U.S. patent number 3,844,514 [Application Number 05/339,473] was granted by the patent office on 1974-10-29 for car retarder control system.
This patent grant is currently assigned to General Signal Corporation. Invention is credited to John J. DiPaola, Richard A. Dobson, Charles W. Morse.
United States Patent |
3,844,514 |
DiPaola , et al. |
October 29, 1974 |
CAR RETARDER CONTROL SYSTEM
Abstract
A control system for a railway car retarder to be used in a
classification yard. The control system includes a digital computer
which is programmed to compute a braking pattern for each car or
cut of cars which is to pass through the retarder. Taking into
account the characteristics of the cut and its velocity in entering
the retarder, the computer determines a braking pattern which will
slow the cut of cars down to the desired exit velocity so that the
cut will traverse the remaining portion of the classification yard
and couple to the preceding cars without excessive speed. The
braking pattern is that which will decrease the velocity of the cut
from its entering velocity to the desired exit velocity using the
maximum amount of retarder length as is possible. In this manner,
the braking force the cut is subjected to is minimized consistent
with decreasing its velocity by the desired amount.
Inventors: |
DiPaola; John J. (Penfield,
NY), Morse; Charles W. (Rochester, NY), Dobson; Richard
A. (Caledonia, NY) |
Assignee: |
General Signal Corporation
(Rochester, NY)
|
Family
ID: |
23329153 |
Appl.
No.: |
05/339,473 |
Filed: |
March 8, 1973 |
Current U.S.
Class: |
246/182A;
104/26.2 |
Current CPC
Class: |
B61K
7/12 (20130101) |
Current International
Class: |
B61K
7/00 (20060101); B61K 7/12 (20060101); B61k
007/12 () |
Field of
Search: |
;246/182A ;104/26A |
References Cited
[Referenced By]
U.S. Patent Documents
Primary Examiner: Wood, Jr.; M. Henson
Assistant Examiner: Libman; George H.
Attorney, Agent or Firm: Kleinman; Milton E. Vande Sande;
George Wynn; Harold S.
Claims
What is claimed is:
1. In a car retarder control system for a railway classification
yard having a multi-position car retarder operable to different
braking positions as each cut traverses said retarder for adjusting
the speed at which cuts being classified couple on their assigned
storage tracks, by reducing the speed of the cut from its entering
velocity to a preselected exit velocity, the combination of:
means responsive to the velocity of said cut when entering said
retarder and to the preselected exit velocity for said cut for
providing a manifestation representative of each braking force to
be exerted by the retarder on each cut at each instant of its
travel through the retarder to reduce its velocity to said
preselected exit velocity at the exit end of the retarder,
means for providing a manifestation related to the position of each
cut as it passes through said retarder,
and means responsive to said first recited means and said second
recited means for varying the braking force exerted on each cut by
said retarder as it passes through the retarder to reduce its
velocity to said preselected exit velocity at the exit end of the
retarder.
2. The retarder control system of claim 1 wherein the retarder is
of the type which is operable to a plurality of discrete braking
positions,
said first named means computing, for each cut, the length of
retarder required to reduce the speed of said cut to said
preselected exit velocity at the exit end of said retarder in
response to the application of light braking force by said
retarder, said means controlling said retarder to apply a braking
force greater than light only when said computed retarder length
exceeds the length of said retarder.
3. Apparatus for controlling a discrete position retarder in a
railroad classification yard in order to control the coupling speed
of a cut when it reaches its destination track and couples to cars
preceding it, comprising,
means for determining the required length of retarder to decrease
the velocity of a cut as it enters the retarder to a preselected
exit velocity with light braking only,
means for comparing said previously computed length with the
effective length of said retarder,
means for storing the results if the required length is less than
or equal to the effective length,
and control means responsive to said means for storing to control
said retarder in accordance with said stored results.
4. The apparatus of claim 3 which further comprises,
means for selecting, if said required length is greater than said
actual length, a point intermediate the entrance and exit of said
retarder and selecting a velocity for said cut which is
intermediate the entrance and exit velocity,
means for computing the required retarder length to decrease the
velocity of the cut from said selected velocity to a preselected
exit velocity with light braking only,
means for comparing said last previously computed length with the
length of said retarder from said intermediate point to said exit
end,
and means for transmitting the results, if the required length is
less than or equal to the actual length, to said means for
storing.
5. The apparatus of claim 4 which further comprises,
means for incrementing the location of said intermediate point if
the last computed length is greater than the length of said
retarder from said intermediate point to said exit end and
decrementing said selected velocity for said cut from said last
selected velocity,
means for determining the required length of said retarder to
decrease said selected velocity of said cut to a preselected exit
velocity with light braking only,
means for comparing said previously computed length with the length
of said retarder from said last selected intermediate point to the
exit of said retarder,
and means for transmitting the results, is the last computed length
is less than or equal to the length of said retarder from said last
selected intermediate point to the exit end, to said means for
storing.
6. The apparatus of claim 5 which further includes means for
determining when the distance from said intermediate selected point
to the entrance of said retarder is equal to or greater than the
length of said retarder,
means responsive to said last named means for determining the
required length of retarder to decrease the velocity of said cut as
it enters the retarder to a preselected exit velocity with higher
braking only,
means for comparing said previously computed length with the
effective length of said retarder,
and means for transmitting the results, if the required length is
less than or equal to the effective length, to said means for
storing.
7. A method of controlling a discrete multi-position retarder in a
railway classification yard comprising the steps of,
computing the necessary retarder length to decrease the velocity of
a cut from its entering velocity to a preselected exit velocity,
with light braking only,
comparing said necessary length with the length of said
retarder,
computing and storing the time elaspsed and distance covered in
reducing the velocity of said cut only if the necessary length is
less than or equal to the length of the retarder,
said controlling said retarder in accordance with the computed and
stored time and distance.
