Masters of Manufacturing James B. Bryan
This is the sixth annual installment in an article series we call Masters of Manufacturing. In these articles, Manufacturing Engineering magazine honors a distinguished figure in manufacturing technology. By doing so, we hope to remind readers that a career of great achievement in manufacturing is still possible.
By Jim Lorincz
James B. Bryan has been called "the founding father of modern precision engineering". He was honored by the European Society for Precision Engineering (euspen) in 2000 for "his tireless promotion of precision engineering philosophies, principles, innovations, practices, and standards through design, research, and teaching." A registered mechanical engineer in the State of California since 1954 (No. 10970, he proudly notes), Bryan began working at Lawrence Livermore National Laboratory (LLNL) in 1955, under Nobel laureate and founding director, Ernest O. Lawrence. By the time he retired in 1986, Bryan was chief of metrology with contributions to precision engineering so diverse that one colleague has commented that picking the most notable one would depend upon whom you asked. Another colleague (Ben Taylor of Renishaw) called Jim "the world's most practical purist."
While at Lawrence Livermore, Bryan studied the principles that enabled him and his team to design and build an ultraprecise 120-ton (190 t), 96" (2-4-m) horizontal diamond turning machine and, as a result, advanced engineers' understanding of how to control variables in automatic CNC machining. Bryan was the recipient of a US patent for the Telescoping Magnetic Ball Bar Test Gage in 1986. It was commercialized, principally by Renishaw Inc., under license, and is used all over the world to profile the accuracy of CNC machine tools. He has been author and co-author of more than 50 papers on precision machining and quality, and author and co-author of standards that have been developed by ASME for Roundness, Flatness, Temperature and Humidity, Axes of Rotation, Surface Finish, NC milling machines (1992) and NC lathes and turning centers (2000). In 1977, he received the Society of Manufacturing Engineers' SME Research Medal for advancing the science of measurement through his innovative work in precision engineering. Since retiring, Bryan has headed his own consulting firm, Bryan Associates (Pleasanton, CA), and in 2006 was awarded a patent for the Slow Tool Servo (as compared to the Fast Tool Servo).
Manufacturing Engineering: What in your early experiences prepared you for a career in precision engineering and metrology?
James B. Bryan: I grew up in Alameda, CA, across San Francisco Bay from San Francisco and have lived in the area my whole life. In 1944, I shipped out as an able-bodied seaman on a US Merchant Marine ship, which meant that at the age of seventeen I was able to pass a test on seamanship principles that I learned in the Alameda Sea Scouts. In the Merchant Marine, I was a wiper, oiler, Junior Engineer, Second Assistant Engineer Steam Vessels, and Third Engineer Diesel Vessels. As a machinist in the engine room, I became adept at making metal parts on a lathe for repairing and rebuilding valves, pumps, and anything else needed to keep that ship's engines running. After the war, I shipped out on merchant ships during the summers, and earned my degree in industrial engineering from the University of California at Berkeley in 1951. Before I joined the Lawrence Livermore Laboratory in 1955, I had no experience or training in precision engineering.
I've had a sailboat since I was 11 years old and have had my 30' (9-m) sailboat, Venga, for the last 34 years. In Spanish Venga means "come along," and it wasn't uncommon for my wife, Edie, son, Bill, and daughter Susan and me to do just that and sail Venga on San Francisco Bay and San Pablo Bay all the way up to the Sacramento-San Joaquin River Delta, a trip of more than two hundred miles, for two weeks at a time.
ME: What was it like working at Lawrence Livermore National Laboratory?
Bryan: Lawrence Livermore National Laboratory under Ernest O. Lawrence was a place of challenge to the imagination and curiosity of its scientists and engineers. Though the primary work of the lab was physics, specifically nuclear physics, it was engineering that moved theory into the development of practical nuclear weapons. The laboratory's missions have broadened today to include prominent roles with DOE in energy, science, and the environment.
Early in my career at Livermore, I was challenged to improve the quality of our machined parts by a factor of ten, as well as our ability to measure such close tolerances. That level of tolerances was almost unknown at the time. At first, I didn't believe it was possible. I then found a book written in the 1920s by Dr. R.H. Rolt—head of the Metrology Division of NPL in England—about measurement techniques and theory that, like many of the ideas from the past, have great relevance to solving today's continuing quality manufacturing and metrology problems. One of the topics in the book was the measurement of gage blocks. Gage blocks were invented and patented by C.E. Johansson in Sweden in 1901. They continue to be the basis for gaging techniques that are the foundation of modern Precision Engineering.
