ChE Computing Becomes Cyberinfrastructure

8/12   in the series We Are ChE: Entering a Golden Age
The third reason I foresee ChE entering a new Golden Age is its ever-increasing use of computing and communication networks as a ubiquitous "cyberinfrastructure." It powers both our understanding and our actions, shaping how we collect and use data and how we simulate, design, and predict processes and products.

Computers and more

Cyberinfrastructure has been with us for a long time, but thinking about it using that term has been just in the past ten years. "Infrastructure" refers to networks of structures and systems that underpin a society or economy, such as roads, water pipelines, the electrical grid, or telephone systems. Our daily lives and work are now underpinned by a cyberinfrastructure of computers, the Internet, portable devices, wireless and wired networks, software, videoconferencing, technical support, and professional "social-media" interactions. Computing wasn't part of ChE in the 1950s, which is often regarded as the previous golden age of ChE. Thomas J. Watson, head of IBM, apparently was misquoted as saying in the 1940s, "I think there is a world market for maybe five computers." Certainly at that point, computers were individual behemoths of vacuum tubes and patch wires. Scientists and engineers could nevertheless envision how such machines could automate tedious calculations. Getting access to computers was another matter, though, as companies could more easily see how computers would automate tedious financial record-keeping.

Shift to the present

From the start of the Space Age, technical advances started tumbling in. Semiconductor microelectronics brought miniaturization of processors and memory. Analog-to-digital and digital-to-analog converters linked the physical world and the computer, capturing vast amounts of data and signaling physical actions. Wired, fiber-optic, and wireless connections created fast networks for data exchange. Liquid-crystal displays and graphics processors brought spectacular visualization.

Software has been another key to this transformation. FORTRAN and other languages made computing logic and programming accessible. New algorithms made numerical methods and simulations practical. Hypertext mark-up language led to a World-Wide Web of shared information, collaborations, and commerce. The analytics field was created as statistical analyses became coupled with models to be used for discovery and decision-making. Two profound developments of recent years are parallel processing and portable computers that aren't perceived as computers at all. Parallel computing splits tasks to run on multiple processors at the same time, ideally(!) reducing the time for a calculation to the single-processor time divided by the number of processors. There's quite a potential for speed-up on 560,640 processors in Oak Ridge's Titan, the current fastest computer! Meanwhile, computers have become devices and computer codes have become apps. The evolution to minicomputers, then to desktop machines, and then to laptops took a dramatic turn when powerful computing and networking capabilities were built into smartphones and tablets. Wherever you are, by using a handheld device it has become possible to check and use your email, your experiment, your plant operations, your company's stock quotation, the news, the weather, movie trailers, and sports scores.

ChEs fit in

Warren Seider

For ChE, two landmark transitions into computing were in 1969. The book Applied Numerical Methods by Carnahan, Luther, and Wilkes (Wiley, 1969) opened the door for many ChEs. About the same time, a group of 15 academic and industrial ChEs formed the CACHE organization (Computing Aids for Chemical Engineers), headed by Warren Seider of Penn and Larry Evans of MIT. CACHE initially emphasized design of large-scale chemical processes, providing property databases, educational case studies, equation solvers, and educational access to the early process simulator FLOWTRAN. It later added molecular modeling and simulation, computational fluid dynamics, computational systems biology, and conferences focused on ChE computing. Two other important steps were in 1977, when AIChE created the Computing and Systems Technology Division and the journal Computers and Chemical Engineering began publication. An important byproduct was Evans' and Seider' leadership of the Advanced System for Process Engineering (ASPEN), funded in 1976 by the US Energy Research and Development Administration (later DOE) for design of fuel-conversion plants. This effort was later spun off as the process-simulation company AspenTech. A parallel international activity has been setting the CAPE-Open standards for computer-aided process engineering (CAPE). A key aspect is the CAPE-OPEN Laboratories Network (CO-LaN), a user-driven organization for testing and managing the standard.

A new era of simulation and data

For ChEs today, computing power has taken two directions, toward powerful simulations on one hand and powerful use of data on the other. The next post will reflect on these changes and on their potential to shape a new era of "Smart Manufacturing."


@PWstmrlnd's picture

ORNL Titan = 17.59 petaflop/s As a rough comparison, the iPhone 4, the 1994-vintage Pentium, and the 1984-vintage Cray X-MP all are about 20 megaflops. Mighty fast, yet Jack Dongarra has extrapolated the top end of the Top 500 supercomputers to 2020 for the first exaflop system (1000 petaflop/s).