top of page

The mass transit model

Atlanta Map Autonomous Cars Traffic Symmetry

Above: Model of Downtown Atlanta with Connected and Fully Computer-Controlled Traffic

Green: Parks and Pedestrian-Only Greenways----mostly made of existing streets, or existing railroads

Thick Pink Line: Main Route to/from the Freeway----the most prioritized route in this part of downtown

Narrow Pink Lines: Tributary Routes----the second most prioritized routes

Thick Black Lines: Ordinary Local Traffic Routes----default level of priority

Narrow Black Lines: Entrances/Exits to Parking Structures

Gray Boxes: Buildings (Some Existing, Some Proposed) with Parking

Yellow Boxes: Buildings (Some Existing, Some Proposed) without Parking

Pink Boxes: Autonomous Minibus Stations/Hubs----the vast majority of suburban commuters can use freeway--based mass transit in this model


In 20 years, partly-automated and connected vehicles could be ubiquitous enough for wireless technology to control all vehicles on strategically important roadways and freeways. Cities can already begin using mathematical standards to map out how to re-imagine themselves with automated traffic plans. Fast-growing metropolises can still to avoid the wrong development patterns, new bottlenecks, threats and possibilities presented by automated traffic.

This Mass Transit Model is based on math and guided by the the design standards and priorities laid out by Ryan Gravel and Time Keane in Atlanta City Design: Aspiring to the Beloved Community.

Single-Purpose Freeway Exits


With an automated traffic system, all exits in a downtown area could, overtime, be replaced with new exits which process arrivals only from the freeway or departures only into the freeway. Old exits could be replaced by new exits which are designed to process traffic from all directions in the same proportion (ie: 30% northbound, 30% southbound, 20% eastbound, 20% westbound ). Consistent with this, point A in the map highlights a proposed freeway exit next to Grady Hospital, which is only for vehicles exiting from the freeway into the city. This one--way exit powers a circuit--like route which eventually feeds into an exit which processes only departures into the freeway. This model's exit is designed to process equal amounts of traffic from northbound and southbound (although a more complex alternative could process more traffic from its south-side to include commuting vehicles from I-20).

The Prioritization Algorithm


In this model, system-wide traffic is controlled by what I call a prioritization algorithm. A prioritization algorithm is made up of the rules and preferences for which streets and routes are prioritized and which vehicles are prioritized (autonomous mass transit could be prioritized over single-rider vehicles, for instance). Among many other things, this model's prioritization algorithm targets the capacity of arrival-only exits on Atlanta's downtown connector freeway to behave with a mean of no more than half of route capacity during peak demand times. This allows momentary surges using up to 100% of the exit’s capacity while still using variable pricing (potentially) or targeted congestion to limit the route’s volume to the preferences written into the prioritization algorithm. This can be done to target traffic flow on a route to be advantageously high, but crossable and tolerable for pedestrians.

A Most Prioritized Freeway


A recent report from the Georgia Department of Transportation shows roughly one-third of the rush-hour vehicles on Atlanta's downtown connector freeway are actually bypassing downtown. My model's prioritization algorithm directs traffic elsewhere, like Atlanta's perimeter freeway, and could incentivize riders to use freeway-based mass transit (autonomous mini-buses) to reach their destinations in the core of the city.

If a worker needed to commute from Jonesboro to Buckhead, their fastest option could be a ride on a mini-bus from Jonesboro to Atlanta's downtown minibus hubs (below in blue) to connect to a minibus route destined to Buckhead mini-bus route----a trip which would take much less time than it does today, despite needing to connect in downtown. My Mass Transit Model below assumes Atlanta's central business district evolves to become dependent on minibuses over time, allowing Atlanta to wield one of the busiest central business districts on earth, which, along with a handful of other cities, would dwarf central business districts as we know them. The blue mini-bus hubs in the graphic below are spaced a half-mile from each other so commuters won't need to walk more than five minutes from their station to their downtown destination. Every yellow mini-bus hub has a unique route to every blue bus hub, as well as routes connecting yellow bus hubs and other regional business districts further away from central Atlanta.