ME: You say you aren't a researcher, yet your curiosity led you to find books on this subject and talk to the men behind them, and you're quick to credit them for their help in making possible your own insights into precision engineering.
Bryan: In one sense, I began interviewing them out of curiosity, and soon realized that you shouldn't reinvent the wheel every time you're looking for a solution. The knowledge of the past is far more relevant to the present than most people seem to think. In the case of Rolt, I did arrange several interviews with him after his retirement from NPL. I also visited Richard Moore [founder of the Moore Tool Co.] and John Loxham of Cranfield Precision, among many others, to consider realistic ways of achieving precision in machining and measurement. Moore had pioneered the use of hardened slideways and advanced metrology in his highly precise jig-boring and measuring machines.
ME: How did these insights shape your understanding—and that of like-minded individuals—of the nature of automatic manufacturing processes?
Bryan: John Loxham, founder of the Cranfield Unit for Precision Engineering, was an astronomer, as well as an educator, precision engineer, businessman, and a pioneer in understanding thermal effects on manufacturing systems. He deserves the credit for formally introducing the Deterministic Point of View to manufacturing engineers. This view states that automatic machine tools and measuring machines are perfectly repeatable, just like the stars and the planets. He contended that there is no known error in any natural law, and that industry has failed to recognize and benefit from these very reliable, 100% perfect laws, as they operate in the engineering workshop.
As a result, automatic machine tools and measuring machines obey cause and effect relationships that are within our ability to understand and affordably control. There is nothing random or probabilistic about their behavior. Everything happens for a reason, and the list of reasons is small enough to manage by common sense, good metrology, and a reasonable investment of resources. A determinist believes that all of the nonrepeatability that may be observed in the performance of automatic machines is caused by systemic sources, and that there is no such thing as random behavior of an automatic machine.
Determinists use the term "apparent nonrepeatability" to emphasize this belief, because each of the sources of apparent nonrepeatability is itself repeatable, if examined closely enough. The magnitude of apparent nonrepeatability depends on the time, money, and skill of the user in creating the proper environment for the machine. Statistical measures of apparent nonrepeatability have limited usefulness, because they cannot be used to predict future machine performance if the time, money, and skill of the user change. On-line monitoring and control of the variables, such as temperature, are a better alternative. Statistical measures can be counterproductive in somehow implying that nonrepeatability is the fault of the machine rather than the user. A determinist believes that any automatic machine can be made to repeat to a value that is close to its resolution, and that the cost of doing so is not unreasonable when compared to the benefits.
ME: What are the practical ways in which engineers can address this need for a suitable arrangement of mechanical systems?
Bryan: In the days before numerical control machines, performance depended on making a part, checking it for defects, and adjusting the machine to make it as close to spec tolerance as possible. The variables were numerous. They included the variations in skill of the machine operators, and others that were yet to be fully appreciated, such as thermal effects, both internal and external, and vibration. These questions were so basic that they had to be answered before today's precision measuring devices and manufacturing processes could emerge.
My quest began as a simple challenge, early in my career at Livermore, to improve the ability to machine to increasingly closer tolerances. I observed an amazing degree of repeatability in the behavior of the machines that I tested. In the beginning, it was short-term repeatability. But as my colleagues and I learned to control temperature, it became long-term repeatability as well. The first machines tested were tracer lathes, followed by NC lathes, NC measuring machines, and diamond turning machines.
ME: In the late 1980s, you and your LLNL team designed and built a 96" (2.4-m) diam diamond horizontal turning machine that took eight years to build and weighed 120 tons. How did it exemplify the ways in which you were able to control the variables in the machine system?
Bryan: The Portas Principle [named after its originator Jeff Portas, managing director of Cranfield Precision Industries at the time] enunciated one day at lunch with John Loxham and myself, states that: "Random results are the consequence of random procedures." The Portas Principle says everything that needs to be said if interpreted broadly enough. The following list indicates the typical sources of apparent nonrepeatability of machine tools operating at zero load. It is a list with a finite length. When these sources are brought under control, the machine becomes deterministic and does exactly what you tell it to do. The sources include: random procedures; thermal effects; hysteresis from loose joints, drag of cables, and hoses; vibration; inadequate resolution; dirt; Coulomb friction; sliding oil films; change of position of rolling elements; and variations in the supply of electricity, compressed air, vacuum, and electrical ground loops. If we include the repeatability of the cutting process, the list is extended by the addition of tool wear, tool geometry, tool deflection, material variability, and workpiece clamping.
