[beresford]

Will OR model the differences between freight engines and passenger engines? Freight engines tend to have greater rolling resistance due to a larger number of coupled wheels, and often don't respond well to cutoff settings of less than 45 because the valve events are set up to deliver power at low speed. One of the less convincing aspects of MSTS was how every engine responded well to low cutoffs.

[copperpen]

The current development is to get the basic steam code working properly. The rest will follow, but exactly how these differences are to be achieved is not yet worked out. I do know that increasing the friction values will make the engine work harder, but you need to experiment to find out how high to take them.

[cjakeman]

Hi Anthony,

beresford, on 14 January 2014 - 09:35 AM, said:

Will OR model the differences between freight engines and passenger engines?

It's a good point. Do you have any data we can use to quantify that?

We're still working on our development of the Doug Jones steam physics to get realistic steam consumption and the power curve. So far all our attention has been on express locos, where there has been more data published.

[beresford]

cjakeman, on 14 January 2014 - 11:50 PM, said:

Hi Anthony,

It's a good point. Do you have any data we can use to quantify that?

Who is Anthony?

I don't think that much data was gathered at the time BUT I have found a guy on the Net who is analysing sources of resistance in steam locos. This may be rather more than you'd hoped for:

http://freespace.vir...0Resistance.pdf

As far as cylinder cutoffs go, David Wardale makes the point that freight and passenger engines were different and there is an intrigueing reference to somebody constructing a mathematical model of a steam engine!:

http://5at.co.uk/ind...valve-gear.html

For my part I can only draw on my footplate course experience and upon reminisces of old enginemen. For example Terry Essery says that Midland 4Fs were not advanced beyond 35% because the exhaust port events became restricted, and I have read that other engines would produce groaning noises from the front end if advanced beyond 45%. This is presumably because with no lead it requires a lot of effort at higher cutoffs to get the steam out of the cylinder during the exhaust cycle. On the other hand BR Standard 9Fs would attain high speeds with 15% cutoff, but they seem to have had positive lead.

[Matej Pacha]

It's your nickname what confused Chris - there is another guy, Anthony Brailsford, who is really enthusiastic about steam locomotives.

I'd prefer not to make any difference between passenger and freight locomotives in the code. There is no logical (and programable) connection between physical behavior and purpose of the locomotive. I believe our steam physics guys are working on pretty detailed model where all the necessary parameters can be set. So, if I may respond to your answer: No, OR will not difference these two types of locomotives in the code, but YOU will be able to do this by setting appropriate parameters in the ENG file (or similar config file).

[steamer_ctn]

When considering the steam locomotive there are a huge number of variables which impact upon the performance of the locomotive. Often the performance of locomotives within the same class will even be different. Added to this are all the different locomotive variations, such as simple, compound, saturated, superheated, oil-burning, etc. The type of valve gear and how it is set up, ie when the valves open and close, etc, has a huge impact on the performance.

To model all of the above will require a very sophisticated mathematical model, as well as an intimate knowledge and detail for the locomotive being represented. Sadly the information required to do this is not readily available in an easily understandable form on the Internet.

The current model has attempted to represent a "generic" performance for a simple saturated and superheated locomotive using information that is readily available from the Internet .

Like all things in life, we are on a journey, and where that journey will end is difficult to predict.

For my part, as a first step, if we can achieve some of the different types listed above, and have them operating in a similar fashion to published test reports, then we will have made a good start.

So any information that you (or others) are able to share will contribute to future outcome.

Cheers

[Genma Saotome]

What I'd like to know is how does the code know when to adjust the cutoff (and by how much) for AI steam locomotives?

For player locomotives I would hope the decision is simply left to the player.

Something else... I have the impression, perhaps incorrectly, that the data being used to test the code is based on small, fast, British steamers coupled to the impression, once again perhaps incorrectly, that the typical large, slow North American steam doesn't... ummm... I'll use the word scale here... doesn't scale up relative to those; IOW, what works well for one doesn't work so well for the other. Again: Impression, perhaps unfounded.

So how wrong am I? And that is an earnest question... I'm quite open to be told 100%... or something less.

[copperpen]

Although we are mainly using British engines to test the code against, primarily because there is hard data to compare with, we do also have one smallish US locomotive that also has data to compare, and in both cases the output from the OR code is pretty close to the real world in both cases. I have also tested with some of the larger US articulated and the OR code scales up well, although as one segment of an articulated is treated as an AI, I am not certain that the correct power is being used for that section of the engine. AI certainly seems to work well though. For the large non-articulated engines, the power is certainly there. It does require that correct data is in the eng file for the key parameters to allow the power to be developed properly, and fuel/water usage to be correctly calculated.

