The past decade has brought much change to IT, process control and automation. Despite the connection of ERP systems with MES and other plant-floor IT, and the fact that most controllers today are PC-based, the long-awaited “convergence” doesn’t yet apply to IT, control and manufacturing groups at many pharmaceutical companies.
This situation may be about to change, according to "Come Together," a white paper published last year by Rockwell Automation. A detailed survey of 22 IT, automation and manufacturing professionals found that senior management was driving IT and automation connections. It also found that companies were generally at one of five stages along the path to convergence, most of them at the earliest. An informal survey of 17 Pharmaceutical Manufacturing readers working in IT and automation reflects a clear disconnect between the two groups.
"Computer science folks do think very differently from engineering groups," writes one reader. “They have different risk models and sets of regulations that they need to satisfy.” Says another, "Goals and objectives are not aligned." The result: hard-to- integrate IT and control systems. “Information systems need to support the pharmaceutical product lifecycle,” says Anjali Kataria, cofounder of Conformia, Inc. (Sunnyvale, Calif.) "Companies gain by developing IT that will support ICH Q8, 9 and 10." Conformia is developing a series of workshops this year that aim to bring various IT and automation groups together.
In general, pharma IT is primarily driven by activities such as sales, purchasing and inventory control, explains Gawayne Mahboubian-Jones, product development manager with UK-based Optimal Industrial Automation, LLP. Automation tends to occur in isolated islands; since it doesn’t provide information that is useful to the higher level systems, no attempt is made to link them.
There are also cultural issues. "As an integration company working from the automation to the IT layer, Optimal encounters these every day," Mahboubian-Jones explains. "Purchasing decisions are often made without even consulting the manufacturing people. The comments we hear about why manufacturing isn’t involved — ‘They don’t understand,’ ‘The new system will change everything for them,’ or ‘It really only gets used by office staff’ — are all self-fulfilling prophecies and all lead to significant problems downstream."
Look under the hood
Pharmaceutical companies need to do due diligence when sourcing their IT and process control systems. "So much really depends on the back-end architecture. Companies need to look under the hood," says Kataria. "Too many pharmaceutical companies don’t think long term. They spend significant amounts to maintain customized software." Reader respondents to Pharmaceutical Manufacturing’s survey noted that closedness was more of a problem for IT than for automation, but the degree of openness varies with each company and facility.
One pharmaceutical company, for instance, may use a data historian in a way that’s not recommended by the manufacturer and face absolute chaos if the system files crash, says Mahboubian-Jones, where another may use open-standard historians that allow easy access to data for all outsiders. "The core problem is lack of understanding within the industry of the interconnectedness of decisions they make when purchasing systems, combined with assuming that the cheapest system that apparently meets a specification is the best deal," Mahboubian-Jones explains.
It’s an IT world
Broadly, control engineers complain that the IT required to operate their systems is still based on systems that were designed for use in the business and financial world. "What’s the most widely used computer programming language in the world? COBOL," says Scott Sommer, automation technology manager for Jacobs Engineering Services, Pasadena, Calif. In the same spirit, he adds, more plant control systems are built to HART than to Fieldbus or Profibus standards.
Even the OPC initiative is designed to work on the Windows operating system, which has a somewhat fragile data communication protocol, Sommer indicates. "If automation continues to be beholden to business systems, we will have a problem," he argues. Today, the most integrated drug companies tend to be the better biotechs, followed by API manufacturers, due in part to the types of systems they have chosen to use, according to Mahboubian-Jones. Lack of integration has many sources. Custom and culture are at the root of the problem. But IT and control vendors may be unknowingly perpetuating the disconnect. Sommer informally surveyed automation engineers, half of them from the pharmaceutical industry, for their views on the "ideal" next generation of control products.
He will present his findings later this month at Interphex and World Batch Forum. Among them:
- Disparate data are kept in too many places
- There’s been a swing back to proprietary systems
- There is a preponderance of PCs, or "too many boxes." Some firms now have twice as many PC servers for their control system as they have actual controllers.
- Insufficient data management capabilities are available at the instrument level
- There is a need for more network health reporting and capacity utilization tools
- Users want simpler, more configurable solutions, for example, systems that would allow some data to be stored directly from a PLC.
- Some users want greater MES functionality in control systems and controllers themselves
- There’s a need for simpler validation of equipment
Mergers often leave drug companies with legacy IT and automation systems. Pharmaceutical companies may also invest in expensive systems, such as MES, only to find that the vendor discontinues them a few years later and moves to a new platform, so they’re not only dealing with closed systems but with obsolete ones as well. "This makes planned upgrading and replacement very difficult," says Mahboubian-Jones. "In addition, all new systems require significant extra work to be made compatible with legacy systems."