ME: What happens when you control these sources of apparent non-repeatability?
Bryan: The results can be dramatic, as was seen in controlling the sources of nonrepeatability in diamond turning a 2" (50.8-mm) diam X-ray microscope to a tolerance of ±1.5 µin. using a Moore machine whose resolution was 1µin.
Random procedures must be corrected as the first step in controlling apparent nonrepeatability. Humans are involved in this step, because they are responsible for setting up the automatic machine. Controlling the behavior of humans can be done by using carefully written procedures, and by performing critical tests to determine that each step in the setup has been correctly performed.
Thermal effects are by far the largest source of apparent nonrepeatability after random procedures have been corrected. The effect of temperature deserves to be re-visited continually, and factored into the design and control of any automatic system. To bring thermal effects under control in 1972, we built a complete box around the machine, which was designated DL-1, and used 40 gpm (151.4 L/min) of temperature-controlled oil held to ±0.01°F (0.006°C) to shower the entire machine. The drift of the structural loop between the tool and the workpiece was held to 1 µin. (0.02 µm) for weeks at a time. Cost of the oil shower system temperature control system was approximately $6000 including the pump, the chilled water-to-oil heat exchanger, the Yellow Springs on-off control system, the filters, the solenoid valves and the piping. This system reduced thermal errors from the largest single source of error to the least.
Vibration was controlled by using self-leveling laminar flow, pneumatic vibration isolators. The problem of dirt was solved by super-fine filtering of the oil shower. These filters are able to filter the oil so that it is cleaner after a few days in the machine than it is brand new. Variations in utility supplies were solved by using commercial pressure regulators and power supplies.
Diamond turning machine number two—DL-2—used a Moore roller way, continuous-path jig grinding machine for its X, Z slideway system. Twenty-pound lead weights eliminate backlash by pulling the X and Z slideway nuts against the screw. A 1000-count encoder mounted to a 100:1 gearbox on the end of the screw gives a resolution of 1 µin. The GE hardwired control system did not know that we fooled it into thinking the feedback pulses were still a tenth of a thousandth. [0.0001"]. The jig grinding machine was designed by Dick Moore and his brother, Bob Moore, more than 50 years ago.
This machine (DL-2) was able to repeatably diamond-bore the same contour on an electroless nickel-plated Xray microscope within 1.5 µin. [0.03 µm]. This repeatability was achieved with different operators, on different shifts, on different days of the week, making new setups for every trial.
ME: How well have some of your concepts been received?
Bryan: It is gratifying to see that the ball bar is used almost universally as a way of testing the accuracy of machining centers, turning machines, and CMMs, and has been accepted as the basis of an ASME standard. In fact, I've been told that every Haas Automation machine goes out of the plant with a ball bar chart.
LLNL's work in developing methods to test the performance of spindles laid the ground work for development of the American standard ANSIB. 89.3.4 "Axes of Rotation–Methods for Specifying and Testing" and ISO-230-7-2006 "Geometric Accuracy of Axes of Rotation."
ME: How have these advances in precision engineering affected the quality of our manufactured products today?
Bryan: The benefits of precision engineering to society are incalculable, and examples are to be found in manufactured consumer products everywhere. I like to cite the example of small model airplane engines. They are machined with such precision, within tenths, that they don't require piston rings. Computer drives, ball point pens, molds for virtually every kind of plastic part, and even pull-top cans are all precision-engineered. Think about it, the pull-top of a soft drink can requires a score at the point of separation that is strong enough to tolerate the beverage's pressure (higher for soft drinks than beer) in shipping, and thin enough to be opened with the pull of a finger. There are so many examples. All you have to do is look at the telecommunications industry and the medical-device industry, and all the products that didn't even exist when we began developing our ideas, and the continuing miniaturization and precision that is required in all manufacturing industries.
This article was first published in the July 2007 edition of Manufacturing Engineering magazine.