[ATSF3751]

Here is a video on how the Milwaukee Road 261 is fired and run! I hope this helps with an idea on how a steam locomotive in America is run. This is the exact locomotive that I volunteer on and building a model of it for Open Rails.

Brandon

https://www.youtube....h?v=JCsJhfdJ9rM

https://www.youtube....h?v=DeFm9On5fuM

[steamer_ctn]

There are two (and a half) principal modes of firing the locomotive, AI (Artificial Intelligence) and Manual.

AI Mode - This mode is designed for those users who just want to focus on driving the locomotive without being distracted by worrying about looking after the fire. The AI fireman focuses only on maintaining steam pressure provided the driver operates the locomotive within "reasonable" operational parameters. The code for the AI fireman is already reasonably complex just to achieve this goal. He cannot see "ahead", and instead is reactive. So consequently he cannot increase firing in preparation for a station start or a gradient, nor can he determine when an impending station stop is to occur. To add these types of features, and others would increase the complexity of the fireman code considerably.

Manual Mode - This mode is designed for those users who wish to to focus on both driving and management of their fire. It would allow a fireman to look ahead, and adjust firing controls in accordance with various operational requirements.

One point that has been raised is the operation of steam pressure release valves in AI Mode. Because the AI fireman is reactive, then often he will arrive with a full fire at a station, and when the train stops the heat generated by this fuel will cause the boiler pressure to rise above the maximum. Often the pressure relief valves will operate. It isn't good, or prototypical to see the pressure relief valves operate at station stops. It wastes water, and also lowers BP, thus making the locomotive difficult to start again. Hence the pressure relief valves are disabled in AI mode.

To try and achieve some simple capability for the AI fireman to "look" ahead, controls were added to allow the driver to turn the fireman on and off depending on the conditions of the route ahead. (See around page 105 of the manual). When these controls are used the pressure relief valves are operational as an encouragement to manage the firing correctly.

If the pressure relief valves operate, then they will do so in a series, is as suggested above there usually are more then one, and they let go at slightly different temperatures.

If drivers want to manage the firing as well, or wish to see firemen type features operate, then they should operate in Manual mode rather then AI mode.

The Manual mode to the best of my knowledge has not been tested fully in OR, so I am uncertain how well it functions. I did try and get a test team together to look at and make appropriate tweaks, but that fell apart. Any testing would need to be done in small steps (focusing on one item at a time), and would require significant research to ensure the accuracy of any changes.

[ATSF3751]

I get what you are saying Peter but not everyone is going to want to run in Manual mode just to get those features to work.

We need the AI fireman to act more prototypical even in AI mode.

The only difference from Manual mode and AI mode should only be that the AI does the firing for you and in Manual mode you do the firing.

I never run in Manual mode because I don't want to deal with the aspect of firing the locomotive.

Just because you run in AI mode shouldn't mean some options are turned off or left out all together!

That is not how the real world works and it should not be that way in the game.

The Engineer and Fireman work together to keep the locomitive running down the road so I believe it should be the same way in the game!

Open Rails has a lot of potential with Steam Locomotives but still has a very long ways to go until it gets there.


In some areas it is way better then the original MSTS that is now 20 years old but in other ways it is worse and still needs work.

I appreciate all of the hard work you have been doing Peter and the rest of the guys but just hear us out with what we have to say and some constructive criticism that we may have.

Quite a few of us actually know what we are talking about and have been around real steam locomotives. Know how they operate and behave in the real world.

Brandon

[ATSF3751]

I am sorry if my last post was a little harsh but being around steam locomotives most of my life and knowing how they operate all I was trying to do is help out and giving some pointers and advice.

I know it would be difficult to code all of it and I fully understand the complexity of the coding. I know it would take time and wish there was more I could do to help Peter out.

I do want to thank Peter for all of the coding he has already done and the hard work he has put into steam locomotives for Open Rails already.

I am hoping to learn how to do coding eventually so I can help out in that respect but until then I hope Peter keeps at it and I will try my best along with others to help out any way we can.

