I got your hookup......

I've been having a nice set of email exchanges with a new Australian friend, who is working on a pretty intensive overhaul such as we are planning for 110, and to similarly include invisible modernizations for sustainability over the long term........and he mentioned their plans include permanently plumbing in connections for transducers such that should the performance decline, it would be a simple and fast process to connect a couple transducers to any critical location to get data that could procide for a better and faster diagnosis and repair.

This dovetails with our desire to build a digital testing program for 17, and to fine tune 110 through digital "indicator cards" once she's up and running using what we learn from working on 17.

Dave Dick and I kicked this around a bit and we both think there's no reason we can't expand our current plan for temporary testing of the smokebox and back pressure with analog gages on 17 (which we will do and pretty soon) to, over time, include building our own permanent system of hookups and develop a collection of transducers and programs for an onboard laptop to graphically record results. We'd like to build a partnership with the Engineering Department at the University to start with, but I think having our own set of transducers is going to work best for us in the long run.

So, we now need to decide what points we want to tap for measuring what parameters, what transducers we either do want to use or don't want to use for taking what kind of measurements, and what the value of each parameter masured might be for either establishing benchmarks or plotting against each other for diagnostic purposes.

From my POV, the obvious is cylinder pressure at either end and a linear measurement of the stroke for indication, and back pressure and smokebox vacuum for front end performance.

What else might we want to include for what purposes?

Dave

Oh Lord, there are a lot.

Linear measurement of stroke can be done by a number of methods, but these are all external (for example, reference surfaces machined on the crosshead and guides). I would design for a transducer that captures both absolute position and relative movement (rather than something like an LVDT that only gives you differential) as it will minimize the signal-conditioning board requirements.

You will need MULTIPLE points in the smokebox, both for vacuum and for relative flow. I would also put some bungs near the front tubeplate, to insert sensors for gas speed and mass flow. (Those can be as simple as a combination of the devices used for MAF sensing and a simple temp sensor/thermocouple which gives you the gas temperature offset necessary)

You will want instrumentation in the gas volume around the stack, which may be a bit involved if there are no plates or screens, in the blast stand (fast-acting) and in the physical regions in the stack. For doing some of the CFD stuff, you will want the ability to introduce streams of material near the front tubeplate (using the instrumentation bungs) and then be able to follow what happens to that stream as it is induced and then ejected. I have some thoughts about this, some of which include RFID or polyspectral tracking.

Conventional measures of indicating only give mass flow indirectly. It should be possible to derive this using an appropriate intelligent cybernetic system (mostly, I think, in software, from fairly simple sensor and reference inputs) so it comes up as a hard number, even if a calculated one with some interpretations or simplifications so long as sufficient data are also recorded to permit more precise calculation...

Sensor in the cylinders that measures nucleate condensation.

Sensors that measure compression (and its rate of rise)

A number of sensors in the firebox (some of which will double as semiautomatic firing-control sensors if you go that route). Some of these will likely be optical, and may require active cooling to work. Pyrometers on some sort of directable probe will also be useful (think of a tlexible staybolt head through-drilled with an extensible stem through it...)

Valve pilot like sensors on the running gear. Inertial accelerometers for wheel balance and augment, Suspension deflection and inertial chassis motion under thrust.

Design the signal-conditioning interfaces to be modular (and both weather- and temperature-tolerant) as far as possible. It is possible that there will be an aftermarket for such a thing, if it is not already marketed, and your academic partner(s) should see the advantage in that (but be SURE to keep a nonexclusive share of the rights, and be SURE to be named on any patents that issue... ;-} )

I'll think about this some more in a bit.

OK Robert, thanks....but for those of us who don't have access to the NASA Jet Propulsion Lab equipment, what should we prioritize? If we could only afford to measure 10 parameters, what would they be?

Dave

Dave,
I had blast pressure, vacuum, 2 cylinder end pressures and displacement, valve chest pressure and found that you can only get a few data per second. This means that electronic indicating is a fully separate circuit which probably needs synchronizing with the other equipment. Electronic indicating has been done in the UK with cheap equipment, picoscope. What I would like to see is a row of pressure sensors on the stack exit, could just be a row of copper tubes with U tubes connected, a photo  of the U-tube array would be enough. It is the stacks exit velocity that counts!
kind regards
Jos

None of this stuff needs to be very high tech. The key is a fast, capable computer system with modern-grade "interface bus" capability. Fortunately, just about any Pentium or laptop falls in this category -- even USB 1 is fast enough to mux multiple signals with low latency. Picoscope is an example of a computer-simulated oscilloscope.