8. The method of claim 7 which further comprises,
selecting, if the necessary length is greater than the length of
retarder, a point intermediate the entrance and exit of said
retarder and selecting a velocity for said cut at that point which
is intermediate the entrance and exit velocity,
computing the required retarder length to decrease the velocity of
the cut from the selected velocity to the preselected exit velocity
with light braking,
comparing the last previously computed length with the retarder
length from said selected point to the exit end,
and computing and storing the elaspsed time and distance covered in
reducing the velocity only if said computed length is less than or
equal to the retarder length from said selected point to the exit
end.
9. The method of claim 8 which further includes the steps of,
incrementing the location of said selected point as measured from
the entrance end and decrementing the sleected velocity if the
computed length exceeds the length of said retarder from selected
point to the exit end,
computing the required length of retarder to decrease said selected
velocity to said preselected exit velocity with light braking
only,
comparing said computed length with the length of said retarder
from said selected point to the exit end,
computing and storing the time elapsed and distance covered in
reducing the velocity to said preselected exit velocity if the
computed length is less than or equal to the length of retarder
from said intermediate point to the exit end.
10. The method of claim 9 which further includes the steps of,
determining where the distance from said entrance end to the
incremented intermediate point is equal to or greater than the
length of the retarder,
computing required length of retarder to decrease the velocity of
said cut as it enters the retarder to a preselected exit velocity
with higher braking,
comparing the previously computed length with the length of the
retarder,
computing and storing the elapsed time and distance to decrease the
velocity to the exit velocity if said computed length is less than
or equal to the length of said retarder.
11. The method of claim 8 which further includes the steps of,
defining the length of the retarder as the distance between the
entrance of said retarder and said intermediate point
defining said preselected exit velocity as said selected
velocity,
computing the retarder length necessary to decrease the entrance
velocity of the exit to said preselected exit velocity at higher
braking,
comparing said last computed length with the length of the
retarder,
computing and storing the time elapsed and distance covered in
reducing the velocity to the preselected exit velocity if the
computed length is less than or equal to the length of the
retarder.
12. Apparatus for controlling a retarder to decelerate a cut
passing therethrough to a preselected exit velocity comprising,
first means for determining if light braking, only, is sufficient
to decelerate said cut to said preselected exit velocity and for
computing the extent of said light braking if light braking, only,
is sufficient,
second means, responsive to said first means, for determining, if
said light braking only, is insufficient, the extent, if any, of
light braking to decelerate said cut to said preselected exit
velocity with braking force transitions along a desired
velocity-distance profile and for also determining an intermediate
velocity of said cut at said braking force transition,
third means, responsive to said second means, for determining a
minimum braking level, above light braking, and the extent thereof
required, to decelerate said cut to said intermediate velocity with
braking force transitions along said desired velocity-distance
profile,
fourth means, responsive to said first, second, and third means for
storing parameters associated with the respective braking levels
determined by said first, second, and third means,
and control means, responsive to said fourth means, to control said
multi-position retarder to its various braking positions to the
extent stored by said fourth means.
Description
BACKGROUND OF THE INVENTION
In a classification yard, a train of railway freight cars is pushed
over the crest of a hump, and each car is then allowed to roll by
gravity down the hump and over a number of route-selecting switches
to a particular one of a number of destination tracks. When several
successive cars are to go to the same destination track, they are
usually left coupled together and allowed to roll together to their
destination track; such a group of cars is called a "cut". In this
way, the cars of a train are classified according to their intended
destination. Hereinafter a cut will be referred to even though it
may consist of only one car.
The grade of a hump is made sufficient so that the car with the
hardest rolling characteristics can reach the most remote
destination in the classification yard and couple onto other cars
in that same destination track. Easier rolling cars must,
consequently, be decelerated so that they too will reach their
destination tracks at a suitable coupling speed. This deceleration
is accomplished by providing car retarders along the track rails
whose brake shoe beams apply a controllable braking force to the
rims of the car wheels.
The earliest car retarders used in classification yards were
manually controlled by an operator who visually observed the cut
proceeding through the retarder, and, with knowledge of the cut's
destination track, selected an appropriate braking force to be
applied to the cut. Difficulties with properly estimating the
amount of deceleration required has led to the use of automatic car
retarder control apparatus. In the past, a variety of schemes have
been used, some of which employed radar speed measuring apparatus
to measure the speed of the cut through the retarder and computing
apparatus to compute the desired cut velocity through the retarder.
The apparatus would compare the desired cut velocity with the
actual cut velocity as determined by the radar speed sensing
apparatus and approximately control the retarder. The complexity of
this apparatus and difficulties found in its use have led to the
desire for a digital approach to the control of a car retarder
along with the elimination of the radar speed measuring apparatus
in the control loop.
In some prior art systems for control of car retarders, the
retarder has been preset, prior to the entry of a cut into the
retarder, to provide a predetermined braking force dependent
primarily upon car weight, with a greater degree of braking force
being applied to cars which are classified as being heavy. In
addition, computations have been made for each cut taking into
account a plurality of different parameters with the objective of
determining an appropriate relase speed from the retarder.
