Designs of physically large and complex (PL&C) systems, nowadays, are achieved through the use of evermore capable digital computer-based techniques. Thus, the process of such designs might be characterized as the practice of science rather than that of an art. The article commences with a consideration of art and science in design. It then addresses the particular nature of the design of such systems and how this is not just an issue of complexity, but also a consequence of large physical size. How computer-aided design is applied early in the designing of such systems, the crucial aspect of the choice of style by the initial designer and the advent of computer-based simulation techniques, applied early in design, are all considered pertinent to the role of art and science in design. A series of high-level fundamental issues are discussed in the belief that they are changing the nature of the design of PL&C systems and ought to be considered by practitioners of such designs. In this way, the power of computer-based techniques, both numerical and graphical, can then enhance the scope of design innovation, given designers' increasing dependency on digitally based practice.
This article follows an earlier article on design methodology for ships and other complex systems (Andrews 1998). A subsequent article then addressed the manner in which simulation could now be undertaken very much earlier in such designs than is currently generally practiced (Andrews 2006a). The latter proposal was seen to be possible because of the realization of the earlier methodology emphasizing the physical architecture of such systems in their initial design. The methodology has been demonstrated through this ‘Design Building Block’ approach to such designs over the last decade and so provides an opportunity to take stock. In particular, there is seen to be a series of issues fundamental to the practice of designing such physically large and complex (PL&C) systems. These issues typically encompass the nature of design synthesis, the use of the techniques of simulation and optimization and how best to exploit the graphical as well as the computational power of digital computers in design. It is considered convenient to address such issues by exploring the nature of art and science in the design of these PL&C systems.
In considering the nature of science and art in design, all three terms need some clarification. It is ‘often said that engineering is as much an art as a science’ (Rogers 1983). In addressing creativity and engineering design, Rogers noted that the Greeks had no word for ‘art’ considering the range of craft practice to include the arts. By the Enlightenment, a distinction existed between the ‘fine arts’ and the ‘useful arts’, raising artists above mere craftsmen and women. Rogers further clarified that ‘craft is the power to produce a preconceived result by consciously controlled action’, whereas ‘art is seen as more than pure technique’. Conversely, it has been recognized that there is an aesthetic or artistic element in general creativity, which can include the creation of a scientific or mathematical breakthrough. This insight was caught by Miller, in considering imagery and creativity in science and art, such that ‘Artists as well as scientists seek aesthetic representations of nature and at the nascent moment of creativity, boundaries between art and science dissolve’ (Miller 1996, p. 442).
Turning to art in design, Lawson (1997) sees the distinction from science as ‘rather hazy and easily blurred’. Yet, Archer (1979), as the Royal College of Art (and Design) Professor of Design, sought to distinguish design from both the sciences and the humanities. Thus he saw design, with modelling as its mode of communication, contrasted with the sciences having distinct notations and with the humanities, having their languages. His iconic representation of ‘the three cultures’, each at an apex of a triangle, led to placing technology (and hence engineering design) on the triangle side linking design and science and placed between the ‘useful arts’ and the physical sciences. He makes a clear distinction between the fine arts (on the design-humanities side nearer the humanities apex) and the useful arts, which is reinforced by Lawson's emphasis that ‘designers unlike artists cannot devote themselves exclusively to problems which are of interest to themselves personally’. However, among the design community, while architects have to nominally satisfy clients, they still retain a strong sense of artistic freedom. This is something that engineering designers—like craftsmen—do not see they have the luxury in which to indulge. For engineering designers, this is due to the constraints arising from the appropriate engineering sciences, together with the imperative to achieve cost-effectiveness in their designs.
In the author's field of designing ships, which can be seen to be archetypically complex and physically large, the characteristics of science and art have traditionally been associated with the application of those engineering sciences and with the ‘art of ship design’, respectively. Thus, the naval architect applies fluid mechanics, whole body dynamics and structural mechanics to analyse the ship's resistance and propulsion, its motions and structural response. In contrast, the art of producing a new ship design has been seen as a creative act, akin to that of the artist or, at least, an architect of the terrestrial built environment. Both these facets of ship design practice have been considerably enhanced in recent decades by digital computation. However, it is considered that the ability provided by graphical representation of design information, owing to very recent advances in computer graphics, continues to substantially change the art and science of the design of such large systems, in particular, and it is that change which this article explores.
That the impact of computer graphics on initial ship design has parallels in the design of other large-scale systems has been confirmed by various architectural theorists. Oxman (2006, 2008) and McCullough (2004) see digital design as altering the manner in which the designers of such large systems ‘represent, present, communicate and materialize’ their designs. Oxman considers it increasingly important ‘to conceptualise this emergent field’ (of digital design thinking) and sees there to be an effect on conceptualization and materialization, as well as in form and material (Oxman 2008). Even the traditional art of the architect to hand-sketch ideas is being questioned as to whether it is still appropriate, given the creative resources provided by digital media (Edwards 2008).
