Millennium Challenges

Machining into the new Millennium

An independant consultant working at the intersection of science and business.

By Paul Sheldon
Contributing editor

[Editor's note: The author uses phrases such as "chunk of stuff" to encourage the reader to think beyond current materials and processes.]

Predicting the future is inherently impossible. Yet speculating about the future is fun and a useful exercise. And as long as we don't extend our thoughts too far into the future, we can construct a set of likely events.

Speaking of too far into the future, by the year 2050, many people may choose to abandon their bodies and download themselves into artificial brains. This could significantly reduce the sales of toilet seats and brake rotors. In the nearer term however, there will still be a significant demand for these sorts of items, and that is good news for those of us who make things for a living.

We basically take stuff and make it into things. We're shape-changers, taking steel and applying well controlled physical or thermal vi olence to change its shape into useful bathroom accessories and such. Our well publicized headlong rush into the information economy has not changed this need for things, and in fact, information technology will continue to enhance our ability to convert stuff into things.

A true 3-axis parallel kinematic machine of extraordinarily high stiffness, developed and patented by Paul Sheldon.

A word about computers: Despite the amazing advances so far, computer and communication technology is still in its infancy. The desktop and control computers with which we are currently so impressed will be reduced to unobtrusive pinheads consuming nanowatts. Embedded computers will become implanted computers, residing in bandages, clothing, tools, and even our bodies. We will also be able to effortlessly select what is truly important from the sea of information available to us at all times.

In the near term, we will view production schedules and cutter forces through the virtual displays built into our safety glasses. A little later we will avail ourselves of the wireless transmission of pertinent information directly to our skulls. Computers will become increasing useful, ubiquitous, and unobtrusive. A contemporary example is your new car, which operates courtesy of a number of computers about which you can remain blissfully unaware.

Currently, machinists are a lot like Michaelangelo. When Michaelangelo was asked how he managed to take a large chunk of marble and convert it into a sculpture of exquisite beauty he said "it's simple: I take the block of marble and remove everything that's not the statue."

Likewise, we currently take a chunk of metal and remove everything which isn't the jet fighter wing, often removing more cubic inches than we keep.

This is subtractive fabrication. Additive fabrication, most usually referred to by the conceptually limiting label of rapid prototyping, is right around the corner. Progress will be made in the direct addition of useful stuff, like steel and aluminum alloys. Picture a printhead which spits minuscule drops of molten metal, building up, rather than removing, layers of material to achieve the desired shape. This is intrinsically simpler than subtractive machining, because complexities of tooling and raw stock geometry are minimized or eliminated.

Significant progress in speed, accuracy, and usable materials must be made before these additive machines become a reality in the production environment. But at this moment, hybrid machines which perform both additive and subtractive fabrication are technically feasible, and may be commercially desirable in the near future. Imagine a tool-changing machining center on which one of the tools in the chain adds more of the metal you are machining: you mount your chunk of metal, carve out the overall shape, make a few pockets, build up some reinforcing webs, drill and tap a few holes, build up a few bosses, then machine them flat and counterbore.

A low cost high precision 3-axis continous probe, patented by Paul Sheldon.

Speed of manufacture will remain a top criterion, so the additive operation can be allowed to be fast but relatively coarse geometrically, because dressing it up to tolerance by conventional subtractive tooling is a relatively quick operation. You build up the feature using the additive tool, then make it precise in a couple of passes using your conventional subtractive tools.

The field of material science is poised for a protracted and accelerating series of advances. Unfortunately for us shape-makers, this will not necessarily make our lives easier. New materials are engineered for end-usability, not machinability. The example most common now is aerospace alloys which exhibit beneficial strength/weight ratios, but may be extremely difficult to shape into useful parts. This challenge to cutting tools, machines, and machinists will continue. A little further out are smart materials, which can report on their physical status and, to some degree, change their physical properties. There are some intriguing shaping possibilities to machining a material that can communicate what is happening to it. Out much farther are the shape-maker's dream: materials which form themselves into the desired shape upon command. We can consider this stuff to be "metamorphic muck" or "co nscious clay." It is currently impossible to even guess when a bucket of this stuff will be available at your local industrial supply house, or what its properties will be. "Reconfigurable maven metal" would be nice to have around­simply command the material to become a valve body or jet turbine blade­but is not likely to arrive in time to eliminate your grandchild's need for metalcutting machines.

Speeds and feeds will continue to increase. However, Newton's laws of force and motion will remain in effect, and this will necessitate considerable further evolution in machine design. For example, high speed spindles and axis travels make little sense unless we also have high acceleration (of the machine, not just the motor) to get around curves and corners without the need to slow down excessively. Mass must come down and stiffness must go up­particularly dynamic stiffness, which is too often currently ignored. If our eyes could see in the realm of a few microns, today's machines performing acceleration and deceleration motions would look like a Warner Brothers cartoon, in which everything is elastic and the body takes off before the head and the neck stretches like a rubber band before slingshotting the head into position, where it vibrates for a while. Control algorithms will not be able to adequately compensate for the deleterious effect of this on the final part geometry. Therefore we will see radical new machine structures such as parallel kinematic machines, of which the few well-designed hexapodal machines out there are the first examples. These machines have the inherent potential for much improved stiffness/mass ratios, and are mechanically simpler than conventional orthogonal machines. This will lead to increased productivity and reliability. Thanks to the continuing advances in control hardware and software, programming and using these weird looking machines will offer much higher performance but present no unreasonable burden in programming or operation.

