A Big, Big Fish Story
by Patricia E. Moody, CMC
Quick, what does Darwin's theory of evolution have to do with canned tuna and a robotic fish at MIT's School of Naval Architecture - and supply chain efficiency for that matter? And how are Deere & Co. suppliers in the Nebraska heartland using a very smart and very simple software solution called a genetic algorithm, based on Darwin's theory of evolution, to make their customer happy? It's a big fish story illustrating smart solutions. In fact it's e-manufacturing, the next step beyond lean manufacturing. It's the software-enabled approach to handling exponential complexity, which is everyone's sourcing problem.
Created for Efficiency
Let's start with the complexity challenge and use the automobile as an example. The typical automobile is assembled from 5,000 to 6,000 bits and pieces of plastic, steel, rubber and cloth, all supplied by 300 to 400 strategic first-tier suppliers, who are in turn supplied by hundreds of second-, third- and raw-material suppliers. A network that touches 1,000 companies offers exponential opportunities for complexity, especially one in which the lower tiers supply more than one big customer. Add to the equation the personalized customer order for a made-to-order orange sports utility vehicle loaded with specialized communications, entertainment and off-road gear. The designer and the supplier's blackboard becomes crowded with hundreds of scribbled calculations, more than even the best human minds can handle.
That's where MIT's robotic tuna comes in. For years marine scientists and naval architects have focused much of their design energy on improving the drag ratios, propeller systems and batteries in ships like the autonomous underwater guided vehicles (AUG's) that locate shipwrecks and fallen airplanes.
When fish swim their fin movement leaves little eddies of water, positive and negative space, that propel the tuna forward. But just as airplane propellers are less than 100 percent efficient, some fin energy is also wasted motion. So MIT scientists decided to fast-forward Darwinian evolution, believing that if they could make their mechanical fish better and more natural then they would have solved a series of AUG design efficiency problems. Looking to nature for a better way - better shape, movement, skin - to move the hull faster through the water with less fuel, MIT scientists discovered that the Blue fin tuna is one of nature's most efficient swimming machines, capable of reaching speeds of up to 40 miles per hour.
But observers were puzzled by what made the tuna such an efficient swimmer. Nature is full of surprises (as was the MIT tank), however, as researcher David Barrett discovered. Barrett, who directed the construction of Charlie the robotic tuna, modeled Charlie's 41 ribs, eight vertebrae and flapping fins, all of which were covered with a bluish skin, after the body of an actual prize Atlantic catch. Years of evolution had developed phenomenally efficient propulsion and agility in Charlie's finny ancestors.
Although man-made propellers generate typically 50 to 70 percent efficiency, Robotuna's robotic flippers cranked out 90 to 95 percent efficiency in Barrett's software simulations. Researchers hoped to learn the key to the efficiency gap by studying Charlie's movements, and they were not disappointed. Color coding Charlie's wake as he flapped from one end of the MIT tank to another revealed that his side-to-side tail motion created a thrust jet that moved his body forward. By studying live fish, scientists also discovered that other body parts - backbone, stomach region - contributed to forward movement and the tuna's remarkable drag ratio. So the challenge for Robotic vehicle designers became to determine which of the millions of possible combinations of body movements worked best for Charlie.
An Iterative Routine
MIT's genetic algorithm software served to shortcut millions of years' of tuna evolution in mere minutes. Although the term genetic algorithm may sound complex, the simple computer rules that duplicate natural selection really aren't. The program uses an iterative routine that compares many variables, one to another, until the program finds the best solution to accomplish the objective which, in Charlie's case, was to get the tuna through the water faster with less effort.
Step one of the program defines the population, specifies the variables, and then calculates Charlie's speed under one specific variable - no fin movement, for example. Ask which is better: the solution with no fin movement or the result of the second calculation, which is movement with one fin flapping a little. The computer (evolution) selects the value that represents "one fin flapping a little," and the selection process continues through a list of possible movements, or "children," in the software evolutionary process. Finally the perfect solution, a dream "child" that meets all the parameters and travels 40 knots through heavy seas for eight hours straight, pops out of the parent's software code. It is Charlie, the perfect swimming machine.