Section 2 addresses the nature of designing PL&C systems, of which ships are only one manifestation. This is followed by consideration of the impact made by advances in computer-aided design (CAD) on such design practice. The specific issue of the choice of design style is seen as a crucial early designator in such designs, with a relevance to the science and art of design. Finally, the future course of such design is addressed through discussion of several fundamental issues drawing on the theme of the role science and art play in the design of PL&C systems.
2. Physically large and complex systems
In considering some fundamental questions germane to the practice of engineering design, the author has drawn a clear distinction with regard to not just complex but also physically large systems and their design (Andrews 2010). Given Bruce Archer's distinct ‘third culture’ for design, it might seem perverse to start distinguishing between different types of designs. However, in recent design studies, distinctions have been made such as in ‘primary classes of design’ (e.g. ‘original’, ‘adaptive’ and ‘variant’ referring to different design outputs (Howard et al. 2008)) and in describing the design process (see table 1 of Howard et al. (2008) with 24 examples, each summarizing the process across a wide set of applications). Joseph (1996) pointed out that standard texts on design ‘advocate their own approaches, based on the practice of their authors’, which seems to confirm the acceptance of distinctly different approaches. These can be said to be applicable to a diverse set of products ranging from the design of mass-produced consumer products (such as that addressed by Howard et al. and, for example, Pahl & Beitz (1984)) through to the views of a wealth of architectural design theorists (such as Hillier et al. 1972; Broadbent 1988; Brawne 2003). Even so, Visser (2009) insists that design is one type of cognitive activity, albeit with different forms.
So, it is considered appropriate not just to draw the distinction between the design of consumer products and of complex major projects (such as is the remit of the Major Projects Association (http://www.majorprojects.org)), but also between generally complex systems and those that have the added dimension of being physically large. This then distinguishes such large systems from both complex software and hardware products. Thus, the latter complex systems can be said to encompass extensive electronic-based systems (such as air traffic control and military command systems) as well as (say) high-performance aircraft. However, both these examples of complex systems lack the further aspect of being physically large to the extent that generally applies to ‘made to order’ or ‘bespoke’-designed systems (Hills et al. 1993). The latter systems, typically, consist of civil engineering constructions, large chemical process plant, ships and offshore facilities. Given that such products are very large and often one-off, they do not have prototypes. Secondly, the facilities to manufacture and assemble them are not factories, which are extensively redesigned for each new product run, rather the end-product is made or assembled on site or in a piecemeal manner. This has significant consequences for the design task, especially if, in addition, the product also has to provide extensive human habitation for considerable periods. This habitational need is considered to pose quite a different design problem to accommodating personnel for short-term transportation, as for example in cars and planes. In producing such artificial environments, the design of PL&C systems can therefore be seen to approach that of designing a whole system and not just the main artefact within it (Levander 2009).
The Royal Academy of Engineering recently published a report, produced by a committee of senior systems engineers (Elliott & Deasley 2007). This largely focused on complex systems, such that the six principles for integrated system design detailed in the report were seen to be both most applicable and necessary, when managing projects costing billions of pounds. Such projects have to be designed to achieve satisfactory performance, within a cost ceiling and to meet a specified timescale for introduction into service. Systems engineering can be seen as essentially a project management discipline, with its origins in post Second World War defence acquisition. Such large-scale projects, with applications beyond defence into civil construction, the process industry and space exploration, now also have significant software complexity (often heightened by safety critical consequences). However, the issue of the end-product being of a physically large nature is seen to be a further complication beyond just the computational magnitude required to achieve their design description.
Because few initial designs of PL&C systems are finally realized and the end-products are also immensely expensive, they are designed to have long operational lives. So adaptability, to uncertain future roles, often figures as a design objective, along with a desire to accommodate already perceived multiple roles. Thus, at the front-end of the design process, without the comfort blanket of at least one full-sized prototype before production commences, working out what is wanted and what is affordable is also complicated. Design theorists have coined the term ‘the wicked problem’ to encapsulate this issue, namely, determining the requirements can actually be more challenging than the subsequent design work to meet those emergent requirements (Rittel & Webber 1973). That this is often unrecognized is perhaps not surprising as the critical early stages of design, which then become a process of working out what are the ‘right’ requirements, are quite often not considered to be truly ‘design’. This is because many non-designers are inclined to reserve the term ‘design’ to describe the large-scale multi-disciplinary team effort of subsequently detailing the design for manufacture, yet it is in the early stages of design that the great majority of the commitment of future cost is made, as this is the only phase when major changes can be made quickly and cheaply. Nowadays, the large-scale design required downstream is managed and executed through Integrated Product Models and Integrated Product Data Environments (IPMs/IPDEs). Thus, for very complex products, managing such detailed design tasks involves ensuring integration and concurrent development is achieved through an alliance of sub-system vendors (Martin 2009).