Thereafter, as the cut progresses through the retarder, its speed
is continually monitored, and the retarder is released when the
speed of the cut has been reduced to a value at or near the
precomputed release speed. Some systems of this general type have
further provided for a reduction in the braking force of the
retarder as the cut proceeds through the retarder. One disadvantage
of such prior art systems has been that the retarder was released
at a time prior to the cut's reaching the exit end of the retarder
with the result that the speed of the cut at the exit end could by
that time have reached a value different from the precomputed,
desired retarder release speed, thereby introducing error into the
system. Also, such a prior art system has resulted in the
application of a higher degree of braking effort, particularly at
the retarder entrance end, than is actually necessary to reduce cut
speed to its desired release speed at the retarder exit end,
thereby resulting in unnecessary wear on the retarder brake shoes
and at times also tending to force the wheels of a car out of the
retarder.
Furthermore, some of the prior art systems which employ feedback in
the control loop subject the retarder to a large number of
different orders during the passage of a cut. This reduces the life
of the retarder mechanism. In particular calling for a low braking
effort and then a higher braking effort subject the retarder to
extreme wear and this type of operation should be avoided
The system disclosed in this application employs a digital computer
to compute a retarder braking control profile for each cut which is
to traverse the retarder. The profile is computed so that the
maximum amount of retarder length can be utilized which minimizes
the amount of braking force applied to the cut at any instant of
time. The cut characteristics, the retarder entering velocity, and
the desired exit velocity are used in the generation of the
retarder braking control profile. The actual retarder control
profile is a series of commands for the retarder, directing it to
one of a plurality of positions to exert one of a plurality of
corresponding braking forces on the cut in the retarder. Since the
present invention is not concerned with aspects of the overall
retarder control problem such as determining the rollability
characteristics of the cut, the weight of the cut, its entering
velocity and computation of exit velocity, the details of apparatus
to perform these functions will not be disclosed herein. Apparatus
to perform these functions are disclosed for instance in U.S. Pats.
Nos. 3,054,983, 3,110,461, 3,217,159, 3,253,141, and 3,268,725, all
assigned to the assignee of the present application.
SUMMARY OF THE INVENTION
The present invention provides a control system for a car retarder
used in a classification yard which maximizes the distance over
which the retarder is used. It is a corresponding object of the
present invention to minimize the amount of braking force applied
to the cut in the retarder at any particular instant of time
consistent with the necessity to reduce the velocity of the cut to
the desired exit velocity.
It is another object of the present invention to utilize a properly
programmed digital computer to compute a set of braking commands
for a railway car retarder in a classification yard so that the
retarder will be utilized for the maximum length and which
minimizes the braking force as is consistent with the necessity to
decrease the cut velocity to a desired exit velocity. It is still a
further object of the present invention to provide a system which
meets the foregoing objectives and at the same time eliminates the
necessity for controlling the retarder in accordance with the
actual cut velocity as the cut proceeds through the retarder.
Furthermore, it is an object of the present invention to compute a
braking pattern for a retarder in which the braking effort, if
caused to change, will decrease as the cut proceeds through the
retarder.
It is a further object of the present invention to eliminate the
necessity for continually sensing cut velocity in the retarder and
thus eliminate the necessity for radar speed measuring devices in
the control loop.
BRIEF DESCRIPTION OF THE DRAWINGS
In describing the invention in detail, reference is made to the
accompanying drawings in which:
FIG. 1 is an explanatory graph of velocity versus distance of a
theoretical cut in a retarder;
FIG. 2A is a flow diagram of a portion of the program utilized in
the instant invention entitled "enter" RCPG;
FIG. 2B is a flow diagram of a portion of the program used in the
present invention entitled "start";
FIG. 2C is a flow diagram of a portion of the program used in the
present invention entitled "next";
FIG. 2D is a flow diagram of a portion of the program used in the
present invention entitled "over";
FIG. 2E is a flow diagram of a portion of the program used in the
present invention entitled "done";
FIG. 3 is a flow diagram of a portion of a program utilized in the
present invention entitled V.sub.en ; and
FIG. 4 is a schematic showing the inter-relationship of apparatus
used in the present invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
FIG. 4 shows the schematic form of the apparatus utilized in
accordance with the present invention to effect proper retarder
control. The portion of the trackway 100 shown in FIG. 4 represents
a portion of the main trackway in a classification yard and, in
accordance with normal practice, the physical track would exist on
a downgrade of known slope. The retarder 104 is preceded by a
number of wheel detectors 105. Each of the wheel detectors provides
an input to the central processing unit 101. The central processing
unit per se is a digital computer of known form which is provided
with the information necessary to compute, in accordance with the
program disclosed in FIGS. 2 and 3, the retarder braking control
profile. The wheel detectors, sensing the time of passage of the
first axle of the cut, enable the central processing unit 101 to
determine the velocity of the cut.
The retarder 104 is controlled by the retarder operating mechanism
103 to apply a controllable braking force to the cut as it proceeds
through the retarder in order to decrease the velocity of the cut
to the desired exit velocity. Since it is the force exerted by the
retarder on the wheels of the cut which is controllable, one of the
factors which must be taken into consideration by the central
processing unit 101 is the mass of the cut. From the well-known
formula that acceleration is equal to force divided by mass, the
central processing unit 101 can determine, for any given mass and
any given force, the acceleration that the force will exert on the
mass.
In addition to the car velocity information received by central
processing unit 101 from the wheel sensors 105, the central
processing unit 101 also receives further information related to
the cut characteristics for which the retarder control profile is
to be generated and receives these characteristics over line 106.
The central processing unit 101 requires information as to the
length of the cut, the length of the first car of the cut if there
is more than one car in the cut, and the total mass of the cut.
Furthermore, the central processing unit 101 receives via line 107
a precomputed exit velocity for the cut. The program discussed with
respect to FIGS. 2 and 3 is designed to control the retarder so as
to reduce the cut's initial velocity to this precomputed exit
velocity. In addition, permanent information stored in the central
processing unit 101 is indicative of the characteristics of the
retarder. For instance, the various levels of braking force
available by the retarder are available. Furthermore, for each of
the different braking positions of the retarder there is a minimum
allowable time in that position and also a transition time, i.e.,
the time it takes for the retarder to move from one braking
position to another.
