Re: Steam Physics

[copperpen]

I have based my Davis figures on the output of Fcalc, which has a section specifically for this. Using a frontal area of 8.3 sq/m I get figures that are far higher than those you are using, indeed higher than I am using. My figures started life as Fcalc output then modified to get the friction curve to match a published one from BR testing. The compound is set up using data from the 1937 edition of British Locomotive Types published by The Railway Publishing Company, with the exception of using 7 foot drivers instead of 6ft 9inch.

This data enables the OR code to set up a boiler capable of generating about 13K pounds of steam per hour. For davis figures, I started out using those from the CTN Hall test locomotive and adjusted those down until I had what I felt was an adequately performing locomotive, running light engine at 22% cutoff and full throttle it will accelerate to about 84MPH without usage exceeding production of steam.

I will insert your davis figures into my BR stock and see what the outcome is using the double headed compounds.

[systema]

That is strange, you and I seem to get vastly different values out of F Calc.

I have set my LMS carriages up using parameters from F Calc and I am happy with the result. The performance I am getting seems reasonable and if anything I go too fast. For an 8 wheel bogie carriage of mass 32t-uk (= 32 * 1.106 tonnes ) I get the following values:

ORTSBearingType ( Friction )
ORTSDavis_A ( 162.6 )
ORTSDavis_B ( 1.075 )
ORTSDavis_C ( 0.02928 )

I have approximated the cross-sectional area to 8m^2 to try to allow for the shape of the body and roof and some open space underneath... For older vehicles built to slightly less width and height I would normally put in 7m^2 (or even 6) for cross sectional area.

As far as I am aware OR calculates resistance using its own formula but approximates a Davis input to its own formula, and OR requires the Davis input to be metric which is based on Newtons and Meters/sec.

Your calculation above looks as though its output is in US tons and mph.

The correct input to Fcalc should be 32UK tons = 32*1.016=32.5 tonnes, area 8m^2, axles 4 which gives

A=732.12 (Newtons)
B=10.6941 (Newtons/meter per sec)
C=0.651680 (Newtons/meter per sec squared)

It is obvious that your resistance is far too low which is why you go too fast.

For a passenger car Fcalc gives the same Davis output for Friction and Roller Bearings as it does for a Steam Loco.
For a Freight wagon it gives lower values for a roller bearing but for friction bearing gives the same answer as a passenger wagon of the same parameters.

Mick Clarke
MEP

[DarwinS]

I am quite happy with the Compound boiler I have set up in ORTS (based on data from OPC "Engines of the LMS") with 1317 ft evaporative surface and a 28.4 sq ft grate it is giving me a maximum steam output of 19700 lb/hr which seems quite good enough.

[Using a formula from an old railway magazine I would have otherwise calculated an absolute maximum output of about 24,000 lb/hr based on 9996 from 147 sq ft of firebox heating surface (68 lb per sq ft) plus 14040 from the tubes (1170 sq ft tube area * (42 / square root of tube length of 12.5 feet) - The eng file calculator that I used with MSTS (and still use for boiler volumes) uses the formula Max boiler can produce (18 lb/sq ft in firebox, 100% of Firebox rate in tubes under 13 ft, then reduced by a factor of 13 divided by actual tube length) and this gives a maximum output of 21096 lb/hr for the Compound boiler.]

What I have at the moment in ORTS is

Comment ( *** Boiler *** )
ORTSSteamBoilerType ( Superheated )
BoilerVolume ( 224ft^3 )
ORTSEvaporationArea ( 1317ft^2 )
MaxBoilerPressure ( 200psi )
ORTSSuperheatArea ( 290.7ft^2 )
MaxSteamHeatingPressure ( 80psi )
ORTSBoilerEfficiency( 0.0 1.0 160 0.45 )
SafetyValvePressureDifference( 5 )

