14. Human Factors in Production Systems

Traditionally, social science research deals with the understanding of the human factor, i.e. explaining, measuring and predicting human experience, affective states, cognitions, and behaviour. This includes the capturing of emotional states and cognitive abilities in different application contexts, against the background of user diversity and the differences in attitudes within and across individuals. But it covers also developmental changes of humans across the life span as well as the systematic understanding, control and change of human behaviours.

The huge volatility of the human nature and its enormous adaptability to different situations necessitated the development and establishment of standardized human factors’ methods and metrics that allow a sound, reliable and valid prediction of the human factor. Hence, on the base of their profound empirical methodology and their sound knowledge base of the human factor, social science can naturally contribute to the understanding of socio-technical systems, as e.g. the current challenges in production technology.

Basically, social science knowledge can contribute to different challenges:

With regard to work environment and design of production systems, usability and user experience research has been shown to exploit a huge benefit for diverse application setting within industrial environments. In many domains (perceived) usability is an established criterion for the quality of a product and is increasingly a deal breaker for the buying intention for consumer electronics. However, usability is not yet widely acknowledged for professional applications in production engineering (Myers et al. 1996; Ziefle and Jakobs 2010) and many interfaces suffer from complicated screen layouts (Arning and Ziefle 2009; Ziefle 2010a), not understandable labels and buttons, undistinguishable and unclear icons (Pappachan and Ziefle 2008), and steep learning curves. Interfaces that somehow work for specific user groups, but fail for the increasingly changing and increasingly diverse labour force (Wilkowska and Ziefle 2011; Calero Valdez et al. 2013).

The reason for investing in usability is simple: Good soft- and hardware usability is the reason for a higher quality of the achievements and increased productivity of the workers (Arning and Ziefle 2007; Ziefle 2010b). Also, good interfaces reduce the time and costs for training and support and increases the users’ work satisfaction, loyalty and commitment (Arning and Ziefle 2010). Furthermore, if usability is considered from early on in the design process and not an added on top in the final stages of the development, it can dramatically decrease the development time and costs (Nielsen 1993; Holzinger 2005). Hence, establishing usability as a core criterion for professional applications can increase the return of investment (ROI) and decrease the total costs of ownership (TCO) dramatically.

This section outlines a few but still the most central usability methods for user-centred design. An in-depth description and additional methods can be found in Dix et al. (2003) or Courage and Baxter (2005).

Methods for developing and evaluating user interfaces can be divided in methods with and methods without involvement of prospected users. Methods without user involvement, such as heuristic evaluation (when experts evaluate if interfaces meet certain heuristics), GOMS (a method for estimating the expected performance, similar to Methods-Time Measurement—MTM), or cognitive walkthroughs (human factors experts identify usability issues by predicting how users would solve tasks) are not covered in this article. They are a valuable addition to every development process, yet many issues will remain uncovered, if prospected users are not included during the design phases.

Wizard-of-Oz This method allows the evaluation of interfaces even if the respective backend functionality is not yet available, by observing the user’s interaction with the prototyping system and simulating its outcome. For example, a paper prototype can “control” a machine, if a hidden observer simulates the user’s interaction with a remote control.

AB-tests To understand if different interface alternatives result in higher speed, higher accuracy or higher user satisfaction different design alternatives can be compared by presenting both to a set of users, either within-subjects (every user uses every interface) or between-subjects (each user evaluates just one interface). Measures may be objective measures, such as task performance and learnability, or subjective measures, such as users’ perceived efforts.

Note that the feedback acquired changes with the perceived completeness of the interface: Paper prototypes are perceived as easy to change, thus users often request fundamental changes, while applications that appear complete are apprehended as difficult to change, hence the articulated feedback is often restricted to wording, colour choices, and other trivialities.

The methods presented are a necessity for building usable, understandable, learnable, goal-oriented, task-adaptive, efficient and satisfying software for production systems. They should as a matter of fact belong to the standard toolbox of all stakeholders included in the technical development process. To identify and obliterate usability pitfalls and to assure that the software captures the users actual needs, these and similar methods must be applied frequently during the design process. However, to unfold their full potentials, human factors experts must also be included in this process, as they are able to detangle individual differences in effectiveness, efficiency and user-satisfaction that are caused for example by motivation, personality, or cognitive abilities. These then allows a fine-grained tuned or individually tailored user interfaces.

