Process optimizing

One of the bigger challenges facing manufacturing companies is shipping products on time, but typically once products are shipped, the manufacturer’s job is done. That’s not the case at Equistar Chemicals LP, a petrochemical company that converts chemicals into pellets used to make plastic bags and other products. In polymer processing, lot-to-lot consistency is important, and customers often need to know the exact conditions under which a rail-size shipment of plastic pellets was manufactured.

Equistar’s complex reactor scheduling and production-level process historian used to run on legacy systems at headquarters in Houston. On a monthly basis, officials at Equistar’s Matagorda manufacturing plant, 100 miles southwest of Houston, would have to manually reconcile legacy data about raw materials and actual products – a process that took approximately three days each month to complete, according to Mark Tibbitts, Equistar’s manager of control systems and electrical engineering.

“The problem was really to … get information into the hands of the people who needed that information to take action on the business process,” Tibbitts says.

Equistar did an analysis of the types of software required by the process industry and decided to deploy manufacturing optimization software from Aspen Technology, along with a planned implementation of SAP’s ERP suite, all running on a Windows NT infrastructure. The solution began as a pilot in 1999 and was rolled out in 18 months.

The technology has saved the company US$20 million – or five times its implementation costs – in the past two years by optimizing production at the Matagorda plant.

“[Plantelligence] enables everyone from plant operations to the business team to make more informed decisions based on accurate and timely information,” Tibbitts says. “Having this information enables them to work more efficiently and frees them up to focus on other areas of the business.” Tibbitts says the reconciliation of material balances alone, which used to take three days, has been reduced to a few hours.

In addition to allowing the Matagorda plant to schedule and track production runs and operate complex instrumentation and data analysis on its four lines, Plantelligence gathers information and provides SAP with data in real time.

“[At one time] the production planners would have to call the plant so the logistics people would know what rail cars to schedule,” Tibbitts says. “Now they report on a daily basis to SAP what they made.”

Plantelligence tracks 15,000 data items every 15 seconds in the Texas plant. This historical data interacts with production data in a relational database and presents the combined information in a usable format so that officials can track details about the railroad-car-size lots they manufacture.

“It is very important to track the genealogy of the product, … what product is going into what rail car,” Tibbitts says. “That’s all dollars in terms of how much less waste we create. [Plantelligence] allows us to be much more agile and to respond very quickly to customer requirements.”

Plantelligence also ties in directly to the instrumentation in the plant so that it can quickly move the plant from one set of conditions to another to make a different product.

Equistar found that focusing on change management was one of the major keys to successful deployment. The company alerted employees to the potential economic benefits that the new software could bring to the company and then tapped representatives from the functional teams that would be most affected to work full-time both on the deployment and as “advocates” of the new system to their colleagues. Officials also audited the system after the rollout to highlight the economic benefits.

In addition, Aspen Technology had business-process coaching classes, presented business problems that people at Equistar could identify with, and worked to solve the problems within the context of the new system, says Scott Breedlove, a director at Aspen Technology in Cambridge, Mass.

As part of its focus on molding the technology to provide business value, Tibbitts says the team never wavered from the critical requirements – more than 130 in all – that it had defined during the planning stages of the process.

“It’s easy to let the scope [of the project] creep, to put ‘nice-to-haves’ in there,” Breedlove says. “It requires discipline on both sides not to do that.”

Equistar is now planning to roll out the Plantelligence solution to several other manufacturing facilities during the next year.

“We expect that a full implementation of Plantelligence will enable us to increase the speed at which we make business, technological, and operating decisions,” Tibbitts says.