LESSONS LEARNED IN PHYSICS-BASED RELIABILITY Fayssal M. Safie, PhD, - PowerPoint PPT Presentation

LESSONS LEARNED IN PHYSICS-BASED RELIABILITY Fayssal M. Safie, PhD, A-P-T Research, Inc., Huntsville, Alabama SRE Monthly, March 10, 2020 A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 1

AGENDA  Definitions  What is Physics-Based Reliability?  Why Physics Based Reliability  Where Does Physics- Based Reliability Fit in a “Reliability Case”  Case Studies - Discussions  Bearing Inner Race – A Design Development Case  Frangible Joints – Understanding Failure Mechanism and Design Margins  Challenger O-Rings – A Design Failure Case  Columbia External Tank Foam – A Process/Design Failure Case  Concluding Remarks A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 2

DEFINITIONS  Mission Reliability prediction is the process of quantitatively estimating the mission reliability for a system, subsystem, or component using both objective and subjective data. Operational reliability prediction techniques are dependent on the degree of the design definition and the availability of the relevant data.  Process Reliability is the process of mapping the design drivers in the manufacturing process to identify the process parameters critical to generate the material properties that meet the specs. A high process reliability is achieved by maintaining a uniform, capable, and controlled processes.  Design Reliability prediction is the process of predicting the reliability of a given design based on failure physics using statistical techniques and probabilistic engineering models.  Reliability Demonstration is the process of quantitatively demonstrating certain reliability level (i.e., comfort level) using objective data at the level intended for demonstration. Models and techniques used in reliability demonstration include Binomial, Exponential, Weibull models, etc.  Probabilistic Risk Assessment (PRA) is the systematic process of analyzing a system, a process, or an activity to answer three basic questions: What can go wrong that would lead to loss or degraded performance; how likely is it (probabilities); and what is the severity of the degradation (consequences). Reliability engineers and probabilistic risk analyst need to understand the terminology and the purpose of each one to help program management to optimize system reliability for better safety, affordability, and mission success. A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 3

HOW DO THEY STACK Process Reliability Design Reliability (process capability, uniformity and (Physics based) control) Demonstrated Mission Reliability Surrogate Data, Test Reliability (based on objective Data, Field Data, (Based on Objective data) and subjective data) Generic Data Expert Elicitation/Bayesian Analysis common cause, human error, external events Probabilistic Risk Assessment (PRA) A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 4

WHAT IS PHYSICS-BASED RELIABILITY?  Physics-based reliability is a methodology to assess a component reliability for a given failure mode using engineering failure models.  The component is characterized by a pair of transfer functions (Strength and Stress) that represent the load (stress, or burden) that the component is placed under by a given failure mode, and capability (strength) the component must withstand failure in that mode.  The variables of these transfer functions are represented by probability density functions.  The interference area of these two probability distributions is indicative of failure. Material Strength Operating Stress Failure Region Note: In general, structural engineers use high safety factor where affordable to achieve high reliability (A deterministic approach)) A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 5

WHAT IS PHYSICS-BASED RELIABILITY? THE NORMAL CASE Failure Region Stress f(s) Strength f(S) µ s µ S Assuming both the stress and strength are normally distributed, the following expression defines the reliability for a structural component. If: Note 1: In general, reliability is defined as the probability that the strength exceeds the stress for all values of the stress. Note 2: Normality assumption does not apply to all engineering phenomena; and, under these special circumstances when the Normal does not apply, different methodology is used to determine reliability. A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 6

WHY PHYSICS – BASED RELIABILITY? Design Reliability Process Reliability Physics-based reliability is key to understanding failure mechanisms and µ s µ S identifying critical design and process variables  The design part is physics-based mainly driven by the loads and environment vs. capability.  The process part is driven by process capability, process uniformity, and process control.  The chart shows that critical design parameters (on the left) are mapped in the process (on the right). The result is a set of critical process variables which are assessed for process capability, process uniformity, and process control. A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 7

WHERE DOES PHYSICS-BASED RELIABILITY FIT? Reliability Program Management & Control Reliability Contractors and Reliability Reliability Failure Review Program Plan Suppliers Monitoring Program Audits Progress Reports Processes Process Reliability Selected Design Reliability Requirements Reliability Elements Identification of Design Reliability Requirements Failure Modes and Effects Analysis Reliability Drivers Analysis Identification of Critical Reliability Requirements Physics-based Process Parameters Allocation Reliability Analysis Process Characterization Reliability Prediction Human Reliability Analysis Process Uniformity Reliability Accelerated Testing Process Capability Case Sneak Circuit Analysis Process Control Reliability Testing Process Monitoring Parts Derating A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 8

CASE STUDIES - DISCUSSIONS A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 9

BEARING INNER RACE – A DESIGN DEVELOPMENT CASE  During rig testing, the High Pressure Fuel Turbo-pump (HPFTP) Bearing of the Space Shuttle Main Engine (SSME) experienced several cracked races. Three out of four tests failed (440C bearing races fractured). As a result, a study was formulated to:  Determine the probability of failure due to the hoop stress exceeding the material’s capability strength causing a fracture.  Study the effect of manufacturing stresses on the fracture probability for two different materials, the 440C (current material) and the 9310 (alternative material). The hoop stress is the force exerted circumferentially (perpendicular both to the axis and to the radius of the object) in both directions on every particle in the cylinder wall. A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 10

BEARING INNER RACE – A DESIGN DEVELOPMENT CASE The Simulation Model Operating Stress Material Strength A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 11

THE ANALYSIS RESULTS Test Race Failures in Failures Configuration 100,000 firings** 440C w/ actual* 3 of 4 68,000 mfg. stresses *Ideal + abusive grinding 440C w /no mfg. N/A 1,500 **Probabilistic Structural Analysis stresses 440 C w/ ideal mfg. N/A 27,000 Manufacturing stresses Induced stresses 9310 w/ ideal mfg. 0 of 15 10 effect stresses  The results of this analysis clearly showed that the 9310 material was preferred over the 440C in terms of the inner race fracture failure mode.  Manufacturing stresses effect for the 440C material was very significant.  Material selection has a major impact on reliability.  Physics-based reliability analysis is critical to perform sensitivity analysis and trade studies for material selection and testing. A-P-T Research, Inc. | 4950 Research Drive, Huntsville, AL 35805 | 256.327.3373 | www.apt-research.com ISO 9001:2015 Certified T-20-00200 | 12