Conceptualisation and formalisation of technical functions

In the engineering design research community, function-based descriptions have become a popular means to describe products and systems using abstract building blocks. A claimed advantage is that they allow designers to synthesise and describe their designs early in the design process without having to consider physical principles, material selection, and geometric shapes and features. In addition, they can be used for critical investigation of both unfinished designs and existing products, in which case functions are supposed to facilitate decoupling of perceived needs of end users from concrete solution elements. This solution independence has been said to encourage out-of-the-box thinking and thus increase creativity and innovation, both in novel designs and in redesign. In order to facilitate a comprehensive investigation of the functional capabilities of products, the technical functions are often taken into consideration together with the utility functions. In the case of the development of new products or components, technical functions are the building blocks or elements of a solution independent functional synthesis, which is typically followed by the investigation of possible realisations of functions and converting the functional schemes into feasible component structures and numerically testable virtual and physical prototypes.

Since the mid-twentieth century a large number of publications have appeared on technical functions. Figure 1 illustrates that the number of scientific articles referring to technical functions has dramatically increased since the early 1990s. Academic debate on functions in design and engineering has also been the topic for several special issues of various journals. These have been edited by Sticklen and Bond (1991), Kumar (1994), Chakrabarti and Blessing (1996), Chittaro and Kumar (1998), Stone and Chakrabarti (2005) and Kroes and Meijers (2006). Despite the growing attention that technical functions receive in academia, they are not widely used by designers and have hardly been incorporated in computer tools developed for industrial users. As Birkhofer (2011) stated, ‘there is an urgent need to close the gap between conceptual approaches in design methodology and the involvement in embodiment and detailed design in design practice’. In a recent survey of design synthesis systems by Chakrabarti et al. (2011), the state-of-the-art computer tools for function-based synthesis were reviewed and it was concluded that, among other things, (i) the systems need to be scaled-up to support problem solving at the levels of complexity expected in practice and (ii) a better understanding is needed on how the systems can be integrated into current design processes in practice. In addition, it was found that rigorous means for evaluating the systems are missing, and that still no commonly accepted terminology is available to foster exchange of research models, methods and results.

Figure 1. Frequency of the occurrence of the phrase ‘technical functions’ in publications, based on keyword searches in Google Scholar.

The extensive criticism that the whole concept of ‘technical function’ has been receiving in recent years underpins and explains the critique and reservations towards computer-based functional design systems. Horváth (2000) identified 12 areas where scientific understanding is missing, among which the most important ones are: (a) mapping requirements onto functions, (b) matching targeted functions to first principles and physical processes, (c) mapping functional arrangements to structural arrangements and (d) coping with the abstraction, incompleteness, and uncertainties in conceptual modelling. Warell (1999) criticised putting too large an emphasis on transformations that technical functions are supposed to perform, and therefore failing to take into account human-centred aspects of use. Crilly (2010) argued that, by focusing on expression of physical goals, technical functions fail to address the underlying connection between seemingly remote uses of products (e.g. using a car for transportation vs. using it to express personal values) and that they thus hamper the exchange of ideas between domains related to such different uses. Maier and Fadel (2009) stated that ‘function has in effect been accepted as the de facto fundamental concept in design without theoretical justification’ and that ‘there is no theory to guide us as to the proper use of function in design, what its limitations are, and what underlying assumptions there might be’. As an alternative approach that would resolve the above issues, put forward by Warell and Crilly, Maier and Fadel have proposed functional affordances as a different form of abstraction, which can be used by designers instead of technical functions.

Despite the critique and the proposed alternatives, research in technical functions does not seem to suffer from lack of attention. For this special issue on technical functions, the inspiration came from the Eighth Symposium on Tools and Methods of Competitive Engineering (TMCE) in Ancona, Italy, in 2010, where several presented papers indicated that new insights into technical functions are emerging. It appears that, nowadays, the majority of authors investigate topics such as the diversity in terminology among approaches and the exchange of models and knowledge between different function frameworks and different application areas, rather than introduce yet another framework of functional modelling. These insights are interesting in that they appear to shed a light on the prevailing problems that prevent technical functions and functional design from getting widespread practical use.

