The School of Hard NOx
What: Public utility JEA used neural network technology to create an artificial intelligence system it implemented last fall. The system automatically determines the optimal combinations of oil and natural gas the utility's boilers need to cost-effectively produce electricity given fuel prices and the amount of electricity required. It also ensures that the amount of nitrous oxide (NOx) emitted during the generation process does not exceed government regulations.
Why: JEA needed to decrease operating expenses, in particular fuel costs, as oil and gas prices began their precipitous ascent in 2002. Forty percent of JEA's $US1.3 billion budget goes to the purchase of oil and gas to power its boilers, so a small change in the way electricity is produced could send millions of dollars to the bottom line.
Technology: Neural network technology models the process of producing electricity. Optimization software from NeuCo determines the right combinations of oil and gas to produce electricity at low cost while minimizing emissions.
Cool quotient: JEA, which serves over 360,000 customers in Jacksonville and three neighboring Florida counties, is the first utility in the world to apply neural network technology to the production of electricity in circulating fluidized bed boilers. It built a system that makes decisions based on historical operating data and as many as 100 inputs associated with the combustion process, including air flows and megawatt outputs. The system learns which fuel combinations are optimal by making adjustments to the boiler in real-time; it also forecasts what to do in the future based on specific fuel cost assumptions. "We had issues with oil prices. At the same time, gas prices went from $US4 a BTU to over $US14. We need to use gas because it decreases emissions. This solution helped us balance all of those items," says Wanyonyi Kendrick, JEA's CIO.
ROI: The project, which IT drove, cost $800,000 and paid for itself in eight weeks. The system reduced the quantity of natural gas that's used to control NOx emissions by 15 percent, an estimated annual savings of $4.8 million. With natural gas prices at $11 per BTU, JEA expects to save $US13 million on fuel in 2006. What's more, JEA has discovered it can use the new technology applications for its water business.
Where the rubber meets the road
What: Working with Sandia National Labs, Goodyear's IT department developed software to design and test tires virtually. In the past, the company built physical prototypes and tested them by driving thousands of miles on tracks. Using a mathematical model, the software simulates tire behavior in different driving conditions so the designer can see how the tire gets pushed, pulled and stretched as it rolls down a road, hits bumps, turns corners, screeches to a halt, and grips the road in wet, dry and icy conditions.
Why: Accurately testing prototypes in the physical world requires driving on the tires for thousands of miles; Goodyear wanted to shorten that time to get its products to market more quickly. Three research and development employees advanced the idea of testing prototypes using computer simulations, which could do the job faster. The company had never done simulations but figured initial investments and subsequent maintenance costs were worth the payoff. Goodyear's cost of goods sold as a percent of its sales decreased by 2.6 percent from 2003 to 2004, the year its first fully simulated tires hit the market. Meanwhile, the R&D budget for tire testing and design decreased by 25 percent.
Technology: Custom-built software runs in a massively parallel computing environment, on hundreds of processors on hundreds of Linux computers. Goodyear invested more than $US6 million to build this high-powered computing environment. It plans to expand and upgrade its Linux clusters to meet business demands for new tires and to improve the fidelity of its virtual tests.
Cool quotient: Goodyear believes it is the first tire maker to use computers to design and test its wheels. Although the auto industry has done computer-assisted design work since the 1980s, the technology had not been applied to tires because their malleable materials made simulation difficult. The math required in order to model a tire on a computer is as complicated as that needed to simulate a nuclear explosion, which is why Goodyear partnered with weapons researcher Sandia. "This is freakingly cool stuff. It's way more power than any IT guy should have, and we have it! Ha-ha-ha-ha!" fiendishly laughs CIO Wernet.
ROI: Designers can perform 10 times more tests, reducing a new tire's time to market from two years to as little as nine months. Goodyear attributes its sales growth from $US15 billion in 2003 to $US20 billion in 2005 to new products introduced as a result of this change.
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.