question about correcting error introduced by using straight slide in valve gear

[Chris KaminskiParticipant @chriskaminski64716]

I have two drawings from K.N.Harris setting out how to correct error
introduced by using straight slide.
Both of them based on Greenly and mostly identical,
but for one dimension (that I have highlighted in red)

as far as I can see in one (from article about JMG) – “L” appears to be vertical, when vibrating lever is at FDC…

in another (from article about Radial gears) – it appears to be in line with vibrating lever…

which one is correct?

please (and why?)

[duncan webster 1Participant @duncanwebster1]

Which articles, without context can’t say?

When he says ‘correction’ it needs to be taken with a pinch of salt. Better than a straight slide, but curved can be better. KNH didn’t have the benefit of computer models, so couldn’t fully see the effect of various dodges. In any case, Heywood used the offset top joint when Greenly was still a boy in short pants.

[Chris KaminskiParticipant @chriskaminski64716]

Thanks Duncan,

1st article (Introducing the JMG Valve Gear) – starts in No 3205 vol 127 (13th Dec 1962)

2nd article (Radial Valve Gears) – starts in No 3432 vol 138 (7th Jan 1972)

Yes, I am taking it with a large pinch of salt.

Yes Heywood was first.

My question is – do I just fiddle with simulator until I get what I want (for straight slide), or…

do Heywood, Greenly, Harris, Webster can offer some suggestions 🙂

[SillyOldDufferModerator @sillyoldduffer]

What I know about valve gear comes from Duncan and following his leads.

As I understand it the problem is that no valve gear correctly translates the sinusoidal movement of a piston into the sinusoidal movement of the valve needed to move the piston sinusoidally.   So what valve gear designers did was to get as close as they could without adding too much mechanical complexity.

The early designers approximated what their gear would do on the drawing board and then tested the linkages with wooden or cardboard mock-ups.  These were fiddled with to get a reasonable compromise.   Today it could be done better with a computer simulator, but it’s still only a compromise, possibly no better than Victorian cardboard!

Matters on full-size locos because wasted energy costs money, as did valve gear unreliable in service or needing lots of maintenance. Less on a scaled down loco because they don’t burn much coal, have many other scale related inefficiencies, and don’t clock millions of miles.    I don’t see anything in the IMLEC trials or in old magazines confirming advantage in minimising the error.     My feeling is either KNH answer will do.  Anyone know different?

Dave

[Chris KaminskiParticipant @chriskaminski64716]

This is not a question of merits of one or other design.

I can easily set out valve gear in question graphically as described by Greenly, Harris and Duncan.

Simplification in model (straight slide) introduces additional errors.

The “correction” to minimise errors has been described before, but the info about it is not very clear.

This is the only “clarification” I am seeking.

Of course, I can fiddle with simulator to get what I want…. 😉

 

[duncan webster 1Participant @duncanwebster1]

I wrote a more mathematical version way back, I think it went into design of the offset top joint. I’ll try to find it, but I’m having computer issues at present, waiting for #2 son to make me a new one. I might be able to get there via back up drive and laptop, try tomorrow.

[duncan webster 1Participant @duncanwebster1]

Which gear are you using?

[Chris KaminskiParticipant @chriskaminski64716]

I know you have computer problems 👍

I am talking about Greenly’s gear, and yes I am interested in more mathematical approach

Thanks a lot!

[duncan webster 1Participant @duncanwebster1]

Have sent you a pm

[Chris Kaminski]

On Chris Kaminski Said:

This is not a question of merits of one or other design.

I know!

I can easily set out valve gear in question graphically as described by Greenly, Harris and Duncan.

Of course.

Simplification in model (straight slide) introduces additional errors.

It does.

The “correction” to minimise errors has been described before, but the info about it is not very clear.

This is the only “clarification” I am seeking.

Which is why I tried to explain the underlying problem and the practical consequences.

Of course, I can fiddle with simulator to get what I want…. 😉

 

A simulator is the answer.  As the question is solved by approximations, I suggest the clarification Chris wants is a Unicorn!  And chasing them is a waste of time.

Root problem is there isn’t a straightforward clarification, there are multiple answers.

The designer starts by finding an arrangement of linkages that open and close the valve at about the right time.  This is done using geometry.  Then, testing his design, he discovers it doesn’t produce perfect valve events!

If the error is serious, he looks again at the geometry of the linkages and might have to start again from scratch.   Rule-of-thumb, simple reliable linkages are high error, whilst low error linkages are complicated and unreliable.

