University of Oxford, MSc Major Programme Management
Course 3: System Design
Course Leader: Dr. Janet Smart
Paper: Summative Assignment
Date: March 2018
___________________________________________________________________
Case Study Assignment Option 1: The application of advanced systems engineering techniques to resolve the warranty impact of entropy and inter-connectedness within vehicle seating systems
___________________________________________________________________
Abstract
Over the past century significant achievements in technological development have transformed vehicle production in the automotive industry. A vast increase in electronic and mechanical systems, coupled with related demands on power and design, has created an array of new engineering opportunities and challenges; both of which have subsequently influenced the release of more features and greater options to the market. From the vehicle manufacturer and supplier perspective, the inclusion of greater system functionality within the associated systems of vehicles has grown in terms of complexity and inter-connectedness. Arguably, the increase in warranty concerns would suggest that the sheer rate of change and interdependence of these systems is now putting at risk the ability of engineers to fully understand and predict system behaviour. As an evolving means of better understanding the functional relationships of systems and sub-systems, the methodical approach to problem-solving referred to as systems engineering (Wymore, 1993) has become a recognised means to better comprehend the transfer function of performance for critical and complex systems. This paper leverages the pattern theory works or rather the “signals of failure” of Alexander et al. (1977) and the anti-pattern approach or “problem-solving” process advocated by Buschmann et al. (2007). More specifically, systems engineering in this paper is used in a post-production context as a mechanism to react to complex and intricate warranty concerns at the component-level design of vehicle seating. The case study leverages the applied pressure distribution and early ergonomics research from Zemp et al. (2016), and builds on the early biomechanical studies of Giuseppe et al. (2002) to enhance the design process to account for the resulting interactions of the car driver body with the front-seat backrest sub-systems.
An introduction to system design
As applied in this paper and case study, pursuant to Wymore (1993), the principal functions of systems engineering are:
• to develop statements of system problems comprehensively, without disastrous oversimplification, precisely without confusing ambiguities, without confusing ends and means, without limiting the ideal in favour of the merely practical, without confounding the abstract and the concrete, without reference to any particular solutions of methods;
• to resolve top-level system problems into simpler problems that are solvable; and
• to integrate the solutions to the simpler problems into systems to solve the top-level problem.
By performing these functions, systems engineering becomes a simple conduit between the super-system (vehicle) problems observed by customers, the investigations at the system (seat) level, and the solutions provided by technology and engineering at the sub-system (seat-back bolster) level. The methodical errors in process, policy, education and perception that typically lead to upstream engineering omissions and downstream warranty concerns can be vastly ameliorated through the application of a total system-level view that:
• takes account of the interactions of complexity and use;
• accounts for life-cycle use patterns;
• adopts a rigorous thought process to problem definition and solution design; and
• transfers the engineering solution to other similar products.
With regards to problem identification patterns and resolution thought process, Buschmann et al. (2007) offers a simple and substantive flow from definition and outcome impact:
• Identification: Name and classification to identify the pattern, for example, a particular type of failure effort defining a reoccurring warranty concern.
• Context: A general description of the circumstance, such as a time-series measure, a geographic state, or a physical condition of use that the failure effect is observed under.
• Problem: The situation that gives rise to the concern, as in the case of an interaction effect of product use and the component-level failure.
• Intervention: The action taken to address the problem, such as a technology or engineering design solution to resolve the issue at the component-level.
• Consequences: The impact of implementing a resolution, or in such cases, the impact of doing nothing.
An introduction to engineering complexity
In automotive seat engineering, the word “complex” is often used to describe a sub-system or component-level interrelationship that is difficult to understand or to solve. As stated by Edmonds (1995), there are many ideas that surround the concept of complexity. Indeed, the size of the system provides an indication of the difficulties to understand the behaviour. In addition, the inclusion in the design of components with different functions, increases the difficulties in engineering and performance modelling. In general, a complex system is the one that cannot be reduced to a simple scheme without diminishing performance behaviour (Kolmogorov, 1965). Moreover, the general definition of complexity given by Simon (1962) can be translated to vehicle engineering, in that a complex system such as a seat can be defined as one comprised by a large number of component parts that interact in a non-simple way with the occupant. In such systems, the different contributions of each part donate an effect to the whole system (Ay et al., 2006), such that the performance of the overall system is not simply the sum of component or sub-system parts.
