Monday morning, I headed down Singapore’s spotless streets and, for Asia, an eerie lack of chaos, up to the National Computer Board to meet with Lew Yii Der, an engineer with Singapore’s Public Works Department. On my last visit, I was told that a contract had been awarded for an electronic road pricing system, and I was eager to learn more details.

Since 1975, Singapore has attempted to reduce congestion in the central business district by charging a fee to enter the area during morning and evening rush hours. There are 26 entry points to the central area, known as the Restricted Zone.

Before entering the Restricted Zone, a motorist pays SG $3 (U.S. $1.76) for a license and, when passing the entry point, ensures that the large pass is prominently visible on the left side of the windshield. Failure to prominently display results in your license number being noted by the police patiently waiting by the side of the road and a ticket being mailed to your home.

On a typical day, 300,000 vehicles enter the Restricted Zone, of which 74,000 enter during the controlled periods. While the system has certainly reduced congestion by keeping those without the will or the resources away during rush hour, it has some drawbacks.

The biggest problem is the manual nature of the system. Cars must stop to buy a license and then slow down to wave it around. Fifty-two officers must be present at the 26 entry points to enforce the system.

Many cities throughout the world are investigating ways to reduce congestion in central business districts. Lew Yii Der explained the two classes of solutions: passive and active. In a passive system, people maintain accounts with the government. When you enter a restricted area, a charge is made to a person’s account. Passive systems thus require central billing, are difficult to maintain, and, even for Singapore, are considered too intrusive. An active system uses a stored value card. Buying a card for a subway and then running it through a reader to activate the turnstile is an example of an active system.

While I had understood, perhaps wrongly, on my previous visit that a contract for an active system had already been awarded, a bit of further discussion turned up the information that not only had the contract not been awarded, but the RFP was not yet out. An RFP had been issued previously, but it was worded so vaguely that vendors were not sure what to bid.

Lew Yii Der gave me an outline of what the new RFP would specify. Smart cards would be mounted on all car windshields. Value for the cards would be added at sales outlets, where the motorist could also review the transaction log stored inside the card.

When a motorist entered the Restricted Zone, the car would pass under a microwave beam, which would debit the appropriate amount from the card. If the card didn’t have enough money on it, a camera would be activated to take a picture of the rear license plate.

Since the system didn’t exist, it was a bit hard to get more technical details, so I spent the rest of my allotted hour trying to learn more about other traffic control systems.

Like most big cities, Singapore has computerized its traffic lights in the busy parts of the city. The original system, installed in 1983, was a fixed system, meaning that the lights changed on some prearranged timetable. Each light had a phone line to a PDP computer which controlled the lights and attempted to form “green waves” so motorists could drive without stopping.

In 1987, Singapore installed an Australian dynamic traffic management system called SCATS. In addition to the traffic light, each intersection in this system has magnetic detector loops under the pavement, used to measure the flow of cars. A local controller at an intersection can use the absence or presence of cars to change lights. The data from the detector goes to the regional PDP computers, which use the traffic data to adjust light patterns while trying to establish green waves. Data from regional computers goes to a VAX 11/750, which is used by headquarters staff to monitor traffic status.

The traffic subsystem is also tied into the emergency fire system. Every major building in Singapore has a fire detection system tied into the fire stations. When there is a fire, trucks are automatically dispatched from fire stations to the fire location and the traffic system is used to provide a green wave for the trucks.

Between the parking control, the Restricted Zone, the traffic light management, the incredibly clean streets, and the gum-free subway system, it was certainly clear that Singapore was able to control the flow of people and vehicles in the city. I had to ask myself, however, if there wasn’t some tradeoff between efficiency on the one hand and, on the other, an atmosphere that promotes a vibrant, creative, and (yes) fun place to live.

On my previous visit, I had learned about Tradenet, an EDI application for clearing customs paperwork. I had heard that not only was the customs work computerized, but that the Port of Singapore Authority (PSA) also had extensive systems. As with all things computer related, the National Computer Board was the place to start.

Singapore is one of the world’s busiest ports. Since the opening of the Suez Canal in 1869, the island has been a major entrepôt port, serving as a clearinghouse between destinations in Europe and Asia.

That afternoon, Chew Keng Wah of NCB brought me down to the Tanjong Pagar terminal where container ships are loaded and unloaded. We met with Ang Chong Hoat, an MIS manager for the port.

Running a terminal as large as Tanjong Pagar is a highly complex scheduling problem. When a ship arrives, containers have to be unloaded and other containers loaded. Cranes mounted on rails are moved up and down the length of the wharf and are used to move containers on and off ships.

Incoming containers are then placed on trucks and run over to an adjacent storage yard. There, containers are picked up by an even bigger crane, a transtainer, a device that constructs huge piles. When the ship arrives that is to carry the container to its destination, the container makes the reverse journey back onto a ship.

The trick is to pile the containers so they are readily available when needed. You hope that the container you need at a given time is not buried under a pile of other containers that are not needed until later. It also makes sense to keep containers for similar destinations in the same area so the transtainer does not have to be moved around.

