Masters of Manufacturing: Patrick J. Hanratty
This is the ninth annual installment in a series we call Masters of Manufacturing. In these articles, we honor a distinguished figure in manufacturing technology, and by doing so, we hope to remind readers that a career of great achievement in manufacturing is still possible.
By Patrick Waurzyniak
An early pioneer in the fields of NC and CAD/CAM software, Patrick J. Hanratty, PhD, discovered his passion for computing and programming almost by accident, answering a newspaper ad seeking programmers in his hometown of San Diego after returning from service in the Air Force during the Korean War. After being hired at General Dynamic's Convair Division, Hanratty picked up programming quickly, heading up a class on the topic without any formal training. Often lauded as "The Father of CAD/CAM" for his later inventions of CAD/CAM software programs, while working at General Electric in 1957, Hanratty developed PRONTO, or Program for Numerical Tooling, an early NC programming software, along with MTD, the Machine Tool Director, the first two NC programs for machine tools used in production manufacturing.
In 1962, Hanratty left GE to join General Motors Corp. (Detroit), where he was part of a team at the GM Research Laboratories, located at the automaker's Tech Center (Warren, MI). At GM, Hanratty developed graphics for GM's DAC (Design Augmented by Computers) project while working on experimental machining, solving problems involving three, four, and five-axis surface machining. After leaving GM, in 1970 Hanratty started Integrated Computer Systems (ICS) where he and his team developed a CAD/CAM system for Redcor minicomputers using TPL (The Programming Language). While highly regarded, the ICS INTERAPT software didn't become a commercial success, leading Hanratty to found another company, Manufacturing Consulting Services Inc. (MCS) in 1971 that developed a new machine-independent software called ADAM (Automated Drafting and Machining), which at that time was said to be the only commercially available drafting and machining system. MCS went on to develop more mechanical CAD and manufacturing software, first with its AD-2000 package and later with the PC and UNIX-based ANVIL CAD/CAM systems.
Today, Hanratty, 79, continues working on programming problems related to CAD/CAM and manufacturing. In particular, he sees no real innovation in CAD/CAM since the debut of Parametric Technology Corp.'s (Needham, MA) parametric Pro/Engineer CAD/CAM package back in 1988, and Hanratty forecasts trouble ahead for manufacturers. He contends that current and future CAD/CAM systems will be unable to provide the precision required to accurately create the components required by mission-critical aerospace/defense applications. In an exclusive interview with Manufacturing Engineering, Hanratty discussed his career in manufacturing research and the CAD/CAM world.
Manufacturing Engineering: Tell me about your early years. How did you get started in programming?
Patrick Hanratty: At the very start, when I came out of the service, I had no idea what I was going to do, since I was trained to be a singer before that. My vocal chords and lungs were scarred as the result of a B-29 crash I survived. I had to find another profession. One of the classes I had to take, as a remote-control turret gunner, was a computer course. Computers in those days meant analog computers. When I saw the ad in the San Diego newspaper 'Wanted, Computer Programmers' at Convair, General Dynamics' division in San Diego, I decided I'd give it a chance. I was lucky enough to get hired as a programmer trainee, and sat in a class which, at the end of three days, I was asked to take over. The IBM instructor teaching the course had all the manuals, but she had no idea what programming was conceptually, and I had an idea almost instantly.
Here I am at Convair, and the very first thing I had to do was the section properties—bending stresses and shear flow for a multi-celled aircraft. The program was successful, but it was pretty interesting because initially I had had no mathematics for the project, but fortunately I was a quick study. This was 1954, 1955. My next assignment was the program for doing field-length takeoff calculations—when an engine goes out on a plane during takeoff, the pilot must determine if takeoff can continue or be aborted. So this was my experience in programming. The IBM 650 was the first machine I worked on. And, in parallel with that, I learned how to program the Univac Scientific 1103-A from Sperry Rand. I loved programming that machine—it was like a toy!
ME: Where did you go next?
Hanratty: A sequence of things made me kind of successful when I joined General Electric, which ran an ad in the paper saying they were starting a computer company in Phoenix. I applied for a job, and in January of '57 I was hired as the seventh employee here in Phoenix for GE's computer department. I had three big assignments out of GE. One was the IRMA project, the MICR magnetic character reader. What we had to do was come up with the character set that could be read in with heavy ink or light ink, fast or slow, or skewed a little bit, or crumpled. I wrote a program that would analyze each character set. That program was then run against each of the character sets that had been submitted by a large number of GE employees, and the character set, which is currently on your checks today, is the one that popped out as the best to use. It was accepted by the Bank of America, and several months later the American Banking Association accepted it as the standard for MICR characters on checks.
ME: When did you get involved in NC programming?
Hanratty: That would be '57–'58 when I developed PRONTO, Program for Numerical Tooling Operations. What I did was develop the instruction set on how to control a Milwaukee-Matic [Kearney & Trecker; West Allis, WI]. Then, the next assignment at GE was developing the MTD, the Machine Tool Director software; PRONTO was a 2½ axis, and the MTD was full three axis. To my knowledge, since APT wasn't in production at that point, I believe that was the first commercial pair of NC programs that existed.
ME: Those programs came before APT [Automatically Programmed Tool] by Doug Ross?
Hanratty: Doug Ross was the principal, as far as APT was concerned. It was a team headed up by Doug Ross, and the principal companies were Boeing and GE. There were altogether about eight or nine companies that worked very closely together. APT was started before the GE computer department was formed, and continued to grow over several years, but it was not a commercial product per se.
