Steam loco test in OR

[lindsayts]

I have been conducting some testing trying to gauge how accurate the steam loco simulation is. The testing method is to pull a known load up a known accurate grade, also pullling a consist at high speeds (60 to 80MPH) in which the total resistance has been calculated by other means.
The power is worked out by noting the velocity, the pull at the draw bar and also at the wheel rims and calculating the power from there.

The loco chosen is a Victorian (Australia) Railways locomotive known as "Heavy Harry" (H220). This is the only machine built of three planned, the 2nd world war preventing the others being built. This loco was used as the dynamometer car tests results are availble so it is one of the few steam locos whos perfmance is known with any degree of accuracy. These tests results show 3300 drawbar HP at 47.5MPH and another 3600 HP at 50 MPH. The boilers maximum output being in the order of 38,000 pounds per hour (results of separate workshops tests).

In the current version of OR after playing around with the eng files I managed around 2000 drawbar HP at 45 MPH with the rated steam consumption. One had no trouble getting higher powers but it was at the cost of higher steam consumption. A test done pulling a heavy train at a hard slog up a 1 in 48 (the grade has been checked as being accurate) resulted in ascending at 10 kph Draw bar HP being 1200 at the rated steam consumption a performance in line with reality, good.

Overall the results are very promising, the simulation correctly varying in efficency at slow speeds (something MSTS greatly lacks). At this stage no amount of playing with eng values particularly "CylinderEffectivity" will improve the situation, a value of 1 producing the best result. Fractional values, ie 0.7, appear to have no effect. It looks a bit like the algorithiom used (I have not looked at the source) is not using the steam as effieciently as the machines could do in real life.

Lindsay

[lindsayts]

Another loco that can be used for these sort of tests is the Pennsylvania Railroad K4. An MSTS model is availible from the file library on this site and the graphs of the indicated Cylinder HP and steam consumption verses speed have definitely been published as I have a copy of them. I just downloaded the model and will be setting it up latter.

Lindsay

[longiron]

Community,

We are in the middle of a major update to the OR physics for all ENG types. Some will require adding additional parameters to the ENG file to be enabled, which will NOT impact backward compatibility.

There are a number of new capabilities around the steam thermodynamic model we are planning to add.

1. Cold startup: Arrive at the Engine with the fire banked and steam pressure at minimal levels. Spend your time getting the locomotive up to operating temps. You will quickly discover why diesel locomotives replaced steam.
2. Controls: Addition of blowers, dampers, grate shaking, manual vs stoker feeds, water & coal consumption, and much more.
3. Designs: realistic simulation of compound (hi/hi & hi/low pressure cylinders), geared and other steam locomotive designs
4. Effects: linking the thermodynamics to the steam/smoke effects so they change with the physics
5. Power: realistic simulation of the uneven pulses of how steam engines apply force to the rails, we have some of this in the model today but we will calculate the inertia, mass and momentum of the valve gear and driving wheels in real time.

and there's more on the drawing board, but I'm not sure what will make the next release.

In addition, the OR team will be publishing the key parameters and their impact on the physics model, so there will be no mysteries as to why stuff works the way it does. The team is about 70% done with the documentation task.

We've had some feedback from realworld steam engineers that the full blown new physicss may to too realistic for a train sim, as they had to pay too much attention to all the operational tasks and not enough on driving the locomotive, as many tasks would be handled by the Fireman. So, we're investigating some simplifications and automation of tasks - think very intelligent AI fireman for those who want it.

[twpol]

At this stage no amount of playing with eng values particularly "CylinderEffectivity" will improve the situation, a value of 1 producing the best result. Fractional values, ie 0.7, appear to have no effect. It looks a bit like the algorithiom used (I have not looked at the source) is not using the steam as effieciently as the machines could do in real life.

I had a quick search of the source code for you, and I can't see any reference to "CylinderEffectivity" so it would appear OR makes no use of this data currently.

[lindsayts]

We've had some feedback from realworld steam engineers that the full blown new physicss may to too realistic for a train sim, as they had to pay too much attention to all the operational tasks and not enough on driving the locomotive, as many tasks would be handled by the Fireman. So, we're investigating some simplifications and automation of tasks - think very intelligent AI fireman for those who want it.

Thank you for the reply.

