I’m trying to get a better understanding of horsepower and train acceleration. looking for quantitative confirmation.
the plots shows train speed v. time for 4 loco horsepower values: red 3000 HP, orange 2000 HP, cyan 1500 HP and green 1000 HP. Raw results are also listed.
horsepower is a measure of work, lb-ft, per unit of time, minutes. This is equivalent to the product of force, lb, and speed, feet per min. My understanding of this is the tractive effort, the tangential force of the wheel on the rails is equal to the horsepower / speed. This means the force decreases with speed (the work is the same because a greater distance is traveled in the same amount of time). (The rectangular hyperbola mentioned in previous threads).
the plots also account for a maximum tractive effort of 70,000 lbs assumed for 150 ton locomotive
the plots indicate speed (vertical axis) in mph vs time in minutes (horizontal axis). The red curve, 3000 HP case, quickly reaches >30 mph w/in 5 min. ~7.5 mins. for 2000 HP (orange) and ~27 mins. for 1500 HP (cyan). The 1000 HP case (green) never exceeds ~24 mph because of insufficient HP.
the simulation also accounts for train resistance due to friction and aerodynamic drag as described in Armstrong’s book, “The Railroad”. His tables indicate that resistance increases from 2.3 lbs/ton @ 10 mph to 10.4 @ 70 for fully loaded cars and 4.5-19.8 lbs/ton for empty cars. This sim uses resistance for a fully loaded cars.
since the resistance increases with speed and the force from a constant horsepower decreases with speed, there can be a p
Hmm, I don’t see why this couldn’t be implemented with DCC. But not int he throttle. The decoder can measure the load ont he motor via BEMF, and a limit could be established to simulate both the maximum horsepower as well as tractive effort (often the limiting factor when starting out). For older locos you would need feddback to the throttle to light up the wheel sli indicator, whereas in many modern locos you can just (assuming indestrucbile drawbars) just open it wide up and the control system will push as much power to the wheels as it can without slipping.
Hmm, good reason to build a DIY decoder, so you cna experiment siwht using the BEMF data, and a couple of the available CVs to enable/disable this feature and to set the limit. You’d basically want to program in somethign that reprsents toe HP of the loco, as well as the maximum TE - a small 1200HP switcher can move a huge cut of cars that might require a pair of 3000HP road units - but the switcher can’t get that cut much above 15MPH while the pair of 3000HP road units can move it along at 60MPH. Low gearing and a higher TE to HP rating help the little guy do its work.
I think you could also do this in the throttle, but it would be more of a preset simulation and you’d need to change it if you drop the entire train and then move the same engine(s) light. You can do it today in the loco by implementing a speed curve that is not linear, but again then it is fixed no matter what the loco is going - moving light, moving a small train, or moving a train close to its maximum capacity. Actually, this has been somewhat mentioned in the past, using 3 step speed curves (or even the full 28 step tables) to configure some engines for faster response at the low end of the throttle and others for slower response at the low end. Uually discussed in terms of a drag freight loco vs a passenger loco. One example - lots of railroads used Trainmasters on commuter trains because of their high power and abilit
Wow that’s one large burst of technical data, … I will give you that.
In scale modeling, I will just call it momentum. My brother and I had this capability in the later part of the 70s when we were still teenagers.
You get set up with the right Transformer, or build one like we did. Then you just crank the knob on your power pack. Your locomotive or locomotives start to creep very slowly and accelerate up to your command in a realistic time frame.
Careful though, they stop doing the same in a more prototypical fashion like your data in reverse I suppose. It takes some getting used to. They have them with a break option but would be unrealistic at this point[:-^]
If you ever over shoot your destination, you might as well back up and try again.
Locomotives take awhile to start and take even longer to stop.
You wouldn’t want to use a dcc setup that replicated the real acceleration rates of locos hauling long and heavy trains. You’d be pulling your hair out waiting for the thing to get up to normal speed – if you’d get there at all.
Running the Conrail SEOP (Selkirk-Oak Point) some nights, with a longish train (say, 120 loads), on the table-top flat Hudson line going south, the engines would be in the 8th notch continuously and it would take 15 miles to begin to edge up close to track speed (50mph). That was with 3 B23-7’s, all they gave you for that run.
After dropping cars at Poughkeepsie and Croton – down to, say, 60 cars – they’d run much better!
yes, a more logarithmic speed curve, where each speed step gets smaller might model this behavior better than constant acceleration rate programmed into CV3.
