I spent the past weekend at the Puffing Billy Railway and had a good look at their newly delivered all welded NG/G16 boiler.

It caused me to ask why they had gone down the route of purchasing the boiler and in turn selling the original to the Welsh Highland Railway.

Primary reason is that the Australian boiler regulations allow for four yearly intervals for boiler inspections of welded boilers V's two years for riveted.

Additionally, there is now no time limit on boiler tube life (same in New Zealand) they are now pulled on condition, not time. With modern boiler treatments, both the Australians (and the New Zealanders) are expecting decades of life out of newly fitted tubes.

TH

Interesting differences - can you tell us more about state of the art water treatment programs down under that provide such good results?

dave

I suspect that improvements in NDT, borescopes, and methods are at least partly responsible for this. If you can see and measure the inside and outside of the tubes, measure their thickness at different points, and determine if their metallurgical structure is sound ... it makes sense to use tube integrity as the basis.

What I wonder is whether the same advances in 'robot' NDT sensors might make it possible to do an appropriate visual and metallurgical test on a riveted boiler shell while the tubes are in. Surely there would be some level of documentation and methodology (perhaps a synthetic video showing the entire path taken by the robot into and around the 'masked' areas of the inside of the shell) that would satisfy the ESC...

Whether this would be cost-effective, of course, is another story -- but even if the equipment is expensive, some of its cost could be recouped by loaning it out, conducting training, etc. As with many aspects of biomedical engineering, even very expensive technology can be costed-down comparatively quickly if there are enough alternative uses with reasonable return from each.

Fascinating stuff gentlemen! Thanks for sharing.

Sincerely,
Glenn

I've often wondered whether a robot could crawl around the interior of the shell. Naturally, thickness measurements could be made (as they could from outside) but would images of seams and other areas of interest be sufficient if the area could not be made as clean as desired without direct access? Xrays and ultrasound could see through scale but maybe you still need a trained eye examining things after cleaning. Then again, maybe among the robotic players could be equipment to clean surfaces. Sort of a Roomba for the boiler. Or Roombas, one that does media blasting, one that does wire wheeling, and one that does needle chipping.

I had in mind an inspection device, probably with rolling magnetic securement (the boiler shell and firebox being ferrous on American power) equipped with appropriate lights and high-resolution cameras (which are now very cheap) along with the sensors and transmitters needed for NDT testing. A separate set of units would be used for media blasting, chipping, chemical treatment, etc., using the same basic 'chassis'

Note that the device could indeed be programmed to follow a set path through areas of the shell and water legs, looking and recording information without requiring a trained operator for the full time of investigation. Naturally the skilled human operator could check the video and data 'feeds' at any time, spot anomalies, request more intensive or different scans, etc.

I had been thinking that an appropriate X-ray sensor array could be provided on the robot, in lieu of a physical film magazine of some sort, with the actual EM provided from outside (and appropriate shielding/blanketing in other areas). You'd coordinate the visual imagery with the X-ray and NDT via some proper method of data fusion, both for examination and record-keeping; it should be easy to set up the data keyed to whatever your examination 'grid' is. (I'd expect the FRA and ESC to work out how the integrity of the data would be assured, how it would be related to other measurements, etc.)



A Roomba is entirely the wrong paradigm for what a 'boiler robot' would have to do.

While the iRobot devices do have some directional controllability, they are primarily designed to follow a random path over the 'area of work', eventually (statistically, as it were) going over every point on the floor at least once. This cuts down dramatically on the equipment and programming to make a consumer-level device, and might have some applicability if you were just cleaning the surface with a 'dumb' process like blasting or chemical etch, but is DEFINITELY not what is wanted for systematic boiler inspection of the kind wanted at a 1472-day inspection.

It would be possible, of course, to set a 'random' path for the robot and then associate the video and NDT data 'virtually' into an electronic representation of the boiler ... eventually. I don't like that approach, and I wouldn't even think of suggesting it to the ESC, as there's no real certainty that the areas would accurately be mapped in the final result, that some form of distortion didn't occur in the mapping, or that some spots might have been missed or done at insufficient resolution. As far as I'm concerned, the robot will be doing, more or less exactly, what a human would do inside the boiler in terms of working to a known pattern ... just with a smaller carrier that can work around the tubes.

I suspect there is also a 'robot' design that can do tubes, and perhaps one for inspection of elements without having to break connections at the header and extract them.

What do pipeline companies use to inspect pipelines? I don't know anything about the type and level of testing they do, but they've often spoken of their crawlers'contribution to safety.

Some of the tiny quadcopters right now are smaller than my handspan, and getting smaller all the time (the alphabet soup agencies reportedly have some that are bumblebee-sized.) One model is controllable from a cell phone app and runs off the guts from an older phone, including its camera.

There's already a very small, light medical ultrasound attachment for a smartphone, built for developing countries and portable in a shirt pocket.

Someone out there doubtless has an X-ray receiver. I'm not an engineer, nor do I play one on TV, so my imagination may fail, but if a device can accept one type of radiation input, it might be able to accept something further down the spectrum.

At the risk of further diverting this thread (my fault in the first place) I think that needing a robot to clean out the boiler is evidence of bad water treatment and washouts in the first place. Surely we have the ability to do a thorough cleaning job through the use of solvents, heat, etc before the final washout before inspection, leaving it in clean and ready to document condition.

dave

Paul, you going to design a removable dome for the 219? :-)

Becky, pigs are good for pipeline service (they can often negotiate severe changes in direction and size, and Williams famously adapted them to string fiber-optic backbone through that company's gas pipelines in order to become a major data-network player. But a pig is an inherently longitudinal device, not what you want in a boiler with both tubeplates intact. While you could run a 'tethered' pig-like device in from a removed dome, you still need multiple-direction drive to cover the whole of the surface area.

As a start on the X-ray NDT here is a promising technology

http://pubs.acs.org/doi/abs/10.1021/nl803386y

This can be made conformable to the curve of the shell if desired.

Correct, I didn't mean Roomba in the randomness sense. In fact, a completely manually driven device, that is not autonomous at all and therefore not really a robot is fine.