With the aforementioned information and the programs discussed with
respect to FIGS. 2 and 3, the central processing unit 101 can
generate a retarder braking control profile for a particular cut.
The profile consists of one or a series of discrete braking
position commands along with the time duration during which that
command is effective and the computed distance the cut will travel
with the retarder in that position. The profile is stored in table
102 so that the profile information can operate retarder operating
mechanism 103.
Table 102 is a schematic showing of a portion of a memory device to
store retarder operating controls and, physically, may be a part of
the central processing unit 101.
The clock, 109, is started when the car enters the retarder and
comparator 108 determines, based on the times stored in the tables
102, when a particular braking command is to be effective. At such
time the comparator 108 causes the command to be transmitted to ROM
103 by gate 110.
Since tables 102 store, in addition to the time span for each
braking command, the distance traveled by the cut during that
command, the control could also be on a distance basis. That is a
track circuit and A/D converter could be substituted for clock 109.
The comparator 108 would then perform a distance rather than time
comparison but the control would proceed in the same manner as
discussed above.
Prior to describing, in detail, an embodiment of the present
invention, it will be helpful to describe first a simplified
embodiment to illustrate some of the basic principles of this
invention. FIG. 1 is a graphical representation of theoretical cut
velocity versus distance through the retarder.
As this description proceeds it will be apparent that some
relationship must be assumed for the variation in cut velocity with
distance through the retarder. The linear relationship shown in
FIG. 1 has been chosen as practical and easy to mechanize.
Furthermore it meets one of the objectives of the invention, i.e.,
to use as much of the retarder as possible within the practical
limits of the problem. It should be stressed that the linear
relationship is used only as a starting point for the computations
and the variation of actual cut velocity with distance through the
retarder is shown by the non-linear portions of FIG. 1.
To obtain a linear cut velocity representations as shown would
require constantly varying the braking force exerted by the
retarder on the cut. The expression
V.sub.e =.sqroot.V.sub.i.sup.2 + 2AX
expresses the relationship between the initial velocity (V.sub.i)
and exit velocity (V.sub.e) under constant deceleration (A) over
the braking distance variable X. Since this is not a linear
function, any constant deceleration will not produce the straight
line relationship between car velocity and distance as expressed in
FIG. 1.
The present invention attempts to maximize the distance over which
the retarder is effective by determining the degree of retardation
which must be applied to the cut as it travels through the retarder
to obtain the desired objective, i.e., a cut velocity at the exit
end equaling the precomputed exit velocity. Since the apparatus of
this invention is particularly adapted for use with retarders which
are operable to provide a plurality of discrete levels of
retardation, the apparatus first computes the distance necessary to
reduce the car's initial velocity to the desired exit velocity on
the assumption that only the minimum braking force is to be used
throughout its travel through the retarder. If this distance is
greater than the effective retarder length, it is apparent that the
minimum braking force will not be sufficient if applied over its
entire length. However, to maximize the retarder length which is
utilized, the present invention then proceeds down the theoretical
profile, i.e., the straight line 6, and attempts to fit minimum
braking to the remainder of the constraints. A brief example will
suffice. The numerical values used here are chosen for illustrative
purposes only. Clearly some of the values chosen would not be met
in practice. Assume, as shown in FIG. 1, that V.sub.i = 60 feet per
second and V.sub.e = 10 feet per second, and that light braking
corresponds to a deceleration of 20 feet per second per second. The
first computation employing the formula D = (V.sub.i.sup.2 -
V.sub.e.sup.2)/2a shows a required length over 87 feet. Since this
is greater than the 50 feet we have assumed for the retarder
length, we will move down the theoretical velocity-versus-distance
profile at equal distance increments (i.e., such as one foot
increments) and repeat the computation. When we reach point 4, at
30 feet per second, some 30 feet through the retarder, the
computation will be performed with v.sub.i = 30 feet per second,
V.sub.e = 10 feet per second and will indicate that an effective
retarder length of 20 feet is required. Since this is exactly (the
equality between remaining length and required length is not
required, it is only necessary that the required length be less
than or equal to the remaining length) the amount of retarder left
at this point, the computation has shown that it is permissible to
use light braking, corresponding to a deceleration of 20 feet per
second per second, if we can achieve a car velocity, at a point 30
feet through the retarder, of 30 feet per second. Thus, a portion
of the problem has been tentatively solved.
A prerequisite for the foregoing solution was the deceleration of
the car from 60 feet per second initial velocity to 30 feet per
second intermediate velocity through 30 feet of the retarder. To
determine if this is possible, we can use the above formula once
again, using a higher braking force, of say 40 feet per second per
second. The first trial computation at point 1 indicates that over
30 feet of braking length would be required, and, since we have
only 30 feet to work with, that will obviously not be sufficient.
After a number of unsuccessful trial computations a computation
will be made at point 2, 10 feet from the beginning of the
retarder, with an initial velocity of 50 feet per second, a final
velocity of 30 feet per second and a deceleration of 40 feet per
second per second. This will show that 20 feet of retarder length
is required. This is exactly (as above, the equality between
remaining retarder length and required retarder length is not
necessary) what is available and therefore a second portion of the
problem has been tentatively completed.
The only remaining portion is to determine if a still higher degree
of braking will slow the car's initial velocity of 60 feet per
second to our required 50 feet per second in 10 feet or less of
retarder. Again using the same formula, we fine that 71/2 feet will
be sufficient at the higher braking lever of 60 feet per second per
second. Since this is less than the 10 feet we had available, the
third and final portion of the problem is completed.