The goal of all this is to minimize the amount of time a ship stays berthed at the terminal, thus making more effective use of scarce resources. This minimization is done in an environment of scarcity, where cranes, ships, transtainers, and berths are all finite resources.

Planning the loading and unloading of ships and the subsequent placement of containers in the yard was once the province of a few highly skilled experts churning out plans by hand. A plan for a single ship could take many hours to work out. To automate the process, the Port Authority and the NCB developed an expert system to assist in the process, cutting the amount of time to develop a plan for a ship in half. The expert system was developed on Sun Workstations in Objective C and Lisp and took over 20 person-years.

The ship planning system is the first of six that the PSA will eventually deploy. Other systems will try to place ships in berths adjacent to the correct yards, or try to take ships in a yard and place them in a correct berth, or make available the proper number of resources for the unloading and loading of a ship.

The planning system thus tries to solve a complex optimization problem a level at a time. The results from one planning exercise, such as putting ships in the correct yards, is moved into a Sybase database. The next planning process takes that data out of Sybase and uses it as input for the next level of decision making.

Since a yard has many simultaneous ships, it is still possible for one ship-unloading plan to interfere with another ship-unloading plan. The last planning module is a terminal operations simulator, used to take all the separately developed plans and see if they conflict. If so, adjustments are made back up the chain.

The planning network at Tanjong Pagar terminal consists of two mirrored Sun servers with approximately 5 Gbytes of disk space. Planners use 13 workstations and various plotters and printers. Some of the modules, such as the ship planning system, were already deployed. Others were just going online.

Planning is one of three major clusters of computer applications in use at the terminal. A large mainframe is used for documentation and a third cluster of applications runs on a Stratus computer for fault-tolerant control of yard operations.

The documentation system contains the raw data, such as when a ship is expected to arrive and the contents of that ship. This information is used to feed the planning process and also is used as input for back-office functions such as billing.

Data for the documentation system is fed in electronically by shipping agents. Small companies emulate a 3270 terminal and dial in to fill out forms. Larger organizations use an APPC-based program to transfer data. Documentation data resides in an Adabase/Natural application and an APPC-based program running on a SunLink gateway transfers data over to the Sybase server for planning.

The output from the planning process is used to feed the yard operations system. This system, which was preparing to go online when I visited, run on a Stratus platform. The Stratus is linked to the IBM using SNA APPC programs and to the Sun workstations using TCP/IP.

The output of the planning process is a series of loading and unloading sequences, specifying container ID numbers, yard locations, ship berth locations, and other identifying data. This information is translated into a series of micro instructions, such as a command to a transtainer to go to a certain location and expect to load a container with a certain label. Another command might give a truck instructions to move to a certain berth.

All this is computerized, with custom PC-based systems on cranes and transtainers. PSA trucks get transponders to track their location and simple display terminals give truckers instructions. Trucks entering the terminal from local shipping companies are given a window during which they are allowed to enter the yard, and at the gate the truck is fitted with a transponder.

As we were driving back from the port, Chew Keng Wah told me that yard operations were only half of the port automation program. The other half was a marine-based control system that would automate the deployment of tugs, pilot allocation, and utilization of deep sea channels in and out of the terminals.

Having finished my whirlwind tour of Singapore, I went back to my hotel to meet a friend for a drink. Bob Cook (not his real name) is a salesman for a large computer company. Salesmen for this company are assigned territories by country. One will work in Thailand, another in Singapore. Bob gets whatever is left. His beat includes places like Brunei and Bangladesh.

I had brought Bob a copy of Stacker for his laptop PC. In many places, including Singapore, the only way to obtain legitimate software locally is to pay three times (or more) the U.S. list price. Even then, many U.S. companies refuse to provide technical support or updates.

Lack of legitimate software drives many otherwise honest users into the arms of the software pirates. Bob always gave me a list of software he needed to purchase and I brought it over for him on my trips. Updates and other luxuries he had to get for himself from the pirates.

After taking his software, Bob looked around the bar furtively. Everybody was watching the lounge lizards doing an Elton John tune. The audience was captivated, although the fact that the performers were clearly out of the Barry Manilow school just made me thirsty.

The coast clear, Bob told me he had something for me, giving me a conspiratorial wink and pulling out a package of Big Red chewing gum. Singapore had just banned all chewing gum on the island, a reaction to the fact that somebody had discarded their gum in a couple of subway doors, jamming them. The fine for importing this vile substance with intent to deal had been set at SG $10,000 and a year in jail. The government had just arrested four gum dealers and was threatening swift retaliatory action to any others fool enough to chew and talk about it. A government official was quick to point out, however, that only the importation and sale of the substance was banned, not its chewing.

Bob’s frequent travels to free trade meccas like Bangladesh had given him plenty of opportunities, and he had become a key gum trafficker, supplying his teenage daughter and even occasionally offering a stick to friends.