DIA - Complexities of Large-Scale Product Development

This is a continuation of my earlier article on Navigating the Complexities of Large-Scale Product Development. One of the better examples of complex, large-scale product development I managed to research was Denver International Airport (DIA) automated baggage handling system, and although it is not a fully software product development example, there is much to be learned from it.

So in this post I shall solely talk about the airport, the needs, the design choice and lessons learned.

Background

Conceived in the late 1980s to replace Denver's aging Stapleton International Airport which had become increasingly congested and outdated, unable to handle the growing demand for air travel in the Denver metropolitan area. The decision to build DIA was driven by several key factors:

1. Capacity Constraints: Stapleton International Airport had reached its physical limits, with only 5 runways and limited space for expansion. DIA was designed with 5 runways, with the ability to expand to 12 runways in the future, providing much greater capacity.
2. Operational Efficiency: The new airport layout and advanced baggage handling system were intended to improve operational efficiency and reduce delays, a chronic issue at the old Stapleton airport.
3. Passenger Experience: DIA was designed with the passenger experience in mind, with features like automated people movers, spacious terminals, and advanced amenities to improve the overall travel experience.
4. Economic Development: The new airport was seen as a critical piece of infrastructure to support the continued economic growth and development of the Denver metropolitan area. It was envisioned as a major hub for both domestic and international air travel.

The scale of the DIA project was immense, with the airport covering over 53 square miles - making it the largest airport in the United States by land area. With an initial budget of $1.7 billion, a timeline of five years was defined to achieve this vision.

Designed Solution

We shall focus on the baggage handling bit of the project. This was the step-by-step solution:

1. At check-in, agents stick glue-backed bar code labels on baggage, identifying the bag's owner, flight number, final destination, and intermediate connections and airlines.
2. The check-in agent then puts the bag on a conveyor belt. Since no baggage can move without a telecar (imagine a cart on rails to hold the bag, you will see this in action later) holding it, a system exists for dealing with telecar allocation. Empty car management software is the heart of the allocation system, dispatching empty telecars to where the tracking computers anticipate they will be needed. The computers sense changes in demand by measuring the flow of passengers throughout the airport. During peak times, all 3550 telecars are available for moving baggage.
3. When an empty telecar arrives, the conveyor belt holding the bag advances.
4. Then a type of high-speed luggage bowling machine flings the bag at a T-intersection just as the telecar moves by, catching the bag in its fiberglass tray. Each telecar has a tray for this purpose that tilts into three positions for automatically loading, carrying, and unloading its baggage.
5. Telecars do not stop for loading or unloading, they only slow. This type of "Dynamic loading" increases handling capacity and saves energy as well. Before the telecar speeds away, a laser scanner similar to those used in grocery stores reads the bar code tag on the bag's handle and associates the bag with its telecar. These laser scanners are triggered by photo-electric sensors that detect a telecar's presence. Telecars pass photo-electric sensors every 150 to 200 feet of track.
6. The computer that scans the bar code tags then sends information to a sortation computer that translates it by using a look up table to match the flight number with the appropriate gate.
7. A tracking computer guides the telecar to its destination by communicating with the hockey puck-sized radio transponders mounted on the side of each telecar. The telecars are able to move on the tracks by linear induction motors, or LIMs, which are mounted periodically on the tracks, and push the telecars along. A metal fin on the bottom of each telecar slides through each induction motor gaining impulse as it goes.
8. Telecars merge with other telecar traffic and exit to unload stations by computers which control PLCs, or programmable logic controllers. The computer tracking a specific telecar directs it by communicating with PLCs that are responsible for causing track switches.

Summary - No more human intervention needed besides dropping off the bag with proper label. Everything was supposed to be automated.

Outcome

Its best to see this news coverage, rather than write..

Almost immediately, the project encountered financial and scheduling challenges. The initial budget ballooned to over $4.8 billion, more than double the original estimate. Delays plagued the construction process, pushing the opening date from 1993 to 1995.

Lessons

The key bad decisions were:

1. Underestimating the Complexity: The project team severely underestimated the technical complexity of building an automated baggage handling system of such a large scale. The system was designed to handle 60,000 bags per day across a 21-mile network of conveyors and autonomous carts.
2. Relying on Unproven Technology: The system relied heavily on cutting-edge, but unproven, technologies like barcode scanning and RFID tracking. These innovations were not mature enough to reliably handle the volume and complexity of luggage at an international airport.
3. Inadequate Testing: There was insufficient testing and prototyping of the system before the full-scale deployment. A small-scale prototype in a warehouse was not enough to uncover the real-world issues the system would face in the airport environment. Rigorous end-to-end testing should have been a prerequisite.
4. Mismanaging Changing Requirements: As the project progressed, the requirements continued to evolve, such as the addition of automated handling for oversized baggage. These scope changes were not properly managed, leading to cascading delays and cost overruns. Effective change control processes were lacking.
5. Dismissing Expert Advice: When the initial bids for the integrated baggage system were rejected, the project team approached the contractor BAE Systems directly, disregarding the earlier warnings about the system's feasibility. This dismissal of expert advice proved to be a critical mistake.
6. Absence of Backup and Recovery Plans: The baggage system was designed as a single, integrated solution, with no backup or fallback options. When parts of the system failed, there was no contingency plan in place, leading to the airport's prolonged 16-month delay in opening. Robust backup and recovery strategies were not implemented.

Ultimately, it was deemed unsalvageable, and after running it for 10 gruelling years, paying $1 million per month in maintenance, the system was ultimately abandoned.

References

• https://www5.in.tum.de/~huckle/schloh_DIA.pdf