The below Metro-Atlanta Model uses 60 blue minibus hubs (including alternatives) which each have 60 unique routes to the 60 yellow minibus hubs (not all shown), creating a matrix of 3600 minibus routes with all routes being serviced every 5 minutes----using 61% of the downtown connector freeway's daily capacity for automated vehicles (43,200 slots of 70,400 per peak-traffic hour ---- assuming half-second reaction times,15-foot vehicle slots, and 12 active freeway lanes).

autonomous mini-bus system map

Screen Shot 2019-09-23 at 1.09.31 PM.png

Mini-buses could be driverless, cheap, & Effective

The most primitive automated traffic standards alone could lead to a doubling of freeway flow per lane. Combining the savings of rerouting downtown connector traffic bypassing central Atlanta with automated freeways needing less space between vehicles (doubling freeway capacity), could even mean the most primitive automated traffic standards could allow for nearly three times as many vehicles to use the downtown connector freeway into the core of Atlanta.


If reaction times drop to a half second and vehicles become more compact, freeway traffic flow capacity could double againAt six times the current flow capacity, this model divides the downtown connector freeway's flow equitably to its east and west sides. This model assumes the most outlying exits (connecting the blue minibus stations) in a central business district would not be further apart than five minutes in order to achieve a largely consistent proportion of commuter origins across a city's downtown, which I base on my observations of existing central business districts.


This model assumes the downtown connector freeway is re-aligned slightly in parts to allow for comfortable 90-mile-per-hour travel without causing curve acceleration to exceed one meter per second squared, in line with current standards for mass transit. These mini-bus hubs each anchor a part of town, typically a quarter of a squared mile (to prevent walking more than 5 minutes from the mini-bus hub). For example, one bus hub anchors Midtown Atlanta's Arts District, another anchors Tech Square, and 3 are found along the current periphery of Georgia Tech's main campus.

autonomous mini-bus system hub map

Screen Shot 2019-09-23 at 1.11.38 PM.png

(The Math for an Automated Freeway's Capacity)


A freeway having connected traffic means its traffic could be designed to always go at top-speeds and at optimal efficiency. I will start with the capacity of a single lane of freeway traffic. It should be carrying a constant capacity per lane at a constant rate of:

((distance traveled per second)/(distance for vehicles to maintain their standard reaction time + standard vehicle length)) = vehicles processed per second


If the downtown connector freeway were left as it currently is, and if the system is to protect passengers from experiencing speed changing at more than 2 meters per second squared (the common industry standard), then the current layout of the Grady Curve alone would be unlikely to allow speeds of more than 45 miles per hour (66 feet per second). The industry standard reaction time for autonomous vehicles is 1 second. The standard vehicle length could vary slightly, but here, let’s assume it is 20 feet. Based on these values, the capacity of a top-priority freeway lane would be:


          ((66 feet)/(66 feet + 20 feet)) = vehicles per second


          ((66)/(86)) = vehicles per second


          0.767 vehicles per second


If Atlanta’s downtown connector were to be slightly reconfigured for passengers to not experience momentum shifts greater than 1 meter per second squared, Atlanta’s downtown connector freeway could carry vehicles up to 90 miles per hour (132 feet per second). Atlanta’s reformed downtown connector freeway would carry a per-lane capacity of:


          ((132 feet)/(132 feet second + 20 feet)) = vehicles per second


          ((132)/(152)) = vehicles per second


          .868 = vehicles per second


This example of a slightly reconfigured downtown connector freeway yields approximately a 13% increase in potential traffic flow and 33% reduction in trip time compared to its current set-up which allows only for 60 mile-per-hour travel in at least 3 sections.

(The Math for Freeway Ramps)


Exit freeway ramps need to be long enough for automated traffic to slow down at a rate which does not exceed the maximum allowed acceleration, which for the sake of this model, is 2 meters per second squared (1 meter per second squared is the industry standard for mass transit and could be the gold standard for cities wanting to ensure a consistently smooth ride). The following equation stripped of integrals (for simplification) but lays out the logic of how long this exit ramp needs to be if slowing down 2 meters per second squared, or 6.56168 feet per second squared:

          (110) feet + (110 – (2*(6.56168))) feet + (110 – (4*(6.56168))) feet

          + (110 – (6*(6.56168))) feet + (110 – (8*(6.56168))) feet

          + (110 – (10*(6.56168))) feet + (110 – (12*(6.56168))) feet

          + (110 – (14*(6.56168))) feet + (110 – (16*(6.56168))) feet


Longer ramps would ensure more comfortable acceleration and deceleration, while giving the system more space for handling momentary surges to a specific off-loading exit.