Despite large-scale investment in IPM/IPDE systems, it remains the case that it is the front end of the process which really distinguishes the design of PL&C systems from that which is described by the many more general descriptions of the design method. This is because such descriptions show that the engineering design process commences with a clear need or a precise set of requirements (Hubka 1982). Rather, for the design of PL&C systems, the initial design phase is characterized by the need to elucidate what the requirements should be, as has been extensively argued elsewhere (Andrews 2011). The initial design phase then consists of a trial-and-error dialogue between the customer/owner/end-user and the designer, who produces prospective solutions (and the plural is very important here). So, both parties are involved in trying to reveal what is really wanted and, vitally, what is realistically achievable in terms of cost, time and risk (Elliott & Deasley 2007). Thus, the design process is problem-dependent with complex design requiring strategies that depend on the problem structure, designer experience and understanding the problem context (Joseph 1996). Also, this more realistic emphasis on requirements elucidation can then seen to be consistent with the approach of deferred commitment or set-based design, based on Toyota's Product Development System and recently introduced into US Navy ship procurement (Singer et al. 2009).
The traditional view of the preliminary design of the PL&C systems was to see the use of the engineering sciences as assisting in the initially creative ‘art’ of design synthesis. However, the author has suggested that the issue of ‘art versus science’, when applied to ship design, is inherently misleading (Andrews 2007). This is because there is both an art in the choice and application of the engineering sciences, when applied to ship design at the analytical level, and also a rational or scientific basis underlying the most artistic aspect of design, that of aesthetics. This can be seen in the ‘rules’ propounded to achieve visual balance in ship design (Guiton 1971). Furthermore, there is considerable ‘art’ in the initial ship design synthesis, to an extent usually unacknowledged. Yet, science is clearly invoked in initial design, in the sense of undertaking a rational and coherent approach. This is in contrast not just to the art of the artist, who does not need to be conventionally rational, but also to the art of the craftsman making traditional buildings and boats. However, in an era of CAD, for ships and large-scale buildings, it could be argued that the craftsman's art is all but a memory. So, it would seem that what traditionally has been considered to be subjective, both in ship aesthetics and in synthesizing a new PL&C system design, is now amenable to a rational scientific approach. Despite this, art, at least in the artisan sense, can be seen to still exist throughout the practice of engineering design through to detailed design. This is argued by Ferguson, who asserts that there are ‘dozens of small decisions and hundreds of tiny ones’ that every designer makes throughout the whole process of engineering design (Ferguson 1993), even when in front of a computer screen. Furthermore, an additional facet of the designer's art, associated with the creative (synthesis) element, can now be restored to initial design by the facility provided by computer graphics.
3. Computer-aided design applied to physically large and complex systems
Taking ship design as representative of the design of PL&C systems and considering the applicability of CAD, Gallin, as early as 1973, pointed out that ‘ship design without the computer is no longer imaginable’ (Gallin 1973). Six years later, at the first international review of marine technology, Benford saw the changing scene as being driven by both economics and the computer (Benford 1979). Thirty years on Nowacki, one of the principal researchers in computer-aided ship design throughout this period, in reviewing five decades of developments in marine design methodology from five aspects (namely, economic efficiency, safety and risk assessment, rationality, optimality and versatility), saw trends in four fields. He considered those to be economy and safety; the systems approach; parametrization of design variables; and design as learning, where he highlighted simulation and visualization (Nowacki 2009). These last two design topics have clearly been facilitated by recent advances in computer-aided graphics, especially when applied to spatial or architecturally driven ship design. They also echo the specific design approach propounded in the author's previous articles in the proceedings (Andrews 1998, 2006a).
Over the last 50 years covering the existence of CAD, Nowacki's topic of rationality in the design of maritime PL&C systems has been demonstrated through an emphasis on developing appropriate tools and methods, often under the term ‘design methodology’. This can be seen in the three sets of state-of-art reports presented to the tri-annual International Marine Design Conference (IMDC) (Sen & Birmingham 1997; Parsons 2006; Erikstad 2009). Of those reports, the ones specifically addressing design methodology were edited by this article's author. In particular, the 2009 IMDC Design Methodology Report presented some 26 iconographic representations of the ship-design process, which have been proposed over the last 50 years and range across the spectrum of ship types. This selection particularly focused on the early or preliminary stages of the process, given the crucial impact of these stages on requirement elucidation and, thus, the eventual built definition (Andrews 2009).