The usual indicator stuff may be mandatory, but it will need to be EXTREMELY fast if done in digital. That includes any effects of inertial mass in manometers. So I would have at a minimum four sensors for pressure, and perhaps four for steam temperature. Accelerometers ought to be cheap, likewise open-source software to run them. Most of what you're discussing involves BUNGS for instruments, not instruments per se, and I'd expect these to have the same general effect on the boiler structure as small washout plugs. My idea with the gas temperature would be to have one probe, or perhaps a few at most, with the ability to move them in height, and between rows, to establish data under reasonably reproduceable conditions of firing or steaming. (In other words, you'd be taking a lot of serial readings and combining them with the assumption they would apply simultaneously under common conditions).

Jos, would your stack sensor array not have to be Pitot tubes or the equivalent, with compensation for external air/gas pressure, and perhaps some temperature sensing ... not that any of these are hard to do! Bundle of manometers would be interesting -- a high-resolution shot would let you read the meniscus to some accuracy if necessary (it probably wouldn't be!) I would still bet, though, that MEMS pressure sensors would be a better, more easily calibrated, perhaps cheaper, and considerably less unwieldy system. The catch here is to avoid influencing the readings by influencing the flow patterns at the data points in the stack.

No matter what arrangement you have, it needs to be made adjustable in length and smooth in cross-section. You then have pairs of 'access ports' down the length of the stack that allow the device to be inserted at different heights, and perhaps turned to different angles across the nozzle array, so that a 3D flow-field can be built up (similar in approach to axial tomography).

I even forgot my 3 temp sensors, fresh, blastpipe, smokebox. Even with indication thrown out I did not get more than 6 datasets per second. It just takes time to read a sensor. However, since there are 20 dollar processor units these days it may be useful to see if these small clever units can control their own sensors and take care of only extremely fast data collection and uploading.
It would be nice to know a lot about the stack exit flow, but for the foreseeable future I am interested in the flatness of the exit profile only, As long as it peeks we are not doing a proper job!

OK, so we're interested in developing a cross sectional view of the pressures at various points along a line through the diameter of the stack at the top if i understand you correctly. I think we could do this with 7-8 sensors mounted on a narrow bit of strap. The probable limitation would be the effect of the hot gasses on the thin wires burning the insulation away.

Hugh has a traditional indicator if we need to go analog.

Would a reasonable plan be to start with the basic measurements - indicating, back ppressure, basic smokebox vacuum in front of the center of the FTS and then find out where we need to take more careful readings based on these initial findings?

Right now, we're only planning back pressure and vacuum, but that's more than we've ever done so far, and I think it can lead to more down the line. Based on what proves useful, we'll know what we want to gear up to test in the future.

Dave

Dave,
You do not need pressure sensors in the hot gases, a copper tube against the flow
connected to a sensor outside the chimney should do!
Back pressure and vacuum is fine with me as long as is measured with some variations in power output.

OK, perhaps we can come up with a sliding tube that can be run across the stack during a steady part of the trip, with stops every inch or so for a measurement to be taken for a few beats. One sensor, one tube....one guy riding on the top of the boiler?

Do I have any volunteers?

Dave

Dave- I know someone that knows someone that has a traditional indicator; I'll have to see how amenable they are to lending it out or selling it outright if we get to that point.

Hugh, there are usually at least the essential components of indicators on eBay most of the time I look, for no more than a couple hundred dollars. By the time we get to actually needing one, I expect we will have had chances at quite a few. Might be better than borrowing if we want to do periodic benchmarking.

There is a partial Ashcroft on now, closing in about 3 days, that is currently sitting at $12.50 (!). It is probably missing its levers, but those are cheap and easy to fabricate compared to the parts that are there...

I confess that I was expecting to use just sensor input data and software to produce the indicator diagrams and other analyses. Much quicker and shows more detail too...

Robert- A co-worker knew of my interest in steam and brought one by for me to look at (and confirm to him that it was in fact a steam engine indicator) a few months ago. It turns out his neighbor's dad had crewed on tugboats here in Charleston, and the indicator had been used on one of them. I don't recall the manufacturer, but it was built in the early 40's, and was a complete "kit" in an oak box. I have photos here somewhere.

I need to ask him about it again; he wasn't sure if the lady wanted to keep it or sell it. It had been stashed away for many years.

The traditional indicator is a NOW thing we can use......in the future I'd also want to go with something digital. You have to start with what you can get....and 17 needs help.

Dave

Yes, Hugh -- check on this QUICKLY. (If the lady wants to keep it, ask about borrowing...)

I had forgotten about 17.

It should be possible to use rotary encoders to 'digitize' output from a conventional indicator mechanism (in the sense of getting the data into computer-usable form). That would mean that no investment in a conventional indicator would be wasted later, even if the advanced testing is done with add-on devices and networking.