Complexity resulting in warranty
The XYZ project is a privacy codename assigned to an executive and mid-size luxury car and estate vehicle produced by a British car manufacturer based in Coventry, England. It was first unveiled in the Autumn of 2007 as a significant mass-produced design change from its predecessor. To surface the comparative degree of engineering complexity of the new vehicle, over 45 meters of electrical wiring was used compared to 30 meters in the previous model, which is more than 6 times that of vehicles manufactured in 1955 (Leen and Heffernan, 2002). With reference to the front-seat, the released model and the later variants combined 13 fabric types, 4 different foam profiles, and 18 electrical functional options, meaning that where inter-connectedness is expressed as n(n-1), there are c.1,200 possible variations for customers to select from and for engineers to manage. The permutations of sub-systems express the amount of possible interactions stored, which evolves in a non-uniform way when a driver connects with the seat. In the classical thermodynamic definition, inter-connectedness is referred to as entropy, which is defined as the natural force which carries a system from a rest state to a probable or expected condition of performance behaviour. In vehicle mechanics, entropy is essentially a measure of the number of ways in which a mass-produced system may be arranged, which most notably for the XYZ project was c.35% greater than the earlier model. According to Zipkin (2001) mass customisation in this sense is a strikingly contradictory term, as mass implies a uniform product, whereas customisation connotes small-scale craft. It equally fails the logic of standard products reflecting standard processes that has typically underpinned mass-produced engineering solutions (Hayes and Wheelwright, 1979). Nevertheless, vehicle manufacturers as well as other markets have strategised that by combining the best of both promises exciting choices for consumers and new opportunities for businesses. Now, whilst for customers choice and personalisation are most certainly modern and desirable buying behaviours, this is often taken to be a measure of disorder by engineers: the higher the entropy, the higher the disorder.
As shown in Diagram 1, as a comparison of project XYZ with its predecessor,
Diagram 1: The warranty impact of increased entropy (XYZ Project, 2012)
disorder is elegantly exhibited in the ratio between warranty concerns and the degree of entropy over a 24-month period. As earlier referenced, the difference in entropy between the earlier model and the XYZ project is c.35%, which based on the rejected units per 1000 produced (r/1000) translates to a c.54% increase in warranty concerns requiring a post-production engineering intervention. Therefore, as an expression of Alexander’s pattern theory (1977), for every unit increase in the number of component-level interactions explained by the introduction of new features and functions, there is a corresponding increase in warranty concerns by 0.3.
Case study: Problem-solving with systems engineering on project XYZ
The specific case study references one such warranty concern of project XYZ. In April 2012, a warranty report was raised by a car dealer where the customer had complained of seat discomfort during long periods of driving. On review, the car dealer observed that the driver’s front-seat outer bolster area of the seat-back was exhibiting a large protrusion. To explain the broader pattern, from a variety of different geographic markets, 168 vehicles exhibited the same conditions of failure for both the inner and outer seat-back bolster, for which the average time-to-failure was c.18 months and the r/1000 was found to be 1.18 for returned seat units requiring replacement or engineering intervention.
Investigation
Once the returned seat units were disassembled, beneath the leather trim a tear in the polyurethane foam was identified at the point of deformation (Diagram 2). A comparative analysis of the warranty data confirmed that the issue was isolated to the foam variants where the front-seat incorporated a heating system or a comfort climate system. Notably, the sport variant eluded the error (Diagram 3).
Diagram 3: Comparison of foam variants and warrant data to isolate the source sub-system
A parameter diagram was used to analyse the inputs and outputs of system performance as a means to isolate the noise factors, or potential causes and mechanisms of failure; and to better understand the control factors in the design that could influence the desired system performance (Diagram 4). 22 noise factors were considered, for which 7 were force-ranked and prioritised for further study. Similarly, 5 control factors were evaluated and 3 were determined to be influential component properties that could explain the variation in warranty data between the sport and normal cushion seat-back foam.