The Atlanta model - route 1

Screen Shot 2019-06-04 at 3.11.31 PM.png

Merge, Merge, Merge! (point A)

Here (Point B on the map), the exit flows from I-75/85 North and I-75/85 South turn to the West and merge with each other. All routes could have an expanded right-of-way at their beginnings and ends to facilitate merging, but how would vehicles most efficiently merge at the start of a route?

For this route, traffic could move as a wave of platoons on four lanes, using the right two lanes in concert with each other, and the left two lanes in concert with each other. I will demonstrate this later. For now, traffic needs to organize such that each section could be timed exactly right for chunks of its left-two or right-two lanes to turn off to another street in tandem. I can also account for given lanes with different allowed speeds. If I assume vehicles are turning on from a street as they are turning off—these new vehicles can add to a platoon as other cars turn off in parallel pairs. If these flows do not move simultaneously when it is possible to do so, it could make it more difficult for pedestrians to cross at intersections between the primary route (which carries the wave of platoons) and its distributor routes. Traffic on primary routes like this could have sufficient space in between intersections with distributor streets and streets carrying local traffic flows and sufficient gaps between pedestrian-only routes to reshape themselves from each intersection with their distributor streets.

Here’s how I break this down:

The Math of Merging:

If vehicles are traveling 16 feet per second (as explained below) as the base speed and are capable of shifting speed at 6.56168 feet per second squared, and the space needed per vehicle is 20 feet for the standard vehicle length and another 16 feet for the speed per second, and lanes are no more than 9 feet, then the amount of time it takes 2 lanes at half capacity to merge and space out can be determined by:


(Again, stripped of integrals):


          16 feet (increasing towards (16 + (2*6.56168 feet))) = Second One

          (16 + (6.56168*2)) feet =Second Two

          (16 + (6.56168*2)) feet =Second Three

          (16 + (6.56168*2)) feet (increasing towards 16 feet) = Second Four


To create open space of 32 feet, it takes approximately five seconds + error


The time and space needed for merging:


          Increasing towards (6.56168*2) = Second One

          (6.56168*2) decreasing towards 0 = Second Two


To merge vehicles in lanes with 9 feet of width, it takes less than 2 seconds + error


If a route has 2 lanes of targeted traffic capacity spread over 4 lanes (its maximum capacity), and 8 lanes at the beginning of its route, then half of the lanes can be for faster moving traffic moving to the front of a platoon as it needs to, merging into the 4 slower lanes of traffic which were not sped up in order to fall into place.


Platoons and Waves (Point B)

With this process for merging, the larger a platoon----the more space needed for vehicles to initially fall into formation. These bottlenecks could be substantially mitigated by creating smaller platoons which form into larger waves and slow down or stop briefly to allow for a larger wave of platoons to build. Because the system may not deliver vehicles to this exit so precisely to create adequate symmetry in every small platoon, this strategy would result in some vacant spaces in a wave. The larger a wave, however, the more time in between waves—and the easier it could be for pedestrians to cross the street—not to mention the intersecting streets carrying local car traffic.


Here is all of that again, but described with math:


Platoons on this route may form into 4 by 4 formations. This would require:


The time and space needed for merging 4 lanes of traffic into 8, I need spacing for 2 new lanes in between the existing, and 2 new lanes which form to the outside. Vehicles may need to arrive on any of these lanes, so  need 4 new lanes:


          (4 lanes) * (9-foot width) = 36 feet:


          Increasing towards (6.56168*2) = Second One

          (6.56168*2) = Second Two

          (6.56168*2) = Second Three

          (6.56168*2) decreasing towards 0 = Second Four


To merge vehicles across as much as 36 feet, it takes approximately 4 seconds + error

At 16 feet per second, this requires 64 feet of distance + error


Now for the spacing a platoon, which is 4 deep, the difference in potential placement is:


          (4 vehicles) * (20 feet standard slot size + 16 feet for reaction time)