My understanding of the use of acceleration rate is to delay each speed step change by an amount proportional to CV3.
The plots show a relatively sharp rise in speed and then each increase is substantially delayed for lower HP locos.
BEMF isn’t necessary unless you want to account for grades. The current speed step is known to the decoder.
I think there should be separate values for tonnage and HP. HP could be programmed into the decoder. Tonnage should be set for each train in the throttle and would need to be communicated to the decoder.
If this were implemented in a decoder, the loco would presumably start from zero if it lost power, a glitch that causes the processor to reboot. If this were implemented in a throttle, if the loco lost power it would start at the speed dictated by time since the throttle was increased. A keep alive would help in both cases.
[quote user=“doctorwayne”]
Al Krug did a nice article
I’ve watched and listened to operators using the Proto Throttle, and they say the start up, moving, acceleration, breaking, all feel much more realistic. One of the participants was an engineer.
I’ve never operated one, and probably won’t, but the way they describe it, it reminded me of the IC Hogger power pack with teathered walk around throttle. It had momentum and breaking features that took some time to get used to. I still have it, haven’t used it since the last plywood central.
It was nothing like the conventional power pack.
With my small layout, I’m happy with what I have, but watching the demo of the PT was impressive.
My reasoning for using BEMF is that the decoder could automatically know the ‘tonnage’ of the train it was pulling. Versus the somewhat unrealistic action of keying it in the throttle (which could just program the decoder)
Realistic yes, but there’s going to be a limit. Imagine you had unlimited room and really could put 50, 100 scake miles of track between towns. How many people would actually do such a realistic op session where at proper scale speed it takes you an actual 8 hours on your feet walking along with the train to get from the starting point to the destination? That would get old, real fast.
Hence a fast clock, even a moderate one, compressing the 15 miles of a real train getting up to speed to fit within what can be realisticlly built - even then it’s too much to try and simulate that accurately. At a 4:1 clock ration, it woult take 1/4 the time, 1/16 the distance - but how many layouts have even that much room even. At best, you’ll just reach speed when it’s time to brake for the next stop. At worst, you’ll be in the next town before even getting close to speed. Which is why I think using momentum, and maybe a non-linear speed table, is about as good a simulation as you are going to get int he confines of model space.
That does result in decidely non-linear acceleration. While the accel and deccel CVs are linear, setting a static time for changing from one speed step to another, if the speed curve is non linear so that each step is not an equal increment in speed, then you have made a non-linear acceleration curve. Because of the single factor nature of the momentum CV, the speed curve might have to be more exaggerated than it would be if no momentum were in use. Short of adding a second table that controls the momentum for each speed step, that’s the only current option and probably close enough, given all the other limitations in the model world.
The largest layout in my area is ironically part of this territory and has maybe about 10 scale miles of track. That means that with “prototypical” acceleration, the train would NEVER reach full speed. There is less than 15 miles, plus at some point the train would have to start to decelerate because it would run out of model railroad and would have to stop.
There is also the consideration of the the dynamics of the actual model train. A model car weighs lets say 5 oz. If it represents a load it weighs 5 oz. If it represents an empty it weighs 5 oz. A real car weighs maybe 30 tons empty and 130 tons loaded, roughly 4x the weight of an empty.
The speed curves are neat, but they only work when you match them to the weight of the train. What weight are you going to use for the train? If John runs 3 SD60’s on the point of his 15 car “loaded” grain trains because that’s the sizee for his layout space and Jim runs 3 SD60’s on the point of his 25 car “loaded” grain train because that’s the size for his space, are you expecting them to operate the same? They both represent the SAME train, a loaded unit grain train. What train weight are you going to tell the system to use to figure the acceleration? 15 model cars? 25 model cars? 15 real cars? 25 real cars? 100 real cars?
