Wednesday, 10 August 2016

D4R Project Management

 Management of a Design for Reliability project should be the same of any engineering project, it should include a target, a scope, time and resources, both human and financials.


 Using agile project management methodologies, as Scrum methodology. is a good recommendation to manage this type of projects.

 Scrum allows an effective management of complex and high risk projects, ensuring results; in order to do it Scrum splits the project in short milestones, or Sprint, to work in parts of the project large enough to be considered a deliverable. 

 A Design for Reliability project could include the following parts:

a.  Definition of the level of reliability required by the customer, to set the reliability program goals. The product user is who should define the level of reliability, if the level is too high the cost is also too high and customer doesn't appreciate it; if the level is too low we will have claims and lost of trust, so we damage our brand. 

We can know our customer opinnion by surveys and by studying claims and warranties.   

b.   Product reliability assessment, in working conditions and for estimated operation time. The easiest way is to perform a qualitative analysis by Failure Modes and Effects Analysis (FMEA), it allows us define the failure modes, their causes, how to prevent / detect them, and their effects for users, and assess them by a Risk Priority Number (RPN). This methodology allows us to define a ranking of level of reliability of the product. 

c. Reliability modelling, showing the weakness of product and improvement opportunities. The modelling could be done by Reliability Block Diagram (RBD) and Fault Tree Analysis (FTA), these methodologies provide quantitative results and allow us to identify weakness of design and try new elements and settings.  

d.    The reliability functions estimation, they allow to perform a quantitative reliability analysis. When we have prototypes or real products working we could analyse real failure data and define the reliability function, failure function, probability density function (pdf) and failure rate function; with this functions is possible to calculate life data, warranties, etc. 

There are several probability distribution that allow to define these functions, Weibull distribution is the most common, it requires to calculate three parameters: shape parameter, scale parameter, and location parameter, that usually has a value of 0 in this type of analysis.



e.  Performing accelerated life tests, as Highly Accelerated Life Test (HALT) and Highly Accelerated Stress Screening (HASS), to confirm data and study possibilities of improvement. Testing prototypes in real conditions is too slowly and expensive, an alternative is to design an accelerate testing, increasing the stresses to induce a failure, most common factors are temperature, vibrations, electric parameters, humidity,... then testing the prototype to failure, and use power relations (as Inverse Power Law Relationship), exponential relations (as Arrhenius relationship or Eyring relationship) or mixed relations (as Temperature - Non Thermal relationship) to estimate the life of product under working conditions.

This methodology allows to modify the design and test the result in a faster and cheaper way, but require a right failure modes identification process to ensure the results are reliable. 

f.  Performing a reliability growth program, to reach the reliability target based in customer requirements. The Reliability Growth program should include the components discussed inside this post, development of test could be modelling by Duane model, Crow-AMSAA model, Lloyd-Lipow model, Gompertz model or Logistic model.