Now just in case,
my post was not meant in any way to be critcal of the work the OR people are doing. It was just to state the current steam loco physics model is not efficent enough. A point I will bring up at this stage is in these machines small changes in the way the loco was built could have quite significant effects on the final machine. It is for this reason each loco had it own personality. So the phsyics model needs to be able to take these changes between machines into account. To this end the eng file needs to be able to be tweeked, with parameters simliar to "CylinderEffectivity" to account for differences (such as size of the steam pipe from boiler to the cylinders) between simliar machines of differing desighn.

Now a comment on the quoted text.
I have had numerous arguments in various trainsim forums about always using AI firing, being accused of "not playing the game". I have always taken MSTS as a serious sim and in real life one CANNOT drive and fire any steam loco at the same time (Note 1). One of the rules in the book of the railways in Victoria (Australia) is a driver must not let anything distract him/her (Note 2) from the task of driving, firing being a definite distraction.

Note 1: Although this has been done in Australia, usually during shunting operations in places where there is no danger any management would show up. Also driver only is the only way can work on "live steam" locos as a consequence rear end collisions have occured on various club tracks because the driver had his "head in the fire box" and was not watching the road ahead.

Note 2: There is female drivers in AUS i have been told.

Lindsay

[CaptainBazza]

G'day mate, from TLOTLWC. I though he was called 'Fat Harry'?

Your idea of using a known loco, with (rare) known performance values is a good idea. I do the same testing for narrow gauge on the US 3ft Cumbres and Toltec RR, on the old DRGW, on the section between Chama and Lobato. There is a lot of anacdotal evidence on prototypical performance over that section, such as locomotive type and typical loadings for single and double hauled, etc, consists. So far the results, even with early OR steam physics, is very promising.

Chris' remarks re the forthcoming upgrade to the steam physics is exciting......oh, but really - a five hour startup from cold to go.....LOL. They had patience, in those days.

Cheers Bazza

[longiron]

Thank you for the reply.

Now just in case,
my post was not meant in any way to be critcal of the work the OR people are doing.

None taken

It was just to state the current steam loco physics model is not efficent enough. A point I will bring up at this stage is in these machines small changes in the way the loco was built could have quite significant effects on the final machine. It is for this reason each loco had it own personality. So the phsyics model needs to be able to take these changes between machines into account. To this end the eng file needs to be able to be tweeked, with parameters simliar to "CylinderEffectivity" to account for differences (such as size of the steam pipe from boiler to the cylinders) between simliar machines of differing design.

My only comment is that if you don't know the equations and the impact of parameters on the output, then parameter changes may not achieve the desired results. The result will be a bunch of postings that changing "so and so" doesn't make steam creation increase/decrease. Let the team finish it's work and publish what the parameters mean and how they impact the steam physics thermodynamic model.

[lindsayts]

None taken



My only comment is that if you don't know the equations and the impact of parameters on the output, then parameter changes may not achieve the desired results. The result will be a bunch of postings that changing "so and so" doesn't make steam creation increase/decrease. Let the team finish it's work and publish what the parameters mean and how they impact the steam physics thermodynamic model.

One could probably have an interesting discussion on knowledge and people at this point...........................

As I understand it OR is trying to (how will I put this) develop using only "real world" physics. A sugestion would be to at least consider "impossible" paramaters as this may allow someone to do something that was not on the developers radar. An example in MSTS is one can make a vehicle not slip under any conditions thus allowing one to simulate a cable operated incline, something clearly the MSTS devlopers never thought of.

Lindsay

[lindsayts]

G'day mate, from TLOTLWC. I though he was called 'Fat Harry'?

No its "heavy Harry", the legend goes that one of the workers building the machine at the Newport workshops wrote the name one some part of it, sort of as a form of grafitti. The loco was well loved by the crews (this being a real good indication of good performance) and even fairly late in the piece VR management was talking up the machine. "Heavy Harry" still exists and is the ARHS railway museum. There is an excelent poem writen by a station master on the Mansfield line of the machine in the waiting room of Seymour station, gives me chills evrytime I read it.

Your idea of using a known loco, with (rare) known performance values is a good idea. I do the same testing for narrow gauge on the US 3ft Cumbres and Toltec RR, on the old DRGW, on the section between Chama and Lobato. There is a lot of anacdotal evidence on prototypical performance over that section, such as locomotive type and typical loadings for single and double hauled, etc, consists. So far the results, even with early OR steam physics, is very promising.