The Call for Papers went out in April 2010. We originally intended to also incorporate articles on design grammars, but we did not receive a sufficiently large number of quality papers on that topic and therefore, at some stage, decided to concentrate on technical functions. Out of the 13 originally submitted articles, five survived the rigorous reviewing process, which involved at least three (mostly four) reviewers participating in three consecutive evaluation rounds.

The first article by Albers et al. (2011), entitled ‘Different notions of function: results from an experiment on the analysis of an existing product’, presents an exploratory study to pinpoint the main confusing issues related to function-related terminology. In their experiment they asked designers to draft function decompositions of an existing hydraulic pump. The results show little consensus on what functions are and how they should be described. The subjects were however consistent in not distinguishing function from behaviour, which obstructed their reasoning about the necessity of functions. Based on the results the authors conclude that functions indeed provide only subjective descriptions of products, and that different designers cannot reach the same understanding of a product just by going through the assigned exercise. To make these outcomes useful as a starting point for improved function-based approaches and their implementation in engineering, it would be commendable to confirm the findings in a more rigorous study with more subjects and more different sample products.

The next two articles deal with conversion between different representations of functional models. In his article ‘Supporting communication across functional frameworks by converting models of functional decomposition’, van Eck (2011) discusses a conversion strategy, which is based on three decomposition mechanisms that he has identified as archetypical among all approaches: (i) behaviour-based decomposition, (ii) effect-based decomposition, and (iii) purpose-based decomposition. He hypothesised that any conversion could be achieved either by following the path (i) → (ii) → (iii), which is from less abstract to more abstract, or the other way round. The author demonstrated a (i) → (ii) conversion and a (ii) → (iii) conversion, identifying conceptual differences between approaches and, based on those, typical information elements that might get lost in abstraction. To preserve this information, which is mainly related to function–means couplings, additions to representations are proposed. In its current form, the strategy is not suitable for new product development, where no product structure and solution elements are given. This is left for further research, as is elaboration of the other conversion direction, rigid formalisation of the strategy and scientific validation.

The second article on conversion entitled ‘Two ontology-driven formalisations of functions and their comparison’, by Garbacz et al. (2011), advances a formal take on conversion. As a basis for developing automated translation algorithms, the authors propose comparative ontological analysis of functional modelling approaches. They showed for two approaches, namely for the functional basis approach and for the functional representation approach, how ambiguities become explicit by formalising function-related notions and by categorising these according to the principles provided by the dolce ontology. The procedure includes the construction of algorithms to translate functional descriptions from one ontology-based (ontologised) approach to the other. It has been proved through the presented formalisms that the procedure works for the two given, manually ontologised functional modelling approaches, and that functions taken from a sample functional model could be successfully compared. However, assessment of the practical usefulness requires further empirical validation.

The fourth article also addresses translation of function-related descriptions, but in this case, the conversion aims to enable exchange of knowledge on solution elements for high-level functions between engineering and biology, which is a completely different domain from the perspective of using functions and functional representations. The ‘Proposal of a technical function grammar oriented to biomimetic’ presented by Rosa et al. (2011) aims to remove the barriers that hamper designers in their reasoning with analogies by stimulating and inspiring them with knowledge available in another domain. Towards this end, the authors present an extension of the standard verb and object type of function formulation and show how their conceptualisation is operationalised as a database-driven computer system with interfaces tailored to engineers and biologists, respectively. The database is currently at an operational stage that allows data entry and demonstration, but the authors intend to extend and improve its structure and interface before it will be empirically evaluated.

