buckminster fuller
5th December 2011, 12:27
"Summary:
Swift PRT Inc. (“Swift”) proposes a new transportation system designed to replace or radically augment the use of cars in urban and dense suburban areas. An ideal transportation system is a combination of speed (200+ km/hr), ubiquitous coverage (stations within a 3-5 minute walk max), on-demand departure (minimize waiting for a vehicle or train), near-noiseless operation, safety, privacy, and cost.
One potential fit is a magnetically levitated system, driven by a linear synchronous motor, and comprised of vehicles that carry only two people. A two-passenger restriction minimizes weight, while fulfilling over 90% of all journey requests – after all, the average car ride in the US is only 1.2 people. Larger groups can be accommodated by virtually coupling two person vehicles such that the group arrives at the final destination simultaneously.
By radically reducing the weight versus trains (100 tonnes) and light-rail (45 tonnes) by a factor of 100, the resulting elevated track structure becomes flyweight: a nearly invisible 20cm tall single track supported on columns no bigger than telephone poles. Vehicles hang below the track for aerodynamics, and so that the vehicle banks perfectly going around curves. Using such a lightweight system, it is possible to achieve rapid acceleration, and top speeds of 216 km/hr (134 mph) with only a 100 kW (134 hp) motor. Energy efficiency at car speeds (100 km/hr, ~60mph) can be 800+ mpg owing to weight, size and streamlined shape.
A full system-level simulator was created. This starts with an editor tool built on top of Google Maps to rapidly layout and prototype realistic networks. Next is the construction of a formal graph, and calculations of the physical limits determined by network geometry, namely centripetal force and track length. An end-to-end routing system was created which ensures velocity continuity of the vehicle and enforces a minimum 10m spacing between vehicles. Traffic shaping algorithms were designed to minimize total travel time of all vehicles in the system. Such algorithms are difficult given that NP-hard nature of the problem which reduces most closely to a multi-commodity flow problem with integer values. Ultimately the problem can be reduced to a time-slot packing algorithm similar to the bin-packing (or knapsack) problem.
Applying traffic patterns to the simulator allows system-level phenomenon to appear. In a Stanford University plus Palo Alto downtown test network consisting of 18 stations, 25km of track, a top speed of 216 km/hr, and acceleration rates of 3.0m/s2, capacity peaks at around 2000 vehicles per station per hour. Trip times are 3-4x equivalent drive-time comparisons between the same points.
Both speed and network capacity were discovered to be ultimately limited by two factors: radius of curves/turns in an urban setting and the “single lane road” problem of any track based system.
Urban and suburban settings are filled with 90-degree right-angle turns. While vehicles can travel 200+ km/hr, they must slow down to 10-30 km/hr around corners to such that centripetal force (velocity2/radius of the turn) is limited to at most 0.5 gees. This in turn requires large buffering times (5-10+ sec) to be placed between fast and slow vehicles, such that a fast vehicle does not overrun a slower, turning vehicle. Such buffer time requirements can reduce theoretical carrying capacity 10x versus the naïve assumption that vehicles can be spaced just a few meters apart.
To avoid this “single-lane-track” problem and decrease vehicle spacing to a few meters requires the use of long acceleration and deceleration on-off ramps. However, such on-off ramps effectively double network costs in urban environments where station spacing on the order of 1-2km is desired: it takes nearly 600m to accelerate to 200km/hr, and another 600m to decelerate.
Costs were analyzed with the help of experts in structural engineering and magnetics, along with calculations of raw commodity costs, and comparison to ski-lift construction costs, which the elevated guide way most closely resembles. A cost of $5-7m per kilometer was estimated, before stations and vehicles, and perhaps $8m per kilometer with stations and vehicles. While this is far better than light rail systems ($30-50m per kilometer), and approaching the cost of interstate highways ($4-8m per lane per mile), it is still far more expensive than non-highway roads. These can be built for $100-300k per mile on flat terrain (i.e. 50x cheaper), excluding land acquisition, tunnels, and bridges.