Yes accurate data is VERY difficult to come by, this is the primary reason for the post. I do have though an excellent technical steam library which did cost a "packet". I will take every advantage these days in observing steam loco peformance to get a better grip on the problem. Using stories of performance deeds is as you say another excellent source. It is quite clear a lot of people these days simply have little idea of the power these machines put out.

Chris' remarks re the forthcoming upgrade to the steam physics is exciting......oh, but really - a five hour startup from cold to go.....LOL. They had patience, in those days.

Cheers Bazza

Absolutely, Cant't wait for releases the sim now is good and clealy getting better. Be nice to see the steam gauge working though as one now has to correctly drive the boiler.

It was not really a patiience problem the shed crew lit up and prepared the machine, a lot of walking between the wood stack up into the cab and throwing it onto the fire and back to the wood stack again (been there done that AND the wood stack was on the other side of the yard). The operating crew usually only arriving a short time before the machine was required.

Lindsay

[Train-a-Mania]

We've had some feedback from realworld steam engineers that the full blown new physicss may to too realistic for a train sim, as they had to pay too much attention to all the operational tasks and not enough on driving the locomotive, as many tasks would be handled by the Fireman. So, we're investigating some simplifications and automation of tasks - think very intelligent AI fireman for those who want it.

This is just a random thought I had. Perhaps multiplayer (once there is time to work on it) could allow one OR user to be the engineer and another OR user to be the fireman.

[ED_4]

We've had some feedback from realworld steam engineers that the full blown new physicss may to too realistic for a train sim, as they had to pay too much attention to all the operational tasks and not enough on driving the locomotive, as many tasks would be handled by the Fireman. So, we're investigating some simplifications and automation of tasks - think very intelligent AI fireman for those who want it.

This would be indeed great for us that want to drive Steam without having to worry about firing the engine. The only reason I really fire the engine myself is the AI fireman in MSTS is very lousy at performing to give the engine the best output, without sacrificing how much it gobbles up in coal/water.

And speaking of steamers, will we also see the fruition of a possible variance between coal fired, oil fired and wood fired steamers too? MSTS had this in mind from the possible options in the engine file parameters. But never got implemented.

[longiron]

And speaking of steamers, will we also see the fruition of a possible variance between coal fired, oil fired and wood fired steamers too? MSTS had this in mind from the possible options in the engine file parameters. But never got implemented.

The new thermodynamic model does account for differences in the type of fuel based on the following parameters:
(1) caloric value of the fuel; coal, oil and wood have very different caloric content which impact temps and steam production
(2) mass of fuel; again this goes to steam production and firing requirements along with firebox calculations
(3) burn rate of fuel; determines firing rate related to steam production
(4) mass of fuel in firebox; with coal and wood you have a fire mass, while oil has "zero" effective fire mass. So steam production calculations are very different depending on throttle position. With oil, closing the throttle immediately impacts steam production, while wood/coal lags substantially.

[ED_4]

(4) mass of fuel in firebox; with coal and wood you have a fire mass, while oil has "zero" effective fire mass. So steam production calculations are very different depending on throttle position. With oil, closing the throttle immediately impacts steam production, while wood/coal lags substantially.

Really? I thought that the fireman in an oil-burner still works some valves to open up or close down the flames for that missing "firemass" to be adjusted. And again the engineer still has to do his/her controls as usual.

[longiron]

Really? I thought that the fireman in an oil-burner still works some valves to open up or close down the flames for that missing "firemass" to be adjusted. And again the engineer still has to do his/her controls as usual.

The point I was trying to make regarding oil burners is that there is no fire mass, the steam production is a product of the AI fireman and the throttle positioning generating draft.

[CaptainBazza]

Steam generation vs steam usage

Recent tests with non-blower equipped locomotives show a natural disparity between steam initial generation and usage at lower speeds. I must say I can't vouch for blower equipped locomotives because I haven't tested them, or else this feature isn't yet implemented in Open Rails. Steam use outstrips production until the locomotive gains enough speed to cause an increased draft throught the grate and tubes. You really have to pay attention to your driving, and steam management, because the BP drops as you use more and more stored steam. Once draft and thus steam production increases, the BP shows an increase in favour of production over use.

It's a nice change from MSTS physics, but it might be initially off-putting for those who are inexperienced with steam locomotives.

Cheers Bazza

Note: The test locomotive(s) is/are small 2ft gauge types, without blowers (that I'm aware of) in the spec's.