An example of ship-design process representation is given in figure 1. This is an updated version of the process flow model, incorporating the architectural element in initial design synthesis that was given in the 1998 proceedings (Andrews 1998). Importantly, this sequence shows the major choices that a ship designer or design organization, not a CAD toolset, has to make in order to proceed to the later phases of the design. Those phases have been summarized as evermore detailed design iterations, in the last three steps of figure 1. It is worth highlighting that this is a top level representation and that while designer decisions (‘Selection’) have to be made before their related step in the process, sometimes the sequence of the steps may be different when specific aspects drive a particular design.
Prior to each computational and configurational design activity, the decision-making steps in figure 1 are themselves design activities, which the designer makes, hopefully, in a conscious manner. Incorporating these selection choices distinguishes this diagram from most representations of the design process, which just sequence the direct designer's activities, such as synthesis and exploration of features. Among the designer's choices is the selection of the ‘style’ of the design solution (an issue discussed more fully in §4) and selection of the synthesis model (or often just a sizing model). Such choices are often not questioned by the designers or, worse, by their design organization and end customer, yet they are likely to have major impact on the eventual solution.
Other decision points, shown in figure 1, such as the basis for decision-making on an initial synthesis output, criteria for acceptance of the design and evaluation (against a selected set of criteria of acceptability), might be considered, to some degree, by the design organization. However, all too often, they are laid down by the customer or an acceptance agency, or, yet again, just adopted from previous practice. Such acquiescence was, perhaps, understandable before computers enabled a more scientific approach both to analysis and synthesis, even though the latter may just be numerically based. However, this now can be seen to be an unacceptable stance, given Nowacki's justifiable emphasis on rationality in the development of ship-design practice (Nowacki 2009). It is important that the facility of option exploration through computation is also tied to graphically modelling the design, if innovative options are to be investigated more comprehensively in early design (McDonald et al. 2011).
Having said that the initial stages of the design of PL&C systems constitute the most vital phase in the design process, this section concludes with a recent architecturally or graphically descriptive example of a set of preliminary ship-design studies. These studies adopted the design approach outlined in the 1998 proceedings article (Andrews 1998), which as the Design Building Block (DBB) approach was subsequently incorporated, via the SURFCON module, into QinetiQ GRC's PARAMARINE CASD system (http://www2.qinetiq.com/home.grc.html), which is now widely used for ship design worldwide.
(a) The UK mothership studies
The UCL design research team used the PARAMARINE–SURFCON tool to produce a series of novel ship concepts. These concepts had different means of deploying relatively small naval combatants, which resulted in several distinct ship configurations to meet the same operational concept of a fast ‘mothership’ to carry small combatants. It was intended to transport the combatants over long distances so that they could then be deployed in a littoral environment, avoiding the need to provide large and costly ocean going combatants. Each of the ‘mothership’ configurations addressed a different deployment and recovery method, namely, well dock; heavy lift; crane; stern ramp; and stern gantry. The origin and management of this set of studies was comprehensively described by Andrews & Pawling (2004) and so, for the purposes of this exposition, what is relevant is whether such a range of concept designs could be investigated, to such a level of design fidelity, without an architecturally driven design approach. The range of design solutions produced can be appreciated from the three-dimensional representations shown in figure 2 of the seven vessel types conceived, with summary characteristics of each technically balanced design provided in table 1. Each new ship concept was driven not just by the carriage and deployment of the small combatants but also, in most instances, by the extensive water-ballasting arrangements required and the considerable bunkerage for the stowage of fuel, necessary to propel the vessel at speed of some 10 000 nmiles. It was found that the architecturally based synthesis gave a higher degree of confidence in the realism of each of the distinct solution types, compared with an approach using just a conventional numerical synthesis. In particular, the integrated representation of ship architecture and the usual technical/numerical aspects of naval architecture were found to mitigate errors in the modelling. This was particularly relevant to the interface between the spatial and numerical representations, which in any subsequent design development might otherwise have been shown to be sufficiently erroneous to render a configuration unworkable.
The physical complexity of combining the architectural and engineering science in initial design required the facility of computer graphics, alongside analytical modules, for the designer to adequately explore configurationally diverse potential solutions. This demonstration effectively unified the art of spatially based inventiveness with the necessary engineering sciences to achieve a range of novel and balanced design concepts.
4. The issue of style in large and complex system design
The mothership studies outlined above can be seen as investigations of different design solutions, which nevertheless essentially adopted the same design standards and style. The latter characteristic is the first design decision shown in figure 1. The term design style was proposed to distinguish a host of disparate issues distinct from the classical engineering sciences, such that many of them could be seen to be on the ‘softer’ end of the scientific spectrum drawing on the arts and humanities. When the nature of ship costs was being considered, the term ‘S5’ was coined to denote the naval architectural aspects of Speed, Stability, Strength, Seakeeping and Style (Brown & Andrews 1980). While the first four terms are, historically, the principal naval architectural (engineering sciences) disciplines associated with a ship's technical behaviour, the term ‘Style’ was devised to summarize those other design concerns, which for the example of the naval ship are listed in table 2. The overall term of Style was adopted to cover this very disparate range of issues, which table 2 then seeks to categorize under some six headings that (ship) designers understand. Thus, for example, concurrent engineering concerns (or Agile design, in software terms (Brooks 2010)) are encompassed by the topics under Design Issues in table 2.
