But a fully automated system could, at least in theory, move bags quickly and efficiently enough to make the entire enterprise work. Indeed, United Airlines would not sign a lease to be the principal tenant at Denver until it was assured that the airport would have an effective automated baggage handling system. (de Neufville, pp. 2-4). The designers of the system focused on speed as its signal characteristic. They promised to deliver a system at which the bags would be moved at speeds up to 24 miles per hour so that the bags from a narrowbody jet could be unloaded and sent to their destinations within twenty minutes. For larger, widebody jets, the delivery time was promised to be thirty minutes. The designers boasted that the system could move a bag from one part of the airport to any other part within ten minutes. (deNeufville, p. 4).

Boeing Airport Equipment, which later changed its name to BAE Automatic Systems, was initially responsible for the design of the automated baggage handling system. BAE's proposed design was the most complex automated system ever designed. BAE systems convinced, Walter Slinger, Denver's Chief Airport Engineer that such an ambitious automated system would work by building a prototype automated baggage-handling system. BAE built this prototype in a 50,000 sq. ft. warehouse near its manufacturing plant in Texas. (Donaldson, 1998).

Denver's system included an impressive collection of technology. It used over 300 desktop computers, a large server that hosted a database essential to running the system, a high-speed fiber-optic network, 14 million feet of wiring, 56 laser arrays, 400 frequency readers, 22 miles of track, 6 miles of conveyor belts, 3,100 standard telecars, 450 oversized telecars, 10,000 motors, among other things. (Schloh, 1996). Each track could, in theory, carry 60 DCVs per minute. The DCVs were controlled by "radio frequency identification" or "RFID." (deNeufville, 1994, p. 3).

The design and installation of the automated baggage handling system did not go smoothly. Problems with the system caused substantial -- and expensive -- delays in the airport's opening. It was originally scheduled to open in October 1993, and delays initially pushed that date back to March or April of 1994. But continuing problems with the baggage handling system prevented the airport from opening even at that later date. The system was not ready and the airport did not open until March 1995. These delays added $500 million in construction and interest costs to the total cost of the project. (de Neufville, 1994, p. 2).

The problems began at the very outset the guiding hand behind the automated system at the outset of the project was Chief Airport Engineer, Walter Slinger. Unfortunately, Slinger died six months into the project. His replacement had a different management style and little knowledge of construction. Moreover, he lacked Slinger's keen commitment to making the system work effectively. This change at the top was indicative of how the rest of the project was going to go.

The problems continued through the construction of the principal facilities for the airport -- its terminals, concourses, runways, and the internal infrastructure that served all of those facilties. In the course of the design, construction and testing of the physical facilities, individual airlines made numerous changes to the facilities that would affect the operation of the baggage handling system, such as adding ski-claiming devices and odd-size baggage elevators.

In addition, the design of the automated baggage handling system at Denver was undertaken after the terminals and runways had already been planned and after their construction was underway. Consequently, the physical specifications of the baggage handling system had to be shoehorned into existing spaces. In many circumstances, the area provided for the baggage handling system was simply not adequate for the system's own requirements. In addition, the contract for the system's design and construction was awarded only twenty-one months before the original opening date. The short timeframe precluded the designers from undertaking any simulation or physical testing of the full-scale design. (deNeufville, p. 4).

Communication problems made things worse. No-one effectively managed the lines of communication connecting city government, the managers of the airport project itself, the designer of the automated baggage handling system, and the airlines themselves. Consequently, coordination among these constituent groups was lacking. This multiplied the problems associated with managing information. And these information management problems were particularly vexing, given the fact that an automated baggage handling system is largely an enterprise in managing information.

There were problems...

The original contract awarded to BAE did not comply with municipal rules promoting contracts with minority-owned businesses. BAE had to hire outside contractors to meet this requirement at an additional cost of $6 million. BAE later lost a contract for providing maintenance because it was unwilling to meet union demands for wages for maintenance workers. (Bartholomew).
There were many mechanical problems at Denver, even before the airport opened. Many of these problems were mechanical. The DCVs jammed in the tracks. The conveyor belts were misaligened with DCVs. Baggage was shredded. (de Neufville, p. 4).

These manifold problems exacerbated the inherent challenge in designing a system that included so many novelties. The system required synchronization between the conveyor belts and the DCVs. This was not something that had been done before in ordinary automated baggage handling systems. In ordinary systems, the conveyor belts ran continuously, delivering luggage in a constant stream. In Denver's automated system, the conveyors only moved when there was a DCV in position to receive the leading bad on the conveyor. Thus, the effective operation of the entire system depended upon the efficient and timely delivery of DCVs to the proper locations. (deNeufville, 1994, p. 3).

The software that controlled the movement of DCVs sent empty cars back to the waiting pool instead of terminal building. When a DCV became jammed on the track, the software's initial design shut down an entire portion of the network instead of merely shutting down a section of the track behind the jammed car. In addition, the optical sensors did not work as they should have. When they were dirty, they malfunctioned, causing the computer system to recognize a section of track as being empty when it was not.

Although there had been automated baggage handling systems that served parts of airports, Denver's system included many innovations that had never been tried before. It was the first system that would serve an entire airport. It was the first in which the DCVs would only slow down but would never stop to receive bags. It was the first to be operated through a network of desktop computers instead of through a single mainframe computer. And it was the first to have the capacity to handle oversized bags. (Myerson).

Inexplicably, there was no provision for a back-up system, which would employ conventional baggage handling techniques. There were no tugs for hauling bags in the event that the conveyors and/or railway system broke down. Even more remarkably, there was no system of access roads over which such tugs could run. (deNeufville, p. 4).

The system's problems seemed to compound each other as efforts to handle some of these problems encountered problems of their own. Once problems in the system began to be glaringly apparent, a German consulting firm, Logplan, was hired to evaluate these technical problems and suggest solutions. Logplan developed some effective approaches to these problems. Problems in misreading tags were addressed by adding more laser readers. Problems with the speed and control of the DCVs on the rails were address by adding more controllers, which also made it easier to avoid misalignments with the conveyors. Despite their effectiveness, these solutions added to costs, slowed the performance of elements of the system, thereby reducing the system's overall cost efficiency.

Eventually, the airport abandoned the effort to provide a fully automated system, even before the airport opened. United Airlines only provided a fully automated system for one of its concourses. It used conventional systems, with tugs and baggage carts, at other concourses. When the airport first opened, the baggage handling systems at Denver served three concourses where covering two main terminals where passengers check and claim their luggage. (de Neufville, pp. 8-9).

In the end, the misadventures with the system resulted in grave cost overruns. The city originally budgeted $193 million for the construction of the automated baggage handling system. The final cost was nearly $311 million. Adding a conventional system contributed an additional $80 million to the project cost, and this did not include the $100 million that had to be spent to re-design the terminals to correspond with changes in the design of the baggage handling system. The entire airport project was notable for its cost overruns, and the automated baggage handling system was the source of the vast majority of those increased costs. (Dubroff, 1994).

Persistent Problems with Automated Baggage Handling Systems

The problems in Denver were, in large part, caused by idiosyncrasies…