Brandon

[steamer_ctn]

Hi Brandon,

 ATSF3751, on 22 January 2021 - 06:52 PM, said:

I am sorry if my last post was a little harsh but being around steam locomotives most of my life and knowing how they operate all I was trying to do is help out and giving some pointers and advice.

Thanks for this, I guess that I can put my voodoo dolls away now (LOL).

Each of us has many gifts and talents to bring to OR regardless of whether we are associated with real locomotives or writing code for computer models of them. There is a place for us all, and it is best if we can all work towards a common outcome.

Certainly my ideal is to achieve a train simulator that reflects real life as closely as possible. My code changes since becoming in involved in OR reflect this approach where I do extensive research to find suitable information to build computer models from. For example, I don't believe in using anecdotal information to build code models.

However we all need to understand that it is unrealistic to ever expect a simulation that is 100% accurate when compared to real life. Therefore the challenge is to work out what is worth putting effort into modelling (and coding) accurately, and what we accept is "as good as it gets". The other challenge is that different people will have different thoughts about what is important to add/change. I am sure that you already make these sorts of decisions in regard to how much effort you put into the different components in your models to "please the rivet counters".

I have worked extensively on the AI fireman code, and over time have attempted to achieve the best possible working model for it. However without a major code change, and extensive effort to address the proactive/reactive nature of the firing model, I think that for the time being it is the best that can be achieved.

I am however always open to consider suggestions (other then changes to the base firing code), which I will attempt to implement, if practical to do so.

When doing coding I work only on a single area of consideration that is identified as being of the most interest to a number of people. I find that if I am given a long "to do" list that this can be daunting (and hard to track) for me, and dampens my interest in doing any changes. So if there are any requests that fit this criteria, let me know, and I will provide some feedback on whether I am in a position to implement them or not.

In regard to pressure relief valves, can I suggest that you try the AI firemen On/Off functions, and this might help to see why they are disabled in the BASE AI model.

[NickonWheels]

Hey Peter, what about a 'minimum coal consuption' which may can be cofigurated and should obviously have a value of higher than 0?

[darwins]

First of all thanks to Peter for all the work he has so far put in to the steam model.

I think out of the small team that are building OR that he is the only one specifically coding for steam at the moment and is therefore a VIP!

What I am noticing here is that having a good AI fireman actually depends a lot on having a good model of manual firing. So if we want to improve the AI fireman, I think we need to start by reconvening the now defunct "manual firing" sub-group.

Some time ago Peter suggested to me that if I could put together an algorithm for an improved AI fireman, that he would have 'one attempt' at making it work. Knowing that nothing to do with computers or code ever works at the first attempt, I have not yet submitted my proposals! Having now spent about three years working with Peter (and others) on brakes - a topic that I previously knew very little about, I can see that development is much more effective if you work on one small area at a time and keep coming back to tweak little bits.

With that in mind I would like to suggest that we break the topic down and try to deal with improvements in the behaviour of the fire, manual fireman and AI fireman into smaller chunks or sub-routines.

Maybe taking a look at safety valve behaviour could be a starting point.

I have attached the first few drafts of my suggest AI fireman algorithm. (They are of course biased towards the British fireman.) Please share / edit / discuss / make improvements and anything specific to US or elsewhere.
One thought here - in UK the fire doors may be left partially open to provide secondary air - in USA I believe they have holes or ports in the doors that can be adjusted to provide secondary air - is that correct?

 AI Fireman.zip (133.15K)
Number of downloads: 300

Quote

Hey Peter, what about a 'minimum coal consuption' which may can be cofigurated and should obviously have a value of higher than 0?

There is already a minimum consumption set by the burn rate curve. There is also a minimum feed rate of 10 lb/h with the AI fireman. Calculations on heat loss by radiation suggest that small locos sitting in the yard might only require 3 or 4 lb/hr so where this is set in a burn rate curve, we need to make sure that the AI fireman is not putting on any more coal - in fact it is perfectly reasonable at times for the feed rate to be zero, letting the fire gradually reduce in size and heat.