Importantly, these style issues can make a substantial difference to the final outcome of a design, so that their relative impact ought, in the case of a complex system, to emerge from a proper dialogue between the designer and the client (or in the naval ship-design case, the operational requirements owner). Furthermore, most of these issues have been difficult to take into account early in the design process because, usually, initial design exploration is undertaken with very simple and, largely, numerical models summarizing the likely eventual design definition (Andrews 1994). That dialogue can now be informed by also having a graphical representation of the system's configuration and internal architecture (figure 2). At the critical early design stages, this can then enable the designer to take account of many of the significant issues, such as those listed in table 2.
The nature of PL&C systems and the need to emphasize the importance and difficulty of early representation of ‘style’ issues is seen to be a further complication. This is due, additionally, to there being a wide range of design practice arising from the degree of design novelty, as is indicated by table 3. This shows a set of examples, not exclusively taken from the field of ship design, where the sophistication in the design undertaken ranges from a simple modification of an existing product, through evermore extensive variations in design practice, to designs adopting, firstly, radical configurations and, beyond that, radical technologies. Although in first of the latter two categories current technology is often adopted, such options are still rarely attempted, owing to the risk of unknowns (usually exacerbated by the lack of a real prototype), while radical technology solutions are even more rarely pursued. In part, this rarity arises because such radical technology solutions require recourse to design practice much more akin to the aerospace approach. Thus, new major aircraft projects, typically, require massive development costs (including full-scale physical prototypes, some tested to destruction) and additionally need tooling and manufacturing facilities to also be specifically designed and then built, before extensive series production can commence. Such distinctions as those of table 3 for PL&C systems design practice suggest any attempt to systematize general engineering design practice needs, at least, to recognize that there is a spectrum, not just from component and consumer product design right through to PL&C system design but that there is also a range of design practice for the latter, as is proposed in table 3. This again has a bearing on the art and science issue, where the relative balance can be seen to differ for different points on this spectrum for PL&C system design.
Also of relevance to the nature of art and science in PL&C system design and the importance of style to this are recent insights from incorporating simulation techniques early in design. Such simulations have only recently been able to be addressed in initial design through the facility provided by computer graphics. The latter has expanded the depth of the representation of the PL&C systems possible in initial synthesis, while integrating that with the traditional numerical representation. The graphical facility directly fosters a creative and exploratory approach to initial PL&C design, which it has been argued marries art and science in the design process for such complex systems (Andrews 2007). Having said the early three-dimensional description of the configuration can now enable many of the design style aspects to be considered early in the design of PL&C systems, an example of this needs to be briefly outlined.
The wider design impetus behind recent research in simulation-based design was covered in the 2006 proceedings article (Andrews 2006a) and the eventual results and implications of a major demonstration of integrating personnel movement simulation were subsequently published (Andrews et al. 2008). That exercise showed that in concept design, such simulations could inform specific style issues, such as access and damage control (table 2). There were some significant lessons acquired in this demonstration, not least being the two issues of the level of design granularity, needed to support a given simulation technique at the early exploratory level of design description, and the presentation of the output of simulation data to the designer (Casarosa 2011). One example of the form the latter might take is shown in figure 4. It is hoped that further developments in digital media will help to resolve both these issues, while still leaving the designer able to creatively explore options and their technical and style implications.
5. Managing the explosion in computational capability
The above example, of simulation integration in the preliminary design of PL&C systems, is further evidence that the designer can exploit the ever-greater computational and graphical capability of modern computer systems. A leading initiative in assessing the likely impact of further developments in high-performance computing (HPC) applied to engineering is the US Department of Defense's CREATE programme (Arevelo et al. 2008). In a comprehensive review of the challenges and risks in large-scale computational simulation, Post et al. (2006) consider that such simulation tools are currently at the third of Petroski's four steps in the maturity process of ‘Design Paradigms’ (Petroski 1994), namely ‘continually more ambitious designs that pushed to the limits of the existing technologies until large-scale failures occurred’. Thus, the pure speed and capacity of computational devices, or more likely multiple platforms comprising such devices, will continue to grow, assuming that some major advances will occur to maintain the growth rate seen to date (Barrett 2006). However, translating this into appropriate analytical and simulation tools is seen by Post et al. to be problematic. Such tools will have to possess the rigour to be usable by engineering designers, without substantial time lags in their application, and this is considered to require considerably more systems engineering practice than has been seen to date. Post et al. (2006) identify three challenges in this regard: those associated with development efficiency, programme accuracy, and application focus and resourcing.
