Batch to Continuous

For the aforementioned brewery fermentation processes, loop control of four fermenter tank levels was a particular challenge. "In a typical process, a loop response time might range from a few seconds to a few hours, but this project often had loop cycles of 24 hours or more. Inventive advanced process control techniques were developed to deal with these significant time lags," says Coker. "These techniques consisted of using much smaller gains, and then adjusting the bias for the PID loop based on the differences in the control variable (CV) output to the valve for the upstream and downstream fermenter level controls. Since the flow cascades through the four fermenters to the beer well, the flow out should match the flow in. Flow metering was not available between fermenters, so the differences in the CVs were used to approximate flow and the loop was biased based on that difference."

More Precise Measurement

Batch processes can be paused when things go wrong, but continuous processes often cannot. This means that instrumentation needs to be more reliable, often through the use of redundant transmitters. (See "Continuous Processes Need Redundancy," at the end of this article  for more information on how redundancy increases reliability.

Most batch processes use off-line measurements made in a lab to control quality and production. With continuous processes, these measurements now need to be moved on-line. This presents numerous instrumentation challenges, up to and including the use of soft sensors that infer process parameters when no direct measurements are possible.

"A soft sensor is a mathematical model based on first principles and empirical data that can approximate measurements based on various process parameters," explains Ric Snyder, senior product manager at Pavilion Technologies (www.pavtech.com), a Rockwell Automation company.

Hron of Kraft Foods details other measurement problems that can arise with continuous control. "There can be problems when going from a single scale in a batching system to multiple flowmeters in a continuous system. Often, extra calibration work is required along with more data monitoring. Most seriously, continuous processes may need sensors that just don't exist."

Not all processes can be converted from batch to continuous, but for those that can the benefits are substantial. The automation challenges involved with making the switch are significant, but can be overcome. The batch manufacturers that make the switch for all or part of their processes will forge ahead, while those that remain mired in the batch world will fall behind.

Dan Hebert is Control's senior technical editor.

Reactor Goes Continuous

In 2000, Bayer MaterialScience started up the world's first commercial continuous process for the production of polyether polyol at its Channelview, Texas, facility. The process had previously used a batch reactor, one of two reactors on-site.

"Continuous processing is inherently more efficient than batch processing," observes William D. Wray, PE, engineering consultant with Bayer. "However, in many processes, including ours, you can't simply decide you want to manufacture continuously―the technology has to allow it. For us, the switch was made possible by a new catalyst with kinetic properties that allowed continuous processing,' continues Wray.

Bayer had a number of specific drivers for going continuous:

1. Increased throughput – The continuous reactor is about one-third of the size of the remaining batch reactor, yet can produce 1.5 times as much product.
2. Increased energy efficiency – The continuous process uses about 25% of the energy used for the batch process.
3. More consistent product quality – Although automation has made Bayer's batch products very consistent, continuous production quality and consistency is even better.
4. A greener process – The new catalyst technology eliminated the need for refining the reactor effluent into saleable product. In the conventional batch technology, the catalyst must be removed from the product. The new catalyst is only present in parts per million, has no deleterious effect on the product and need not be removed.

"The continuous process produces no waste, so every pound of feed is a pound of product," notes Wray.

Aside from mechanical changes, including beefing up the agitation in the reactor and adding cooling capability, instrumentation changes were needed.

Feeds previously batched into the reactor with a weigh vessel are now metered using redundant Coriolis flow meters and control valves. "We had previous experience with Coriolis meters for some reactor feeds, such as propylene oxide, but not all. The relative quantities of feeds not only control product quality, but also affect process safety by ensuring adequate catalyst feed to the reactor," explains Wray.

A second change affecting instrumentation is in level measurement, as the level instrumentation now needs to handle over-ranging caused by liquid full operation.

"Our plant was built in 1988 with a strong commitment to ISA-88 style modular batch automation, and the operations staff insisted we apply the lessons and techniques learned in batch automation to the converted continuous process. As it turns out, applying batch automation principles to continuous processes is pretty logical," feels Wray.

"There have been several papers presented at WBF - The Organization for Production Technology (www.wbf.org) conferences on this topic, and WBF continues to lead the expansion of ISA-88 principles into continuous operations. I recommend WBF as a resource for those contemplating a switch from batch to continuous. From an automation philosophy perspective, it might not be as big a change as expected," advises Wray.