[darwins]

Here is a first response regarding safety valves from a gentleman who has fired a good variety of locos:

Quote

First, the easy one - the two safety valves are set differently - I think they are different sizes as well - I seem to remember on one show about steam railways there being a test to check that the fire can’t make more steam than the safety valves (presumably when both lifted) will lose. One common myth is that safety valves have to lift on the red line. There is a margin of error - I don’t know what it is, but it might be 5 or 10 % of working pressure, or maybe 5 or 10 pounds above or below the red line, which is perfectly legal. I think Wells may even have four valves.

The easy question first - when the safety valves lift, how can you stop them?

The most common answer is to put on an injector - this puts more water into the boiler, which cools it down, reducing the pressure - the deciding factor here is of course the water level, which if it gets too high could risk the loco priming, or carrying water over into the cylinders - very bad! Other things you can do are closing the dampers (restricting the flow of air, so the fire doesn’t generate as much heat), opening the firehole doors (allowing more cold air in from above the fire, which will cool the tubes, reducing the heat), turning the blower down (pulling less air through the fire, reducing the heat) or if you are really struggling, you could rake a hole in the fire (air will naturally choose the path of least resistance and go through the hole in the fire, so it won’t be heated properly, so reduce the heating effect of the fire).

Of course you may have noticed the contradiction there - less air will work, as will more air. Primary (or bottom) air comes through the firebed so it is heated on its way through the firebed. Secondary (or top) air comes through the firehole doors, doesn’t get heated as much - so more primary air will make the fire hotter, more secondary air will cool it down.

On a similar note, the blower will generally heat the fire up by drawing more air through the hot firebed and along the boiler tubes. However, if there is a thin fire, or a hole in the fire, and the firehole doors are open, you will actually cool the fire with the blower. For example, we have a two-hour layover at Porthmadog on the Welsh Highland Railway, so we let the fire die back quite a bit to avoid blowing off for two hours. If you have just a little bit of fire in there (perhaps just a bit of coal burning in the back corners) you can (although it’s not good practice) fill the firebox with coal and whack the blower open.

The first thing you will notice is the pressure drops back as the blower is pulling cold air through coal that isn’t burning and down the tubes. As the coal starts to catch, the draught heats the fire up and the fire is heating the air up enough for the blower to pull hot air along the tubes and heat up the boiler, increasing the pressure.

One other thing that complicates the issue is that the boiling point of water increases with pressure. So as the pressure increases, so does the boiling point, which means the fire has to produce more heat to boil the water off and increase the pressure further. Conversely, when the safety valves lift, the pressure drops, which means the boiling point of the water drops, so the amount of heat the fire is producing (which doesn’t change quickly, so is effectively constant) will boil more water off and increase the pressure, keeping the safety valves lifted. This is why, in some situations, a loco can run for miles with the safety valves lifted.

So, when the safety valves lift what happens?

It depends mostly on the state of the fire. If your fire is small, for example sitting on-shed in a morning, they may lift, the pressure will drop back a bit, then they will stop, the pressure begins to creep up again and the valves lift and it gets into a cycle, with the valves lifting for shorter periods each time, staying closed for longer periods.

If you are racing up the hill and (for example) you see a p-way gang in the four-foot and you have to shut off steam and brake, you have a raging fire, both valves will lift and you will sit there with the valves roaring for some time. Chances are you’ve been running with the injector on anyway, so they will make little difference initially.

Even with (for example) a stop at Damems Junction. From a fireman’s point of view, you have to be ready to go straight through at Damems Junction if the other train is already there, or to stop if it isn’t, so it can be very difficult to keep the loco quiet if you are there for more than a minute or two.

Mostly, when the safety valves lift, you will see the pressure gauge flinch slightly, as the pressure drops initially, then return to normal (pressure reduces, boils faster), followed by a slow reduction in pressure to five or ten pounds below working pressure where they will seat again and the pressure will start to climb. Obviously how slow the pressure change will be depends on how hot your fire is. Some locos the valves will ‘feather’ before blowing properly, others (rarely) will have a very definite open and shut. It’s not uncommon to see the pressure sit at up to three or four pounds above the red line with the safety valves open.