Historically, engineers have been prepared to use the available computational tools without an adequate verification and validation process, which to some extent has been justified by recourse to traditional experimental benchmarking. Such benchmarking is less likely to be available in the future, both owing to cost pressures and to the limits of empirical capability to match the growing computational sophistication. In the case of PL&C systems, this is seen to be a further justification for enhanced use of simulation techniques early in the design process. Given the implications and risks of a growing reliance on computational tools, which are either not validated or even not capable of being validated, then some risk reduction could be achieved through as early an assessment as possible in the design process. Such early identification would then highlight the potential consequences and enable any mitigating action be taken before too great a design commitment has been made. This can be seen as a further example of the ‘art’ of applying science to PL&C systems, in particular, given the scale of their inherent complexity.
In addition, the issue of design-sketching is one that, with computational power becoming ever more available to engineering designers, could make a further substantial difference in the practice of the preliminary design of PL&C systems. This issue is particularly relevant, along with the use of simulation tools, to the style-related issues, identified in §3, and hence to broadening both the practice and scope for creativity or art in preliminary design. It has been recognized by architectural practitioners and theorists that ‘sketching’ is primarily a means of design communication ‘nearly always three-dimensional doodles … as a clarification for oneself and for spreading the notions’ (Brawne 2003). In design theory, the traditional view has been that sketching is ‘simply as a way to externalise images thought to be already present in the mind of the designer’ (Fallman 2003). However, it has been proposed that the concept of sketching, when termed ‘prototyping’ in the field of human–computer interaction, is more fundamental to design work, in typifying design as ‘a kind of dialogue; a reflective conversation’ (Fallman 2003).
Given this extension of the traditional ‘pen and paper’ into computer-generated visualization is still seen to be essentially ‘sketching’, then the question arises, in the field of PL&C systems design, as to whether both the architect's and industrial designer's traditional view of sketching can be facilitated in future engineering CAD tools. These would then need to be ‘flexible, visually rich tool(s) with integrated technical analysis’ (Pawling 2007). What this would then give designers of PL&C systems is a direct sketching capability, alongside the growing availability of decision-making techniques (such as genetic algorithms, Pareto plots and data-management systems (see Vasudevan (2008) and McDonald et al. (2012))). This would then integrate a creative sketching ability and decision-making techniques, through the medium of computer graphical interfaces already allied to the existing synthesis tools used for designing PL&C systems (Pawling 2007), and thus integrate the elements of art and science in such design. A future possibility of a sketching-based approach for preliminary ship design was recently proposed and an indicative concept, suggesting how rapid ‘sketching on the screen’ might be employed, is reproduced as figure 5.
6. The future of physically large and complex system design
The integration possibilities suggested in the previous section, as a realization of an amalgam of art and science in design, lead on to a broader set of fundamental questions raised by the manner in which digital media continue to alter design practice. These issues have emerged in part from an awareness of a wider philosophical debate, which was presented to the maritime engineering design community (Andrews 2006b). From several approaches attempting to scope design philosophy, Galle justified the study of the philosophy of design by seeing it as ‘helping, guiding, suggesting how the [designer] comes to understand what he is doing and not simply how he comes to do what he is doing’ (Galle 2002). The relevance of all this to PL&C system design is that it leads to questioning the belief held by many engineers that an extensive abstract functional analysis is all that is required before any materially descriptive synthesis can commence (Andrews 2011).
In summary, the following issues are seen to be pertinent to the future practice of general engineering design and to PL&C system design in particular. They are also pertinent to the nature of art and science in such design practice:
— How, at least for complex and large products, can designers best synthesize new designs?
— Should future engineering designers see enhanced techniques applied to optimization in engineering design, primarily, as a means of informing solution search rather than the means of decision-making for solution selection?
— Do most engineers have a mistaken belief in a Functionalist philosophy (given that it is argued below that, contrariwise, ‘Form does not follow function’)? Rather this Functionalist belief can be seen as just another design style and, furthermore, is one that ignores the ‘wicked’ nature of requirement elucidation.
— How can it be ensured that computer-based techniques enable designers to adequately tackle the integration task in complex design?
— With the advent of simulation-based design and virtual reality tools, how can engineering designers ensure that the key word in CAD remains ‘Aided’ rather than potentially ‘Automatic’ as the latter is seen to be unobtainable for safe and efficient PL&C systems for the foreseeable future?