Hi,
I also have such a unit. However it is totally blocked, these things seem to seize when not in regular use. Please note that you have to have an operator close by, that is why these wooden shacks in front of the locomotive were built. Electronic indicating is a lot more user friendly!
Kind regards
Jos

The seizure is, in part, a consequence of the really precise lapped bearing tolerance required to have a mechanism with minimal 'slop'. I wonder whether a good grade of watch oil, or one of the modern lubricants (like those with nanodiamond) would help prevent this problem.

I had thought the reason for 'close by' attention was the need to keep replacing the physical indicator cards with each measurement. It occurs to me that it might be easy to 'remote' either some method of duplicating the cards on a ground-based printer, or moving physical media in appropriate 'steps' under the indicator pens to take sequential readings. I have to confess that given the top speed under which I expect 110 will be tested, a full enclosure might be easier to construct -- perhaps using a deer blind? ;-}

Southern shortline, people....duct tape a guy to the footboard.

Two goals here: immediate, gain information useful on improving 17 during the next several months. Long term, build a testing program with hardware and hard-wired hookups for sensors for long term troubleshooting and benchmarking.

I'm still trying to finish gathering the boilerplate support documents I need to send out in grant applications. Paperwork isn't in one place, different board members have papers from their terms in their possession, so archiving is now a job to be done so we can write grants to help us do these jobs.

Unfortunately, this sort of job doesn't appeal to many volunteers.

Dave

I thought on Southern shortlines you duct-taped the cheeseburgers and beers to the running boards, and the volunteers would cling on of their own volition...

I'd be helping with this if I were there. Will help from here if there is need.

Fried chicken or BBQ and onion rings.

Now back to the regular programming.........

Dave

Dave,
What you need to know right now is whether there is blow by in the cylinders. This could be established by indicating. If you are sure there isn't,(how?) I would like to suggest to do just the blastpressure - vacuum measurements.
Kind regards
Jos

Actually, I'm reasonably certain there is - and in the valves also. I was up there on a damp cold January day disassembling the right side valve chest a couple years ago with no tools or rigging on a track bordered by a ditch full of muddy water because there was a whistle believed to be coming from it. I found the cylinder cocks were worn, and the balancing plate had a fair sized divot knocked into it. We didn't take anything more apart because we could barely do what we did, and ended up bolting it back together with mud for a lapping compound on the surfaces. I think we made whatever it was worse.

Until we can set up for more work not done under stupid conditions, indicating is our best bet. It might also tell us what to look for and where.

Dave

Safe to assume there is blowby; the question is whether it is so excessive as to be a meaningful waste of water and fuel energy...

Note that to some extent, you can determine blowby by motoring the engine and measuring the pressures during "compression", in the spaces on either side of the piston.  Technically you could do this with just two pressure transducers per side, because you don't care as much whether the end-to-end blowby occurs with the piston at any particular longitudinal zone in the cylinder.  But you'd still need either an analog plot of the changing pressures, or frequent enough/fast enough sampling and recording of pressures to give an accurate plot for calculating rate-of-change.  So I'd bet manometers are neither fast enough nor amenable to rapid accurate reading.  What I think I would do is use a video camera set to quick shutter and very rapid frame rate, step through and record the individual readings, calculate the indicated rate of pressure changes, and make any corrections to the physics of manometry (air compression, inertia of the fluid column, etc.)

Problem is that, if I recall correctly, using traditional indication for steam-consumption purposes, rather than power, is that the actual mass flow of steam was difficult to determine accurately. either for inlet or exhaust.  I have done some preliminary investigation of methods for measuring actual mass flow ... and the problem appears so far to remain as sophisticated as I remember it.  So blowby testing that gives you additional information over a simple port-to-port leakdown test, or a leakdown test with the wheels being moved, compensated for expected pressure, might not be worth the effort if done only with a typical performance/IHP indicator...

Anyway, much of this is premature to the issue I thought we were discussing, which was the type and location of the PORTS to be provided for instrumentation. Those don't cost that much, unless you discover at test time that locations you need or want can't be accessed without machining, welding, etc.

Imho opinion blowby can be established a little easier. You have to uncouple the cylinder drain valve rods. Put the Johnson bar into full gear and roll the engine in such a position that the foremost port is open. The position is critical since you have to be sure that both ports at the other side are covered by the valve. Brake on. Open backside cylinder drain and give some steam with regulator. If there is blow by it will come out of the backside cylinder drain. Repeat for other cylinder sides. You can establish this way whether there is a massive or just a little blow by. Probably best done when cylinders are hot.
Kind regards
Jos

He's right, of course.