Something to consider is that safety valves don’t always open bang on the red line - a pound or two light is common - occasionally five pounds even. Likewise, often the valves will not seat until you are maybe ten-pounds below working pressure. This is why firemen don’t like blowing off:

1) It’s a waste of coal - you’ve shovelled coal into the fire to make steam you are now wasting.

2) It’s noisy - the cab of a loco with the safety valves lifting is very loud. Especially on some narrow gauge locos where the valves are actually in the cab.

3) Locos will generally steam best when the pressure is kept consistent. Maintaining a pressure is much easier than knocking it back and having to built it up again. Imagine cycling along with a friend. Every time you have to slow down and drop behind them to let somebody pass, you’ve got to work harder to make up the speed you’ve just lost. In an ideal world, running at a constant speed, with an injector just holding back the pressure (and maintaining the water level), firing little and often, is by far the easiest for a footplate crew. If only it were that easy in real life.

Finally (and not related to safety valves) on the Worth Valley in particular, the size of the loco plays a huge part.

If you are using the WD, with four coaches, you’re not really working the loco anywhere near its capacity. That huge boiler holds a lot of steam, so if the loco goes off the boil, you’ve actually got a boiler full of steam that will tide you over for a bit while you bring the fire round, so in many ways they are easier for the crews - you have a bit of insurance if you make a mess of it - however, the flip side is that when she does go off the boil, you’ve a lot of water to heat up to bring her round again. If this happens on the KWVR, chances are she’ll just be warming up again by the time you get to Oxenhope and you’ll sit there blowing off all the time.

Smaller locos are being worked much nearer their limits, so you don’t have that insurance policy. However, they tend to have a better ratio of heating surface to water capacity, so it is possible to bring them round again much more quickly than with a larger loco.

Steam locos are like that - full of contradictions - you would think thrashing a loco hard would kill the fire and the pressure - which it does long-term, but in the short-term it pulls lots of air through a fire, making it burn hotter and actually steam better - as long as there’s a good fire in there to start with. I’ll never forget (excuse me digressing) firing on the Welsh Highland from Porthmadog up towards Caernarfon. We were trundling across the Traeth and I was trying to warm her up for the hill - I must have been looking a bit concerned because the driver turned round to me and said “Is the fire in there?” (i.e. have you got a good thick fire of the right shape)
“Yes” I replied.
“Don’t worry - I’ll get it burning” he said, promptly opened the regulator and sure enough she came round quickly!

Sorry - forgot one thing - I was out as a cleaner on the Middleton with Sir Berkely. I think they are Salter-type valves, but they have a lever which protrudes into the cab. If the loco was about to blow off, you could actually gently lean on the lever to keep them quiet - within reason. But that’s not something you really should do, and it’s not something I think you could incorporate into your simulator!

[ATSF3751]

I completly agree with Darwins and Peter's idea of breaking it up into small chunks.

Here is my proposal.

I think we should first gather up the ones of us that know the most about steam locomotives. Both from the UK and US and forum a Beta Team to help Peter out.

We can then discuss what is most important that we would like to get done first or improve what we have already coded into the game.

Ask Peter if it is possible to fix whatever we would like tweaked and do beta test for him after he has tweaked or added something.

I think our first task is NOT to add any more features but try to fix or tweak what we already have in the game so far. From what I can see in the forum we already have a lot to work on that Peter already has implemented into the game so let's start there.

Then once we have all of that stuff figured out we can then start adding more features.

I also think it is a good idea to get Manual Firing figured out first so we can then transfer that data into the AI Fireman in the future.

We need to work as a team here and throwing out random ideas and stuff for him to add and modify is not helping the cause one bit.

Obviously this is a daunting task for one person to be coding all of this info into the game and us with the knowledge of steam locomotives need to try and take a step back and think in real terms.

All of us could maybe work on a chunk to get info on a specific aspect to give to Peter instead of saying we want this or that changed.

