Sie sind auf Seite 1von 6

Train Resistance Calculations

1 of 6

http://www.trainsimitalia.net/atdownload/guide/din_traz/train_resistan...

Train Resistance Calculations


by Joseph T. Realmuto
The purpose of this document is to detail how and why I made the modifications to the resistance values in the .eng and .wag files that came with Microsoft
Train Simulator. For the uninitiated, these are the files that control the physical behavior of the locomotives and cars that run in Microsoft Train Simulator. A
detailed description of all the values present in these files is included with the game in the files Eng_and_wag_file_reference_guide.doc and
Eng_and_wag_file_reference_guidev2.doc which are included with the game. They are not installed with the game, but can be found by going to the folder
Techdocs on CD1 and running the program Techdocs.exe. This program is a self-extracting executable which will unzip all the technical documentation for
the game to a folder of your choosing.
First of all, I need to start with a little background. I have had an interest in trains from the time I was young, and felt that eventually we would reach the
point where either models and/or computer simulations could perform prototypically. I acquired some of the tools needed to further this end during my
college education, although of course this was not my primary goal of going to college. I majored in electrical engineering, and also took a course in
transportation engineering, of which one of the subjects was individual vehicle motion. During my last 2 years I commuted to school daily, and this journey
included 8 miles by New York City subway and 51.5 miles on Amtrak's Northeast Corrider(New York to Princeton Junction, shuttle from Princeton Junction
to Princeton), so I am intimately familiar with the behavior and operation of real trains. This commuting sort of rekindled my interest in trains, which had
remained dormant for a number of years due to lack of time.
Upon graduating I made electronic projects in my spare time, and remotored HO scale trains with the goal of obtaining more life-like performance. I made a
number of control circuits to this end, some of which I plan to patent and market eventually, and did in fact reach my goal of making model trains accelerate
and decelerate realistically. This necessitated being able to measure scale speed accurately(another device for possible future patenting and marketing),
and to determine actual train performance based on weight, number of cars, etc. so that I would be able to adjust my control circuit accurately to reflect
various types of trains and motive power. These tests were tedious, and necessitated many calculations of train resistance. I used the equations present in
one of my texts(Introduction To Transportion Engineering and Planning by Edward K. Morlok, 1978 edition) for these calculations. Initially, I did many
calculations using paper and a hand calculator(I graduated in 1985, way before PCs were common). After I acquired my first PC in 1990(secondhandactually little more than a word processor) I wrote programs to do the calculations for me. This reduced maybe a half hour of work to a minute or so of
entering data, and with more accurate results. I made refinements to my control circuits as time permitted, although I still lack the space, time, or money to
build a scale layout.

During the last 2 or so years, my main hobby has been PCs. I acquired my first real computer(a 386-40) about 2 years ago from a friend, and have
"inherited" some of his newer machines also, the most recent of which is a Pentium II 450 which I upgraded to 512 MB of RAM. After going online about a
year ago, I was amazed at the quantity of train simulator programs, and tested some of them. Since I have been "out of the loop" for a while as far as
computers were concerned, I was amazed how good some of these simulations were. When MS Train Simulator came out, I eagerly purchased it to see
what a "state of the art" railsim had to offer.
To make a long story short, I was not disappointed with MS Train Simulator. My few criticisms of the program are more nitpicking than anything else. One
thing I noticed as I became familiar with the game was that the trains just didn't feel right when coasting, especially the passenger trains, which seemed to
lose speed too quickly. At very low speeds, the opposite was true. It was IMPOSSIBLE to let a train roll to a complete stop without applying the brakes. A
few quick calculations and use of the stopwatch on level track(i.e. NE Corrider) confirmed this. The next question-was it possible to correct this minor but
annoying deficiency? In short, the answer is yes, thanks to nearly every aspect of the program being controlled by simple text files.
Here I must digress to describe the equations used for calculating train resistance. A typical resistance equation looks like:

R = A + BV + CV
In this equation, R is the train resistance(friction), A, B, and C are constants which depend upon the type of train, and V is the velocity. Note that we have
one term which is a constant. This is bearing and contact friction, and varies with the weight of the train and number of axles. The second term is
proportional to velocity, and is mostly due to increased rolling resistance at higher speeds, although it also includes some components of laminar air flow.
The last term is proportional to velocity squared, and this is aerodynamic drag, or friction with the atmosphere.
The actual equations that I used are:
Locomotive:

R=1.3T + 29N + 0.03TV + 0.0024CAV


Passenger Car:

R=1.3T + 29N + 0.03TV + 0.041V

Standard Freight Car:

R=1.5T + 72.5 + 0.015TV + 0.055V

Trailer On Flat Car(TOFC):

R=0.6T + 20N + 0.01TV + 0.20V

6.8.2013. 12:17

Train Resistance Calculations

2 of 6

http://www.trainsimitalia.net/atdownload/guide/din_traz/train_resistan...

In all equations, R=resistance, T=weight, N=number of axles, A=frontal area, C=drag coefficient, V=velocity.
In these equations, resistance is in pounds, velocity is in mph, area is in square feet, and weight is in U.S. tons.
For those who prefer to work in the metric system, the metric versions are:
Locomotive:

R=6.4T + 129N + 0.033TV + 0.573CAV

Passenger Car:

R=6.4T + 129N + 0.033TV + 0.91V

Standard Freight Car:

R=7.4T + 325 + 0.165TV + 1.22V


Trailer On Flat Car(TOFC):

R=2.9T + 89N + 0.11TV + 4.45V

In these metric equations, resistance is in Newtons(4.45 Newtons= 1 pound), velocity is in meters/sec(1 meter/sec=2.237 mph), area is in square
meters(1 square meter=10.725 square feet) and weight is in metric tons.(1 metric ton = 1000 kg = 2207 lbs)
Since MS Train Simulator uses metric values, I was forced to use the metric equations. Note that in all these equations the first two terms together are
actually equal to the constant A in my general equation. This makes sense intuitively since more axles and /or more weight should give increased rolling
resistance. Please remember that these equations are NOT meant to be exact representations of a particular car, but are based on empirical
measurements of large numbers of cars, according to the source(Introduction To Transportion Engineering and Planning by Edward K. Morlok, 1978
edition). For more accurate analysis one could probably input the actual shape file into an finite analysis program. However, I have neither the software to
do so, nor a machine powerful enough to run such software on. In any event you would have to analyze an entire train at a time, since, for example, a box
car behind a flat car would have more drag than a box car behind another box car, and MS Train Simulator is not set up to have multiple sets of constants
for each car, so such calculations would be useless as far as the game is concerned. I'm not even sure if real railroads would bother to go through such
analysis when calculating how much motive power to put on a train.
MS Train Simulator uses a series of values to represent friction. The lines found in the .eng and .wag files look like:
Friction (
100N/m/s 1 -1mph 0 1
5.1N/rad/s 1 -1rad/s 0 1
)
The first set of five numbers in red are what is used to represent velocity dependent friction while the second five numbers represent friction on curves. It is
the first set of numbers that we are interested in. These numbers represent the constants C1, E1, V2, C2, E2, respectively. In this case, C1=100, E1=1, V2=
-1, C2=0, E2 = 1. These numbers are used by the program to calculate friction as follows:

E1

F=C1* V
F=C1* V
F=C1*V2

E1

E1

if V2< 0
if V<= V2

+ C2*V2 + C2*(V-V2)

E2

if V > V2

In these three equations A means A to the power of B. F is the friction in newtons(4.45 newtons = 1 pound), and V is the velocity in meters/sec(1
meter/sec = 2.237 mph). Since V2 is -1, or less than 0, then this means that with the values given above the program calculates friction as:

F=100*V =100*V

Therefore, by default MS Train Simulator calculates only the term proportional to velocity. This accounts for the near zero friction at very low speed.
Unfortunately, there is no way to make the program use three terms(A+BV+CV) to do a calculation, but it is possible to use non-integer exponents, which is

6.8.2013. 12:17

Train Resistance Calculations

http://www.trainsimitalia.net/atdownload/guide/din_traz/train_resistan...

nearly as good. I chose a very low value(1 mph) for V2, so that at speeds over 1 mph the program would use the more complex equation with two terms
proportional to different powers of velocity. I decided to choose 0 for the exponent of the first term, thereby making that term a constant since any number to
the zero power is 1, and found by experimentation that 1.8 gave the best fit over a wide speed range for the second constant. The new values(in red) in the
.wag files for a 100 metric ton freight car were as shown below:

Friction (
1065N/m/s 0 1mph 3.47N/m/s 1.8
5.1N/rad/s 1 -1rad/s 0 1
)

By plugging these numbers(C1=1065, E1=0, V2=1, C2=3.47, E2=1.8) into the generalized equations we end up with:

F=1065 if V<= 1 mph


F=1068.5 + 3.47V

1.8

if V > 1mph

For illustrative purposes, in the second equation I made (V-V2) equal to V since V2 is very small and therefore (V-V2) is very nearly equal to V. However, I
used the actual equation when plotting these graphs. The end result is that we have a constant term plus a term proportional to velocity to the 1.8 power.
For the non-mathematically inclined, you can raise a number to a non-integer power. For example, a number raised to the 0.5 power is simply the square
root of that number [i.e. 16

0.5

= 4].

A graph of the results is shown below:

As you can see, the modified equation nearly agrees with the actual resistance over the entire speed range from 0 to 150 mph. For comparison, I have
shown the results of the original resistance equation. You can see that although it doesn't diverge too much between 30 mph and 120 mph, it is way off as
speeds get under 30 mph or over 120 mph.

The original equations produce much worse results for a passenger car and a Dash9 locomotive, as shown by the following two charts:

3 of 6

6.8.2013. 12:17

Train Resistance Calculations

4 of 6

http://www.trainsimitalia.net/atdownload/guide/din_traz/train_resistan...

In the case of the passenger car, the original resistance calculation is way too high at nearly all speeds, and much too low as you get under 10 mph. For the
locomotive, the original resistance calculation is too low all the time. In both cases, my modified equation is in close agreement with the true value over the
entire speed range. The next graph is the resistance of a 50-car freight train pulled by two Dash 9 locomotives:

6.8.2013. 12:17

Train Resistance Calculations

5 of 6

http://www.trainsimitalia.net/atdownload/guide/din_traz/train_resistan...

Here something interesting happens. The original equation gives results which are fairly good between 30 mph and 90 mph. Since it underestimated the
locomotive resistance but overestimated the freight car resistance, these two tended to cancel each out. However, the results are again poor as we get
under 30 mph, which is where freight trains spend a good deal of their time. Once again, my modified equation gives nearly the same results as the actual
resistance equation at all speeds. Note that on this graph I have included the tractive effort curve as well. The point where this curve intersects the
resistance curve is the maximum speed of the train on level track. In this case, the train can reach 80 mph, in theory. In practice, the gearing on a Dash 9
will cause the tractive effort to rapidly decrease to 0 above 74 mph, so an actual train would only reach about 75 mph. On a locomotive geared for
passenger service, the tractive effort would start falling off at much higher speeds(110 mph for the Amtrak Genesis diesel, 165 mph for the Acela power
car).
I have included a program, Fcalc_10.exe, for those who wish to calculate the correct values to use when they add a new car to MS Train Simulator. For
most types of rolling stock, this program calculates the new values of follows:
1)The resistance of the car is calculated at a speeds of zero and 100mph(44.704 meters/sec) using the set of metric equations presented earlier in this
document.
2)The constant C1 is then set equal to the resistance at 0 mph.
3)Since V2=1, E1=0 and E2=1.8, the only unknown is C2. The program finds the value of C2 that satisfies the following equation:

C1 + C2*V2 + C2*(V-V2)

1.8

=calculated resistance at 100 mph(44.704 m/sec)

Since V=100 mph(44.704 m/sec) and V1=1 mph(0.44704 m/sec) this reduces to the following:
1.8

C1 + 0.44704*C2 + C2*(44.257)
Since (44.257)

1.8

=calculated resistance at 100 mph(44.704 m/sec)

is 917.841 we are left with:

C1 + 918.288*C2=calculated resistance at 100 mph(44.704 m/sec)