The following three exemplary cases present “success stories” in which the methodology of user-centred design and human factors research was applied in the area of production systems.

The first case quantifies the influence of poor usability on efficiency while interpreting large data sets in supply chain management. The second case outlines how human factors relevant for good job performance in supply chain management can be identified. The third case describes how an adequately designed worker support system can relocate the focus between speed and accuracy depending on the task.

Managing the follow of material in supply chain management depends on both, the ability to perceive, understand and interpret given data correctly, as well as the presentation of the data. To understand how insufficient data presentation and bad usability impacts the decision quality, we conducted a formal AB-test. We measured the decision speed and decision quality in dependence on human information processing speed and the presentation form (poor and good usability constructed as small or medium font sizes in a tabular presentation of supply chain data). The study revealed that poor usability obviously decreases the overall performance. Strikingly though is the finding, that poor usability disproportionately impairs people who have a lower information processing speed, while faster information processors can compensate poor information presentation. This finding highlights the necessity of user-centred and participatory design, as software developers, interfaces designers and contributing mechanical engineers usually do not realize the negative effects of poor interfaces on decision speed and decision quality, as they are able to compensate the negative effects, while the end users often are not. Frequent user tests with methods as presented above will reveal these barriers even in early stages of the design.

Supply chains are sociotechnical systems with high dimensional and nonlinear solution spaces. The performance of a supply chains is not only determined by technical factors (e.g. shipping times, replacement times, delivery strategies, lot sizes, order costs, etc.) but also by the abilities of the human operators who needs oversee the possible choices and make good decisions in this complex solution space. To identify the factors that contribute to a better understanding of the supply chain and to develop methods that help supply chain managers to make better decisions in shorter time we developed a series of supply chain games (Brauner et al. 2013; Stiller et al. 2014). The business simulation games are virtual test beds with multiple uses: First, they are a flexible tool to identify and quantify human factors that contribute to efficiency and effectivity in managing information in logistics and supply chain management, by systematically varying the difficulty of the game and investigating how different human factors (capacity, processing speed, motivation, self-efficacy, personality traits) relate to game performance. Second, through experimental variation the user interface and/or provided decision support tools, the benefit or costs of these can be evaluated and quantified within the test bed before the proposed changes are implemented in commercial applications. Third, the relationship between in-game performance and performance in the job as supply chain mangers can be used to develop interactive methods for personnel selection with an increased accuracy.

The design of this research and evaluation framework followed the design principles presented above and several of the usability methods presented above were applied. The following sections describe the iterative development process and some of the methods used during the process.

The development of the business simulation game was a collaborative effort between the four disciplines mechanical engineering, communication science, computer science and psychology. Each of the four disciplines contributed methods in order to on ensure the simulation model’s validity on one side and the good usability and suitability for psychometrical evaluations on the other side.

At the beginning of the project the experts form each discipline discussed the game model and the relevant indicators for inferring the simulated companies status. Then the paper prototyping technique was used to arrange the indicators to form the user interface of the game. In a second step a low-fidelity software prototype of the game simulation was realized and the previously selected indicators were populated with data from the simulation model and the experts evaluated the suitability of the indicators and the simulation model. Third, the game was implemented as a web application and the design of the user-interface ware strictly based on the earlier prototypes and influenced by technical considerations. Forth, during one user study feedback from external usability experts was gathered and their suggestions were integrated in the game. A subsequent user study attested that the user interface refinements led to an increased profit of the simulated company as the users hat a better overview of the performance indicators and were able to make better decisions. Throughout the design process feedback was gathered from other experts and test users and the user interface and the game model was refined accordingly. Figure 14.2 shows the development progress of the game across three prototype levels.

The third test case of including human factors research and design methodologies in production engineering is the design and evaluation of a work support system for carbon-fibre reinforced plastic manufacturing (CFRP) (Brauner et al. 2014).

The CFRP production process relies on manual production steps in which multiple layers of carbon fibre cloth have to be aligned in specific orientations. The overall stability of a CFRP part is prone to misalignments and a mismatch of 5° reduces the mechanical stability of a component by 50 %. These defects can only be detected late in the production process, which yields in extra costs for the process steps between origin of the error and the detection.

