Hoping your products are too good to fail? Find out using JMP® 18 Constant Stress Accelerated Life Test (CSALT)

[#1]

To predict the reliability of some parts or products, the time and resources required to test under normal conditions isn’t practical, because it could take years for a failure to occur. Instead, engineers accelerate the time to failure. That is, they test units under conditions that are more severe than normal use conditions, causing the product to degrade faster and fail more quickly than normal. They then use this accelerated failure data to predict product reliability at normal use conditions.

This technique, accelerated life testing (ALT), has been included in JMP for several releases. Based on customer feedback, JMP 18 offers a revamped, improved accelerated life testing design interface and approach.

I recently spoke with JMP Senior Research Statistician Developer Caleb King, who led the effort to create the new JMP 18 Constant Stress Accelerated Life Test (CSALT) capabilities.

What is an accelerated life test plan?

Our new capability, Constant Stress Accelerated Life Test (CSALT),  uses experiment design principles to determine the number of units to test at a variety of stress levels to obtain failure data that can be used to predict reliability under normal conditions. It is a new option under Special Purpose on the DOE menu.

What do you mean by accelerated conditions, and who would use them?

An accelerated condition is one that is more severe than normal everyday use of the part or product. Common accelerators include temperatures, voltage (electrical devices), speed (mechanical devices), humidity, a corrosive environment, or UV exposure (paints). Engineers use accelerated conditions to test product reliability.

In a nutshell, what are the JMP 18 CSALT steps?

Users interactively specify details for up to three accelerating factors for an ALT test. Next, they specify the characteristics of the ALT test plan. These characteristics include model assumptions. Then, JMP generates three alternative plans to consider. Users choose a plan and then simply press a button under the desired plan to make a data table for data collection. After the units have been tested and the data collected, users can proceed to use the table script to build a model for lifetime prediction.

Additionally, CSALT makes is easy to investigate alternative factor variations and plan settings within the same JMP window. Each set of factor variations is contained in a tab. Each variation of plan settings is also contained within tabs. Within each test plan, JMP generates three testing options for the user to consider.

CSALT generates three test plans automatically. Why three?

Users have specific resources and goals, so having three plans gives them options. The Optimal Plan optimizes the large-sample approximate variance of the quantity specified in the plan settings. It typically has the fewest test settings and is the most dependent on the validity of the assumed model. The Compromise Plan also attempts to optimize the large-sample variance, but with the additional constraints of a mid-point test setting with a minimum proportion of units allocated to that setting. The Balanced Plan places units at every combination of test settings, based on the number of levels per factor provided by the user. It typically has the most test settings.

Constant Stress ALT Design is a new option on the DOE Special Purpose DOE Menu.

Users identify up to three acceleration factors to include in their accelerated life test plan.

Users specify the details and model assumptions for their accelerated testing.

JMP generates three test plans, or designs, for the specified factors, distributions, number of units, and model assumptions.

Below, Caleb discusses the new CSALT, explains how a large local landmark might have benefited from CSALT, and talks about some of his JMP patents.

Caleb and JMP Senior Analytics Software Tester Jacob Rhyne demonstrate and explain the new capability on May 2 at the Developer Tutorial: Constant Stress Accelerated Life Testing. If you register for the live webinar, be sure to bring your questions for the integrated Q&A session at the end.