Comment ( *** Fire *** )
ORTSGrateArea ( 28.4ft^2 )
ORTSFuelCalorific ( 13000btu/lb )
ORTSSteamFiremanMaxPossibleFiringRate( 3000lb/h )
SteamFiremanIsMechanicalStoker ( 0 )
ORTSBurnRate( 1000 10000 5000 40000 )

The problem for me is with the power output at different speeds. For the hill climbing I need to sustain 20 mph up a 1 in 54 gradient with a trailing load of 166 tons. (Based on recorded Compound performance between Girvan and Stranraer). but the same eng file that achieves that gives me too much power and speed on more gentle gradients and level and downhill stretches. (Based on a log from a St Pancras to Leicester service with a load of 11 bogies 329 tons).

[DarwinS]

I am now getting very confused. This is the output that I get from FCalc. The values for Davis A B and C are given here in imperial units rather than metric.



Furthermore, whilst A and B values from FCalc are the same for passenger cars and locomotives, I get different C values for steam locomotives. (Reading the theory document with FCalc a factor is added in for the valve gear and rods), Similarly for MU trains there are different values for front, mid and rear cars because of different surfaces exposed to the 'headwind'.

[systema]

I am finding that using the Davis figures generated by Fcalc causes too much resistance when moving. So I use them as the starting point, having set up all the other engine parameters with correct data. I also use the CTN OR test track which has a large level oval, a section which is all curves, and several gradients in the inner sector.

So, using the Fcalc figures I set off with a reasonable load on the hook and see what the maximum speed is in that configuration. Usually it falls way short of expectations. The Compound for example, double headed with 9 BR MK1s would just about creep past 40mph. By reducing that in stages The consist will now accelerate up to 55MPH in a reasonable distance and continue to increase speed up to 63.3mph and maintain that speed on 90% throttle and 24% forward gear. With a lighter, shorter train speed would be higher.
The Davis figures I use on my coaching stock are
ORTSDavis_A ( 482.68 )
ORTSDavis_B ( 11.1877 )
ORTSDavis_C ( 0.4952 )
ORTSBearingType ( Roller )

I have been testing various locos on the ctn test track over the past months using Fcalc Davis Input for resistance.
I calculated theoretical maximum speed by estimating the power that gave an actual OR Simulation Max Speed and then checked this with various train loads to see what speeds could be achieved. The results were remarkably consistent, though I would agree that the max speed seemed a bit low. The results are for a Rebuilt Scot on dead flat track using 25% cut off and max regulator. It took about 9 miles depending on load to reach the top speed and generally about 15 minutes. This is from memory as I did not note these figures during the tests.

8 Coaches = 78mph (Actual and Calculated)
10 Coaches=74mph
12 Coaches=70mph
14 Coaches=67mph

I also checked performance on the 10 mile 1.3% incline from Settle to Ribblehead. I noted the total journey time for this and the minimum cut off required to maintain reasonable speed. Again for the Rebuilt Scot.

8 Coaches, 30% cut off, 18 minutes, 38mph actual, 38mph calculated
10 Coaches, 35% cut off, 21 minutes, 33mph actual, 34mph calculated
10 Coaches, 35-40% cut off, 22 minutes, 29mph actual, 30mph calculated
10 Coaches, 40-45% cut off, 26 minutes, 25mph actual, 27mph calculated

Its hard to find information on what real world performance would have been. Anecdotal reports imply better performance than these tests indicate. The Rebuilt Scot was supposed to be one of the star performers, so maybe there is something not quite right with OR's calculations.

Mick Clarke
MEP

[systema]

I am now getting very confused. This is the output that I get from FCalc. The values for Davis A B and C are given here in imperial units rather than metric.

If you use Fcalc.exe you can use either US units or metric units (File Menu) and get a direct answer for Davis (View Menu)

To covert the spreadsheet result to metric you need
A (lbs)= A*4.44822 (Newtons)
B (lbs/mph)= B*4.448222*2.23693(Newtons/m/s)
C (lbs/mph²) = C*4.448222*2.23693*2.23693 (Newtons/m/s^2)

You can add these formulae to your spreadsheet but Fcalc.exe is much easier.