Given a cost of $8m per kilometer, stations must serve about 8,000 people each to be economical. This in turn means stations must be placed at least 2.6km apart even in dense suburban areas (2000 ppl/km2). At this spacing, average time to walk to a station is around 10 minutes. At 10 minutes walk to a station, plus 10 minutes walk to a destination, it is faster to drive a car parked right outside your home directly to your destination for any distances under 32km (20 miles) even if the maglev system averages 4x better velocity (as Swift simulations show is possible). Time to get to a station (walking, or drive plus park) kills the effectiveness of most personal rapid transit and light-rail systems until you have a population density of 5000+ ppl/km2.
While a 3-4x velocity improvement, 800+mpg energy efficiency, and 3-4x reduction in cost over light-rail systems, it’s not enough to displace cars. The future of transportation is still the concrete and asphalt road, for the simple reason that at $17 per tonne of asphalt versus $900 per tonne of steel or $10,000 per tonne of copper, roads are the only thing cheap enough to be ubiquitous in lower density areas. Not even the ultra lightweight track of Swift looks cheap enough to displace roads outside of dense urban areas where, because it uses a much smaller rolling stock (number of vehicles) and can have higher throughput, mass transit is likely a better approach.
If roads are the future, then so is the self-driving car, functioning like an on-demand taxi system, connected into a centralized traffic database to avoid congestion and minimize time, and driven by computers in a platoon formation of 5-10 cars to minimize aerodynamic drag by up to 35%.
It is quite possible the same system-level traffic engineering algorithms developed at Swift can be used in such a self-driving system. That said, Swift PRT as a maglev concept will be abandoned. No track-based system (not maglev, not light-rail or metro systems) can compete with the cost and ubiquity of roads for population densities below 5000 people / km2."
download the full doc in pdf format (http://swiftprt.com/blog/wp-content/uploads/2011/12/Future-of-Transportation.pdf)
source (http://swiftprt.com/blog/2011/12/the-future-of-ground-based-transportation-systems/)
Swift PRT Inc. (“Swift”) proposes a new transportation system designed to replace or radically augment the use of cars in urban and dense suburban areas. An ideal transportation system is a combination of speed (200+ km/hr), ubiquitous coverage (stations within a 3-5 minute walk max), on-demand departure (minimize waiting for a vehicle or train), near-noiseless operation, safety, privacy, and cost.
One potential fit is a magnetically levitated system, driven by a linear synchronous motor, and comprised of vehicles that carry only two people. A two-passenger restriction minimizes weight, while fulfilling over 90% of all journey requests – after all, the average car ride in the US is only 1.2 people. Larger groups can be accommodated by virtually coupling two person vehicles such that the group arrives at the final destination simultaneously.
By radically reducing the weight versus trains (100 tonnes) and light-rail (45 tonnes) by a factor of 100, the resulting elevated track structure becomes flyweight: a nearly invisible 20cm tall single track supported on columns no bigger than telephone poles. Vehicles hang below the track for aerodynamics, and so that the vehicle banks perfectly going around curves. Using such a lightweight system, it is possible to achieve rapid acceleration, and top speeds of 216 km/hr (134 mph) with only a 100 kW (134 hp) motor. Energy efficiency at car speeds (100 km/hr, ~60mph) can be 800+ mpg owing to weight, size and streamlined shape.
A full system-level simulator was created. This starts with an editor tool built on top of Google Maps to rapidly layout and prototype realistic networks. Next is the construction of a formal graph, and calculations of the physical limits determined by network geometry, namely centripetal force and track length. An end-to-end routing system was created which ensures velocity continuity of the vehicle and enforces a minimum 10m spacing between vehicles. Traffic shaping algorithms were designed to minimize total travel time of all vehicles in the system. Such algorithms are difficult given that NP-hard nature of the problem which reduces most closely to a multi-commodity flow problem with integer values. Ultimately the problem can be reduced to a time-slot packing algorithm similar to the bin-packing (or knapsack) problem.
