Long complicated article stating Weibull analysis of predicting track defects ia severely limited. Maybe MC can comment?
https://www.rtands.com/track-structure/ballast-ties-rail/research-future-of-weibull-defects-analysis-in-the-railway-industry/
After reading the article, it seems that Weibull analysis was being used to predict the end of life of the rail. I would think that the Weibull parameters will vary with axle loading and percentage of wheels violating "WILD" limits (wheel impact due to flat spots, etc.).
Main use for the Weibull analysis may be scheduling more frequent rail inspections for rail nearing end of life.
The old roadmaster would see that the frequency of changing out broken rails was starting to increase, and suggest it might be time to think about a new rail program. He would get louder as conditions worsened. My feeling is that the Weibull analysis more or less parallels this, but in a more quantified way. Make it look scientific and it becomes more convincing to those deciding to spend the money.
And while predicting the "end of life" is nice, the reality is that budget planning for the more distant future is mostly theoretical. When the "scheduled" time comes it is often found to be possible to defer the changeout to a future year to help this year's budget.
Spam filters at home and at work are not allowing me to see the article for different reasons. General comment is this:
(1) Track life is dependent on variables and random occurences that no software program, no system and no person will ever determine. You aren't operating in a vaccuum.
(2) Used properly, the analysis is a tool - but not an absolute.
(3) Pencil geeks, financial buttonpushers, and control freaks are out of their minds if they think they can program it all. Boots on the ground trump office analysis every time.
CX: You forget that the operating budget is dominated by the operating department and not the engineering department. Track budget gets cut before the less durable sexy shiny toys do.
mudchickenCX: You forget that the operating budget is dominated by the operating department and not the engineering department. Track budget gets cut before the less durable sexy shiny toys do.
I certainly can't dispute that. Where the engineering department is vulnerable is that half way through the year it is the only one where work programs, especially those scheduled for the second half, can be cancelled because the revenue came in lower than forecast. The shiny toys, produced by contracts with 3rd parties, are delivered or underway and there is rarely flexibility to cancel without penalty.
Marketing and operating fall short, engineering gets penalized......
mudchicken (1) Track life is dependent on variables and random occurences that no software program, no system and no person will ever determine. You aren't operating in a vaccuum.
Makes sense, Weibull analysis would be best thought as giving guidelines for expected rail life as opposed to predicting failure in specific pieces of track. The "variables and random occurences" sound like what the EE world calls 1/f noise.
I've had similar experiences with software used for predicting failrue rates in electronics, noting that failure rates can be much higher than what the model says should be. One reason is that these failures are often induced by "random occurences" that are really difficult to model. However, there are some fun situations covered such as electronics shot out of a cannon (e.g. proximity fuzes, laser guidances, etc).
Our community is FREE to join. To participate you must either login or register for an account.