In the last article, entitled ‘A protocol to formalise function verbs to support conservation-based model checking’, Sen et al. (2011) explore opportunities to check mass and energy conservation within flow-based functional models. They propose a formalisation protocol that they have applied to three elementary function-building blocks (verbs) selected from one particular functional modelling approach. The authors demonstrate how the resulting formalisations can be used to (i) check for correctness with respect to conservation laws, (ii) identify and handle residual flows efficiently, and (iii) check the feasibility of realising the requested output when the available input is given. Conservation laws have limited relevance for signal flows. The authors aim to include conservation checking of the mass and energy flows that carry signals in the future, but one may wonder how conservation issues related to information carriers may actually present a bottleneck to designers in practice. Considerable effort is still needed towards software implementation. The selected modelling approach (or any similar approach to which the protocol is claimed to be applicable) consists of more than the three addressed verbs, and validation in practical situations awaits further research.

If the five articles selected and presented in this special issue are representative of the latest research efforts and insights into functional modelling, then the impression arises that the ambition to support designers in their reasoning from the abstract to the concrete with computer tools has been fading. Nowadays, much more modest goals are pursued. It is generally accepted that there exist many alternative approaches, but none of them is perfect, and that this abundance comes with a confusing amount of terms and notions, with the same or similar notions being used in different approaches with slightly different meanings. Instead of addressing the issues of a (semi-)automatic conversion from abstract representations to embodiments, automated conversion between representation approaches at the same level of abstraction and conversion across domains appears to be a preferred (and a more rewarding direction) for current research. However, as long as designers do not use computer systems that help them devise functional models, most of the initiatives towards automatic conversion between different function representations remain on the level of academic exercises. Tools supporting functional knowledge exchange with areas outside engineering design, such as biology, are perhaps closer to practical needs, feasibility and applicability. Such a tool can be noncommittal and maybe less threatening to designers since it can be realised without having to enforce functional modelling and decomposition on designers. However, designers have to make abstraction from the physical manifestation of a given solution element (e.g. a part for which an alternative physical manifestation is needed) towards its functions, in order to be able to synthesise alternative manifestations that fulfil the same functions in another domain.

Some fundamental problems with functional modelling and decompositions can be demonstrated by taking the last article of the special issue as an example. In this article, the proposed formalism is applied not only to support automated reasoning from the abstract to the concrete, but also to specify functional models for enabling a particular type of analysis. The concretisation, which is required in order to achieve enough detail through function decomposition, is done manually. Nevertheless, it remains abstract in that it does not reach the level of physical realisation. It can be seen however that even when staying in the realm of abstract descriptions, the required level of detail forces the decomposition to go beyond solution-independent elements that purely describe what the product has to do. At some point of detailing, conservation-based model checking requires specification of ‘unwanted functions’ such as ‘export thermal energy’ for a light source, of which the designer's intention has been to produce light, not heat. Such ‘functionality’ can only be modelled with specific solution elements in mind, since checking will give different outcomes for, say, a tungsten lamp and a led source. However, if these elements are known, which is true for the existing product presented in the article, typically more sophisticated simulation tools are widely available, and function analysis does not seem useful. This raises fundamental questions like ‘how far can we decompose if we want an unbiased analysis of a product?’, ‘how long can solution independence be maintained during decomposition?’ and ‘when can concrete solution elements no longer be disregarded?’.

As a final remark, we would like to point out one other shortcoming in current work as presented in this special issue and in other recent publications on functions. There is still no aspiration among researchers to consolidate their findings based on rigorous validation, involving large amounts of human subjects (designers), sample products and cases in an industrial setting. We think that in the coming years, researchers seriously have to consider such scientific underpinning of theories related to technical functions.

We would like to thank the authors for their contributions and their efforts to contribute and to incorporate the reviewers’ comments, recommendations and critique. Also, we are grateful to the anonymous reviewers who dedicated their precious time to warrant the quality of the end result, and the Editor, Prof. Dr. Alex Duffy, for his support during the realisation of this special issue.