          144 feet = length of platoon


          16 feet (increasing towards (16 + (2*6.56168 feet))) = Second One

          (16 + (6.56168*2)) feet =Second Two

          (16 + (6.56168*2)) feet =Second Three

          (16 + (6.56168*2)) feet =Second Four

          (16 + (6.56168*2)) feet =Second Five

          (16 + (6.56168*2)) feet =Second Six

          (16 + (6.56168*2)) feet =Second Seven

          (16 + (6.56168*2)) feet =Second Eight

          (16 + (6.56168*2)) feet =Second Nine

          (16 + (6.56168*2)) feet =Second Ten

          (16 + (6.56168*2)) feet (increasing towards 16 feet) = Second Eleven


Approximately 11 seconds + error


For vehicles to space out 144 feet, with flow speeds at 16 feet per second and (16 + ((16 + (6.56168*2)) feet per second, it takes approximately 11 seconds + error and 330 feet + error.

When Traffic is finished merging (Point C)

Instead of using connected traffic only to transform Downtown Atlanta, my model uses a connected traffic model to physically expand the central business district beyond its downtown, integrating it with neighbors like Old Fourth Ward, Midtown, and Vine City, while reducing the number of streets needed for cars (not placing them on diets) and increasing the need for mass transit. My intent is to accomplish this without sacrificing the number of vehicles which Downtown Atlanta can receive from its downtown connector freeway.


Regardless of the speed, the capacity of every lane can not be allowed to exceed the speed of the slowest lanes on the entire main route. The slowest speed on this route is for turning at 48-foot radii, which is:


          square root (radius * rate of change) = slowest speed

          square root (48 * 6.56168) = slowest speed

          17.747 feet per second = slowest speed


Now, I need to find the flow capacity of a route with the slowest speed by including the standard length of vehicles:


          ((square root (radius * rate of change))/((radius * rate of change) + standard vehicle length)

          = vehicles flowing per lane, per second

          ((square root (48 feet * 6.56168 feet)) / ((square root (48 feet * 6.56168 feet)) + 20 feet)

          = vehicles flowing per lane, per second

          0.470 vehicles = vehicles flowing per lane, per second


A city street’s traffic capacity is determined by the same fundamental equation as freeway capacity, but now there’s a new caveat: traffic must flow in waves which are broken up to allow for local car traffic and pedestrians to cross the street. In this model, the 50% target for capacity does not only serve to handle momentary surges, but to give time and space in between waves of traffic, even at peak-traffic times.


Described mathematically, this route’s targeted capacity during peak demand should be:


          (vehicles flowing per lane, per second) * (percentage target) * (number of lanes)

          (.470) * (.500) * (4)

          .940 = this route’s average vehicles processed per second during peak demand


          (.940) * (60)

          57.6 = this route’s average vehicles processed per minute during peak demand


If peak demand for this exit lasts from 5 am to 9 pm, then it would process:


          (57.6 vehicles per hour) * (4 hours) * (60 minutes)

          13824 vehicles = this route’s expected number of vehicles processed during peak demand


Assuming a 2-1 ratio between 4-hour peak demand and 4-hour trough demand in a 16-hour traffic cycle, and a 3:2 ratio between 4-peak demand and 8-mid-day demand:


          (cars processed in peak demand) + (cars processed in trough demand) + (cars processed in mid-day demand)

          = expected daily number of vehicles processed by this route


          (13,824) + (13,824 * (1/2) * (4/4)) + (13,824 * (2/3) * (8/4))

          (13,824) + (6,912) + (18,432)

          39,168 = expected daily number of vehicles processed by this route



Compared to the Downtown Connector Freeway


The downtown connector freeway could be slightly re-configured to move traffic at 90 miles per hour, carrying .8976 vehicles per second per lane. If the walls were removed and additional modifications allowed, the downtown connector would be able to rely on the prioritization algorithm shifting the number of active lanes in either direction over the day to adjust to shifting demand, allowing traffic to realistically use the downtown connector freeway at its maximum capacity over the course of the day. This means each freeway would be able to carry:


(vehicles per second per lane) * (16 hours) * (60 minutes) * (60 seconds) = daily flow per freeway lane


(.8976) * (16) * (60) * (60) = daily flow per freeway lane

51706.76 = daily flow per freeway lane


In other words, this route would move fewer vehicles per day than a single lane of the freeway. This represents one of the great challenges for city planners, engineers, and architects in the face of wirelessly-managed traffic: freeways will experience disproportionately bigger improvements in traffic flow compared to their downtown street counterparts.