ME: What did PRONTO do for the Milwaukee-Matic?
Hanratty: It was a 2½-axis CAM system language that allowed programmers to produce the tool movements for machining parts. It generated a punch-paper tape to control the machine, and it won the Joseph Marie Jacquard Memorial Award [in 1982 from the Numerical Control Society].
ME: You're described as the Father of CAD/CAM. How did you get started in CAD?
Hanratty: When I went to GM, I was part of the DAC project, and I was responsible for both the machining output and any drafting output—our drafting machine consisted of a Giddings & Lewis [Fond du Lac, WI] milling machine. It had a big linoleum-covered table with the vacuum pumps behind it, with little holes, sheets of mylar to draw on, and with a ballpoint pen in the place of the tool. And that was, as far as I know, the first mechanical drafting capability. That was the first thing I did at GM in 1962.
ME: What did you do at GM that advanced part manufacturing?
Hanratty: OK, there are several things. One of the largest was the creation of a mathematical surface approximation so that a five-axis milling machine could be used to effectively make any shape. I had done some experimental machining at GM. The five-axis world was one that really interested me, and I came up with an algorithm, a way of simulating the sculpted surface with contiguous developable surfaces.
When the developable surface concept was presented, a mathematics professor at one of the universities GM consulted with told GM not to spend any research money on that approximation approach—it wasn't feasible. I was at a meeting with Fisher Body, which at that time was responsible for the die models for all cars; they had this one cube, and had no way of getting it turned into a die model. I said 'Hey, I can do that! I've been doing some experimental machining at Numac in Cleveland, with their five-axis milling machine. I know that I can get that deck lid done—I tell you what, I'll bet my job on it.' I had one assistant go with me to Numac. When we got there I found out all my surface normals were going in the wrong direction. I did a couple of tests and I reprogrammed the output, changing the normals—and finally got it working! We had to cut part of the toolchanger off from this gigantic machine, because the model was so big. I hadn't been to sleep for more than a total of three hours in the previous seven days, but we got the model finished. Afterwards, when we took it to the research center, the Fisher Body engineer measuring it said that it was the most accurate, finest model produced by General Motors Corp. in the 45 years he'd worked there. He couldn't measure any differences—it was a perfect model.
ME: You founded MCS in 1971, which developed ADAM for automated drafting, and the AD-2000 software package, which was said to cement your CAD/CAM reputation. Please tell us about this period in your career.
Hanratty: ADAM was the first MCS product. It was an integrated drafting and numerical control system. MCS licensed ADAM to Computervision [CADDS], Gerber Scientific [IDS 3], Tektronix [Teknicad], and United Computing [Unigraphics]. The second MCS product, four years after ADAM, was AD-2000. This moved MCS software from a 16-bit word only system to an integrated CAD/CAM system for 24-bit, 32-bit, 48-bit, and 60-bit computers. AD-2000 was licensed to Autotrol [AD380], Control Data [CD-2000], and Kongsberg [CDM-300]. NASA also licensed AD-2000, which they named NASA IPAD [Integrated Program for Aerospace Design]. MCS had many thousands of customers all over the world. Caterpillar tractor, for instance, bought the source code, so they could use it throughout their entire Caterpillar system. This happened with dozens of large companies worldwide. There isn't a company that I can think of that we didn't have at least a few seats in.
ME: Some observers say that 70% of 3-D mechanical CAD systems can be traced to your original code.
Hanratty: At one time, it was 100%. When we sold an ADAM license to United Computing, MCS and United Computing named it Unigraphics, because United Computing had a product called UniApt, and they wanted to have our package to go with that. So Unigraphics was purchased from MCS by United Computing, and of course, McDonnell-Douglas bought it and expanded it.
ME: You said that there has been no real innovation in CAD since PTC announced its parametric Pro/Engineer; why is that?
Hanratty: Parametric Technologies came out with Pro/Engineer in 1988, and that was the first heavily marketed parametric solid modeler. Before long, everyone was convinced that they had to have parametric solid modeling. Up to that time, the MCS solid modeler, OMNISOLIDS, was the strongest solid-modeling package in the world, and MCS won every competition when it came up for a comparison between it and other solid modelers. Unfortunately, OMNISOLIDS was not parametric, and we pulled it from the marketplace, and two years later, we put old OMNISOLIDS down!
ME: You're also concerned about current and future CAD systems' capability to meet accuracy required by precision manufacturing in aerospace/defense. Why?
Hanratty: The current design systems don't have enough accuracy because they are limited to 64-bit floating point in design calculations. There is 128-bit capability in both modern computers and modern programming languages available right now. If CAD companies want to move from the 20th century into today, they can produce a new system using today's technology, a very large addressing space from the 64-bit chips and 128-bit precision—quad precision—to accurately make the 21st century items that are in the minds of today's designer.
What we've done is exhaust what we have in the way of design resources, as far as our surface and solid-modeling world is concerned. Over the past two years, 2008 and 2009, I have talked with most, if not all, the software officers of the major CAD companies. They have all indicated that they are not planning to take the next step that is needed for future design and manufacturing tolerances.
What's needed is a CAD surface/solid modeling kernel that will function within all the major CAD and CAM systems, offering double-precision in and out, with all processing in quadruple precision, and an effectively 'unlimited' object size, for example, blending two free-form surfaces, each over 100-m long, within a micron tolerance.
This article was first published in the July 2010 edition of Manufacturing Engineering magazine.