for column-free space in his building?

All these questions could be answered with reasonable accuracy by investigation, but time and money are limited. Cycling in the design process is most desirable, but there must come a point at which the diminishing returns no longer justify the effort likely to be involved. Furthermore, the alternatives are liable to prove, after several months' investigation, to have led to their own problems and to be no better than the original proposals.

At some stage, therefore, an informed guess must be made that the imperfect solution which has been arrived at is the best that can be achieved under the circumstances. It has to be accepted that there will always be an element of uncertainty in these decisions. Like the old-timer's gold strike, the 'perfect solution' may be just over the next hill!

In view of this, a good chief designer has been described as one who knows when to call a halt to the exploration of a problem and is then able to define his goals and motivate everyone in his team to press forward in the same direction. It is, however, essential that the decision to halt cycling be conscious and rational as far as the inherent uncertainties of the design process permit. There are a number of factors in conventional design practice which tend to restrict cycling even when it is appropriate.

Factors which tend to restrict cycling.

The most common factors which restrict cycling are budget limitations and time deadlines imposed by the client unit. These are in most cases a practical necessity, allowing the client unit to make plans for the financing and commissioning of the project. However they have the disadvantage that they may force the design unit to operate entirely within a sub-cycle and achieve a local rather than a global optimum.

Notice that each unit may act in turn as a design unit, as it receives broad instructions from higher up the organizational tree (Fig. 13.2), and as a client unit when it defines sub-problems and hands these over for solution to units further down the line or in another branch. Figure shows arrangements within a large organisation, but as we saw earlier, similar relationships occur in the less structured co-operation between a client and his consultants.

Fig. 13.2. Formal resolution of disputes in a large organisation may involve several levels of the hierarchy. Co-operation is usually more direct.

Problems of communication arise if the unit which defines a problem is several rungs up the ladder from the unit which experiences difficulties in design resulting from that definition, or is situated on a different branch. The new information concerning the design space may become garbled or be suppressed before it reaches the client unit.

It is difficult under these circumstances to achieve rational Balance, although the greater the problems experienced, the greater will be the back-pressure for re-definition. Further inertia results from the fact that problem definition is a more mentally demanding task than goal-centred design and the tendency is to struggle with the immediate problems in the familiar way rather than retrace one's steps and start again. There is also an attitudinal resistance in that engineers see themselves as people who 'get things done'. Each unit desires constructive progress and at some stage will prefer to go ahead with a reasonably satisfactory answer rather than pursue a more perfect one. Cycling is liable to be seen as 'dithering' regardless of the actual balance involved between effort and potential reward.

Finally, pride, prejudice and mental inertia are liable to restrict cycling by interfering with the proper reception and use of new information and ideas. It is these human factors which the young engineer finds most mystifying and infuriating in the design process.

The twentieth century vernacular.

The pressures listed above normally reduce the amount of cycling in design to a level below the optimum. As we have seen this is achieved by the development of a conventional wisdom about the way in which certain engineering problems should be formulated and solved.

The simplest reaction on the part of the unit presented with a problem is to say 'yes, this is a standard problem and there are several standard possible solutions'. If it has encountered a similar problem before, the unit will look up its previous computations and drawings and adapt them to fit the new situation. If it has not tackled a similar problem, but knows that it is a fairly common one, it will search text-books and journals and possibly make a tour of the surrounding district (or even the world) to see how other people have solved the same problem. This is how the 'twentieth century vernacular' is created. Developments, when they occur, are generally achieved by small modifications to existing practice.

The method has obvious strengths, but some deficiencies. One is a lack of feedback because faults in buildings are often tolerated by users who know no better. Designers are naturally reluctant to publicize their own failures and, particularly when they occur in the serviceability area, are liable to overlook them. Perhaps the major drawback of the twentieth-century vernacular is its inability to respond to rapid developments in technical knowledge and the speed with which the demands of society are changing.

Dissatisfaction with existing approaches has therefore resulted in increased attempts to analyze and model the design process.

Symbolic and verbal models of the design process.

During the past two decades much emphasis has been placed on the development of a design methodology for the fully conscious generation of new ideas and solutions. [Note 2.] The direct impetus for this burst of activity arose in America following the launching of the first Russian space satellite. However, there are more fundamental pressures at work. These are due to the preference of the technologist for logical and repeatable processes. He likes to be able to justify proposals to clients and colleagues; to be sure that, within reason, all likely avenues of approach have been considered; and to prove that the solution adopted is reasonably close to the practicable optimum.

The major difficulty for the individual in conscious creative effort is that of assembling and retaining all relevant items of information in the forefront of the conscious mind, analyzing the inter-relationships between them, and assembling constructive proposals from them without forgetting or obscuring a great deal in the process. It is also very difficult to supplant the first idea to arise or to subject it to proper criticism. This is why designers often 'put aside' a problem for a day or two, so that their subconscious pride in their first idea can subside.