Mick Clarke
MEP

[copperpen]

@Darwin
Your boiler and fire specs are virtually the same as I am using. 19700lbs/hr will be reached if the engine is worked hard, but the theoretical max in the HUD is about 1K lower.

Using your Davis figures on my BR Mk1s, and a similar C figure for the compound, I am able to get a double header with 9 on up to a max of 80MPH on the level using the CTN test track. Not tried a single locomotive yet, but I suspect it will top out somewhere between 60 and 70 with the same number of carriages. One must remember that these locomotives were designed to haul light fast trains, not the heavier trains of later years.

@Mick
During the early days of testing with the MN I felt that OR was being a bit heavy with the application of the Davis figures. There are Locomotive Performance charts that show these are fully capable of 100mph running with 9 or 10 on, but OR was unable to produce that kind of running, so that is when I started to massage the Davis figures until I got what had been reported.

ORTSDavis_A ( 1448.04 )
ORTSDavis_B ( 11.6711 )
ORTSDavis_C ( 1.5 )
ORTSBearingType ( Roller ) is my current setup.

Looking at your results on the Settle climb, I do not think that the Davis figures would have much of an impact at those lower speeds, it is the higher speeds where the C figure has most effect.

[systema]

Copperpen,

Early on I also tried the MN on ctn and felt the same, somewhat disappointed I could not reach 100mph.

However, I have just tried the compound with the calculated (all Friction bearings) metric Davis Figures on all stock in the consist. With 5 coaches (165t) behind the tender I reached 78mph on the flat at 25% cutoff. I could start from a stand on the 2% slope and fairly easily reach 20 mph plus. So this should satisfy Darwin. There was no feeling of the train running away on the flat. It took about 12 minutes to reach max speed.

With 8 coaches (265t) behind the tender I reached 69mph on the flat but struggled up the 2% incline. On the 1.2% slope I could however maintain 24mph fairly well adjusting up to 40% cutoff.

My theoretical TE is 20.7Klbf.

We need to be careful not to end up like MSTS with people making up eng files to create enhanced and even unrealistic performance.

I will retry my own version of a MN and see what happens.

Mick Clarke
MEP

[DarwinS]

@Systema - Thanks for correcting the units on the Davis figures. This does help considerably in preventing the train running away at higher speeds.
The LMS carriages I am using now have figures similar to this:

Comment ( *** Friction *** )
ORTSBearingType ( Friction )
ORTSDavis_A ( 710.4 )
ORTSDavis_B ( 10.04 )
ORTSDavis_C ( 0.6761 )

With 11 bogies (332 tons) I got up to around 60 mph with the Compound on the level part of the test track in about 12 minutes.
It does not now run away so easily down hill, but with steam can get up to a bit over 80 descending from Luton towards Bedford.

It can also take 167 tons (5 bogies and a van) up the 1 in 54.

I am still fiddling a bit with the cylinder initial pressure drop. At present I have this:

ORTSCylinderInitialPressureDrop( 0 0.97 50 0.78 100 0.68 150 0.63 200 0.61 250 0.58 300 0.52 350 0.39 )

But will probably tweak it some more.

For the loco Friction I have:

ORTSBearingType ( Friction )
ORTSDavis_A ( 4753 )
ORTSDavis_B ( 20.30 )
ORTSDavis_C ( 4.485 )

I suspect the larger C figures will work well for older designs to keep them in check. (Opposite problem to the Merchant Navy)

In LMS days the Compounds were allowed 220 tons on XL limit bookings on the Midland Main Line - so in theory they should be able to get that load up to about 80 mph on level track.... next attempt.

[systema]

Darwin,

I am going the other way with Davis. I ran my Merchant Navy with 10 coaches and could only get just over 80mph. I then maxed out everything I could for better efficiency including limiting Back pressure to 4psi. This got me to 84mph. I then started to look at reducing Davis Numbers. I believe the formulae are quite old and based on locos and stock built around the early 20th century.

