Density Lock

Copyright ©2015 by Paul Niquette. All rights reserved.
   

Traffic IlluminationSophisticated solvers will want take into consideration all three variables in the fundamental equation, speed = flux / density.  Increasing separation of stopped vehicles reduces traffic density -- of stopped vehicles only. 

The density and therefore the speed of vehicles traveling in steady-state will not be affected.  Ah, but freed from density lock, vehicles are able to accelerate faster from a standing stop and reach steady-state speed sooner.
  As for the capacity of the traffic system, the instantaneous flux is zero for stopped vehicles at any density, but quicker restoration to non-zero flux would seem to be an advantage of increased separation of stopped vehicles; however... 
Increasing the separation of stopped vehicles means that, while stopped, they are farther away from their ultimate destinations.  That reality flies in the face of elementary -- not to say primitive -- psychology of drivers and passengers.  Back in 1955, objections were loud. 
The idea for increasing stopped separation followed from UCLA's proposal for the earliest ramp metering. People would scoff, "How do you expect to speed up a system by deliberately slowing it down."  Similarly, responses to the proposal for mandating separation took the form, "Reducing density means each vehicle takes up more 'real estate' in its lane."  Which is true enough, but certainly no more 'real estate' than when the vehicle is happily traveling in steady-state at safe following distance.  Here is the question in the puzzle...

Would mandating that vehicles stop at
prescribed separation distances prevent
density lock
?
Solution: Yes, comma, but...

Let us explore some implications of our thought-experiment, using the parameters in the table below.  We see that in steady-state speed = 35 mph, flux = 3,600 / tH = 2,045 veh/hr, density = 5,280 / xH = 59 veh/mi.

Description

Symbol

Value

Speed Limit

vL

35 mph, 51 fps

Headway Time

tH

1.76 sec

Headway Distance

xH

90 ft

Reaction Time

tR

0.75 sec

Reaction Distance

xR

38 ft

Safe Following Time

tF

1.5 sec

Safe Following Distance

xF

77 ft

Separation Distance

xS

2 ft versus xF/2 @ vL

Vehicle Length

xL

13 ft

Acceleration for N = 0

a0

0.5 g, 16.1 fpsps

Deceleration for N = 0

b0

1.0 g, 32.2 fpsps


The base case has xS = 2 ft  and the experimental case has xS = xF / 2 = 39 ft for vL = 35 mph.  We can readily calculate the densities for the stopped separations as 352 veh/mi and 102 veh/mi respectively.  In the base case, vehicles are density locked, inasmuch as there is insufficient space for immediate acceleration.
Routine stops are not common on expressways, freeways, and throughways.  Today's highways do suffer plenty of congestion, however, as lane capacities are routinely exceeded by peak traffic demands.  Any sudden interference in flow can cause a traffic jam on the highway, and the resulting density lock can linger long after the cause of the jam has been cleared up.  This subject will be addressed in the Density Wave puzzle.
Traffic
                        IlluminationNot much can be done about jams -- not by individual drivers, anyway.  The main concern is that braking of each vehicle must be adequate throughout the whole platoon.  The postulation page of the Density Lock puzzle, confirmed that requisite braking bN < b(N-1).  Once those vehicles are stopped, though, the flux for that lane is zero.  Best anybody can hope for is to restore a non-zero traffic flux as rapidly as possible when the lane ahead opens up. 

Here we see graphs of the accelerations and recovery times using
aN = a0 vL2 / [vL2 +2 N (xF - xS) a0].


The recovery of traffic flux is paced by the last vehicle in a platoon.  Thus in the experimental case, vehicles are stopped with separations of 39 feet, and a platoon of 11 vehicles will recover to a flux of 2,045 veh/hr about 14 seconds sooner than the base case with separations of only 2 feet -- that's about 44% quicker.  By the way, a platoon of any number of vehicles will recover about 44% quicker.
As for comparing the overall speed for individual vehicles between the two cases, skeptics might ask, "Since prevention of density lock in the experimental case forces each vehicle to start up from 37 feet farther back, won't that increase time-to-destination?" 
Not so much.  Sophisticated solvers will merely multiply the net recovery time for a given vehicle by the steady-state speed, which represents the extra distance traveled at 35 mph and compare that to the cumulative separation distance for that vehicle.  Those two distances are exactly the same for every vehicle in the platoon.  By the way, that equivalence will apply to any separation between stopped vehicles.  Hmm.
Skeptics will also point out that freedom from density lock requires the stopped platoon with 10 vehicles to stretch back from the stop line 390 feet, about the length of a city block.
So what?  About 32 seconds after the light turns green, all 10 vehicles will have accelerated to 35 mph.  During that time they will spread out along the traffic lane to 900 ft or two tenths of a mile.  That happens to be twice the 'real estate' when stopped and exactly the same as if there had been no red signal at all.


Creep-to-the-Stop-Line

Traffic
                        IlluminationHere is an alternative modeling concept for the base case as mentioned in Density Lock puzzle.  Instead of a constant accelerations aN, each vehicle other than N=0, is assumed to creep forward as allowed by the instantaneous location of of vehicle N-1 ahead.