I know for myself I am the same way when it comes to building models for the game. I am very detailed oriented and if it doesnt look or sound right I do not use it.

If someone is a giveme pig to me as well I am less likely to release my work to them so I fully understand where Peter is coming from.

I have many connections here in the US with some of the top steam locomotive guys who operate, fire, maintain and rebuild them so I can also go to them for help as well to see if any of them would be willing to answer any questions we may have in the future.

It can be very overwhelming at time and this is suppose to be a fun activity for all of us to enjoy so let's start doing this the right way and work together as a team to get things done!

Brandon

[darwins]

I am definitely on board with what Brandon is suggesting above.

Whenever a change is being made it will need to be tested by different members of the group using a range of different locos and conditions. This is both to check that we are broadly happy that any change is an improvement (I doubt that we will all ever be in total agreement!) and also of course to check for bugs.

In the meantime another fireman has responded on the question of safety valves:

Quote

My experience of the past as a stoker, compared to German locomotives like the BR50, 52, 23, etc.

Both (2) safety valves are "equal" adjusted, so to +/- 16Bar. Because there are always small differences in adjustment, one safety can open a little earlier than the other, sometimes even "alternately". But in principle, so simultaneously.
However, both safety devices have a "pre-blow device", so practically the safety valve opens +/- 15,9BAR (I think about 40%) and the "main valve" opens at 16,1 BAR (100%), for maximum opening. (IOW, each safety valve has 2 openings at different pressure)

The boiler pressure no longer increases as example 16BAR, but often the safety valve closes slightly lower, so on example 15.7BAR (Everything of course how well these are adjusted, with a bad safety valve the valve can close on example 14.5BAR or lower , but this is technically not 100% .....)

One safety valve must be able to discharge the entire boiler steam production, so boiler pressure must never increase. (this sometimes happens a bit in OR ........!)



To prevent the maximum boiler pressure from being reached, one can:
1)
Add water with the injectors, but never to the maximum value! Then water could be sucked in when opening the regulator, so that "water" in the superheaters is converted into steam. Since the superheaters are usually behind the regulator valve, this means that the locomotive can become inoperable for some time, so "runs wild". (then reverse immediately to 0% and open cylinder valves immediately !!!)
Once experienced this myself, the locomotive immediately shoots forward about 30-40 meters, luckily this happened on "a free part" .... is not a pleasant event .....
Another danger of high water is "water hammer", so water enters the cylinders, so that the cylinder safety valves open / break ... (or worse .....)
An "emergency bypass" at high water is to use the Boiler blowdown to lower the water level. (and then use the injectors again if necessary).
To be clear, with boiler blowdown the water level drops, not the boiler pressure!


2)
To temper the coal fire (heat). this is possible (depending on the circumstances and "future expectations")
- Temper the air supply (dampers) or close it. Completely closing often has the result that the coal fire starts to "slag" (stick), so that the grate closes afterwards.
IOW, you have to rake it open again or use shaking grate if available.
- Slide a "hole-opening" in the center of the fire on the grate. Usually better than 100% closing dampers, especially when standing still for a long time. IOW, you need to keep a ring or coal fire for firebox protection, but in the center partially colder air is coming in through the air dampers.
- Open the fire door (especially when short standing still) so that "cold air" dampens the heat. Normally this is not allowed, but with a well burning and hot fire this 'moderately' is possible.
- And as crazy as it may sound, throw more coals on the fire ...... When the coal fire is thicker you have less oxygen, so heat is less. You do this especially if you know that you have to drive later (as temporarily wait for a red signal). After all, coal will burn up again later, so that ...... etc. etc. (In OR: "coal mass> 110%")


Regarding OR, my experience.
- Safety devices indeed have "one opening pressure value". (and does not close at slightly lower pressure, which is usually practical)
- In OR, "100% steam discharge" is probably not used or possible, often you can still see the boiler pressure increasing. (sometimes even 1 BAR ......)
- "Steam production" is skyrocketing. (also when opening cylinder valves). This is not 100% correct. You also get "black chuff smoke" which is of course not correct.