The aim of much design methodology is to counter these difficulties by using some sort of symbolic shorthand to record and manipulate the relevant concepts and their relationships on paper, or by means of a computer.

Further impetus has been given to methodology by the need to design across the boundaries of professional disciplines. If inter-disciplinary teams are to work efficiently it is essential that as much of their design strategy as possible be conscious and communicable, so that all relevant information is pooled and all can share in the search for the global optimum. Otherwise the final design will be achieved by an amalgam of local optimums.

Mechanical engineers and architects have traditionally been more concerned with Design Methodology than structural engineers, perhaps because their problems are less 'structured'! For this reason most of the texts and conferences have a mechanical bias or are concerned with architectural and environmental design. There is an unfortunate lack of examples in the field of structural design.

The first step in Design Methodology is to devise a model representing current practice or an idealized version of it. All such models employ the concept of 'phases' in the design process such as 'Problem Definition', 'Postulation of Solutions', 'Evaluation of Alternative Solutions', and 'Choice'. The definition of these phases tends to be imprecise and it is questionable whether they do in fact exist as separate entities. However without these concepts it would be impossible to even begin to analyze the design process.

Early models of the process were often linear. The number of phases varied considerably. Alexander (1964) defined only two; Analysis or 'decomposition' and Synthesis or 'realization'. Krick (1969) listed five: "problem formulation"; "problem analysis"; "search"; "decision" and "specification". Koberg (1980) compares 22 different formulations of the design process.

The concept of the 'decision tree' figured largely in the early theories (Fig. 13.3). The basic problem is supposed to be solved by a number of proposals, each of which gives rise to a number of sub-problems. The solution of each of these in turn produces new, more detailed problems, until finally every sub-problem is amenable to a simple, standard response and a complete solution is achieved. The generation of new ideas was seen to work in the other direction, starting with a number of disparate elementary concepts which were combined into sub-groups and finally into a complete solution. This perception of design as a process of synthesis towards a clearly defined goal has been traced back as far as Aristotle. Despite the linearity of these models, many authors did recognize the importance of feedback, as the designer's efforts provide more knowledge of the nature of the problem and its possible solutions.

Fig. 13.3. According to the 'decision tree' concept of the design process, a problem is split into successively smaller sub-problems until a solution has been found for each.

The term 'cycle' is often used in the engineering context and in others such as administrative 'problem-solving' and architecture. However a circle (or cycle) implies no progress and the term 'spiral' is becoming more popular.

It is surprising that the analogy of the maze is not commonly used in design theory. As illustrated in Fig. 13.4 this can model the false-starts, the back-tracking, the lateral searches, and the existence of more than one point of exit from the design process. It also models the previously mentioned fact that where one commences looking has a strong influence on where one ends up. Lawson (1980) uses the analogy of a person lost in a forest and trying to find his way out to convey a similar impression.

Fig. 13.4. A maze model of the design process simulates 'false starts', 'blind alleys' and parallel solutions.

The early theories of the design process, which assumed that problems could be broken down in a purely objective manner into their constituent elements, led to design methods which studied the relationships between these sub-problems and their possible solutions, and proposed a linear process for the assembly of the complete solution.

These were virtually ignored by practising designers on the grounds that they were unrealistic, took longer than conventional intuitive methods and were no more efficient.

A crisis occurred in the 'design-methods movement' and a 'second generation' of design theories was evolved which took account of the true nature of real problems as described in earlier pages and admitted the possibility of a creative moment of synthesis within the design process. Alexander, whose Notes on the Synthesis of Form (1964) is still quoted as a definitive text renounced the concept of design method.

A large part of the movement was repelled by the totalitarian overtones of comprehensive design methods which they feared would give the 'expert' architect or planner unwarranted power over the lives of tenants and citizens. This reaction contributed to the pressure for 'participatory design' in which the opinions and special knowledge of the users are given prominence in decision-making.

There is now [1984] a feeling that the movement has entered a third phase in which a more mature assessment has been attained concerning the advantages and disadvantages of the methods proposed so far. Some of these will be illustrated in the next three sections in which the process will be assumed to consist of three phases, each of which will be discussed in turn.

Notes.

Note 1. Frank Lloyd Wright was one of those who refused to take part in MacKinnon's study. (Broadbent 1973, p.9 and Barron 1969, p.59.) MacKinnon's paper appears in MacKinnon (1962). [Return.]

Note 2. ["The past two decades" means prior to 1984.] Examples were Bazjanac, pp. 3-20 in Spillers (1974) and Broadbent (1973 Chapter 13), (1979), (1980a), (1980b). [Return.]

[Top.]

The art in structural design navigation:
[Chapter 12.] [Chapter 14.] [Contents] [References]

Site navigation:
[Aesthetics of built form.] [Papers.] [Work of Jörg Schlaich.]
[John Monash's early engineering.]