How does this compare to traffic currently?


12 lanes of freeway carrying 51706.76 vehicles per 16-hour cycle would mean as many as 620,421 using the downtown connector, nearly tripling the freeway traffic arrivals and departures along this congested freeway today, based on GDOT’s most recent analysis. This is without considering redesigns beyond re-alignments which would make better use of I-20.

Currently, this part of downtown is somewhat reliant on commuters using existing mass transit (subways and buses). A 16-hour flow with 39,168 vehicles arriving and departing with this connected traffic model is not enough to satisfy the commuter needs of this part of downtown if people largely arrive in cars. Like the rest of Atlanta's future central business district, I expect the transition towards convenient, clean, and cost-effective mini-buses will be inevitable in this part of downtown.


Stalling Room (Point D)

If a platoon of vehicles is briefly waiting for other platoons to form behind it such that the wave of platoons is targeted to be one minute long, this means I need to calculate the space which will be needed for cars to have stalling room after their initial merging off the freeway. This will require us knowing the flow per second of this route (which I will do later) multiplied by the time length of the wave of platoons. This will determine the maximum number of vehicles which could need stalling room. Then I multiply this by amount of space needed for each stalled vehicle. For this model, I assume the standard space for a stalling vehicle to be the standard length of vehicles (20 feet) plus 4 feet to be in between vehicles. Even for vehicles which are much shorter than 20 feet, it is important they use a uniform amount of space: 24 feet per stalled vehicle.


(vehicles flowing per second per lane * seconds in a minute) * (standing space needed for vehicles)


As will be demonstrated below, the potential vehicular flow per second per lane of this route is .470.



          (.47 vehicles * 60 seconds) * (24 feet) = length of standing space needed for a one-minute flow formation

          676.8 feet = length of standing space needed for a one-minute flow formation


This route specifically has little more than 1,000 feet between the genesis of its merging formation and its first subsidiary intersection. The wave spacing can target 50 seconds, for instance, so the route still has reasonable room to function



subsidiary intersection (point e)


Subsidiary routes, the narrow pink lines, are the routes where all traffic in an area should arrive and leave their destinations. Local traffic routes (thick black) and the freeway-oriented routes (thick pink) accept and deliver traffic to these subsidiary routes. Subsidiary routes must always connect to a local traffic route and a freeway-oriented route. Local routes should not deliver vehicles to their destinations unless their destination is also connected to a subsidiary route. This represents a big departure from city planning and engineering currently: parking structures which cannot connect to a unified subsidiary route will be useless or costly to the efficiency of the system.


How do Subsidiary Routes Function?


All subsidiary routes (slim pink) on this freeway-oriented route (thick pink) are 4 lanes (at least 36 feet) wide. This allows every subsidiary route to accept 2 lanes of turning traffic at the same time it delivers 2 lanes of turning traffic in return.


When there is an excess delivery of traffic from the subsidiary route to the main route, such traffic will move directly before or after a wave to quickly combine with it.


Because subsidiary routes need extra space for vehicles to turn in both directions, the one-way maximum acceptance capacity of a 4-lane subsidiary route from a 4-lane freeway-oriented route would be:


          (flow capacity of freeway-oriented route) * (1/2) * (flow reduction to allow for turning vehicles)

          * (flow capacity needed to move local traffic) = one-way maximum acceptance capacity


The exact answer to this equation is unclear, so I design for subsidiary routes which accept no more than 25% of the main route’s flow. This is conservative enough to avoid problems while giving designers general goals for where to place parking structures.


This means a subsidiary route’s daily maximum intake from the freeway-oriented route would be:


          (expected daily number of vehicles processed by this route) * (.25)


          (39,168 * .25) = each subsidiary route’s daily maximum intake from the freeway-oriented route

          9,792 = subsidiary route’s daily maximum intake from the freeway-oriented route

          9,792 + daily local traffic intake = subsidiary routes daily maximum intake


Solving this equation will require determining the likely daily traffic arrivals, but for now, I know any amount of daily traffic which contributes to this equation will be much smaller than the freeway intake—no more than 2,000 vehicles daily. Therefore, a route’s total daily intake should be expected not to exceed 12,000.