 Creep-to-Intersection
The assumption in the graph is that, after the traffic light changes from red to green, each vehicle is able to progress forward to the intersection stop-line at some average 'creep speed' vC.  At a density of about 220 veh/mi (24 ft/veh) and a flux of 1,113 veh/hr (3.2 sec/veh),  vC = 5 mi/hr (7.3 ft/sec).  Keeping xL = 13 ft, the following-distance-while-creeping averages xF = 11 ft.
Using the constant-acceleration results for the base case, which show a recovery time for the 11-vehicle platoon of 32 seconds, we can estimate the result for a creep-to-the-stop-line by calculating the creep-time tC for the the n=10 vehicle as 10 x 24 / 7.3 = 33 sec.  Assuming an acceleration a10 = a0 = 16.1 fpsps, the recovery time at 35 mph = 33 + 51/16.1 = 36 sec, which is only a few seconds slower than the results from the constant acceleration model.
One is tempted to conclude that the advantage in the experimental case has been slightly strengthened by this calculation; however...

 Escape-Time

Traffic
                              IlluminationSolvers of the Density Lock puzzle have identified a villain in congested flow -- density.  High density means low speed; low density means low flux, the former relevant to traffic-system users, the latter to traffic-system managers.  Both speed and flux put time in the denominator.
As we have discovered in several puzzles, it is generally a good idea to put time in the numerator.  Recovery time does that here in the Density Lock puzzle.  It does something else, too.  We have been made aware of a potentially relevant criterion for judging the results.
So far, 'recovery-time', has been taken as the interval that begins with the instant the traffic signal changes from red to green and concludes when the last vehicle in a platoon has accelerated to the speed limit.  Limited by density lock, the base case vehicles take more time than the experimental case vehicles, which are starting up from separated locations.  Though they have been stopped at greater distances from their respective destinations, we have shown that when the light changes from red to green, the vehicles in experimental case do not suffer a time penalty in reaching those destinations; however...   
Many of the 'recoveries' lie beyond the intersection, where any number of unanticipated interferences might delay recovery.  One can argue that the system being modeled is therefore incomplete -- that all relevant factors for judging the efficacy of mandating stopped separations are situated before the intersection is reached.  That argument leads to a different criterion...
The escape-time criterion simple measures the time for each vehicle to reach the stop line at the intersection. It regards the transportation system as being 'responsible' for managing the movement of vehicles only in their respective approaches to an intersection not beyond.  We can easily calculate the time required for the last vehicle in a platoon to reach the stop line at the intersection -- the escape-time tE
Base Case: At the instant the light changes from red to green, vehicle N=10 is stopped at 10 x 15 = 150 ft from the stop line.  At a10 = 0.6 g (19.3 fpsps), it accelerates to vL = 35 mph (51 fps) in 51/19.3 = 2.6 sec and traveled 512/2(19.3) = 67 ft, leaving 150-67=83 ft to travel at vL to reach the stop line, for an escape time tE = 2.6 + 83/51 = 4.2 sec.

Experimental Case
At the instant the light changes from red to green, vehicle N=10 is stopped at 10 x 51 = 510 ft from the stop line.  At a10 = 0.9 g (29.0 fpsps), it accelerates to vL = 35 mph (51 fps) in 51/29.0 = 1.8 sec and traveled 512/2(29.0) = 45 ft, leaving 510-45=465 ft to travel at vL to reach the stop line, for an escape time tE = 1.8 + 465/51 = 10.9 sec.
Thus, if the criterion for judging recovery-time is escape-time, the model shows that the proposal fails; however...
Base Case with Creep-to-the-Stop Line: At the instant the light changes from red to green, vehicle N=10 is stopped at 10 x 15 = 150 ft from the stop line. The model has been adjusted to accommodate creeping at vC = 5 mph (7.3 fps).  The 150 ft requires tE = 20.5 sec.
...which indicates that creeping-to-the-stop line is not the best way to overcoming density lock. Your puzzle-master, who now lives in France, has observed that French drivers spontaneously stop with xS > 2 ft in approaching intersections -- perhaps as much as xS = 2.0 meters (6.5 ft), which allows for a faster creep speed vC > 5 mph (8 km/hr) -- possibly as much as 10 mph (16 km/hr).  Thus, we might consider...
French Case with Creep-to-the-Stop Line: At the instant the light changes from red to green, vehicle N=10 is stopped at 10 x 18 = 195 ft from the stop line. The model can be adjusted to accommodate creeping at vC = 10 mph (15 fps).  The 195 ft requires tE = 13.3 sec, which agrees with the Experimental Case within a couple of seconds.  Exclamation point anyone?


Over a period of 20 years, France has been appropriating and applying one traffic management technology which implements an elegant stratagem for eliminating density lock.  Sophisticated solvers will learn about why that technology is so successful while solving...

 
rainbow line

Home Page
Puzzle Page
Math and Models
The Puzzle as a Literary Genre

 

Epilog
Traffic
                                        Illumination The founding dean of the College of Engineering at UCLA, L.M.K.Boelter, was the 1957 recipient of the ASME Medal, the highest award of ASME for "eminently distinguished engineering achievement." 
The recipient of the 1955 award from Dean Boelter has personally observed an astounding history of technologies.  Most are considered common-place in 2015: Watching a demonstration of something called a transistor in 1955, driving on one of the first Interstate Highways in 1956, first ride on a jet airliner in 1960, placing the first call using direct distance dialing in the 1960s, using zip codes in 1963, seeing miniature French traffic lights in 1971, and -- oh, right, consulting on microprocessor-controlled traffic lights in the 1980s.