Nature's Assembly Line
Bucket Brigades

Since the dawn of the Industrial Age, humans have used engineers to organize the workers on the line. But ISyE research led by John Bartholdi, Manhattan Associates Chair of Supply Chain Management, indicates that nature may have a more powerful method of balancing the work.

Bartholdi and then-doctoral student Don Eisenstein, now a professor at the Graduate School of Business at the University of Chicago, worked together to harness nature's technique in a system that has revolutionized order picking in warehouses around the world. Known as "bucket brigades," the system mimics the behavior of ants, whose methods of foraging food are designed to give the colony its strongest chance of survival.

Bartholdi's work on bucket brigades was born out of the same impetus as Dr. Craig Tovey's work with honey bee colonies (see article page 13). They were inspired by biologist Tom Seeley, a professor at Cornell and an expert on bees, as was another research partner, ISyE Professor John Vande Vate. "We were all attuned to the issues of self organization and decentralization in the emergence of behavior," says Bartholdi.

Bucket Brigades in Industry

Studying nature was an interesting break for a professor who normally concentrates on industrial issues. But duty called, and Bartholdi was asked to work on a project for the Defense Logistics Agency. The agency was looking for a better way to operate the assembly lines that sew military uniforms. He soon learned of the Toyota Sewing System, a Japanese method which allows workers to move around from machine to machine during production — an assembly line, but not a strictly defined one. "There is little bit of freedom in their movements, but not much," he says. "We looked at this for about a year, and realized it could be improved by sequencing the workers from slowest to fastest, allowing them to move any place on the line, completely cross trained."

In modeling, the extended idea worked well. "We applied the theory of dynamical systems to show that such lines, which we called 'bucket brigades', had the properties to self organize. Once you set them up correctly, they will spontaneously configure themselves so that they are the most productive that they can possibly be. That seemed like a great idea, because if they spontaneously organize themselves in the best way, that means that an engineer doesn't have to do it." Another benefit: if there are disruptions, the system reorganizes itself.

"Another way of thinking about this is that you've set up a force, like the force of gravity, that pulls the assembly line to the very best organization. Like gravity, that pull is always there. Even if something goes wrong, like the machine breaks and throws the line out of balance, once fixed, it will be pulled right back to balance." A sick worker can leave the line to go home, and the remaining workers will keep the line in balance. The workers don't even have to understand the principle, he says. "All they have to do is keep following the simple rule, much as a line ant or a bee follows a simple rule."

Bucket Brigades, as Used by the Ant Species Messor Barbarus J. L. Reyes and J. Fernandez Haeger, in their published paper 'Sequential co-operative load transport in the seed-harvesting ant Messor barbarus," describe the ant as using bucket brigades to carry seeds back to the colony. The smallest, slowest ants forage out farthest. When carrying a seed back toward the nest, such an ant may be interrupted by a larger, faster ant, who wrests the seed from the first ant and continues carrying it towards the nest. After the largest ant leaves the seed at the nest, she goes back out to get another. This raises an interesting question: How do bucket brigades arise? Bartholdi, with C. Anderson and J. J. Boomsa, offer an explanation in their paper, 'Task partitioning in insect societies: bucket brigades." First we make the following assumptions, which are consistent with the observations of Reyes and Fernandez Haeger (and others). Assumption 1: Larger ants are faster. Assumption 2: An ant can take a food item from a smaller ant but not from a larger ant. Now under Assumptions 1 and 2, bucket brigades arise spontaneously if each forager follows this simple, myopic rule: The Foraging Rule: If you are without a food item, run out along the foraging trail until you encounter one and then take it if you can, even if you must wrest it from another ant, and carry it back toward the nest. Consider the experience of a large forager. As it leaves the nest, it is likely that the first returning forager it encounters is smaller, and so our ant will successfully wrest the food item away and return to the nest. It is unlikely to meet a still larger ant on the way back. Subsequent trips are likely to repeat this experience. Similarly, consider the experience of a smaller ant. It is likely to have to travel for a long time before it encounters an even smaller forager that is returning with a food item; in fact, it may have to travel all the way out to the food source to get a food item. As our small ant returns with a food item, it is likely that any forager it encounters will be larger and will take the food item, after which our small ant will return to the food source. Again, subsequent trips are likely to repeat this experience. The result is that the ants will sort themselves from slowest (smallest) to fastest (largest) along the direction of seed movement towards the nest.

