New ideas hatch in process development
Compared with the rapid pace of product introductions elsewhere, the chemical industry can at times appear to lack innovation. As a mature and necessarily conservative industry, it certainly cannot be expected to match the pace of change in, say, consumer electronics and mobile telephony. But appearances can be deceptive. All across all the chemical industry, from the realm of high volume, low margin bulk petrochemicals to high-added-value fine chemicals and pharmaceuticals, new products, and the processes to make them, are constantly under development, albeit more slowly and more methodically than in some other industries, where perhaps fashion is a more important driver than function.
Bipin Vora, senior corporate fellow for process technology development at UOP, Des Plaines, Ill., speaking at July’s World Congress of Chemical Engineering (WCCE) in Glasgow, Scotland, put it this way: “Successful technology development, from concept to commercialization, requires a structured process. It may take anywhere from five to 10 years, requiring substantial expenditures in terms of research, pilot plant construction, development, scale-up, engineering design, and economic analyses.”
Benchmarking
Central to process development in most cases is the pilot plant, although in reality it is but one link in a chain that starts on the laboratory bench and ends at the plant with a fully commercialized unit. At the WCCE, results of an American Institute of Chemical Engineers (AIChE) pilot-plant benchmarking study were presented for the first time to an open audience. Thirty companies from across the commodity chemical, specialty chemical, pharmaceutical and oil and gas industries in North America took part in the exercise, a three-year project completed by the association’s process development division toward the end of last year.
The reasons why companies decide to pilot new and improved processes vary across the industry sectors, noted David Edwards of Zeton, Burlington, Ont., chairman of the division’s pilot plant group. Overall, they pilot either to demonstrate the viability of new processes, to generate design data or to produce market development samples of product. Different sectors have different priorities, of course. Sample production scores high among the pharmaceutical companies, for example. It is far less important to the oil and gas sector, which relies on piloting primarily to prove the viability of a new process and generate reliable design data.
Likewise, sectors showed distinct differences in how they decide which potential processes should progress through to the pilot stage. Approaches include: opting to pilot all processes, using a formalized risk-assessment process, making the choice based on an informal team or individual judgment, conducting a systematic review process, or relying on “stage gates” in which specific stages of the development process have to be completed before moving on to the next step.
The only sector that prefers to use piloting for all process development is pharmaceuticals, with 57% of respondents from that sector saying their company chose this route. The bulk commodity chemical industry, on the other hand, overwhelmingly appears to take the view that the decision to pilot should be based on a formal risk assessment of the process concerned.
Dan Pintar, operations manager at UOP’s Riverside, Ill., facility and a member of the AIChE team conducting the benchmarking survey, says the “gated” approach used by his company is a good way of involving multidisciplinary teams in the development process at an early stage. “Chemists might establish ‘proof of principle’ from their lab work,” he explains, “but to get to the next stage of development you have to pass through a stage gate, which is when you get input from the process development engineers.” And it’s the engineers, along with representatives from the commercial side of the business, who give the “thumbs up or down” to allow the project to move to the next stage.
According to Vora, at this and later stages of the process development, “the statistical design of experiments can and should play an important role, to better understand the results and minimize redundant efforts.” These experiments are likely to be on a small bench scale, often with the aim of screening various catalyst formulations or determining the range of operating parameters. Vora sounds a note of caution here, however. “Because a bench-scale ‘pilot plant’ often does not have product recovery or internal recycle streams built in,” he says, “the results need to be taken with a certain level of healthy skepticism. The results achieved under a perfectly controlled environment may not translate as well, or at all, to real-life situations.”
It’s at the next stage, the actual pilot plant, that issues such as the impact of the various recycle streams and impurity buildups can be fully assessed. “Even though the commercial design of the project may be several years away,” Vora says, “input from all the various branches of engineering design is critical at this early stage.”
High stakes
UOP focuses on developing processes to license and, so, is understandably cautious because it actually will not be running the processes. Some operating companies that do their own process development also see merit in a measured approach. With facilities in Houston, Texas, and Amsterdam in the Netherlands, Shell Chemicals’ Chemical Process and Development Group has delivered many new processes and process improvements to the company’s operating sites around the world. Heading the process engineering and evaluation group in Houston is David Torres, who says: “It’s important to learn early on that an idea has merit before millions are spent building a unit.” His group does preliminary process design and economic analysis to guide research programs and to determine the economic viability of a project.
As to whether companies would prefer not to pilot at all — relying instead, for instance, on process simulation models developed from lab-based experimentation — the AIChE survey (which, after all, was of pilot plant users) probably doesn’t provide many answers. But Zeton’s Edwards says that, while companies may be more selective about which processes they pilot in the future as resources become scarcer, “the need to pilot at a meaningful scale before moving to a commercial scale will always be a requirement. The risk is too great in not piloting a new process, because there are always surprises during piloting — byproduct accumulation, catalyst performance issues, corrosion issues and so on — and models are not too good in revealing such surprises.”