If the standard reaction time for vehicles dropped from one second to, for example, half a second, over time, the same equations for above reveal a 4-lane subsidiary route’s daily maximum intake which could rise to 16,000 a day.

freeway-capping park (point f)


In this model, much of Jessie Hill Jr Drive could no longer be needed for moving cars. This is part of a larger area which could be ideal for conversion into a large park surrounding and extending eastward from Georgia’s Capitol. Over the years, many have proposed parks capping the downtown connector freeway near Memorial Drive. Because this model strips any need for cars to use Capitol Avenue and Martin Luther King Jr Drive (east of Courtland Avenue) this could justify a more ambitious freeway-capping park than most existing proposals.


Given its location on land which was epitomized Atlanta's black ghettos, and given the adjacency to the state capitol, city capitol, the Martin Luther King Jr Center, and Oakland Cemetery, this park could be heavily oriented to Georgia’s history, Atlanta’s history, or a singular part of their collective histories—like civil rights. To do the last one truly well, like the Lincoln Memorial in Washington DC, such a park would need a lot of land to match the gravity of the subject matter.



second subsidiary intersection (point g)


This subsidiary intersection takes and gives from the right two lanes of the freeway-oriented route. It directs traffic to existing parking structures which primarily serve Grady Hospital and some residencies of Georgia State University. Grady Hospital is expanding and the existing parking structures are generally thought of as unattractive and unfriendly to the street.

Over time, the parking structures on this route could be replaced with taller buildings with larger parking structures in their bellies. This subsidiary route is capable of handling much more capacity than its existing parking structures hold, but that does not mean this subsidiary route will achieve its maximum potential, given the limited capacity of the main route and its several subsidiaries.



urban university Park (point h)


Edgewood Avenue and Hurt Park


In my model, Auburn Avenue moves local traffic, while Edgewood Avenue is needed neither for local traffic movement, nor freeway-oriented traffic. Edgewood Avenue could be ideal for conversion into a linear park lined by historic retail and newer mixed-use buildings----an urban walking path in the style of the Beltline. Edgewood Avenue was originally designed as a trolley path, due in part to its relatively flat topography, making it ideal for a walking greenway. An Edgewood Avenue-based greenway also has the potential to connect what is currently Hurt Park and Woodruff Park to form a large rectangle of green space in the center of Georgia State University's campus.


For a university which is trying to re-invent itself from a “concrete jungle” to one with sweeping greenery and connecting parks, this would give Georgia State University an urban greenway to bind its efforts. Students could walk from dorms to class to lunch—all seamlessly through linear parks. The only remaining streets with active traffic in Georgia State University’s downtown campus—in this model—are Peachtree Center Avenue, Decatur Street, Auburn Avenue, and the freeway-oriented route which uses Gilmer Street and the Courtland Avenue Bridge.


Urban University Campus (point I)


If a different route should be chosen for freeway-oriented traffic to move in this part of Atlanta, it would benefit Georgia State University's plans to be sure of where the busiest routes will be. Anyone visiting downtown to see a concert or a game in State Farm Arena, Mercedes Benz Stadium, the Georgia State capitol, the Atlanta city capitol, the massive parks, or tourist attractions around Atlanta’s historic entertainment district—and those coming to have fun in the re-made South Downtown Atlanta, the Gulch, or Castleberry Hill—all of them will see Georgia State University along this route, giving the university an opportunity to make a big impression.

bridge in the street grid (point j)


Courtland Avenue already uses a bridge which passes over Decatur Street, making this section of Courtland Avenue ideal for a freeway-oriented route. This bridge could also make it easier for Georgia State University’s students to avoid its freeway-oriented traffic. If Georgia State’s central quad, currently a concrete courtyard, were redesigned to connect more naturally below Courtland Avenue, as opposed to its current design, which favors connecting by crossing Courtland Avenue Bridge, this part of Georgia State’s downtown campus could feel more like the Beltline’s pass under Freedom Parkway.


Georgia State students do not already use the underpass of the Courtland Avenue Bridge more naturally in part because both sides of its underbelly are used for cars arriving and leaving at its small parking garages. My model calls for the parking garage north of Decatur Street to be eliminated—this parking garage does not naturally connect to other routes in this model and it disturbs the core of Georgia State’s campus. This model calls for eliminating this parking garage and re-situating Georgia State’s quad towards connecting most naturally beneath Courtland Avenue Bridge, relieving students of the area’s most heavily trafficked vehicle route and relieving the traffic of students. My model also calls for creating a larger green quad using land currently dedicated to Hurt Park and Edgewood Avenue to the north.