Although a human could do the optimization problem, with so many variables the calculations would take months with little latitude for change. Software is the only way to handle endless variety and complexity in our lifetime. Barrett set seven parameters for movement in his computer-generated selection process that totals some 282 million combinations. It's these computer-generated combinations that are allowing supply chains to operate more efficiently - and MIT scientists are not the exclusive users of such innovation.
Back to the Heartland
Deere & Co. is a big old company that loves innovation. From the Vermont blacksmith who trekked west and created a self-scouring plow to work the rich Mississippi valley soil, down to today's innovative supply chain solutions, Deere has remained true to its founder's dreams. In May 1997 the Smithsonian Institution honored Deere & Co.'s use of a genetic algorithm-based schedule optimization package on the seed planter assembly line where more than six million combinations of available options produced scheduling problems far beyond human capabilities. The factory's daily assembly schedule is created in minutes, and the impact on the line is manageable. Where early production plants struggled to reach volume, modern lean sites grapple with endless variety.
Bill Fulkerson, the Deere software champion, a trained mathematician and innovation nut, is credited with making the connections that allowed Deere to enter the software evolution game. Fast forward now, however, to a later Deere application and the genetic algorithm solution gets cheaper and is as portable as your laptop.
At Deere the beat goes on. The same genetic algorithm concept has allowed suppliers and planners to crunch massive amounts of data - possible combinations of work centers and process flows - to make up the "ideal" or "fittest" combination of possible routings across production machines. Here Dave Meyer, supply base manager for Deere's worldwide agricultural division, is another manufacturing pioneer who loves innovation. Meyer uses an off-the-shelf package on top of a spreadsheet to set up and run a variety of process flow combinations. In fact the genetic algorithm package has been successfully used inside Deere as well as in the supply base.
Because many manufacturing suppliers have traditional batch runs through functional departments, scheduling supplier production can be a nightmare of unpredictable and long-cycle time schedules. Software solutions not only enable better run combinations, they also allow planners to rethink the functional layouts into patterns of more efficient cellular machine distribution.
And Meyer's work, like Darwin's before him, proves that evolution works. Even in job shops.
Copyright Patricia E. Moody. All Rights Reserved.
|© Copyright @ 2004 Keynote Resource Inc. All rights reserved. Permission is granted for linking to web pages within Keynote Resource
Keynote Resource speakers bureau can help you find the ideal keynote speakers for your next event. We represent inspirational speakers, motivational speakers, corporate entertainment and more.
Keynote Resource speakers bureau will find the perfect keynote speaker for your upcoming event, whether you are looking for inspirational speakers with a message, motivational speakers to set the tone of your conference or a facilitator for your annual retreat.
Keynote Resource speakers bureau works closely with executives, meeting planners and training directors to create events that result in highly productive learning or are just plain fun. We can identify business speakers, inspirational speakers or motivational speakers that are well suited to your event, send you videos and press kits to help you with your speaker selection and we’ll work closely with the keynote speakers you hire to ensure they customize their presentations to meet the specific needs of your audience.
We have speakers in all areas including Arizona, California, Colorado, Connecticut, Delaware, Florida, Georgia, Hawaii, Illinois, Indiana, Iowa, Kansas, Maryland, Massachusetts, Michigan, Minnesota, Missouri, Nevada, New York, North Carolina, Ohio, Oregon, Pennsylvania, South Carolina, Tennessee, Texas, Utah, Virginia, Washington and Wisconsin.
We can book speakers for you worldwide, in Canada, New Zealand, Australia, Singapore, Hong Kong, China, Ireland, England, France, Italy, Switzerland, Denmark, Sweden, Brazil and The Bahamas. Click Here to Search A Speaker
Speaker fees are determined based on a number of factors and may change without notice. Fees may vary based on the speaker’s availability, supply and demand, program length and location of the event.
Each fee range listed on this website is intended to serve as a guideline only. In some cases, the actual price quote may be above or below the fee range stated. For the most current fee, please contact your representative directly.