Diagram 4: A “P” or Parameter Diagram of the signal and noise influence to the ideal system performance
A complete vehicle was acquired as part of the system engineering review, for which a full range of occupants from the 5th to the 95th percentile range were evaluated at the point of ingress and egress from the vehicle. A pressure mapping unit was utilised to record the results and the possible conditions under which the wear and deformation could occur. The initial study revealed that certain anthropomorphic types and load paths were a match for high load saturation at the points of interest (Diagram 5). A similar exercise was conducted in a sport variant which revealed a more benign pressure distribution, further reinforcing the warranty data of a significant difference. A more detailed comparison study was undertaken to identify the differences between the warranty units and the sport variant.
Diagram 5: Overlaid pressure mapping data showing egress load saturation (red hot-spot) at the bolster area
Sections of the sport variant bolster and the standard foam profile were examined (Diagram 6) and the foam thickness compared. The study revealed that the foam formulation was similar; however, the sport variant used a thinner foam profile at the bolster area, which was a contrary finding to engineering intuition (refer to the thought process explained in Diagram 4), in that it fails to explain why the sport variant did not exhibit the same failure effects as the standard foam seat-back.
Diagram 6: Foam bolster section comparison between the sport and standard form variants
At a greater level of detail, a further comparison between the two variants was undertaken (Diagram 7) which revealed that the frame and bolster support plate under the foam were common components. Importantly, whilst the foam profile of the sport variant was thinner about the bolster, it was observed that this variant included an inflatable bolster bladder for automatic bolster adjustment, as determined in the parameter diagram. Surrounding the bladder was a rigid carpet which enabled attachment.
Diagram 7: Component-level comparison between the sport and standard form variants
As shown from the rather subtle pressure distribution in Diagram 8, it was found that the additional carpet in the sport variant was acting as a load distributor for the hard casing of the bolster support plate under the foam, evident from the absence of load saturation points (red hot-spots) as observed without the carpet insert.
A robot load path was developed to replicate the failure (Diagram 9) and for the purpose of validating a design change (Diagram 10). With respect to the latter, incorporating the rigid carpet from the sport variant into the standard model was proven to be a successful warranty intervention and logical explanation of the performance transfer function, completing the investigative study.
The rigid carpet insert was communicated as a global containment, preventing the complete front-seat back replacement cost that had been the modus operandi. Commensurate with Buschmann’s problem-solving approach (2007) and Wymore’s perspective of solution transferability (1993), design guidelines were documented and introduced into the failure-mode analysis artefacts and design review processes for future projects. In parallel, a subsequent review was undertaken to determine whether the warranty concern could surface in other front-seat systems, to which the solution could be applied.
Conclusion
The engineering disposition is compounded by an overwhelming geographic and technology-cycle demand for more frequent releases of new products. Whilst typically a new vehicle release will be developed over a 4 to 5-year horizon, the macro-effects from the Eastern economic expansion (Chu, 2011) through the broadening south-south trade routes has materially altered Western business models to accommodate increasingly growing demands from new economically viable diasporas. Furthermore, technological cycles have increased in frequency from monthly releases to daily batch releases of new code and functions (Cusumano and Selby, 1997). Consequently, engineering product development cycles have materially reduced to complete for revenue and gain market share, and to accommodate the consumer demands to incorporate emerging technology trends. As shown in the case study, the implication of introducing a more complex product in even a comparable timeframe to a legacy product means that all parties in the value-chain are forced to consider a more risk-based approach to engineering and assurance. Therefore, assuming a consistent coverage of testing and given that the probability of warranty as determined by the defect density of error relative increased entropy and greater production volume, the likelihood of warranty is an increasingly common phenomenon that has seemingly become an accepted compromise of progression. Indeed, it could possibly be argued that at the heart of such increases in warranty is what Plato referred to as the “noble lie”: the circumstance in which for the greater business and consumer demand, a decision is made that is incongruent with engineering knowledge, experience and intuition. This would certainly underscore Hirschman’s principle of the “Hiding Hand” (Flyvbjerg and Sunstein, 2016): a resigned position that holds blind faith in the human ability to problem-solve unplanned and unforeseen events. If the contrary were true, or if the total cost of ownership (including warranty costs) were known at the outset, it is perhaps a safe assumption that many recognisable vehicles would not be produced in the absence of a greater obligation. Whilst systems engineering in this instance is shown as an influential pattern and anti-pattern mechanism to resolve complex and intricate post-production problems, arguably the real value as alluded to by Wymore (1993) is in advancing the thought process upstream to better address the increasing complexity of systems and the performance effects of entropy.