Given a promising arrangement it’s possible improve results by fiddling with the details. More art than science.  A curved link might get closer than a straight one, and tuning the position of joints helps.  None of them result in perfect valve events – they are all approximations.

As I said, Victorian designers started with the geometry, and then identified improvements by experimenting with models.   Lot of work and the result was never perfect!   Today, finding a good compromise is best done with a computer simulation because they do the necessary sums quickly, and can run thousands of comparisons.  A goal-seeking algorithm can be applied.  But fundamentally the computer does much the same as Greenly, just much faster.

KNH built on Greenly’s Victorian design process.  There are many ways of reducing but not eliminating the errors in Greenly’s basic design.  I think KNH experimented with the two examples quoted.  It’s not that one is calculated properly and the other is wrong,  more likely, KNH tried alternatives.  Not many – he’s unlikely to have used a computer.

This isn’t about the merits of two designs.  It’s fundamentally about the limitations of mechanical movements in this application and how to reduce the error.  As there’s no single answer, there’s no straightforward way of calculating it.

What works is compromise finding by comparing simulations.   A computer simulation doesn’t do any special maths, it changes parameters, does thousands of sums, and then, with luck, one of the results is obviously the best.   More likely there will be hundreds of similar answers that all work.

Happy to be proved wrong!

Dave

 

 

[SillyOldDuffer]

On SillyOldDuffer Said:

Chris wants is a Unicorn!

….

And chasing them is a waste of time

…..

Happy to be proved wrong!

Dave

 

 

If I was a girl I could accuse you of gaslighting 🙂

I refer you to an excellent articles by Duncan

“Radial Valve Gears Revisited” ME  Oct 1997

I know few model engineers even bother trying to understand, preferring “black magic” approach,

but I enjoy my unicorns – thank you Duncan

[duncan webster 1]

On duncan webster 1 Said:

Have sent you a pm

Thank you – you have one as well

[duncan webster 1]

 duncan webster 1

Managed to dig out the original article by H.Greenly,

and can see now which of the two drawings by K.N.Harris got lost in “translation”.
also, original article by Greenly has instructions/explanation missing from Harris.
Computer is not the answer to everything 🙂 (if you don’t understand what you are trying to do)

I have seen in my working life too many CAD jockeys

who thought that being able to use computer makes them into designers…

For anybody interested (although I doubt there are many/any out there)

here is the extracted bits of original article:

[BazyleParticipant @bazyle]

Why does it matter? The target should be constant acceleration/deceleration on the valve hence constant force to minimise wear on the linkages. Does a sine wave do that? I can’t remember any of this O level maths.

[Bazyle]

On Bazyle Said:

Why does it matter?

It’s not the wear we are concerned about, but valve events.

[BazyleParticipant @bazyle]

Yes but if your aim is to go from 100% admission to cut-off at 50% why does it matter if the valve moves sinusoidally, linearly, hyperbolically etc? Ok maybe you don’t want instant cut-off (except poppet valves?) to avoid hammering but what is so magical about sine? I feel it was a Victorian misconception that didn’t get analysed on a Babbage computer.

[duncan webster 1Participant @duncanwebster1]

If we’re talking slide valves, I’d guess the biggest load is friction due to steam pushing the valve onto the face, so acceleration of the bits won’t make much difference. All conventional valve gears (driven by eccentrics/cranks rather than cams) give more of less sinusoidal travel anyway. I could work it out, but much as I like sums (sad I know) I’m not going to.

The steam load on piston valves is much lower, but I’ve no idea how low.

[Bazyle]

On Bazyle Said:

… why does it matter if the valve moves sinusoidally, linearly, hyperbolically etc? …

The object of the exercise is to have valve events (admission, cut-off, release, and compression) that are close to equal for both ends of the cylinder over a wide range of cut-off settings and often when running in both directions. You have only one valve to do all this, moved by a valve gear that combines two different phases of roughly harmonic motion in varying proportions. Subtle tweaking of the basic geometry can improve the accuracy of the timing.

[Charles Lamont]

On Charles Lamont Said:

The object of the exercise is to have valve events (admission, cut-off, release, and compression) that are close to equal for both ends of the cylinder over a wide range of cut-off settings and often when running in both directions. You have only one valve to do all this, moved by a valve gear that combines two different phases of roughly harmonic motion in varying proportions. Subtle tweaking of the basic geometry can improve the accuracy of the timing.

👍

[Author]

Posts