The interaction of complex technical tasks with organisational and human factors makes industrial product development possible for an archetypical case of complex problem solving. The scientific interest in cooperative product development is a result of the obvious importance of the design work for the later success of a product on the market. Until now, the emphasis of scientific investigations of product development has lied primarily on organisational and task focused questions. Consequently, approaches for supporting problem solving in design are technical- or procedure-oriented, such as in rules for embodiment design (see e.g. Leyer, 1971; Niemann, 1975), design methods for systematic product development work (see e.g. Altshuler, 1984; Andreasen & Hein, 1987; Cross, 1989; Ehrlenspiel, 1995; Hubka & Eder, 1988; Koller, 1985; Pahl & Beitz, 1997; Pugh, 1990; Roth, 1982), and various computer based tools for modelling geometry (“Computer Aided Design” CAD, see e.g. Spur & Krause, 1984) or for information- and product-data-management (see e.g. Meerkamm, 1998). But there are also new approaches which for example try to support the designers’ work on the basis of the analysis of dialogues exchanged between designers in design meetings (see Darses, Détienne, Falzon, & Visser, 2001).

The aim of this investigation was to understand the relevance and the interrelation of different factors influencing the course of the design work and thus the result. The complexity of such an approach makes it impossible to investigate so-called “comparable groups” in several companies. On the contrary, the investigation has to focus on a very detailed observation of “single cases” over an extended period of time. Therefore, we chose a single-case approach with a thorough analysis of few projects in industry. In this investigation, a group of engineers and psychologists observed, documented and analysed design projects without participating in the design work. In two phases of this investigation continuing for four years, altogether 40 weeks of cooperative design work were observed in five companies. The results of this paper are based on data of the first phase, which contains four projects with altogether 28 weeks in two companies.

The dynamic course of the design process requires a detailed description at short time intervals, thus a continuous on-line protocol of the design work was accomplished. Sitting in the same room a mechanical engineer observed the technical activities of the designers (e.g., working steps, sub-functions/components, ideas and solution variants), and a psychologist concentrated his/her observations on the social aspects of the design process (e.g., ways of decision-making and group interactions). Furthermore, video recordings of relevant phases of design work were made in order to review the description of the design process in interesting phases afterwards. The final protocol consisted of a word-by-word transcription of dialogues and a description of the entire process with an average duration of 30 seconds per protocol line. The qualitative and quantitative analysis of the design process (see section III) were based on these protocols.

Our main focus in regard to the individual prerequisites was on biographical data, on the designers’ evaluations of the project and the working conditions and on the individual abilities and competencies in dealing with novel and complex problems. From studies concerning individual and group-action-regulation behaviour in complex situations (Badke-Schaub, 1993; Dörner, 1996; Rasmussen, Brehmer, & Leplat, 1991), appropriate investigation methods are known which yield statements about relevant aspects of individual and group problem solving processes. The biographical data and designers’ evaluations of the working conditions were compiled mainly by means of semi-structured interviews whereas the ability of dealing with complex problems was assessed by making the designers solving three different computer simulated problems. Contrary to design tasks, these computer-simulated problems can be solved without any specified knowledge. They are used to study the ability of subjects to tackle novel problems in a complex, dynamic and in-transparent environment (Badke-Schaub & Tisdale, 1995). Dealing with these standardised problems, which are isolated from external conditions and pre-knowledge as far as possible, problem solving strategies become obvious and allow for making statements on the abilities of a person in dealing with complex situations. In addition the heuristic and social competences of the designers were compiled by a self-assessment questionnaire developed by Stäudel (1987) and by the analysis of the design work. Several studies on heuristic competence indicate that a positive self-assessment of problem-solving abilities supports successful problem solving in complex situations.

The aim of this investigation was to comprehend rules and deciding factors of co-operative engineering design work. Following our investigations, we accumulated extensive data on design work that allowed us to analyse result- and process-related questions. Nevertheless, the evaluation of the data called for a new approach that connected the data from the different fields (design process, external conditions, the individual and the group) and allowed a generalisation of the findings. This section introduces the “critical situations method” which was invented in order to evaluate the data of the design projects in terms of the influencing network of design work.

The basic idea of our method is the reduction of the documented design process to phases of routine work on the one hand and critical situations on the other hand where the design process takes a new direction on a conceptual or embodiment design level. In contrast to a colloquial understanding of “critical” we define a critical situation not as a negative property of a task, but we consider a critical situation as a complex network of interactions between task, subjects, groups and the embedded organisation with a positive or negative outcome.

The phrase “critical situation” sounds familiar with reference to the “critical incidents” by Flanagan (1954) or the “critical moves” by Goldschmidt (1996), but the method of critical situations refer to another theoretical background. We define designing as a complex problem solving process which requires several steps and operations in order to proceed from the starting point to the goal. The general problem solving process as described by many authors (e.g. Dörner, 1996; Ehrlenspiel, 1995) is the theoretical background of the definition of the types of critical situations. Thus, we differ between types of critical situations regarding their aim in the problem solving process, such as “goal-analysis”, “goal-decision”, “solution-search”, “solution-analysis” and “solution-decision” and situations that are important in their social context such as “conflicts” and “disturbances”. These situations require “conflict-management” and “disturbance-management” (see Figure 1). Of course, there are situations in the design process which are part of the problem solving context and the social context. For example there could be a conflict between the leader of the design department and a designer related to a goal decision. Then, both situations have to be analysed separately (see section III . 3).