Applying traffic patterns to the simulator allows system-level phenomenon to appear. In a Stanford University plus Palo Alto downtown test network consisting of 18 stations, 25km of track, a top speed of 216 km/hr, and acceleration rates of 3.0m/s2, capacity peaks at around 2000 vehicles per station per hour. Trip times are 3-4x equivalent drive-time comparisons between the same points.
Both speed and network capacity were discovered to be ultimately limited by two factors: radius of curves/turns in an urban setting and the “single lane road” problem of any track based system.
Urban and suburban settings are filled with 90-degree right-angle turns. While vehicles can travel 200+ km/hr, they must slow down to 10-30 km/hr around corners to such that centripetal force (velocity2/radius of the turn) is limited to at most 0.5 gees. This in turn requires large buffering times (5-10+ sec) to be placed between fast and slow vehicles, such that a fast vehicle does not overrun a slower, turning vehicle. Such buffer time requirements can reduce theoretical carrying capacity 10x versus the naïve assumption that vehicles can be spaced just a few meters apart.
To avoid this “single-lane-track” problem and decrease vehicle spacing to a few meters requires the use of long acceleration and deceleration on-off ramps. However, such on-off ramps effectively double network costs in urban environments where station spacing on the order of 1-2km is desired: it takes nearly 600m to accelerate to 200km/hr, and another 600m to decelerate.
Costs were analyzed with the help of experts in structural engineering and magnetics, along with calculations of raw commodity costs, and comparison to ski-lift construction costs, which the elevated guide way most closely resembles. A cost of $5-7m per kilometer was estimated, before stations and vehicles, and perhaps $8m per kilometer with stations and vehicles. While this is far better than light rail systems ($30-50m per kilometer), and approaching the cost of interstate highways ($4-8m per lane per mile), it is still far more expensive than non-highway roads. These can be built for $100-300k per mile on flat terrain (i.e. 50x cheaper), excluding land acquisition, tunnels, and bridges.
Given a cost of $8m per kilometer, stations must serve about 8,000 people each to be economical. This in turn means stations must be placed at least 2.6km apart even in dense suburban areas (2000 ppl/km2). At this spacing, average time to walk to a station is around 10 minutes. At 10 minutes walk to a station, plus 10 minutes walk to a destination, it is faster to drive a car parked right outside your home directly to your destination for any distances under 32km (20 miles) even if the maglev system averages 4x better velocity (as Swift simulations show is possible). Time to get to a station (walking, or drive plus park) kills the effectiveness of most personal rapid transit and light-rail systems until you have a population density of 5000+ ppl/km2.
While a 3-4x velocity improvement, 800+mpg energy efficiency, and 3-4x reduction in cost over light-rail systems, it’s not enough to displace cars. The future of transportation is still the concrete and asphalt road, for the simple reason that at $17 per tonne of asphalt versus $900 per tonne of steel or $10,000 per tonne of copper, roads are the only thing cheap enough to be ubiquitous in lower density areas. Not even the ultra lightweight track of Swift looks cheap enough to displace roads outside of dense urban areas where, because it uses a much smaller rolling stock (number of vehicles) and can have higher throughput, mass transit is likely a better approach.
If roads are the future, then so is the self-driving car, functioning like an on-demand taxi system, connected into a centralized traffic database to avoid congestion and minimize time, and driven by computers in a platoon formation of 5-10 cars to minimize aerodynamic drag by up to 35%.
It is quite possible the same system-level traffic engineering algorithms developed at Swift can be used in such a self-driving system. That said, Swift PRT as a maglev concept will be abandoned. No track-based system (not maglev, not light-rail or metro systems) can compete with the cost and ubiquity of roads for population densities below 5000 people / km2."
download the full doc in pdf format (http://swiftprt.com/blog/wp-content/uploads/2011/12/Future-of-Transportation.pdf)
source (http://swiftprt.com/blog/2011/12/the-future-of-ground-based-transportation-systems/)