In summary, the braking of our assumed car would begin at a
deceleration of 60 feet per second per second and would continue
until the car has traversed 71/2 feet of retarder at which time its
velocity would have decreased to some 50 feet per second. At this
point, the retarder is controlled to apply a deceleration
corresponding to 40 feet per second per second which would continue
until the car has traversed an additional 20 feet through the
retarder at which point its velocity would now be down to 30 feet
per second. At this point, the retarder control would again be
changed to apply an effective deceleration of 20 feet per second
per second for the remaining 20 feet of the retarder length. The
car would then exit from the retarder with a velocity of 10 feet
per second. Thus, the retarder is to be utilized over 471/2 of its
total of 50 feet and the car has been slowed with the maximum
amount of light braking that is possible. Furthermore the braking
effort, when changed, has always decreased.
The above example is only illustrative, as it obviously uses
speeds, distances, and deceleration factors which might not at all
be met in practice. Furthermore, the above example ignores the
transition effects in changing the retarder effective deceleration
and also ignores the acceleration of the car in the retarder when
it is in the open position. The above example does however
illustrate the manner in which the computation proceeds from the
exit end of the retarder toward the entrance end and attempts to
fit the car and retarder characteristics to an optimal
velocity-distance profile.
FIG. 2A shows the first portion of the routine entitled RCPG,
standing for Retarder Control Profile Generator. As has been
previously explained, this program, taking into consideration the
desired exit velocity of the cut, and the braking effect of the
various retarder braking positions available on the cut, and the
cut's entering velocity, will generate a series of orders for the
retarder control. These orders will determine when, and to what,
positions the retarder is directed to during the time that the cut
is subject to action by the retarder.
The first step 10 in the program is to clear the tables. The orders
which the program generates for the retarder control are stored in
the table and clearing the table ensures that it will accumulate
only orders with respect to the cut which is presently under
consideration. The next step, 11, sets the current acceleration
equal to "light". The term "acceleration" is used in the generic
sense, that is, covering both acceleration and deceleration. The
braking force that the retarder exerts on the car or cut is thus
properly termed an acceleration although in the algebraic sense it
is a deceleration. Furthermore, the retarders which this program is
designed to control have a number of discrete positions available,
such as open, light, medium, heavy, and extra heavy. Of course, in
the open position, the retarder would have no braking effect on the
cut; in the light position some braking occurs, and in the medium,
heavy, and extra heavy positions more and more braking is effected.
The step presently under discussion, 11, merely sets the initial
trial retarder position to "light". This is in keeping with the
general purpose of the program which is to control the retarder to
utilize as much retarder length as possible. In line with this
goal, the program attempts to fit the fixed parameters; entering
velocity and exiting velocity, to a retarder control profile which
utilizes the minimum possible braking effort at any one time.
The next program function, 12, is to set a parameter for previous
acceleration (AP) to open. The computations performed by the
program utilize, in addition to various velocities, parameters for
current acceleration and previous acceleration. This function, 12,
sets the previous acceleration to the open position.
At this point it may be helpful to the reader to say a little about
the term "previous", as previous acceleration or previous starting
velocity (which will come up shortly). The discussion with respect
to FIG. 1 showed how the retarder control profile is computed
separately for the different braking efforts required. After the
tentative solution for light braking (between 30' and 50' on FIG.
1) was arrived at an intermediate braking computation was made
(between 10' and 30' on FIG. 1). In this manner the braking control
profile is built up from the exit end of the retarder. It is in
this sense that the word previous is used. Therefore the light
braking (between 30' and 50' in FIG. 1) is "previous" to the
intermediate braking (between 10' and 30' on FIG. 1) because it is
used first in the computation. In the same sense for the exit
transition computation from light to open the previous acceleration
(AP) is open and the present acceleration (A) is light.
The function "clear repass work", 13, is an internal record keeping
function. Depending upon the results of the various trial
computations that are made the repass word may be set to keep track
of the completeness of the computation. For instance, it may become
necessary to know whether or not a previously computed partial
solution has been rejected for one reason or another. The condition
of the repass word is then significant. However, prior to when it
becomes necessary to set the repass word, this word should be
cleared. This function, 13, ensures that the repass work is
cleared.
The next function "set final velocity" (v.sub.f), 14, sets a
parameter used in the computation (V.sub.f) to be equal to the exit
velocity. The exit velocity is obtained in a manner well known in
the art, based upon the classification track to which the cut is
directed, its distance from the retarder, the profile of the
terrain to that track and the weight and rolling characteristics of
the cut. A typical example showing apparatus to perform an exit
velocity computation is shown in prior U.S. Pat. No. 3,217,159.
Subsequently, the exit velocity is also stored as the previous
starting velocity in function 15. This is another program parameter
that will be utilized as the computations proceed. Finally, at
function 16 the parameter "distance left" (SL) is set to be equal
to the total retarder length plus the total wheel base of the cut
minus the first wheel base of the cut. It is normal classification
yard practice to handle all directly adjacent cars which are
destined for the identical classification track as a unit or cut of
cars. Of course, if there are no such directly adjacent cars
destined for the same classification track, a cut may be made up of
only one car. In those cases the distance left (SL) would be merely
the retarder length since the other functions of adding in the
total wheel base and then subtracting the first wheel base would
cancel out. However, where a cut is made up of a number of cars,
the distance left (SL) will be equal to the total retarder length
plus the total wheel base of the cut less the wheel base of the
first car in the cut. This is the effective length of the retarder
for that cut. Since all cars in a cut are coupled, if any one of
them is in the retarder, the retarder is effective on the entire
cut. That is the basis for computing the distance left (SL) in the
manner just stated.