The first of these questions addresses, at least for complex large products, the process of synthesizing a new design. It was recently pointed out by a philosopher of science that even Karl Popper saw the nearest scientific equivalent to design synthesis, namely the creative process to produce a scientific conjecture, as being more an issue of metaphysics than philosophy (Lipton 2010). Furthermore, the endeavours of design researchers to study design synthesis seem largely to be so artificial, small-scale and hence irrelevant to the practice of complex design, given such attempts to represent the real world are generally through the environment of a student design exercise. Consequently, such investigations are seen to provide very little in the way of insight on design practice for PL&C systems.
However, it is still useful to draw on two studies of design synthesis that were done in the early days of general design research. The first was by Darke (1979), who quizzed architects, rather than design engineers, on how they came up with new designs. Darke suggested that an architect searches for a key design feature or generator to provide the basis of getting the new concept. Interestingly, some parallels have been found from comments by designers of post-Second World War British naval ships (Brown 1983), where a similar approach appears to have been adopted: in each case, some key aspect was identified by the lead designer as driving the style of the new design (Andrews 1984).
The other study was by Daley (1980) whose insight into the creative process suggested that individual designers bring to the process a set of schema to achieve a design creation. The schemas Daley identified were, firstly (unsurprisingly) the visual; next the verbal, acknowledged to be crucial in communicating engineering design, for PL&C systems in particular (Betts 1993); and, finally, a designer's value system, which might explain certain style-related choices. If the conclusions of Darke and Daley can be considered plausible, then both have some quite profound implications given the accelerating impact of computers on engineering design.
Next, there would seem to be a fundamental issue associated with the ever greater sophistication in the variety of numerical optimization techniques in engineering design. Even if the appropriate measure of performance or cost-effectiveness can be identified, a better stance would seem to be to see such optimization techniques as a very powerful means to search for solutions. This can then be achieved by providing clearer insights but not necessarily the results being seen to directly provide the design decision. Such pragmatism is considered essential if engineers are to avoid compounding the quantification of disparate quantities and, often, resorting to specific techniques, such as genetic algorithms, without necessarily questioning the applicability of the specific technique to the specific problem in hand (Andrews 2004). Here, the visualization of alternative design solutions can help in clarifying the insights obtained through different computationally derived data (figure 4).
Something that many advocates of systems engineering, in using simplistic models of the design process, seem to assume is an essentially Functionalist philosophy, despite the fact it has long been recognized by architects as just another style (Jencks 1971). Rather, form does not follow function, to the extent that Mies van der Rohe even reversed Sullivan's functionalist epigram, in his justification for the concept of adaptable, open plan layouts in high-rise offices (Jencks 1971). Furthermore, it is not possible to have a dialogue to solve the ‘wicked problem’ of requirement elucidation through a process that describes requirements in ‘functional’ (i.e. non-material solution) capability statements (John 2002). Given that the major concern of the customer is bound to be affordability and that affordability is accessible only through prospective material solutions, this means that Requirement Engineering's focus on deriving (seemingly abstract) functional statements is flawed (Andrews 2011). Rather, a fundamental design choice is the style of the prospective solution and this should be part of any design methodology, as is caught by the third step in figure 1 and the initial bubble of the ‘V’ diagram of Elliott & Deasley (2007) guide to complex systems design. That choice can best be addressed through more information-rich trial solutions, exploring style and marrying numerics and graphics, which can then inform the dialogue with the client/requirements owner.
Two other related issues, arising from the inevitable reliance on the computer for complex product design, are associated with the need to tackle the integration task and the ability to judge the output of complex computer software. Integration requires an understanding of the total system, as well as the correctness of the individual sub-systems. Of course, there is the engineering truism that the ‘devil is in the detail’ and most engineers recognize that this is crucial, because they know that is where failure is likely to occur, even if the source of failure may originate in choices made through the overall decision process. So, the PL&C system designer should be able to see the totality to comprehend the integration issues. Computers can now provide extensive detailed output as part of (say) a finite-element analysis, but how representative of the material reality is the input to a given computer analysis? How appropriate is the predicted response to that of the actual system's likely behaviour? Such questions raise significant philosophical and educational implications.
So, what do these fundamental or philosophically oriented issues mean for engineering design practice, given the ubiquitous presence of electronic computation, not just for number-crunching analysis but, thanks to computer graphics, also the growing reliance on CAD, simulation-based design and virtual reality? From direct experience in the practice of ship design and more recent involvement in the research and development of computer-aided preliminary ship-design tools, there is seen to be a need to emphasize that the crucial word in CAD is aided. Thus, any recourse to ‘black box’ CAD systems or optimization routines that leave the engineering designer dependent on opaque computer codes and data limitations, often unaware of the basis of the specific synthesis process or the decision-making that may be built into design tool they are using, has to be strenuously avoided, as poor value for money (Andrews 2004).