To increase the stability of the process we designed a worker support system that alleviates the variances in the orientations of the carbon fibre cloths. The system captures the orientation of the currently placed layer and provides auditory and visual feedback to the worker. Later iterations of the prototype may be realized stationary in the assembly cell or on mobile augmented reality systems, such as Google Glass.

An evaluation of a rapid prototype of the system with the Wizard-of-Oz technique (compared four different feedback modalities against each other (no feedback, auditory, visual, and combined feedback). The key finding is that providing no feedback yielded in the highest speed and the lowest accuracy, while combined auditory and visual feedback led to slowest speed and highest accuracy. Hence, a target-oriented design of worker support systems can nudge workers to either increase the speed of the production process or to increase its accuracy (speed—accuracy trade-off).

Summarizing all three cases, utilizing methods from user-centred design and human factors research reveal a deeper understanding of the behaviour of human workers on different levels of production processes. The methods facilitate the development of better systems and applications for production processes in shorter time. Adequately designed system can influence the production processes by shifting the trade-off between speed and accuracy by giving carefully designed feedback. Furthermore, some methods allow an explanation and prediction of individual differences in speed, accuracy, performance and motivational factors that might either contribute to better designed and better targeted software systems or to more specialized recruiting and training processes for the employees.

A common problem among interdisciplinary teams is the lack of a shared language and misconceptions about the other methodologies. Established systems for enabling interdisciplinary communication in teams are pattern languages. Christopher Alexander’s seminal work suggested these languages as a method to enable different stakeholders in urban planning (e.g. architects, civil engineers, city planers and residents) to collaboratively design the living space (Alexander et al. 1977). Each pattern describes a solution for recurring problems, defines a shared iconic name, captures the forces that argue for and against the given solution and refer to other patterns that relate to the solution, either as possible alternatives (horizontal) or superordinate and subordinate patterns (vertical). A network of interlinked patterns then forms a complete pattern language. Other disciplines adopted pattern languages as a tool to capture disciplinary knowledge, but sacrificed the aspiration for participatory design. In computer science the “Gang of Four” introduced software design patterns (Gamma et al. 1995) that quickly revolutionized the communication between experts in software engineering. In mechanical engineering (Feldhusen and Bungert 2007) suggested a pattern language to manage archetypal engineering knowledge, but again, this pattern language captured engineering knowledge by and for experts. It shows that pattern languages exist for various domains, but still Alexander’s original goal to enable all stakeholders to jointly develop holistic solutions got astray.

Thus, a dedicated pattern language for the design of production systems may enable truly participatory design in production engineering and a more efficient collaboration among interdisciplinary teams. The goal of this language must be to empower all stakeholders to understand the constraints of a given problem and overview the set of possible solutions. This language could cover individual competencies and methods of the contributing disciplines and will enable interdisciplinary teams to collaborate more efficiently on future production systems.

Recurring again to the generation Y, the novel attitude of workers might also requests a change within the performance culture in the production and work environment. In this perspective, the quality of “good interface of technology” relies on affective and hedonic aspects of work and production—attributes emphasizing individuals’ well-being, pleasure and fun when interacting with technology and technological systems, the quality and the design of products, but also the well-being of teams, working groups as well as the well-being of society, focusing on social morality, working ethics, work-life balance, environmental justice, or life style. To this end, the relationship of users and technological products and their working environment is of importance and the making sense of user experience. In addition, the work experience and domain knowledge of workers, end users and consumers of technology or technical systems is of high value (and is so far, mostly ignored). It seems indispensable for efficient production environments to focus on human factors in order to enable highly motivated and high performance teams not only steering with the traditional motivators—money, pressure, or competition—but rather to focus on the internal motivation of workers to contribute to the system effectiveness by including their knowledge and their expertise within iterative product development cycles.

Naturally, the relationship between leaders and workers need to be reformatted accordingly. Efficient teams then should be characterized by transparent group communication, a commonly shared information policy and the appreciation of ideas and innovations created by working teams. This team culture though requires trust in both, the team leader and the workers and might be a overdue performance driver of current working environments in enterprises that might ensure sustainable high performance cultures on the long run.