From function 16 the program proceeds to the routine "start".
"Start" is shown in FIG. 2B. "Start" performs some basic
computations related to a transition and also initializes the
program so the computations contained in OVER can be performed. The
explanation of FIG. 1, the simplified description, stated that the
transitions in braking had been ignored. However, to refine the
accuracy of the computation, the transition periods in the actual
program are not ignored. When a retarder changes position, that is,
when it changes from one braking effort to another, the braking
force and the resulting acceleration imparted to the cut will be
somewhere between the value imparted by the previous position and
the value imparted by the new position. Since the program
computations proceed from the exit end of the retarder and build up
the control from that end, this first transition to be computed
concerns the last transition of the retarder operation, i.e., from
some braking position to the open position.
The first function, 17, computes average acceleration (A.sub.ave)
during the transition from light braking, which had been set at
function 11 and the open position, which had been set at 12. The
computation proceeds accordng to the formula: A.sub.ave = (A+
AP).div. 2.
To compute further the parameters in this transition, it is
necessary to know the time delay during which the transition takes
place, that is how long does the retarder take to traverse from the
light to open position. This is a precomputed time which merely
requires reference to a table and is accomplished in function
18.
Although the exit velocity of the cut is a precomputed parameter,
it is now necessary to determine the theoretical cut velocity when
the retarder initiates the transition from light braking to open.
This computation is performed at function 19 according to the
formula V= V.sub.f + (A+ AP/2) T.sub.a, where V.sub.f had been set
previously as the exit velocity, the expression (A+ AP)/2 is the
average acceleration during the transition and T.sub.a is the time
taken by the transition. This computed velocity is the velocity the
cut should have when the retarder begins its transition from light
braking to open.
It is now also necessary to compute the retarder length covered by
the cut during the transition period and this is computed in
function 20 where SL.sub.t is equal to the distance taken during
the transition period and is found by computing the function
(V.sup.2 - V.sub.f.sup.2)/2A.sub.ave. Since this retarder length
must be available for the transition to take place, it is retarder
length which cannot be utilized in any particular braking condition
and therefore must be subtracted from the available retarder length
computed in function 16. Therefore, function 21 updates a new
retarder distance left (SL) by subtracting from the previous value
the just-computed SL.sub.t.
Functions 22 and 23 initialize the program for the computations to
be performed in OVER. In the simplified description, and
accompanying drawing, FIG. 1, it was explained that a linear
relationship between car velocity and distance through the retarder
could not be obtained with a single value of acceleration;
nevertheless, this is the assumed profile which is a starting point
for the computations. It should be understood that other profiles
could be assumed and used in this program. A particular constraint
on the selection of an assumed profile is the time necessary to
compute the parameters of the profile. The time available for these
computations is limited by the time the cut takes to travel from
the wheel detectors to the retarder. It is now necessary to have
available the characteristics of this assumed profile. They are the
velocity intercept, V.sub.en', which is computed by the program
V.sub.en, to be discussed later, V.sub.ex, the exit velocity, the
computation of which has been discussed above, and the slope of the
line which is simply V.sub.en' minus V.sub.ex divided by S.sub.max
where S.sub.max equals the maximum retarder length. Upon completing
the computation of these parameters, the program proceeds to OVER
which is shown in FIG. 2D.
The first function in OVER is to compute a trial point on the slope
of the line, which is accomplished by function 24. The computation
is performed using the equation Y = (AL) X+ BL where AL is the
slope previously determined and BL is equal to V.sub.en' and X
represents distance through the retarder. On the first pass through
this routine, X has been set to zero by function 22. The
computation then will result in y= V.sub.en' since V.sub.en' equals
BL.
The next function is to compute a distance S which is that required
at the assumed braking to decrease the car's velocity from Y to V,
and this is performed in function 25. The computation of S is
determined by the equation (Y.sup.2 - V.sup.2) /2A.
The next function, 26, is to compute the estimated distance left
from the formula SL-X and since on this first pass, X equals 0,
then the estimated distance left would merely be SL. Decision point
27 determines if S is less then SL-X and, assuming for purposes of
discussion, it is not, then the program proceeds to the decision
point 37 which determines whether or not this is a repass. In
accordance with our example, this is not a repass and therefore
function 38 would set X to a new value, higher than zero.
Subsequently, decision point 39 determines if X is equal to SL and,
further assuming that it is not, proceeds to repeat OVER. What has
been accomplished in this loop is a determination of whether or not
light braking will suffice from the initial velocity, V.sub.en' to
reduce the car velocity to the required velocity V in a distance
which is equal to or less than the available distance, SL. Since we
have postulated, by the result of decision point 27 that S is not
smaller than SL, then we have incremented X. What we are doing is
proceeding along the line shown in FIG. 1, the assumed profile, and
attempting to find an intermediate point in the retarder from which
light braking will be sufficient to achieve the desired velocity V.
On the next pass through, OVER function 24 will compute Y with the
new value of X and this will be somewhat less than the previous
value of V.sub.en'. As before, a new distance is computed and the
estimated distance left SL-X is again computed and this value is
compared to the necessary distance.