All rails become trails (point K)


A long-term opportunity to keep in mind—trains, in all likelihood, should eventually fail to be price-competitive with autonomous trucks. While such a change may be three or four decades away, my model keeps options open for a vibrant greenway—Atlanta Beltline in quality—on the existing railway connecting Atlanta to Decatur and beyond.


Georgia State University’s downtown campus is more adjacent to this rail route than any other institution. As trains are phased out for cost-superior automated trucks, Georgia State University can take advantage of this by relentlessly connecting its campus and greenways to what will be a former railway. This is not needed right now, but Georgia State University can give itself room to flex on these opportunities in the future.

What happens when local traffic routes intersect? (point L)

Local traffic intersections could be complicated. If they are to be easily crossed by pedestrians, cars which need to turn from one route to the other need to wait. This forces traffic to stop and wait. This could uniquely choke local traffic routes and require other streets to be activated for cars instead of activated for use by pedestrians only.


This could be resolved if these intersections have sufficiently long turning lanes. Many of these intersections do not have enough existing width for turning lanes, this intersection included. There is additional right of way, however which this intersection could use—an undeveloped block, currently a sidewalk hugging a parking lot. If the edge of this block is preserved from being developed, it would give space for local traffic to operate on fewer streets, opening neighboring streets to being greenways, as my model calls for.

No Turning when local crosses freeway-bound traffic (point M)

In my model, local traffic needs to rarely, if ever, turn directly onto a freeway-oriented route, and a freeway-oriented route is rarely allowed to turn onto a local traffic route. There must be subsidiary routes to make this transition. This makes intersections with local and freeway-oriented traffic routes safe and predictable to cross for pedestrians while one flow of traffic moves—no traffic is turning at these busy intersections. This intersection would be no exception.


When local traffic needs to become freeway-oriented (point N)


In my model, the local traffic moving on Central Avenue moves whenever a wave of one-way freeway-oriented traffic is moving along this section of Martin Luther King Jr Drive. Pedestrians cross at this tempo as well. This is a part of why targeted traffic flow for Martin Luther King Jr Drive is half its potential flow—so that roughly half the time, even in peak traffic, local traffic on Central Avenue can cross, along with Central Avenue’s pedestrian traffic.


Martin Luther King Jr Drive would be the third subsidiary intersection on this route, and the first one which connects in both directions. Pryor Street is wide enough for four lanes, nine feet each, and is a natural candidate for placing a subsidiary traffic route on a four-lane freeway-oriented route.

Underground Walkability (Point O)


Immediately north of Martin Luther King Jr Drive sits a pair of multi-story parking garages straddling Pryor Street. Given the heavy proportion of surrounding buildings which do not have their own parking, these parking garages could eventually be replaced by structures with far more parking, hidden in the bellies of tall buildings complete with street-level retail, further-up functions for office, residential, hotel, or mixed-use concepts.



Parking Structures (Point P)


Parking structures, in this model, represent to any building where vehicles can be parked inside. In my model, parking along the street is not needed for capacity purposes and mostly not allowed in this model----this allows my model to both move enough traffic and convert some existing streets into greenways.


In this model, destinations without parking (yellow boxes) are rarely placed more than 300 feet from a parking structure (gray box). This is the equivalent of not needing to walk more than one typical downtown block to reach a destination from a parking structure for those who still need to arrive downtown in their own connected cars, including people with varying mobility or workers who may need to bring their van for work.


For the Fulton Government County Offices, in particular, because the surrounding area could be in need of a parking structure, and because of the low density of this complex, I recommend eventually tearing down most Fulton County government offices (except the historic nine-story courthouse building) in order to build one (or more) of Atlanta’s tallest buildings, complete with street-level retail and many stories of parking hidden within, overlooking the South Downtown District currently being overhauled by real estate companies like Newport RE US to be a center for celebration and commerce to become Atlanta's best-preserved district. In my model, South Atlanta is a crossroads for pedestrian-only infrastructure. Current roads could be mostly turned into connective greenspace. 

bottom of page