This is easily solved for C2. I have found that by choosing 100 mph as the speed at which my equation and the actual equation are set equal, there is very
good agreement at all speeds between 0 and 150 mph.
Note that I said for most types of rolling stock. For high speed trains such as the Acela I have found that using a value of 1.86 for E2 instead of 1.8, and
choosing 250 mph instead of 100 mph for the speed at which I calculate C2 gives much better agreement over the entire speed range. The program does
this automatically when you choose high-speed for the type of rolling stock. You are then asked to choose locomotive or passenger car for the type of
high-speed rolling stock. Some other changes I made are summarized in the equations below:
High-Speed Locomotive:

R=3T + 90N + 0.011TV + 0.573CAV


High-Speed Passenger Car:

R=3T + 90N + 0.011TV + 0.025LV

6.8.2013. 12:17

Train Resistance Calculations

6 of 6

http://www.trainsimitalia.net/atdownload/guide/din_traz/train_resistan...

Note that these are metric equations. Also note that I included a new factor, L, in the last term of the passenger car equation. This is the length of the car in
meters(1 meter=3.275 feet). A longer car will have more surface area than a shorter one, and hence more aerodynamic drag. In both equations, I reduced
the terms relating to rolling friction to account for the better maintained and lower friction bearings and better wheel balance found on these trains. The last
term in the locomotive equation remains the same, but remember that the frontal areas of high-speed locomotives will be less than conventional ones, and
the drag coefficient will be much less than one, sometimes as low as 0.25(I used 0.3 for the Acela). Also, the trailing locomotive will have much less drag
than the leading one, although it will have more than a passenger car due to the fact that it is dragging along the slipstream. In cases where there is only
one .eng file but two engines(leading and trailing), I assumed the rear engine had a drag coefficient of 40% of the front one, and put the average of the two
sets of values in the .eng file. The aerodynamic drag of a standard-length(85 foot) high-speed passenger car was assumed to be 70% of the drag of a
conventional passenger car due to flush windows and doors, closer coupling, and fairings over underbody equipment. The drag of a shorter car will be
proportionately less due to its shorter length. I don't have much data on high-speed trains, so I welcome any input in this area if my equations are seriously
incorrect. They do seem to give maximum speeds for the Acela of about 225 mph(factoring out the 165 mph gearing), which is about what one can expect
from this type of equipment without modifications. The French TGVs that set the speed records of 236 mph in 1981 and 320 mph in 1990 were shorter
trains with modified power plants. The train which set the standing record of 320 mph in 1990 had double the normal power, 3 trailers instead of 8 or 10, air
dams on the locomotives, rubber skirts between the cars, 43-inch wheels instead of the usual 36-inch wheels on the locomotives, and was on a 2.5%
downgrade when it hit the maximum. I am not taking anything away from this impressive achievement, but simply wish to point out that it was not done with
standard equipment. A standard train did sweep the tracks at 217 mph prior to the test runs, so this gives some idea of what the standard equipment can
do.
The last issue that I will address in this document is whether these modifications were worth the time that I spent on them, and if most users will notice a
difference. I think that the answer is yes. One thing which you can now do that you couldn't before is let your trains coast down to a complete stop. Prior to
my modifications, long freight trains would roll on for miles and would never come to a standstill. Now, you can coast down to a stop from 30 mph in maybe
20 minutes and 5 miles if you want. Coupling is more challenging as well, since you have to apply short bursts of power occasionally to keep a train going at
2 or 3 mph rather than just braking to that speed and letting it coast. You may also need to apply 25% power or more just to start a long freight train rolling
on level track due to the non-zero value of friction at 0 mph, whereas before even 2% would start it moving, albeit very slowly. Passenger trains will coast
much longer than before, but will also eventually come to a standstill. Long passenger consists will reach higher top speeds than before, as well. The
biggest difference will be noticed with the Acela. Since the friction was grossly overestimated before, the train will now coast for quite a while, even at higher
speeds. To coast down from 100 mph to 0 takes about 25 miles due to the train's fairly heavy weight and low drag. You actually need to use the brakes
when you see a reduced speed zone coming up. This gives some insight into why TGVs have four disc brakes per axle. There is simply no other way to
stop these things in a reasonable time frame. To sum it all up, the trains simply feel right now, whereas they didn't before.

References
1) Introduction To Transportion Engineering and Planning by Edward K. Morlok, copyright 1978 McGraw-Hill, Inc.

6.8.2013. 12:17

Das könnte Ihnen auch gefallen