Acknowledging that his views are those of someone working for a company that designs and builds pilot plants for others to operate, Edwards nevertheless acknowledges the importance of the other stages of process development. “I think lab work, computer models and pilot plants all have an important role to play in process development,” he says. “Fundamental lab work will always be needed and a model can screen which potential new processes should be piloted.”
UOP’s Pintar takes a similar view. “There’s always going to be a need for the pilot plant in our current paradigm,” he argues, “because people want to see data. They want to see proof. Although, if you could develop a good kinetic model based on your pilot-plant data, then you might not need to run the plant all the time to generate estimates for customers or to do revamp studies. The problem is that we are always trying to push the units, to push the processing conditions outside of the regime for which the model was built.”
Faster screening
To develop any process model, however, presupposes a process in the first place. And for this we still need the laboratory bench and what was once the laborious work of screening many different compounds and assessing how they react under different catalytic conditions. This is now the realm of combinatorial chemistry — in which large numbers of reactions can be performed simultaneously in high throughput, small-scale systems.
An example is the HTS (High Throughput Screening) system of Symyx Technologies, Santa Clara, Calif., which has just won Frost & Sullivan’s 2005 Technology Leadership Award. “Symyx’s high throughout approaches offer significant advantages over conventional methods of catalysts discovery,” says F&S industry analyst, Anil Naidu. “The systems can rapidly screen materials to achieve the desired properties, delivering results faster and at a much lower cost.”
The success of Symyx’s combination of high throughput experimentation with its proprietary software tools for handling the data produced was highlighted last year with the start-up by Dow Chemical in Tarragona, Spain, of its first commercial plant to produce Versify plastomers and elastomers. These speciality propylene-ethylene copolymers are manufactured using a new catalyst system developed in collaboration between Dow and Symyx. “This is an important milestone for Symyx,” commented chairman and CEO Steve Goldby, “when an innovative discovery coming out of our labs goes into full commercial production.”
Earlier this year, Symyx announced a $120-million five-year strategic alliance with Dow “to effect a broad change in Dow’s R&D capabilities and efficiencies.” This deal follows a similar alliance with ExxonMobil signed in 2003 to run for five years and worth more than $200 million to Symyx.
As noted earlier, companies in the pharmaceuticals sector tend to have different priorities in process development than the bulk commodity chemical producers. According to David Ainsworth of engineering/procurement/construction contractor Foster Wheeler Energy, Reading, U.K., the use of simulation models — such as Batch Plus from Aspen Technology, Cambridge, Mass., and SuperPro Designer from Intelligen, Scotch Plains, N.J. — can help pharmaceutical companies investigate numerous design alternatives quickly and easily. “The computer model adds value at all stages of the design process,” he says, “from early conceptual design through to the ultimate operation of the facility.”
Ainsworth also cites the value that early involvement of an experienced process contractor can add — particularly in the pharmaceutical industry where processes are typically developed by teams of chemists. “Analyzed in a methodical manner, the specific characteristics of each process become evident and alternative processing methods can then be identified.”
Better tools
In its analysis arsenal, Foster Wheeler includes weapons developed by Britest, Cheadle, U.K. This not-for-profit company was set up in 1998 by a group of leading chemical and pharmaceutical companies, including AstraZeneca, Avecia, GlaxoSmithKline and Rhodia, to follow up on new approaches to process technology coming out of the universities at the time and to encourage technology transfer. Using what are known as the Britest tools — a set of proprietary procedures and software programs — is, says Ainsworth, a time-effective way of starting the development process and determining all the potential (and infeasible) process options.
One area the Britest toolkit considers is process intensification (PI), not just in the development stages but through to the commercial stage, as well. At Zeton, Edwards also is seeing a trend among the company’s pilot plant customers towards PI techniques and equipment — “although it’s still very much in its infancy,” he says. “We think operating companies are going to be interested enough to want to try it, but nervous enough not to go full scale until they have tried it on a pilot scale.”
UOP’s Pintar notes that at the same time as “pilot plants themselves have shrunk in terms of reactor size,” there is a growing drive for more data collection and on-line analyses from the plants. Fulfilling both of these goals, a new process analytical tool has recently been successfully trialed by specialty chemicals producer Clariant Chemicals at its plant in Leeds, U.K.
The plant used a patented “constant flux” reaction calorimeter developed by Ashe Morris, Radlett, U.K. The Coflux technology — which is akin to a variable area, rather than variable temperature, heat exchanger — is said by co-developer Robert Ashe to permit stirred tank reactors of virtually any size or type to be operated as precision calorimeters, offering a simple solution for on-line monitoring of chemical and biological processes. The R&D manager at the Leeds plant, Jim Wilson, said Clariant was able to monitor the rate of change (powder dissolution and reaction) throughout the trial experiments and could successfully detect the start and finish of each step in real time.
Real-time monitoring and increased automation were certainly among the trends identified by the AIChE study, as was an increasing emphasis on the safety of pilot plant operations. This highlighted something of a paradox because, as Pintar observes, despite the increasing levels of automation, “the majority of companies don’t allow unattended operation of their pilot plants.”
No doubt this will be a topic for the next benchmarking exercise, expected in three to five years time. Until then, however, the future for the pilot plant at the heart of process development seems assured.