After a few effective simulations, the researchers were confident enough to approach Revco Drug Stores (now CVS) and asked for the opportunity to test the method at Revco's national distribution center in Knoxville, Tennessee. "The director of logistics for Revco had been trained in operations research, so he understood exactly what we were talking about it," says Bartholdi. On a slow Monday morning, they explained the procedure to those responsible for order picking. "Total implementation time was about 15 minutes. They tried it, and we saw almost immediate improvements. We didn't have to buy any new equipment, didn't have to change any software — we just sequenced the workers from slowest to fastest and told the industrial engineer to stop assigning the work. Let them move to where the work is spontaneously, just like the social insects."

Revco soon discovered the method was 34 percent more productive than its own system. "That's a huge cost savings for companies like Revco, which have enormous seasonalities. They're a retailer, they're very busy at the end of the year. They have to supplement their staff with a lot of new and untrained people, and that's not very productive. They also have to pay a lot of overtime, so getting a 34 percent improvement in the pick rate (the rate at which people send product out of the warehouse shelves to be sent to customers) is a huge savings. And it cost them nothing to achieve," he continues.

"I think it's safe to say that most of the big retailers now use this or some adapted form of it for coordinating their order pickers. All of the software systems, warehouse management systems, and similar systems have been adapted to integrate with the bucket brigade style of order picking."

Bucket Brigades in Nature

While Bartholdi and Eisenstein were formatting their ideas around bucket brigades, they continued their study of social insects. "A friend of mine e-mailed me that two Spanish biologists had found a species of ant in the high plateau of Spain which appeared to carry seed back to the nest by bucket brigade," says Bartholdi. "It was the oddest thing. The seed would be picked up by the smallest and slowest ant, carried back, passed over to a larger and faster ant, who speeds it back toward the nest, where it is taken by a bigger and faster ant, until the biggest and the fastest ant of all races back to the nest. The slowest guy goes back to get another one."

Bartholdi was fascinated. The ants don't even have an engineer to issue the basic commands. "They just do it. It turns out that you can explain it very simply. If each ant simply grabs whatever seed it can, the bigger ants are able to grab the seeds from smaller ants, because they are bigger and faster. The biggest ants tend to work at the end of the line. The smallest ants, who can't take anyone's seed, have to walk all the way out to the end of the path and pick up the seed off the ground, because it's the only way they can get a seed."

"The allocation of work emerges spontaneously," he continues. "In the industrial environment, we had to tell people to sequence themselves from slowest to fastest. The ants do it automatically. Everywhere we've been with our research, the workers always know who is slowest and who is fastest. In the world of social insects, even the bucket brigades emerge spontaneously. Once they've emerged, they balance themselves."

For ants, the system has other good qualities. "As the ant carries the seed back to the nest, she is subject to predation, capture by a spider, or to another species of ant stealing the seed. So you can imagine, as the seed gets closer to the nest, it increases in value. You've already invested work in it. At the same time, as it gets closer to the nest it is being carried by bigger and faster ants, so it is more likely to be safe. In addition to being an efficient way of gathering seeds, in the ant context it makes economic sense."

The Work Continues

Bartholdi and Eisenstein continue their work on bucket brigades. "We've found additional uses for bucket brigades, adapting them slightly for different kinds of environments. It appears to be very robust behavior that balances itself strongly, as long as you're careful about how you set it up. If you set it up incorrectly, you induce chaos. Even though your system is very predictable, even mechanical, it can behave as if it were effectively random, which means that there is a high variability in the time to complete a product."

