April 15 2013 Nathan's Blog
Scalability and baggage-handling
Currently there are two small-vehicle PRT systems in the world: the Masdar Institute, with two stations and 10 vehicles, and Heathrow Terminal 5, with three stations and 21 vehicles. By the end of this month, a third should begin operating in Suncheon, South Korea, with 2 stations and 40 vehicles.
Masdar and Heathrow have proven themselves to be extremely reliable -- demonstrating system availability of over 99.5%, which is superior to many long-established transport systems -- and passengers and clients alike have loved the systems. I'm sure that Suncheon will be no different. However the very small size of these systems -- two or three stations -- has led many prospective clients to ask: "is PRT truly scalable?"
"Sure it is," I say, showering them with detailed microsimulations and discourses on packet-switched mesh networking.
"Then why hasn't anybody done it yet?" they ask
"It's still early days for the technology," I reply, "and these small systems are really just pilots, to demonstrate that all the critical elements of a PRT system -- infrastructure, vehicles, communications and control systems, operating concepts, passenger behavior -- can really be put together successfully. And they've proved that they can be. From here, they're theoretically capable of scaling up a tremendously. In fact there is every reason to believe that they'll work even better at large scales than they do at small scales."
"Sure, but lots of things work, in theory. How can you prove that it actually works in practice?"
We can answer that question by looking in an surprising place: baggage handling systems. Specifically "Destination Coded Vehicle" (DCV)-type baggage handling systems.
DCV systems send on-demand, self-steering, autonomously-routed vehicles through complex networks with many origins and destinations. Sound familiar? In many respects, they're essentially PRT systems in miniature.
In America, DCV systems have a dark legacy, thanks to a catastrophically failed project in the mid-90s. The newly-built Denver Airport was intended to host America's first DCV system -- an ambitious network consisting of 17 miles of track serving 88 individual gates, carrying luggage on over 3,500 vehicles operating at 1-second headways. The project failed catastrophically and ultimately had to be removed. By the time the fiasco was over, hundreds of millions of dollars had been wasted and the opening of the airport had been delayed 16 months. No DCV system has been attempted in America since then.
This seems like improbable ground upon which to establish the viability of large-scale PRT networks. Indeed, several anti-PRT activists explicitly claimed that the Denver DCV failure proved that PRT was impossible. But in fact, DCVs have proved exactly the opposite.
At many airports, your luggage already rides the PRT
While the U.S. never again tried to implement an advanced baggage-handling system, the rest of the world has moved on. Complex DCV networks are now operating at dozens of airports throughout the world. Here you can see the Vanderlande Industries BagTrax system at Heathrow Terminal 5, with over 500 autonomous vehicles running at 1-second headways, at speeds of over 30 mph:
And here is the Beumer Autover system in Toulouse:
A look at the Toulouse network shows just how PRT-like these systems truly are:
The PRT -- no, DCV network at Toulouse Airport
These systems are not perfect analogues for PRT. They don't have to deal with the subtleties of passenger experience and behavior, which established operators can tell you is a very important thing to get right. And they operate in a completely different regulatory regime than PRT, which is a life-safety-critical system.
However in terms of the operational principles, the nature and complexity of the mechanical and electrical engineering, the communications and control systems, and the algorithms for optimising line capacity, vehicle routing, queue management, and empty-vehicle redistribution -- these DCV networks are essentially identical to PRT networks. Contrary to the outdated claims of anti-PRT activists, DCV systems have proved beyond any doubt that PRT systems are able to scale up to encompass very large and complex networks.
So why did the Denver DCV fail, then? There have been many detailed reports on its failure, and it is worthwhile for any serious PRT professionals to give them a read. This analysis from Calleam Consulting contains some particularly sagacious points about how the failures were rooted in the strategic planning of the airport, even before the system encountered any technical difficulties:
Because earlier generation baggage facilities were dedicated to individual airlines, airlines had historically built their own systems when a new airport was built. The advent of the integrated airport wide system required a change in mindset. The integrated nature of the new systems meant that instead of airlines looking after their own facilities, airports needed to take control.
The key point the airport’s Project Management team failed to see was that the shift in technology required a corresponding shift in organizational responsibilities. The failure to recognise that shift represents a planning failure that dated back to the very start of the construction project.
Another article from Material Handling & Logistics makes some excellent points:
Dave Bowman, vice president and division manager of sortation and distribution systems at Beumer Corporation, tells me that some dozen cart systems have been implemented since Denver. Says Bowman: "These successful projects utilized basic material handling technologies that range from nearly identical (passive, dumb carts) to radically different (active, intelligent carts) than that implemented at DIA. Most notable in all of this is that none of the new projects were in the U.S.; they were all international projects, with Europe claiming the largest share."
Concludes Bowman: "My point in all of this is that for the U.S. baggage handling industry, a good idea that is badly implemented is too often only remembered as a bad idea."
"Once this is done, it can be a generation before someone comes along who is bright enough or naïve enough to revisit the technology as an option for consideration."
One thing which baffled me, however, was the insistence in many analyses that "line balancing" -- managing a high-capacity flow of vehicles -- was a particularly complex, difficult, and ultimately unsolved problem for the designers of the Denver system. Being very familiar with PRT routing problems, I know that they're not exactly simple, but neither are they overwhelmingly difficult. So what was going on?
The answer finally presented itself in a recent email conversation I've been having with Vanderlande, a modern DCV supplier. They wrote:
Our company is a strong believer in the so called ‘mainline concept’, as opposite to in-line concepts such as the Denver fiasco in the mid-90s. This means that all DCV functions are configured around a mainline or backbone, that has a continuous transport function and designed to have a very high availability. All other functions such as loading, unloading, empty cart storage etc – the more critical processes – are placed offline. Although this sometimes may consume somewhat more space than the in-line concept, the availability of the complete system is strongly increased and a high number of stations (or local areas as we tend to call them) can be connected to the loop.
In other words, Vanderlande has rediscovered the principle of off-line stations. Offline stations are fundamental to the whole concept of PRT -- and the Denver DCV designers didn't understand this subtle but critical point. They were trying to do delicate bag-handling processes with vehicles stopped on the high-speed guideway. So of course it was a disaster -- not because it was similar to PRT, but because it wasn't similar enough.
In any event, we can now say that the success of modern PRT-like baggage-handling systems at airports have proven, beyond any doubt, the fundamental scalability of PRT systems.
Now we just need to go and build them.