Search Results

Automobiles have a high degree of mechanical and electrical complexity. However, product complexity has the accompanying effect of requiring high levels of design and process oversight. The net result is a product creation process which is prone to creating failures. These failures typically have their origin in an overall lack of complete understanding of the system in terms of materials, geometries and energy flows. Despite all of the engineering intentions, failures are inevitable, common, and must be dealt with accordingly. In the worst case, if a failure manifests itself into an observable failure the customer may have a negative experience. Therefore, it is imperative that design engineers, suppliers along with reliability professionals be able to assess the design risk. One approach to assess risk is the use of degradation analysis. Degradation analysis often provides more information than failure time data for assessing reliability and predicting the remnant life of a system.

Warranty forecasting of repairable systems is very important for manufacturers of mass produced systems. It is desired to predict the Expected Number of Failures (ENF) after a censoring time using collected failure data before the censoring time. Moreover, systems may be produced with a defective component resulting in extensive warranty costs even after the defective component is detected and replaced with a new design. In this paper, we present a forecasting method to predict the ENF of a repairable system using observed data which is used to calibrate a Generalized Renewal Processes (GRP) model. Manufacturing of products may exhibit different production patterns with different failure statistics through time. For example, vehicles produced in different months may have different failure intensities because of supply chain differences or different skills of production workers, for example.

The most common used warranty prediction method at component level (non-repairable system) is called Weibull analysis. In Weibull analysis, failure time is assumed to follow a certain distribution such as Weibull, and time is the only predictor in the model for predicting percentage of failures. However, other variables such as design variables, manufacturing parameters, and field use condition also affect warranty. These variables should be considered in the prediction. In this paper, a multiple regression approach is proposed to predict warranty failures of a solenoid switch by considering multiple factors that affect the warranty. A single failure mode caused by fatigue is studied. The failure is caused by out of GD&T (Geometric Dimension and Tolerance) specs. These GD&T variables together with component operation time are used as predictors in the model. The final model is established by integrating physics of failures with statistical analysis results.

Pre-production vehicle validation is a critical step in understanding what potential issues end customers may find. Road profiles used in vehicle level tests are critical in finding failures. Clearly, if all the vehicles are tested only on highway, many failures will not be discovered. Therefore, using the right road profiles is very important. Traditionally, customer survey data is used to identify an appropriate road profile by defining a route that represents the Xth percentile customer. In this paper, a clustering method is applied to group all the customers into several groups. Each group is represented by a single road profile, and the entire customer population can be represented by multiple road profiles. If vehicles are tested using these profiles, then the road test can better represent the field condition, and hopefully failures can be discovered more efficiently.