A 3-axis micromachine that, when manufactured on a silicon chip with flexures for joints, requires no assembly. The resulting machine would be the size of the period at the end of this sentence. The author holds the patent.

We currently bring the part to the machine. With the advent of rapid design of specialized machines, we will see smaller mobile machines, increasingly autonomous, wandering around performing their fabrication duties on parts where they lie.

A little farther into the future we will have macro, micro, and even nano fabrication machines. Consider the work currently performed by an ant colony. It's a form of parallel processing: each ant can't do much, but given a few thousand of them...Artificial ants with mandibles capable of removing or adding a speck of metal will swarm over a chunk of stuff, creating a part. A number of technical breakthroughs, and even a little new fundamental science­concerning items such as swarm intelligence, miniaturization, actuation, position sensing, and energy capture/storage/transmission­are required, but none of the issues is fundamentally unsolvable and the result may be commercially compelling.

Service and repair are all about uptime. Over the past half century, machines have become operationally more brittle. The old machines exhibited graceful degradation; they would slowly lose tolerance, make strange noises, and wear out. Modern machines are more likely to just suddenly stop working. Just picture the difference in failure modes between an old oil-filled 2500 rpm spindle and a new 25,000 rpm spindle with critically metered lubrication. Future machines will continue this trend, for two reasons. As the physics of the machine become more extreme, typically as a result of high performance issues like speed and acceleration, when something goes wrong it often leads to catastrophic failure. The other reason this trend will continue is because of increased sensing and monitoring of machine behavior. When something appears suspect, we automatically stop the machine. This is both to prevent damage, and because our process criteria continue to get tighter. A machine that has quit productive activity needs to get back online as quickly as possible. We can expect to see less time spent on field repair, and more fast module swapping, both electrical and mechanical. With increasingly portable machines, it will become more common to simply replace the whole machine with a spare. Real time and remote diagnostics will become the standard, and advances in prognostics­monitoring and analysis to predict maintenance requirements­will attempt to keep pace with the increasing operational brittleness of the individual physical systems.

Running our shops will continue to become more challenging. JIT, zero inventory, mass customization, immediate response to design changes, complete subsystem supply, and so forth will increasingly make our business lives exciting. Just as the individual machines have become more brittle, some of these demands tend to make the supplier/customer system more fragile. We will need to run our businesses in a more resilient, robust, and responsive manner. Much of this has to do with how we use our resources. When we are making many different parts in varying quantities­all subject to change at a moment's notice while minimizing incoming and outgoing inventories­it is difficult to achieve the best possible productivity in our shops.

The people who conduct operations research study the efficiency with which we can run our shops. They long ago discovered that certain challenges, such as how to schedule mixed and changing production, are often extremely difficult or impossible to optimize.

Because the best possible solution to these problems cannot be easily found, we spend a lot of time and anxiety fighting operational fires when a machine goes down, an order changes, or the delivery of input materials is late. Considerable relief may be on the horizon, courtesy of a new style of computer program known as Evolutionary Computing. You may have run across this under the names of Genetic Algorithms, Genetic Programming, or in Europe as Evolutionary Strategies. But whatever the name, these approaches represent a fundamentally new way of handling these intractable operations challenges.

This technology has been simmering in the university labs for quite a while, but is now emerging in real world applications. As we speak, Deere & Company has Evolutionary Computing-based software actively scheduling and running flexible assembly production lines in a number of plants. They have seen notable improvement in key business metrics such as shop productivity. Evolutionary Computing algorithms do not guarantee the perfect solution, but when properly applied offer better results than can be achieved by our current fire-fighting approach. You have to simply trust the system and stop trying to involve yourself in the minute-by-minute decision making process. This relatively passive role for us in the midst of potential crisis goes against both our training and nature. And it is this very nature which impedes change. It's embedded in our manufacturing culture.

For an improvement in our shape-making art to be adopted it must be scientifically possible, technologically feasible, and economically compelling. Perhaps the most daunting factor is that these improvements must be culturally palatable. Giving up any degree of personal active control (however illusory that control may be) in the midst of a seeming production crisis may be culturally unacceptable to you. If this is the case, you will not allow this sort of Evolutionary Computing to control your shop. This effect is generally referred to as cultural inertia or culture lag, and is blamed by the technologically enthusiastic for the slow pace of change.

However, this hesitancy is part of a valuable system of checks and balances. When new technology becomes available, there are the early adopters, the pack, and the late adopters. Each group plays an important role, and this cultural sieve plays a critical part in our ability to speculate, but not predict, what the future will bring. Consequently what is possible is not necessarily probable, but the opportunities for improving the way we do things are boundless. We must each evaluate the risks and opportunities, and seize that which will allow us to do a better job of making stuff into things. As a group we will create the future.

[Paul Sheldon is an independent R&D consultant working at the intersection of science and business. He enjoys helping people figure out better ways to make stuff into things, and specializes in the development and management of new product design. Email to paul.sheldon@sheldonworks.com

Originally published in the April 2000 issue
of Tooling & Production.
Please Note: some pictures or diagrams are only available through the printed media.