The incremental value by which X is changed during the performance
of each loop depends upon a number of considerations. Clearly, the
smaller the incremental value, the more accurate the computation
will be, and also the computation will take a longer amount of time
to proceed down the line from V.sub.en' to V.sub.ex. Since only a
predetermined amount of time is available for this routine to be
accomplished, the increments added to X must be balanced between
these considerations. In one embodiment of the present invention,
the increment of X used has been one foot. This computation then
proceeds in accordance with the simplified description to determine
whether or not the assumed light braking will be sufficient from
some intermediate point of the retarder to achieve the desired
velocity V. Assuming that it is, then at some point in this looping
process between functions 24 and 39, decision point 27 will
determine that S is less than SL-X and proceed to perform the
computation to determine T (POS) by function 28. The formula used
for this computation is (Y-V)/A. Function 29 determines the minimum
allowable time in the assumed braking position. This again merely
requires reference to a predetermined table. Decision point 30
determines whether this time is sufficient, i.e., whether or not it
is longer than the minimum allowable time. If it is, then that time
in position is stored in the table and a new distance left is
determined by subtracting S from SL in function 32.
Function 33 computes the actual transition time as
(V-V.sub.f)/A.sub.ave. Function 34 stores the Y value at which S
was first less than SL-X as a new V.sub.f and then function 35 adds
the transition time to the previously computed time in this
position and distance in this position. Decision point 65
determines if this is a repass. If not, function 66 increases AP to
the next level. Otherwise function 66 is omitted. Decision point 36
determines wheither or not the new V.sub.f is within the tolerance
of V.sub.en ', the modified entering velocity. There is no
assurance that any particular V.sub.f will be identical to V.sub.en
' and therefore if V.sub.f is within reasonable tolerance of
V.sub.en ' the computation is considered finished. If the computed
V.sub.f is within the tolerance then the program proceeds to
perform DONE whereas if it is not, then the program proceeds to
perform NEXT.
It very well may be, however, that the computation proceeds down
the slope of the line and never determines, at decision point 27,
that S is less than SL-X. Of course, each time this negative result
is reached, decision point 37 checks to determine if this is a
repass. Since we have assumed it is not, function 38 increments X
to a new value. At some point, after a number of iterations of this
routine decision point 39 will determine that X is equal to SL
which will render function 40 operative to replace the transition
distance previously removed at function 21. Thus, the new SL will
be the old SL with SL.sub.t added back in. The same result is
reached if at any time decision point 30 determines that the actual
time in position, computed at function 28, is below the minimum
time determined at function 29. In either event, the program has
determined that the particulur brake setting is not going to be
used and therefore the previous transition distance which had been
subtracted at function 21 must be added back in and recomputed
using a new value of current acceleration.
The program portion NEXT, shown in FIG. 2C, can be entered for one
of two reasons. Either, during a trial computation for a particular
brake setting, it is determined that that particular brake setting
will not be used; then, after completing function 40 (FIG. 2D) NEXT
is entered. On the other hand, if a particular brake setting
computation is completed and decision point 36 (FIG. 2D) determines
that V.sub.f is not within the tolerance of V.sub.en ', then NEXT
will also be entered. In either case, decision point 41 determines
whether or not this is a repass. Since, in the example under
discussion, the repass flag has not been set, we will assume that
it is not a repass and function 42 sets the acceleration level to
the next value.
Decision point 43 determines whether or not this value of
acceleration is allowable for the particular cut now within the
retarder. Assuming that it is, the routine loops back to start and
computes new transition values. If NEXT had been entered when
function 40 determined that a particular brake setting would not be
used, then it will be apparent that the current setting A is two
levels above setting AP. Since the program, prior to function 40,
had eliminated one of the braking levels, the transition now is
between braking levels which are separated by an intermediate
braking level. On the other hand, if the program portion NEXT had
been entered subsequent to decision point 36, function 66, just
prior thereto, would have increased AP another level so that when
the transition computation, in START, is accomplished, the levels
of A and AP would be adjacent.
The second and subsequent passes through START and OVER would
perform computations very similar to those already explained, with
increasingly higher braking levels. This would accomplish the
functions explained with respect to the simplified description of
FIG. 1. That is, for the next higher braking level, the computation
would begin with V.sub.en ' and determine if the distance left (SL)
was sufficient to decrease the velocity of the car or cut to
V.sub.f, as set by function 34. If it was not, then some further
intermediate velocity along the theoretical velocity profile would
be chosen, by incrementing X, and the computation would again
proceed until the remaining distance (SL-X) had been reduced to
zero unless a successful solution is found.
If no successful solution is found, or if the successful solution
indicates a braking time which is below the minimum allowable, then
this braking level will be deleted from the retarder control orders
and NEXT would again be performed to increase the braking effort
further.
In this fashion, the computation proceeds until V.sub.f becomes
equal to V.sub.en ', or within the tolerance range of V.sub.en '
and the program is directed to DONE. On the other hand, if V.sub.f
does not approach V.sub.en ' by the allowable tolerance, then at
some point in the passes through NEXT, decision point 43 will
determine that the acceleration is too great for the car or cut and
function 44 will set V.sub.en ' equal to Y, function 45 will set
the repass flag, function 46 will set A back to one lower level
(thus cancelling out the increment added in at function 42).
Decision point 47 will determine whether or not this braking level
corresponds to OPEN, and if not, function 48 will set AP equal to A
and function 49 will decrement AP one level.
The only reason the program would progress beyond decision point 43
in NEXT, is that the computations performed have indicated that
with allowable braking forces, the velocity of the car has not yet
been reduced from its V.sub.en ' to the V.sub.ex in the allowable
distance. Assuming that the AP set by function 49 is not open, as
checked by decision point 50, then the program will look for a
previous braking position which used a lower than maximum allowable
braking force, and eliminate it and substitute the highest
allowable braking force for the computation. If decision point 52
determines that the particular braking position was not used, the
program loops back to function 49 to decrement the AP setting until
decision point 52 determines that it has discovered a braking
position that has been utilized. Functions 53 and 54 then revert to
the previous distance left (SL) and previous time and distance
totals existing prior to this braking. Function 55 clears the table
of the stored values, thus eliminating all traces of this braking
position and then directs the program to START to perform a new
computation at a higher braking level.