John’s layout has only 3 miles of main track. The grain train will meet 2 other trains along the way. That means it will only go about 1 mile before it has to stop. That means that one third to
OP wrote: “…BEMF isn’t necessary unless you want to account for grades. The current speed step is known to the decoder…”
This is incorrect. BEMF is very much needed for our decoders to meter out the proper voltage to the drive mechanism so that acceleration and deceleration are prototypical, and so that the drive responds accurately to the speed table’s parameters as the drive mechanisms running characteristics change with lubrication, operating temperature later in the session when the drive loosens, and to account for gear lash and wear over time.
i had asked for confirmation of the plots. No one seems to dispute their accuracy. Thanks for the link to the Krug page. I believe the plots are consistent though Krug is more focused on a train going up a grade which is a bigger deal.
the goal is to recognize how prototypical trains actually behave
of course most modelers can’t realistically model the distances and train lengths of real railroads. but the ProtoThrottle suggests that many want to model the behavior more realistically. I don’t believe constant acceleration is realistic.
in terms of time, simply rescale the time axis in seconds instead of minutes: ~5 sec for the 3000 HP loco to get to 30 mph, 8 sec for the 2000 HP loco and ~20 sec for the 1500 HP loco.
on the following plot I added curves using CV3 settings of 5, 9, 25 and 40 where the x-axis for these curves is in seconds. If modelers are willing to use CV3 value similar to these, then the more realistic plot times in seconds should be acceptable on many layouts.
but again, it’s not the length of time, but the speed profile that makes it more realistic, depending on hp and tonnage.
[quote user=“selector”]
BEMF is very much needed for our decoders to meter out the proper voltage to the drive mechanism so that acceleration and deceleration ar
Now graph those same speed curves on the SAME time scale. You end up with a square wave.
The other thing to realize in all this is that railroads power their trains up to certain horsepower per trailing ton based on the TRAIN not necessarily an engine.
If there is a 5000 ton grain train it might get 2500 hp. If its a 5000 ton manifest train it might get 5000 hp. If its a 5000 ton intermodal train it might get 12,000 hp.
If you put 2 GP40’s on a 5000 ton manifest train, and 8 GP7’s on an 5000 ton intermodal train, while a single GP7 has a slower acceleration curve, I guarantee that the intermodal train will out accelerate the manifest train. The acceleration curve of a TRAIN is the sum of the horsepower of the engines (actually its more complicated).
If you program a 1500 hp engine to top out at 30 mph, that’s baloney. Early 4 axle units were (F3-F7-GP7-RS2-RS3-FA1-FA2) were nominally 1500 hp and easily handled trains at speeds up to 60-70 mph. The ATSF used 1500 hp engines on its premier transcontinental passenger trains at very high speeds with rapid acceleration.
A railroad powers a TRAIN to the achieve the acceleration it needs for that TRAIN.
Some railroads have gone away from horsepower/trailing ton (hp/tt) and have gone to a tons per powered axle (TPA). A “powered axle” represents a specific amount of tractive effort and engines are rated at how many “axles” they have. A C44AC might r
you mentioned a square wave in your frist response where all the plots were on the same time scale: minutes.
if you used a 60:1 fast clock, the scale of the plots would be seconds.
if you used a 1800:1 fast clock - 1 sec == 30 mins, then all trains would immediately jump to full speed (i.e. 23 mph for the 1000 hp case). Is this what your suggesting.
not withstanding the case above, I don’t see how time scale affects the profile.
i’m not suggesting modeling prototypical increases in speed that take minutes.
i’m suggesting that instead of using constant acceleration to full speed over the number of seconds resulting from the DCC CV3 value, that for hp limited trains, the speed should increase rapidly at first but then much more slowly.
that profile depends on the hp of the loco (or combined hp of the consist) and the tonnage of the train. I’ve mentioned 3 profiles
…while BEMF can accurately measure speed, it is not necessary for more realistic behavior unless there are severe mechanical issues with the model loco or you’re interested in a very accurate speed profile.
Simply setting the motor voltage for more realistic speed at a more realistic time should be more than adequate.
I don’t really have an opinion here since I don’t use DCC.
Even with the momentum feature turned off, my Aristo throttles have an “acceleration rate” that is not instant. Being a push button throttle you must hold the “FAST” button for a period of time to achieve full speed.
I have never mapped it to see if it is linear or not, but my natural sense is that it is not. And that by default, the pulse with modulation circuit of the Aristo throttle already provides the effect Greg is looking for.
I have never cared for using any throttle with momentum, DC or DCC.
And one of my dislikes of many DCC throttles are encoder wheel throttle knobs.
So maybe pulse width modulated DC radio throttles, with its ability to provide constant brighness headlights before motion, a reasonable acceleration curve, and reliable slow speed is not so “dated” after all.
But I have never been concerned with having the “latest thing”, just with having stuff that works well for my needs.
