10 years/150,000 miles reliability: What does it mean?
Small & Medium Vehicle Centre Vehicle Centre Quality Review September 2, 1998
Note: Since this paper was first written, and filed in the Sept. 2, 1998 VCQR binder, some minor corrections to grammar and wording have been made. These changes are highlighted by avertical bar on the left hand side of the page, similar to the one opposite.
Prepared: Tim Davis (TDAVIS5) SVC Quality Office 8-70-33149 D-MC/PI-1
Purpose: to clarify the definition of 10 years/150,000 miles (10/150) reliability, as it relates to automotive engineering, and to frame the discussion within the context of how we should set reliability targets for vehicle programs. i.e. we tryto answer the question “what does 10 years/150,000 miles reliability mean?”. A good starting point is the AVT definition of 10/150 reliability, developed from pre-Ford 2000 reliability work in NAAO and EAO, and is “Satisfy Customer expectations for reliability throughout a vehicle useful life of 10 years or 150,000 miles (whichever is tougher)” Background: Reliability, as a mathematical concept,represents a measure of probability that a specimen remains free from failure within a specified period of either years or miles or cycles. This concept is sketched out in Box 1 on the facer. Ten years is considered a “reasonable period” over which a vehicle is used, and about 10% of our customers do more than 150,000 miles in this period. An engineering definition of reliability is failure modeavoidance. Avoiding failure modes increases the probability of failure free operation. Types of failure: The definition of failure is crucial to reliability assessment. Failures can be soft (degraded performance to an unacceptable level) or hard (product function ceases). Usually hard failures relate to components, while soft failures relate to systems. Professor Don Clausing (MIT) has categorizedthe causes of failures into two types. These are 1) failures due to mistakes, and 2) failures due to lack of robustness. A mistake is a reliability problem that could have been avoided by the application of current engineering knowledge (for example design guidelines detailed in System Design Specifications), and avoidance of these types of failures is primarily a matter of vigilance. Theoccurrence of these failures in the field is hard to predict upfront, since much of our engineering activity is directed at not having these happen at all. Any prediction of mistakes actually occurring is an admission of failure in our engineering process, which we try to put right before launch (e.g. FMEA/design reviews, Fresh Eyes reviews, and application of the Campaign Prevention process). Failuresdue to 2) are bound to occur since this represents the robustness of the design to combinations of the 5 noises (see Box 2 facer), and there will always be noise combinations in the field that could lead to failure that either we did not foresee, or that we did not test for in the DVP. The objective of robustness engineering is to minimize the occurrence of these failures, but they can never beeliminated completely. This leads to the obvious question of trying to measure the reliability (for example, as illustrated in Box 1) of both soft and hard failures for robustness problems. This problem is discussed on page 2. Noise conditions: Failures caused by robustness problems are dependent on the actual noise condition that is encountered in the field. For example, high speed auto-bahn drivingin EU (noise #3), coupled with excessive variability of the bearing diameter (noise #1) causes needle bearing failures in the CD4E transmission at the rate of about 5% within 1 year. These failures are not observed in the US, where speeds are lower. The intent of the Key Life Test is (or should be) to reflect the appropriate noise condition (using the “noise tree” in Box 2) to generate the...