Since this is a repass, the program portion OVER will only be
performed once, and if decision point 27 determines the S is not
smaller than SL-X, then the program will again be directed to
function 49 to decrement AP to a still lower value and go through
the same routine. In this fashion, previously computed lighter
braking levels are deleted in turn to allow the highest braking
effort to be exerted over longer and longer distances and attempts
to find a solution which will reduce the entering velocity
(V.sub.en ') to the required exit velocity (V.sub.ex) in some
allowable distance. Of course, if the problem is satisfactorily
solved, the computed values are stored and the program is directed
to DONE.
If, at some point in traversing NEXT, either decision point 47 or
50 determines that either the acceleration level is in effect open
or the previous acceleration level under check is open, then the
program is directed to set the distance in the current acceleration
position to the maximum length of the retarder, at function 56, and
the time in position is set to the maximum at function 57 and the
program is concluded. In effect, decision point 43 has determined
that the current acceleration is the highest allowable for the car
and, either this is open or, all previous acceleration levels that
had been selected for use have now been eliminated and therefore
there is only one allowable acceleration force and this will be
used to the maximum length of the retarder.
The program portion DONE is entered when the computations arrive at
an entering velocity which is equal to V.sub.en ' or within an
allowable tolerance thereof. This particular portion is shown in
FIG. 2D, and the first function, 58, adds all the calculated
distances that the car or cut will travel with the retarder in a
particular braking position. Function 59 determines how much
retarder length is left over and function 60 sets the retarder to
the open position for this distance. Function 61 adds the
previously computed open distance to that added in by function 60
and function 62 computes the time period during which the retarder
should be open. This can be computed, knowing the total distance in
the open position and dividing that by the desired exit velocity.
Function 63 stores this time as the time during which the retarder
will remain open and adds it by function 64 to the previous total
time the retarder had been set to open. That concludes the
program.
The conclusion of the program leaves in a table, a series of times
and distances for the different retarder braking positions. The
retarder then can be controlled from this table, either on a time
or distance basis to change from one position to another. That is,
on a time basis, a clock is started when the cut first enters the
retarder and, after the time elapsed reaches the value set in the
table for the retarder at that braking position, the retarder is
controlled to the next braking level which has a time value in it
and the clock is again atarted. In a like manner, the retarder is
controlled to each of its positions for the predetermined period of
time computed by the program for that position. It should be
obvious that if distance sensing information is available, the
retarder can alternatively be controlled on the basis of the
computed table, as the cut proceeds through the retarder.
In the program disclosed in FIG. 2, a velocity V.sub.en ' is
utilized for computations. The program V.sub.en, shown in FIG. 3,
is utilized to compute V.sub.en '. The program V.sub.en also
computes the largest braking force that can be utilized as the car
or cut enters the retarder in line with the other constraints
placed upon the system.
The first function of V.sub.en, 67, is to compute the actual
entrance velocity of the car or cut into the retarder. As shown in
FIG. 4, a series of wheel detectors determine the time it takes for
the first truck of the first car to traverse a known distance, and
from well-known principles, the velocity can thus be
determined.
The next function, 68, sets S, a distance, equal to the distance
between the first and second truck of the car, or this distance
with respect to the first car of a cut. Function 69 sets the
acceleration to the highest possible braking force and function 70
computes a trial value for V.sub.en '. The expression relating
V.sub.en ' to V.sub.i (the initial velocity of the car entering the
retarder) is V.sub.en.sup.2 = V.sub.i.sup.2 - 2AS. Decision point
71 determines that this trial value of V.sub.en ' is acceptable by
determining whether it is greater than the exit velocity plus a
factor for the car's deceleration through the retarder; that is,
decision point 71 determines if V.sub.en '> V.sub.ex + .DELTA.
(where .DELTA. is the minimum velocity loss through the retarder).
If it is, then this is an acceptable entry condition and the
program skips to function 75 to compute the time and distance in
this braking position which is then stored by function 76 in the
tables for controlling the retarder and that will complete the
program. Assuming, however, that the V.sub.en ' is not large
enough, that it is not greater than V.sub.ex by a sufficient
margin, then function 72 sets the braking force to a lower level
and function 73 determines if there is such a lower level. If there
is, then the program recomputes a new V.sub.en ' using the new
braking force and performs the same check by function 71. If this
new V.sub.en ' is acceptable, then the program proceeds as
discussed above. If it is not, then function 72 sets a new braking
level and the program proceeds as discussed above. If it is not,
then function 72 sets a new braking level and the program proceeds
as discussed above. At some point, if an acceptable V.sub.en ' is
not found, decision point 73 determines that there are no further
braking positions available and directs the program to function 74
where the retarder is set to OPEN.
In this fashion, the highest allowable braking level is used for
the time it takes the first car to completely enter the retarder.
The calculated values are stored in the table to control the
retarder and the computed V.sub.en ', the velocity of the car when
the second truck enters the retarder is made available to the
program of FIG. 2 for its computations.
As is well known the braking effect, the acceleration imparted to a
cut varies with the number of axles actually in the retarder. The
program of FIG. 2 is effective only when there are at least two
trucks (at least four axles) in the retarder. Prior to that the
braking effort determined by V.sub.en (shown in FIG. 3) is
effective. Although there will be some variation in braking effort
on the cut when the number of axles in the retarder exceeds four it
has been found unnecessary to take this variation into account.
Balancing the increased precision obtainable against the additional
complexity and time consumed by execution adequate results are
obtained by ignoring variations in braking effort.
* * * * *