For the modern rebuilt MN I reduced the weight on drivers factor to a multiplier of 10 rather than 20, to try and cater for better efficiency of the running gear etc. I also reduced the C figure. I have been calculating rather lazily with the default area of 10m^2. I have concluded this is too high for UK stock and is probably set at this because the author is American where they have much bigger stock. Looking at the front profile in Shape Viewer with the Bounding box shaded area showing you can see that doing a simple height times width calculation for area is quite a bit too high. I also calculated the tender Davis numbers using the passenger input as using freight friction gives a very high output. My estimated conclusions are that the MN area should be 9.0 m^2, the same for the tender. For Mk1 style coaches I used 8.5m^2. These don't reduce the C output a great deal but it is enough to get me up to 89mph eventually. I am reasonably satisfied with this. Steady running with 20% cut-off keeps a speed of about 85mph with Boiler Pressure and fire level all stable.

I am reasonably satisfied with this as 100mph should be easily achieved on down gradients.

Mick Clarke
MEP

[DarwinS]

I have just had a quick play with some drawings of locomotives and some squares and triangles.

Coronation Class - width over cab 8ft 11in height 13ft 2.75in - gives me 9.22 m^2

5P5F 4-6-0 - width over cab 8ft 7in height 12ft 8in - gives me 8.07 m^2

[systema]

It would be good to have a set of standard areas for various stock items.

Mick Clarke
MEP

[DarwinS]

I agree

It would be good to have a set of standard areas for various stock items.

This is probably easier for modern rolling stock (ducks ).

Copperpen has given a value of 8.3m^2 for BR Mk I stock. This would also be good for BR MkII and MkIII stock, most LMS stock (~9ft wide over body - well 8ft 11.75in), a lot of LNER stock (9ft wide), the GWR stock built for through working to other companies and Southern Railway R4 carriages.

The LNER also built 8ft6in and 9ft3in widths. GWR general service stock was 9ft 3in over body. The SR also built R0 8ft 0in and R1 8ft 6in stock and the LMS made a few sleeping cars ~9ft 2in over the waist.

Pre-grouping so many sizes and shapes that we would need a standard for each profile. (I am happy to work out ones for carriages that I have built for MSTS.)

Happy with the Compound, I thought I would try the LMS Princess. I have a detailed log of a journey with the down Royal Scot from January 1937 that gives cut offs used as well as speeds and passing times. The load was 16 LMS bogies - 492 tons tare. With that load and the correct parameters for friction, my "out of the box" Princess can accelerate past 70 mph up a 1 in 300 with 20% cut off - and that huge boiler delivering 36 000 lb/h of steam is barely keeping up with the demand from the cylinders. The log shows the real train got up to just over 60 mph, so I will try adjusting cylinder parameters again.

[DarwinS]

I think I will call myself happy with the Princess performance. This is the engine that I have:

Comment ( *** Cylinders *** )
NumCylinders ( 4 )
CylinderStroke ( 26in )
CylinderDiameter ( 19in )
ORTSCylinderEfficiencyRate ( 1.0 )
ORTSCylinderPortOpening ( 0.120 )
ORTSCylinderInitialPressureDrop( 0 0.98 50 0.85 100 0.75 150 0.67 200 0.64 250 0.63 300 0.63 350 0.63 )
ORTSCylinderBackPressure( 0 0 500 1.2 1000 3.3 1500 6.3 2000 10.2 2500 15.0 3000.0 20.7 )
SteamCylinderCocksOperation ( Manual )

The nice thing about this is that it can also now be used for a GWR King (just with different boiler and firebox!).
As with the Merchant Navy I was a bit slower on recorded downhill maxima - the log shows the train reaching 85 mph at times downhill, I could not get mine beyond 78 mph. However with appropriate uphill speeds and average speeds then all seems well.

