Vextec (United States)

Laboratory testing is essential in understanding the fatigue durability limits of a medical device. Only a limited number of tests, however, are feasible considering the time and cost of manufacturing and testing the device. Many geometries and materials are available as design alternatives and testing is typically constrained to a limited number of devices. Additionally, physical test method capability typically has limitations and uncertainty for such breath of design alternatives. The time and cost is greatly magnified if the designs do not meet the durability test requirements. This leads to an important question, “How do we combine results from the laboratory testing of similar designs and similar materials with state-of-the-art computational methods to create the best possible understanding of the design space?” The Virtual Life Management® (VLM) methodology accomplishes this by creating a Virtual Design of Experiment: a computational method that combines structural analysis (FEA) data, material characterization models and probabilistic methods.For a specific Context of Use, using a computational approach, a) the structural analysis (FEA) models are used to understand the effect of the boundary conditions and loads on the stresses, stress gradients and stressed volumes, b) the material models are used to understand the damage induced by the stresses, and c) probabilistic and Monte Carlo methods are used to extend the design of experiment to high reliability levels and understand the design space at the statistical tails of the durability distribution i.e., Minimum Device Lifetime. The sensitivity of the minimum lifetime to the design parameter will be different than the sensitive at nominal lifetimes. Only by predicting the tails of the distribution can we quantify the durability of the design in the marketplace. This process is pictorially depicted in Fig. 1.Boston Scientific has been working with VEXTEC to create probabilistic microstructural material models for computational strength prediction and then combining these with structural analysis (FEA) and Virtual Life Management to predict risk i.e., probability of failure of a device. Two specific devices and Contexts of Use include: a) pacemaker leads made of MP35N material subject to two amplitudes of bending fatigue and b) endoscopic airway stent made of Nitinol material subject to a crushing load to represent stent deformation depicting a cough. For the airway stent effort, VLM was able to estimate the fatigue lifetime of stent design before stent is manufactured and tested. The simulation was able to show difference in average fatigue lifetime for two Nitinol alloy lots with different inclusion contents (and both complied with the specification (ASTM2063-05)). This was confirmed by the few (∼10) experiments completed. Significantly, VLM was able to predict a large difference for the −3σ fatigue lifetime for the two flavors of Nitinol. (Experimentally this would have required a large number of tests.) For the pacemaker leads project, VLM was able to characterize the lead design as a function of material microstructure (grain size, inclusion size and density) while using residual stress as a calibrating parameter (shown in Fig. 2). This study was able to demonstrate VLM's ability to iterate with material substitution, vendor management, sensitivity analysis, design trade studies and design optimization. To summarize, merging material science with computational methods opens a whole new level of engineering capabilities specifically uncertainty quantification and probabilistic distributions associated with microstructural characterization; a key step in supporting the verification and validation framework being developed by the ASME V&V 40 standard for medical devices.Although the work to date is early stage, the favorable results have provided confidence in the methodology and toolset; and insight to the potential application in the formal medical device product development process.

Many studies have confirmed that electronic failure is not a static or random event, but a process of material changes. These changes may not cause loss of operation for the electronic system, even when the components are technically out of specification. This self- compensating aspect of electronics makes diagnostics of failures particularly difficult and contributes to the ubiquitous problem of cannot duplicate (CND) failures. In addition to self-compensation, degraded failure modes of components manifest in different operating conditions and may offset one another depending on circuit configuration.12 With these considerations, VEXTEC is developing an intermittent fault reduction methodology which considers the material changes in the electronic components on the circuit board and provides a probability of true fault when a fault is detected by built in test (BIT). By modeling the electronics as a system rather than focusing on individual components, VEXTEC can model an entire circuit board with a standard library of models as part of the OEM's normal design cycle. The model is then used with the off board PHM (Prognostics and Health Management) system within a vehicle or other high reliability electronic system to predict the probabilities of intermittent failure. This modeling capability has far reaching effects on the diagnostics and prognostics of an electronic system. The operators can manage the fleet by risk rather than by guessing the level of degradation or when intermittent failures will occur. Organizations responsible for quality, maintenance and repair can minimize the impact of CND faults on the supply chain and on troubleshooting. The order of components replaced during a fault event can even be prioritized according to the probabilities provided by this algorithm.

Prognostics and health management integrated into complex systems and platforms such as the joint strike fighter (JSF) and future combat systems (FCS) will enable autonomic logistics and significant reductions in lifetime operating costs. VEXTEC has developed a system of physics of failure based modeling for electronics components which can predict component reliability without adding weight, sensors or complexity to the system. These models incorporate the already available environmental and usage data to evaluate the stresses on individual components within a system to provide an overall health indication for that system. Reliability modeling allows for timely and efficient planning of spares, maintenance and readiness to meeting operation needs and reduce cost for the armed forces. This paper shows the development and implementation of the process.

This paper presents a method for predicting fatigue failure using a virtual prototyping software tool that allows the simulation of real material behavior. Computer models are used to simulate the three dimensional microstructure in which fatigue evolves. Because grain structure properties are randomly distributed through any macro-sized structure, Monte Carlo simulation is used to give a probabilistic distribution of fatigue failure outcomes over the operating life of the structure. The reliability of structural elements with complex stress distributions is predicted by integrating the fatigue simulation model within traditional structural finite element models (FEM). The reliability analyses results for all the individual elements are combined using system reliability modeling techniques to determine the fatigue reliability of the entire component. This results in prediction of fatigue reliability as cycles to failure probability or the probability of exceeding a fatigue threshold. In making the fatigue reliability predictions, only the material micrographs, stress-strain curves and a single fatigue test data point (at a single stress) were provided. The results show excellent correlation between the predicted fatigue life and experimental fatigue life