Control settings for the player locomotive are player controlled, with both auto and manual firing modes available.

[steamer_ctn]

Genma Saotome, on 15 January 2014 - 01:49 PM, said:

What I'd like to know is how does the code know when to adjust the cutoff (and by how much) for AI steam locomotives?

I haven't investigated the AI code yet, so I can't state how it is adjusted. My suspicion is that it purely works off the line speed, though I cannot be certain at this moment.

Genma Saotome, on 15 January 2014 - 01:49 PM, said:

For player locomotives I would hope the decision is simply left to the player.

It is.

Genma Saotome, on 15 January 2014 - 01:49 PM, said:

Something else... I have the impression, perhaps incorrectly, that the data being used to test the code is based on small, fast, British steamers coupled to the impression, once again perhaps incorrectly, that the typical large, slow North American steam doesn't... ummm... I'll use the word scale here... doesn't scale up relative to those; IOW, what works well for one doesn't work so well for the other. Again: Impression, perhaps unfounded.

As suggested by Copperpen, both British and American locomotives have been used for testing, provided a test report was available. A test page has been set up which has links to test reports, the relevant test reports and also reference documentation which describes the steam locomotive operation. At the moment only the American locomotives are available on the page, but we have tested against a British unit as well.

Given the wide variety of parameters which influence the performance, it is not easy to allow for all variations, however the performance is modeled on "best fit" basis relevant to the core locomotive parameters, such as evaporation area, grate area, cylinder size, etc. Tests seem to suggest that it provides a relative close correlation.

Up until x1942 the locomotive MotiveForce was being calculated based upon the MaxPower and MaxForce values in the ENG file, now it is calculated based upon the core parameters of the steam locomotive.

Cheers

[AuzGnosis]

A little time jumping across various threads.

steamer_ctn, on 15 January 2014 - 12:03 PM, said:

When considering the steam locomotive there are a huge number of variables which impact upon the performance of the locomotive. Often the performance of locomotives within the same class will even be different. Added to this are all the different locomotive variations, such as simple, compound, saturated, superheated, oil-burning, etc. The type of valve gear and how it is set up, ie when the valves open and close, etc, has a huge impact on the performance.

To model all of the above will require a very sophisticated mathematical model, as well as an intimate knowledge and detail for the locomotive being represented. Sadly the information required to do this is not readily available in an easily understandable form on the Internet.
..........................<snip>.................

For my part, as a first step, if we can achieve some of the different types listed above, and have them operating in a similar fashion to published test reports, then we will have made a good start.

So any information that you (or others) are able to share will contribute to future outcome.

Cheers

Then a couple of years later Peter wrote what IMHO reads as a connected idea so;

steamer_ctn, on 31 August 2018 - 10:24 PM, said:

Depending upon the type of steam locomotive some additional data might be displayed.

For example, a compound locomotive will have two lines of steam cylinder pressures, ie the high and low cylinder pressure,
whereas for a simple locomotive normally on one line of information is displayed.

Prompting me to note:

I have been pondering the whole compound Steam question since I wrote to Peter at the Coals to Newcastle website a few months ago. For the ZigZag route the Baldwin J483 Vauclain compound version was notably more efficient (under constant work) than saturated version. However the Vauclain compound was marginally out-performed by the much easier to (maintain &)operate later super-heated version. So while one could calculate high & low cylinder as two lines, in the case of the Vauclain compounds timing events only impact the high pressure cylinder which exhausts competently directly into the adjoining low pressure cylinder.

Calculating High, Intermediate & Low pressure cylinders separately makes perfect sense where the cylinder timings are independently controllable. So I see here the question as before undertaking such efforts how many of the Open Rail users may wish to run such exotic locomotives? If a major slice of the users were modelling Eastern European steam lines then such an effort would be easily justified but not knowing the demographics of modeler I have no feel for my question's answer.

That comment being tempered by my uncertain from not having yet discovered nor investigated the AI code yet. Coming from an Expert System, Knowledge Engineering background is that "AI code" classical Artificial Intelligence knowledge manipulating works {Rule Bases, Inference Engines, Frames, Neural Nets et al} or is it simply a nicely crafted collection of stacked If-Then-Else procedural logic mimicking AI like independent operational features?? If you do indeed have a rule processing inference capability buried in the code that may provide with an alternative option to address the complexities of numerous steam engine configurations.

Just a thought.