The first step in the analysis and evaluation of the design process by the method of critical situations is the identification of critical situations. This was done by two researchers independent of each other, based on the compiled data of the design project. The interrater-reliability was recorded for each project and exceeded for each of the four projects by 90%.

The key to a generalisation of the findings from the single analysed critical situations is the assumption that the frequency of occurrence of the factors and relations in a particular type of critical situation represents their importance in the design projects. Thus, by summing up the models of one type of critical situation, we can identify the main factors and their relations responsible for positive or negative outcomes of the specific type of critical situation. On the basis of this analysis we can answer questions such as: Which are the main factors responsible for a deficient analysis of goals? Or, which are the mechanisms leading to a successful conflict management? Moreover, we can quantify the crucial factors and relations that determine quality, time and cost of the result with positive or negative consequences (Badke-Schaub & Frankenberger, 1997; Frankenberger, 1997). In the following section, we want to introduce and discuss some central mechanisms leading to success and failure in different types of critical situations in cooperative design work. We label a causal chain between central factors leading to a certain result as a “mechanism”.

First of all, we want to give an overview of the basis of the analysis of mechanisms in types of critical situation. Figure 3 depicts how often the different types of critical situations occurred with positive and negative consequences.

As the example at the beginning of this paper illustrates, a deficient goal analysis can be caused by a non-sufficient clarification of requirements. Of course, there are more reasons which may lead to an unsuccessful goal analysis. Figure 5 depicts the mechanisms in 18 situations of insufficient goal analysis. The thickness of the arrows represents the frequency (in percent) of the relations occurring in this type of critical situation. The thickness of the frames depicts the frequency (in percent) of the factors identified in all critical situations of unsuccessful goal analysis. Numbers lower than 10% are not depicted in any of the following models.

Designer “B” is searching for a concept for the fleece surface control system of the height-adjustable equalising-device. He thinks about a swivelling strip of plate and asks his Colleague “C” from next beside him to generate an idea for the strip-support: “... I thought I’d support this near the centre of gravity, and very little force will swivel the strip...” Together they develop two alternative solutions in an engaged discussion: A swivel support by an edge and by a pivot pin. Designer “B”: “... We need a very easy axis. This means the smaller diameter is the better. I’d take a little DU-sleeve, 6 mm...” And Designer “C” warns: “... Don’t make it too small...”

While redesigning a pneumatic fruit press the inexperienced Designer “C” asks the Electrician “K” for the required components of an electrical juice trough drive. The experienced Electrician “K” points out the problems of an electrical drive such as power failure, the blocking of the trough by pulp on the guides and the effort needed for an electrical variant compared to a crank handle and other problems. In the same situation an expert from the Client Consulting Department discourages the Designer “C” from deciding to take the electric drive.

Another episode in another design group:

In the previous section, we illustrated some important mechanisms which result in successful and unsuccessful design processes in different types of critical situations. As one consistent out come, the importance of communication was emphasised in all types of critical situations. This thesis is underlined by the following results, which were gained by analysing the factors “communication” and “information transfer” in critical situations in greater detail.

In our investigations, during about 70% of the whole working time, the designers were working individually, even though nearly 90% of critical situations were performed in collaboration. Obviously, communication between colleagues is extremely important for the exchange of decisive information. In additional interviews in ten R&D departments of major German companies the importance of colleagues as the most frequently mentioned source of information was underlined (Badke-Schaub & Frankenberger, 1998).

When analysing the factor “information availability” in critical situations, we can differ between “active” information requisition and “passive” reception of important information. Figure 9 depicts how often focused questions and individual information search were observed in critical situations. Obviously, the individual search for information was less successful than focused questions. Contrary to the active search for information, one need not wonder that the passive reception of information always goes along with a good availability of information. Passive reception of information can happen unintentionally as a part of an informal conversation between colleagues or can be intended by a colleague who wants to transfer information.

In this paper, the concept of critical situations was presented as an instrument for analysing cooperative design processes. The method of critical situations seems to be a suitable instrument for detecting the different fields of influences in the complex network of cooperative design processes. But this method is not limited to the area of product design; it is moreover a method for analysing complex working environments in general. In hightech disciplines such as aviation, chemical industry and medicine daily work consists of repetitious routines but also of different types of critical situations. In particular, the results of critical situations that end up as disasters appear highly visible and create great interest. In a Lufthansa research program on flight safety (Kemmler, 1999), 1,718 pilots were asked about their latest safety-relevant event in a very detailed questionnaire. On the basis of this huge data pool, a lot of interesting results were found which are very close to some of the results we found in the reported investigations: The data revealed that social relationship factors played a greater role than human error, operational or technical problems in the occurrence, risk, and mastery of safety-relevant incidents. The most important factors were poor quality of communication, inadequate information management and a deficient relationship climate. Furthermore it was shown that organisational conditions have an obvious influence on the safety behaviour of pilots. And last but not least, the data revealed that incidents are not due to one reason but the result of a complex and multi-level process (see also Reason, 1999).

The research project “Teamwork in Engineering Design Practice” was supported by the Deutsche Forschungsgemeinschaft (DFG).