Consider the variables needed to produce a gallon of gasoline. Thousands of factors across the supply chain need to be weighed. Crude oil from different regions of the world comes in a variety of grades. Each grade includes different elements. Crudes have different prices. Refineries are configured to handle various types of crude to make different products. With prices of crude oil and natural gas constantly in flux, calculating equations based on all these variables simultaneously is beyond the abilities of any human brain.
To meet this challenge, Chevron engineers designed a remarkable system called Petro. Analysts trained in running the software work in refineries and next to traders in each of Chevron's trading hubs in Houston, London, San Ramon and Singapore. Petro considers a long list of dynamic variables and helps the analyst advise traders on whether a particular bid makes sense at that time. And Petro helps analysts decide which processing equipment will best take advantage of every part of the barrel of oil.
Chevron was the first to develop this type of technology and, in the 1980s, to move the program off mainframe computers and onto personal computers. Since then, Petro's challenge has grown to be as broad and varied as Chevron Downstream—product refining, transportation, trading and marketing—itself.
The Supply and Demand Equation
The crude-to-customer supply chain begins in oil fields around the world. Crude oil, some of which is Chevron's production, goes to Chevron refineries where it is made into a spectrum of products ranging from gasoline, diesel and jet fuel to base oils that are later blended into lubricants, additives and other products. Much of the refineries' output is sent to terminals, where it is then sent on to industrial customers and retail outlets. Natural gas, also purchased on the market, is needed to fire refinery furnaces and produce hydrogen for refining.
Petro allows analysts to explore more planning alternatives, arrive at the best recommendations and do all that faster than with other refinery planning technologies.
For good results, Petro requires good data—provided courtesy of Chevron's testing labs. Chevron's refineries accept crude of different types from all over the world—about three-quarters of it from oil fields outside the company. Raw material from each field has a different molecular makeup. And as the field matures, the chemical composition can change. So every day, samples arrive at Chevron's labs for testing.
Petro processes 400 to 500 variables associated with crude oil characterization, along with thousands of refinery-capability variables, to arrive at the best solution—and all within a few seconds. Planners analyze the results before recommending a course of action.
“Ultimately, it's people who make Petro work. As a tool, Petro is only as good as the analysts who use it.”
Process experts at each refinery are constantly updating Petro to reflect changes in crude oil properties, the refinery's processing capabilities, and government and industry specifications. Typically, these analysts are chemical engineers with about 10 years' experience—usually in refining—who have a strong understanding of the manufacturing business.
Petro's Scenarios
Factoring in market prices for product is important as well. "The market is dynamic for natural gas, gasoline and crude oil," said analysis team leader Mark Davis. "Petro helps you shift your operations to capture the best economic point, whereas if you operated under an accepted paradigm, you'd miss opportunities or be slow to capture them."
Analysts also use Petro to see what happens when key variables change. Recently, Ted Kutz, Chevron's refinery optimization expert and Petro's chief architect, did just that. In a work session with California's Air Resources Board staff, Kutz presented scenarios based on different fuel mixtures that the agency was considering for future regulations. He explained how certain fuel recipes could result in lower gasoline production and higher operating costs, which could lead to higher consumer prices. Based on Petro's report, Kutz showed how other formulations had less impact on supply and cost while meeting the board's environmental targets.
In the early days, one analyst would spend a half-hour setting up a question and wait two or three hours for a paper report an inch and a half thick. Only then could the analyst mull over the results. Now, the program is used by up to 200 people at once while processing as many as 10,000 equations. And thanks to ever-increasing computing power, it takes only three minutes to set up a report and two or three seconds to see the results on screen. No longer data entry clerks, analysts now spend most of their time actually analyzing.
There's also an element of social networking to Petro. The ability to quickly share information has allowed for more collaboration.
Ultimately, it's the people who make Petro work. As a tool, Petro is only as good as the analysts who use it.
Kutz used an analogy: "In boxing, a sparring partner helps train the boxer, and the same goes for Petro," he said. When looking at a Petro report, "a good analyst will ask himself, 'Did Petro confirm my expectation? If not, why not?'" According to Kutz, "Petro supports the premise from a good analyst 80 to 90 percent of the time. But sometimes, nuances reveal something unexpected and point to a better course of action. It's one way we're using technology to differentiate Chevron's performance."