The next stop for bucket brigades is the military. Bartholdi, Eisenstein, and former ISyE doctoral student Kevin Gue are in discussion with the Pentagon to adapt some of these ideas for what they refer to as "sense and respond" logistics, Bartholdi says. "In the military exercise, the effort would function almost like an organic creature. Social insects again are a good metaphor here. We know that ants fight wars with other ant colonies, for example. And there is clearly no general in charge. What we would like to do is find some context in military activities in which one can embed this form of self-organization, so that a colonel doesn't have to enforce it. We'd like to have individuals make decisions right there at the front, in a way that this coordination emerges spontaneously."

Bucket Brigades: How They Work By John Bartholdi and Don Eisenstein What are 'bucket brigades'? 'Bucket brigades" are a way of organizing workers on a flow line so that the line balances itself. Here is how it works. Products on a flow line are progressively assembled as they move down the line toward completion. An assembly line is a familiar example from the realm of manufacturing; but flow lines are found in all types of industries, wherever 'products" may be imagined to move along, from worker to worker. A classic difficulty in the management of flow lines is to balance the line so that it will be maximally productive. This requires precise and time-consuming identification of the work elements and estimates of standard work-content. For example, assembly lines are balanced by teams of engineers, who define task elements and then conduct time-motion studies so that the work can be divided equally among workers. Because bucket brigades are self-organizing, the need for centralized planning and management is reduced. This idea may be found in the social insects, such as ants or bees, which are highly effective at organizing themselves even though without blueprint or management. Instead, global coordination emerges spontaneously, through the multiple interactions of many simple components. Similarly, when workers on a flow line are organized into bucket brigades, they can function as a self-organizing system that spontaneously achieves its own optimum configuration, without special equipment, time-motion studies, work-content models, management, or control systems. The operation of bucket brigades is simple: Each worker carries a product towards completion; when the last worker finishes his product he sends it off and then walks back upstream to take over the work of his predecessor, who walks back and takes over the work of his predecessor and so on, until after relinquishing his product, the first worker walks back to the start to begin a new product. If, in addition, workers are sequenced from slowest to fastest, then we call the system a bucket brigade and the workers will, we have proven, spontaneously gravitate to the optimal division of work so that throughput is maximized. Notice that workers must maintain their sequence: No passing is allowed and so it can happen that one worker is blocked by his successor, in which case we require that he simply wait until he can resume work, after his successor has moved out of the way. (This waiting is not necessarily bad because it is the means by which the workers migrate to their optimum locations.) Benefits There is a reduced need for planning and management because bucket brigades make the flow line self-balancing. Production becomes more flexible and agile because bucket brigades 'tune' themselves, without time-motion studies or the other cumbrous endeavors of assembly-line balancing. Throughput is increased because bucket brigades spontaneously generate the optimal division of work. Secondary labor is reduced and quality improved because bucket brigades operate with the absolute minimal work-in-process. Training and coordination are simplified because it is easy for workers to know what to do next. Who is using bucket brigades? Bucket brigades are used mostly in distribution warehouses to organize order-pickers, in the apparel industry to organize garment-sewers, and in simple assembly processes. We believe bucket brigades to be more widely applicable but feel that the greatest economic significance is in order-picking, which is very labor-intensive. A typical high-volume distribution warehouse employs hundreds of workers to pick orders and the work must be rebalanced daily, and sometimes more often. For more information, as well as a web-based demonstration of the bucket brigade concept, visit https://www.isye.gatech.edu/~jjb/bucket-brigades.html.

• Spring 2004 Engineering Enterprise Table of Contents
• Engineering Enterprise Home Page



Web Site © Copyright 2020 by Lionheart Publishing, Inc. and ISyE, Georgia Institute of Technology. All rights reserved. No portion of this publication may be reproduced in any form without the written permission of the publisher.


Lionheart Publishing, Inc.
34 Hillside Ave
Phone: +44 23 8110 3411 |
E-mail:
Web: www.lionheartpub.com

ISyE / Georgia Institute of Technology
Atlanta, GA 30332-0205
|
Web: www.isye.gatech.edu