Adding new functions to a vehicle that are useful for the customers is one of the challenging tasks the DaimlerChrysler developers face. After all, premieres in automotive engineering highlight the company’s technological leadership. However, innovations also harbor a number of risks. No customer accepts “teething troubles” in technical advancements. To address this problem, a joint project has been set up between DaimlerChrysler’s Advanced Development unit, the Mercedes Car Group and the van business. The objective: Reliability has to become predictable from the very start.

We don’t measure the success of our work by how many accolades the scientific community heaps on us for it. Our work has been successful if we succeed in integrating our methods into the vehicle development process and if the users — that is, the development engineers — let us know that they can work better from now on thanks to our method,” says Alexander Bodensohn, department head in the Reliability and Diagnosis Laboratory at Advanced Development.

Although Bodensohn isn’t aiming to reap any scientific awards with the “Planned Reliability” project, engineers around the world are following the progress made by the industrial researchers with great interest. As far as they are concerned, the reliability project counts as one of the “hot topics” in the sector.

In addition to Advanced Development, the Mercedes Car Group’s Series Development and Quality Management areas as well as the Total Vehicle Test/Reliability Vans unit are also involved in this project.

Expressed more simply, the aim is to be able to assess at a very early stage within the vehicle development process whether a planned innovation functions reliably in accordance with requirements. After all, the customer is supposed to be satisfied with the product in the end, and the service area shouldn’t be confronted with unpredictably high costs resulting from warranty or goodwill expenses. Reliability must become more predictable, in the same way that advance assessment of the costs of creating a new product has become increasingly precise since the 1990s.

Although determining the reliability of a new function after it has been introduced requires a significant amount of work, it is by no means a complicated undertaking, because experience has already been gained with the component in question.

The numbers of customers who are satisfied or dissatisfied with it can be as easily determined as can the repair frequency and the costs that have already been incurred for the company within the scope of various warranties and goodwill payments.

However, the empirical knowledge gained at this late date leaves only a bitter aftertaste if, in fact, there was still a gremlin in the works of the new product in spite of all the tests.

The method used in the “Planned Reliability” project therefore focuses on the initial phase of vehicle development. “We want to evaluate concepts for product innovations with regard to their technical reliability, feasibility and the cost and time required, and to do that already in the strategy phase,” explains Bodensohn. The important decision concerning which functions are to be implemented in a new model is made at that point.
However, only the function itself has now been decided on. In most cases, the way it will ultimately be implemented has not yet been definitively determined. Generally, a whole range of alternative technical concepts is available for that purpose.

“Take, for example, the function for protection against something getting caught in electrically moved parts,” clarifies Bodensohn, using an example to illustrate the point. “The impending threat of something catching in a sunroof, a sliding door or a window can be registered by various sensors. In actuator engineering one can choose between hydraulic or electromechanical drive systems, and various approaches are possible for the controls.”

But which concept harbors the least development risk, can also be implemented with the least amount of time and expense, and works the most reliably — all at the same time? These questions must be responded to with clear answers based on a comparative concept evaluation within the scope of the reliability analysis. “The question we face is: How can we reach a substantiated assertion about the reliability on the basis of the little information that is available in this early stage?” says Ali Bilgic, the responsible team leader from the department, in outlining the key challenge of this project.

A proven method for arriving at a sound decision in such a situation is the expert discussion. Ultimately, it is the experts who can best judge where an idea can, for example, fall back on tried and proven technology components, which hurdles must still be overcome in order to transform a prototype into a production-ready product, and which technical aspects might possibly turn out to be an Achilles’ heel.

> The first step: Qualitative evaluation of the concepts

Such expert discussions have already been successfully held within the scope of the project in various pilot projects within Mercedes Car Group and Vans vehicle development.

“The experts from the respective departments go through each single partial function of a product concept with their colleagues from Quality Assurance,” says Joerg Mahrle, project manager and reliability expert, explaining the modus operandi. During the discussion, all aspects defining the concept’s reliability are examined and evaluated in a structured manner. At the end of this qualitative examination, the discussion participants summarize their individual assessments of the partial elements and the group subsequently reaches a comparative judgment with regard to the reliability and development feasibility of each individual concept. “Dissent among the experts is very rare; the divergences between the assessments are minor. And there is never a case where they are totally at variance,” says Mahrle.
“The exact numerical value that characterizes the overall development potential of a concept later on isn’t important at all,” adds Bodensohn. “The crucial thing is the discussion process itself, in which the experts with varying specialist backgrounds and experience contribute their knowledge and so ensure that the partial aspects that they believe are critical to a concept are addressed and jointly evaluated.”

> The second step: Defining reliability quantitatively

A further advantage of the concluding, documented concept evaluation is the decision’s transparency. Later on it is easy to see why, on the basis of certain characteristics, Concept A was preferred to the competing concepts or which disadvantages led to the discarding of Concept B. “What’s more,” Bilgic adds, “the comparative arrangement of the concepts on a bar that symbolizes their development feasibility from zero to one, also permits the evaluation of the concepts’ distances from one another.” In the end, it makes a difference whether the experts give a very clear reliability advantage to a certain concept or whether they judge two concepts to be very close together.