It was necessary to change ORTSSteamFiremanMaxPossibleFiringRate( 3000lb/h ) to ( 3500lb/h ) even so boost was used a couple of times and the final fire mass was still a bit low, however there is no way 9 or 10 tons will last for a 7 or 8 hour run if I increase this any further. As with the real thing a higher level of performance could have been achieved if there were a mechanical stoker (or two firemen!) and a larger tender.

[steamer1960]

Based upon some of the comments in this thread, and further research on resistance (friction), I have updated the information on the CTN website ( http://www.coalstonewcastle.com.au/physics/resistance/, http://www.coalstonewcastle.com.au/phys ... tance-set/, and http://www.coalstonewcastle.com.au/phys ... nce-fcalc/ ).

I am hoping that this will provide more background information and clarify some of the questions raised. If anybody has time to read it, and provide any feedback (errors, or areas where more detail could be added?) I would be happy to consider to it.

I am also wondering if it would be useful to make a spreadsheet available for users to compare curves, as I believe that modifying parameters needs to be done very carefully to ensure that any curves used align with reasonable railway practices? For example, depending upon the type of stock the Davis curve should lie somewhere between the upper and lower limit formed by the original and adjusted Davis equation. Naturally this would be overridden where information has been published by the railway company or researcher.

I agree that the C value is the critical value that impacts the resistance at higher speeds. For the time being, I think that the taking the values in the Size statement should give a reasonable estimate of the cross sectional area. Has testing to-date shown otherwise?

The nearest I can readily find to a worked example for British rolling stock is this:

Carriage rolling resistance: R = (1.1 + 0.021v + 0.000175v2) kg/tonne - from Koffman applying to BR coaches.

from http://www.5at.co.uk/index.php/definiti ... tance.html

............ SNIP ..........................................

[Would it not be easier to have OR calculate the values for the Davies equation rather than the content creator? - type of bearing, number of axles, and mass or all specified, for area it might not be too far off (other than for high speed trains) to calculate it as a rectangle from x and y in the Size () parameter.]

The answer to this question is in the first half of the post. The Koffman formula for example does not provide any references to the parameters mentioned in the second half, so therefore OR would be unable to calculate the Davis curve aligning with the Koffman formula if it tried to use the mass, axles, etc. But by having the ability to enter the ABC values, it is possible for the user to get OR to replicate the Koffman curve once the ABC values are entered.

Having looked at the CoalstoNewcastle website OR only models a non receiver 4cylinder compound at the moment. It would be great for us in the UK if a Midland/LMS style 3 cylinder compound could eventually be modelled.

I would be quite happy to try and help Peter on this in some way.

Unfortunately it is correct that only a non-receiver compound is currently modeled. Maybe in the future this might change, however as they say, I would not hold my breath while waiting.

Is there one compound model that is representative of all British models?

I think I will call myself happy with the Princess performance. This is the engine that I have:

Comment ( *** Cylinders *** )
NumCylinders ( 4 )
CylinderStroke ( 26in )
CylinderDiameter ( 19in )

...... SNIP .............

Your post suggests that you have now got a "good" working model, but this seems to be at the expense of adjusting the cylinder size, and stroke. Whilst I understand your wish to do this, it saddens me a little, as this was one of the principles that I had hoped to ensure, ie that the ENG would be defined by simply cutting and pasting "real" design information into it. To me moving away from this approach has the risk that it will mean that ENG files may become like some in MSTS, and bear no resemblance to the real locomotive design parameters. I suspect that your problem is caused by the way that TE is calculated, as there are many different formula for TE for the different types of compound locomotives. At the moment I do not have an answer for this issue.

I would however be interested in looking at the models, log sheet that you have referred to, etc at some stage out of interest. Would you be prepared to package it all up, and send it to me when you have finished working on it? You can contact me by email through the contact page of the CTN site.

Thanks again to all for an interesting thread.