There seem to be two facets in considering the scope of a philosophy of engineering design: those related to design and those to engineering. Design methodologists for several decades have considered the philosophy of design. The 2002 exercise referred to at the beginning of this section seemed to conclude that the philosophy of design was an immature union of philosophy and design research in ‘the pursuit of insights about design by philosophical means’ (Galle 2002). So, if philosophical practice is concerned with ‘the principles underlying any sphere of knowledge’ (Chambers 1971), this would all seem to come down to investigating the principles that lie behind the practice of (engineering) design, and hence the relevance of key issues, such as those listed above. Given Galle's cautionary remark, it would seem unwise to expect an early consensus to emerge from current philosophical considerations of design and this then cautions against prescriptive approaches to the design of complex engineering systems. Rather practitioners and researchers in the design field should focus on approaches that reveal understanding and enlightenment, not least with regard to the nature of science and art in the practice of the various domains of design.
The second element in a philosophically oriented consideration of engineering design is the engineering dimension, where again the principles behind engineering practice are pertinent. The topics listed above were raised at the 2007 Royal Academy of Engineering seminar (Andrews 2010) as they were regarded as highly relevant to the future practice of the engineering discipline and, specifically, engineering design. Given that engineering, in the UK in particular, still lacks the status it ought to have and fails to attract and retain the brightest young people in sufficient numbers, it is considered that one element in that situation is a certain feeling or belief that engineering practice is inherently intellectually inferior to the ‘pure’ sciences. Understanding what engineering practice is and articulating this with an intellectual rigour, which conveys the worth and creativity of its art and science, could go some way to counter this misapprehension. Such an endeavour needs to be done without simplifying and straitjacketing that practice, and hence the emphasis on the need to recognize clear distinctions across the spectrum of engineering design, as has been articulated in this article.
So what might be the emergent constituents of a philosophy of engineering design? Without being prescriptive and trying to reflect the creativity and sophistication of that practice, it is argued that the earliest stages of complex engineering design need a more inclusive approach to the computer-aided practice of design. Thus, the basis for the initial design synthesis of PL&C systems should itself be a synthesis of the physical architecture of the artefact, as a set of three-dimensional building blocks (Andrews & Pawling 2006), together with the existing numerically based synthesis, which itself was summarized (for ships) by figure 5 of the 1998 Proceedings article (Andrews 1998). This integration of a PL&C system's architecture and of the engineering science, describing its physical performance, is a way of fostering a designer-led approach that is creative and holistic in terms of engineering design. The approach recognizes the complexity of the initial design of such engineering products, as can be discerned from figure 6 and exemplified by the mothership design examples summarized in §3. Like most engineering practices, producing such designs requires domain knowledge and experience; it is thus a challenge to ensure that the development of design tools, to be used in the early stages of the design of PL&C systems, produces tools that truly assist the designer in dealing with complexity. This then will provide a more philosophically robust approach to complex design, sensibly marrying its art and science.
This article commenced with the issue as to whether, with the dominance of computational-based tools and methods, the current design practice for PL&C systems was no longer an art but had become a science. The article has argued not just that numerical computation but also CAD and, now, computer graphical methods have changed the nature of such design, so that it can either become more ‘black box’ like or, preferably, use computer graphics to open up early design synthesis to the use of simulation and visualization. The issue of art and science in design can be seen to be explained in that the scientific approach assists the ‘art of designing’ to enable creative ideas to be produced alongside obtaining rational decisions, which are themselves consistent with a scientific approach. Simulation in initial design will then change the nature of the design of PL&C systems enabling the designer to be more innovative and exploratory. If this approach is grasped by the engineering design discipline, then art and science can be seen to be complementary in design.
Archer's vision of visual representation or modelling as design's medium, supports the view in this article, that with spatial modelling at the heart of the design of PL&C systems, the creative art in design can be sensibly integrated with the science provided by growing computational power. However, this can only be achieved through an approach that makes engineering design both more visibly rational and fosters an innovative milieu, such as the CAD sketching proposal, at figure 5, would provide. Given that engineering design is already the most scientifically based of the design disciplines, it is argued that the computer-based graphical integration of an architectural synthesis with the engineering science-based numerical description, at the earliest stages of design, would ensure a fundamental clarity. This article has been motivated by a belief that there are shared features in arts and science, given that ‘the visual very often has the central role’ (Kemp 2000). This is seen to be most pertinent to large and complex design, since ‘Having a visual, geometric representation of a design process is crucial, for designers are spatial thinkers’ (Brooks 2010).
The author would like to acknowledge the contribution of the late Emeritus Professor Louis Rydill to his three proceedings articles on complex design and to Charles Betts, FREng, for comments on the current article. The very helpful input provided by the reviewers has also greatly aided achieving the final form of the current article; however; the views expressed remain solely those of the author.
- Received September 29, 2011.
- Accepted November 7, 2011.
- This journal is © 2011 The Royal Society