“The third result of this procedure is that there is also the question of whether the experts are ultimately really satisfied with the evaluation of the anticipated reliability of the favored concept,” says Mahrle. Maybe the “winning concept” in this relative evaluation comparison was not so great after all. Although it may have been the most promising of the tested alternatives, it was still not able to convince the committee. This too is a very important analysis result, because it means that the vehicle development team has no option but to rethink the technical implementation of the function. In this case the experts’ analysis of the various weak points helps to show the developers the direction in which alternative solutions can be found that could make a concept distinctly more suitable.

This qualitative concept evaluation is only the first of two steps in the project to make reliability more predictable. The next step is to make it possible to analyze the favored concept quantitatively as well. Stated more simply, a numerical value for the expected failure probability will be the result of such a “reliability calculation” of a product concept that has not yet been implemented.

“This subject area is still uncharted territory from a scientific perspective,” says Bodensohn, “and we still have a lot of work to do here.”

Thanks to a long-term cooperation agreement with the University of Stuttgart, the team was able to secure the expertise of the Institute for Machine Elements. Its director, Prof. Bernd Bertsche, is deemed to be one of the leading experts worldwide for reliability technology in automobile design.

The approach of quantitative analysis consists of putting the error rates and error behavior of a technical system’s individual components into a mathematical relationship with one another in order to arrive at an overall failure probability. Basically, this is also possible in the case of innovative products. After all, they also contain hardware and software components whose error behavior in existing products is known.

“The long-term goal of this project,” adds Bodensohn, “consists not only of calculating failure probability, but also of predicting how high the costs for product innovations’ warranties and goodwill settlements will be.”
But for all the scientific demands, the research team knows only too well that the “reliability analysis” tool will only earn a place in the vehicle developers’ day-to-day work if the users see an immediate benefit in it. The experience gained from the pilot projects gives Bodensohn grounds for optimism: “Of course we sensed skepticism at first from the developers, who feared that they would be saddled with more work in addition to the usual workload to be dealt with. But they soon realized that this procedure enabled them to work more efficiently.”

Bodensohn speaks in this context of “reduced complexity”. Even without a quantitative reliability analysis system, developers are of course already trying to predict the reliability of new components or functions. An important tool for doing so is the FMEA method. The abbreviation stands for “Failure Mode Effect Analysis,” which involves the systematic investigation of all possible failure cases and their causes as well as the effects of a malfunction on the vehicle.

In practice, these four letters often represent months of work by the entire development team — depending on the complexity of the product being studied. “We don’t replace the FMEA with the reliability analysis by any means, but we design the qualitative part of this analysis in such a way that a lot of data and parameters from it can flow quickly into an FMEA,” Bodensohn explains.

He sums up: “During the pilot projects, our colleagues in Development found out through experience that reliability can be assured more systematically, more transparently and more efficiently with the new procedure. Thus the method certainly fits into the existing landscape.”










Planned reliability in the development process



In computer-aided simulations the characteristics of components or even entire vehicles can be tested before they exist in reality. The basis of every simulation is a mathematical model of the physical behavior of the tested components — such as the stability of a vehicle upon lateral impact, the air flow behavior and heat distribution of an air conditioning system, or an object’s aerodynamics.



The advantage of test rig testing lies in the fact that individual components, such as a driver’s seat, or vehicle systems, such as an engine, can be checked in great detail. Such inspections, however, assume that the tested components are at least available as prototypes. The systemic behavior of a component in the overall vehicle is also difficult to check on test rigs.



In computer-aided simulations the characteristics of components or even entire vehicles can be tested before they exist in reality. The basis of every simulation is a mathematical model of the physical behavior of the tested components — such as the stability of a vehicle upon lateral impact, the air flow behavior and heat distribution of an air conditioning system, or an object’s aerodynamics.



The key test for evaluating functionality and reliability is the driving test in the experimental vehicle, the roadworthy prototype or a pilot series vehicle. The systemic behavior of the components to be tested can now show its true colors under near-real-life conditions or even in normal road traffic. The disadvantage is that if conceptual weaknesses become evident at this late stage, a lot of time and money is required to make changes.



The advantage of test rig testing lies in the fact that individual components, such as a driver’s seat, or vehicle systems, such as an engine, can be checked in great detail. Such inspections, however, assume that the tested components are at least available as prototypes. The systemic behavior of a component in the overall vehicle is also difficult to check on test rigs.



In computer-aided simulations the characteristics of components or even entire vehicles can be tested before they exist in reality. The basis of every simulation is a mathematical model of the physical behavior of the tested components — such as the stability of a vehicle upon lateral impact, the air flow behavior and heat distribution of an air conditioning system, or an object’s aerodynamics.


Copyright © 2006, DaimlerChrysler AG