The reality of continuous processing
Huw Thomas, of Foster Wheeler Energy, argues that continuous processing will always be more economic than batch processing, regardless of the scale of the operation.
Huw Thomas, of Foster Wheeler Energy, argues that continuous processing will always be more economic than batch processing, regardless of the scale of the operation.
In large scale chemical production, continuous processing is the norm. This is because simple economics dictate that, for a given process, batch operation cannot compete. On the other hand, in the small scale chemical processing of fine chemicals and API manufacture, batch processing is more common.
It has been taken as fact that there is a cross-over point where, as the scale of manufacture comes down, the economic pendulum swings from continuous to batch processing, but I am proposing that, depending on the basic chemistry of the process, there is no cross-over point and continuous processing will always be the more economic solution.
The question needs to be asked: 'is there a better way to design my process?'; and correspondingly: 'does a reagent molecule know whether it is in a stirred tank reactor, a continuous loop reactor or a microreactor?'
The answer is, of course, no - the molecule simply responds to the variables - kinetics, thermodynamics and fluid dynamics - that it experiences.
fundamental principle
Given this, why start the development process by confining the chemistry to a stirred tank? Why not define the process, then design the equipment to provide optimum physical processing conditions?
This fundamental principle is the basis of the 'fit the equipment to the process and not the process to the equipment' school of thought. The concept has been developed as a series of process design tools by Britest, a not-for-profit organisation combining academics, fine chemicals and pharmaceutical manufacturing companies, and engineering contractors.
This Britest Methodology - a step-by-step guide to process design and equipment selection - forces the user to develop a greater understanding of the whole process, and thus to adopt the optimum solution.
A wide range of 'process intensified' (PI) equipment has been developed that can give rates of mass and heat transfer several orders of magnitude greater than that of the standard batch stirred tank. This does not, however, agree with the holistic approach needed to produce the most economically sound design (see 'An example of a 'Process Intensified' approach', which is linked at the end of this article).
In April 2003, Foster Wheeler carried out a detailed review of small scale continuous processing equipment, concluding that all items required to continuously carry out unit operations in a typical pharma plant are available in industrially proven, packaged equipment. In many cases this had been designed for use in industries with similar requirements, such as food and semiconductor fabrication, thus minimising the need to 'pharmaceuticalise'.
A continuous plant should be designed to make the optimum return. Process plants are not exempt from the law of diminishing returns, and when one is moving into new technology the process designer needs to assess the cost-benefit-risk of their decisions, and keep in mind the question: 'how much of this plant needs to be continuous to give the step change in performance I need?'
The solid raw material for the plant is most likely to arrive in batch form, and the finished product will probably leave the same way. It is between these two batch operations that the process designer must decide where the interfaces between batch operations and continuous operation lie.
A Foster Wheeler design study in June 20031 where the design for an existing batch API plant was converted to operate continuously, revealed moving to continuous processing to be inherently more efficient as:
•it opens up chemical and physical processing opportunities that cannot be achieved in batch equipment;
• it entails more efficient processing as reactors can be designed to give more degrees of freedom to segregate competing reactions, thereby giving higher yields and selectivities;
• the variation of conditions is greatly reduced, thus reducing the variation in reactor product;
•counter-current extractions maximise extraction efficiency while minimising extractant solvent usage;
•where the process returns to operate batch wise for crystallisation, isolation and drying, the scale of the equipment is greatly reduced, thus reducing process variation and giving a more robust process.
Figure 1 shows the process flow diagrams (PFD) for the existing batch plant. The process stages in green are the stages where value is added; those in red are carried out only because of the stage-by-stage campaign operating regime of the batch plant. Figure 2 is the PFD for the integrated continuous plant. Not only are there far fewer stages, but the scale of the equipment for each stage is greatly reduced (the filter-dryer is replaced with a continuous pressure filter with a hundredth of the volume).
The batch-to-continuous design study1 quantified the utility and warehouse reductions, and showed that for a new plant the utility requirements were a fraction of the required cost, space and duty compared with a batch plant of the same throughput. It also showed that for an existing plant the process capacity could be extended with no cost for additional utilities as the continuous process can be tied into the existing systems.
Modern API facilities typically have multiple process floors, using gravity to transfer batches of solid and liquid between equipment. Reactor charging is typically done on the top floor, with reactors, filters and dryers on lower floors and the pack-off area on the lowest floor. There may be a separate utility building and a tank farm for bulk storage of solvents.
This rationale does not apply with a continuous process, and it was seen that the API plant could be fitted into a space little bigger than a tennis court, incorporating the material handling and utilities within this footprint.
Given the small size and potential mobility of the process equipment, plus the small utility requirements and the lack of structural or gravity flow constraints, one foreseeable scenario is to install numbers of continuous API production lines in an existing batch API plant.
In conventional batch processing, material traceability is relatively simple, with product output from a process stage readily correlated with input materials; and raw materials brought in as batches, processed at each stage as individual batches, and despatched in discrete containers as a batch. Such material traceability requires redefining for a continuous plant.
traceability redefined
Traceability is therefore based on a time period of production, not a specific batch. The time period chosen to represent a 'batch' is based on an assessment of the risk of producing out-of-specification (OOS) material. For continuous plants, this risk is mitigated by the steady state operation of the process, and the control system, including PAT systems, continuously monitoring and controlling the critical control parameters. This approach is in line with the current FDA 'Science Based Approach' and its PAT Initiative.
However, there are greater benefits to be reaped from changing a plant to continuous operation. With batch processing, a plant usually requires a shut-down period for cleaning and re-configuration between each stage, meaning that campaign length is selected to balance the inventory of material required against production time lost during shut-down. The ideal situation is obviously minimal inventory and maximum operating uptime, and while at some point there will be an economic optimum, this will inevitably entail a compromise of some sort.
economic benefits
Continuous processing can make the equipment sufficiently small that a dedicated plant, which eliminates down-time from plant re-configuration, may be economically justified. Where a multi-purpose plant is required, the plant can be rapidly re-configured by, for example, taking a piece of equipment that is difficult to clean and dedicating it to a particular process, or installing equipment that can be taken out of line easily, allowing cleaning out of place rather than less time efficient cleaning in place (CIP). Continuous production also provides a reduction in lead-time, removing the need for large inventory and enabling a move to just-in-time (JIT) manufacturing with all its attendant benefits.
lower costs
In the drug industry, the main driver towards continuous processing is the reduction in manufacturing cost, which can account for up to 28% of drug cost; and reducing the cost and time of every activity in the drug-to-market cycle. The major stumbling block it faces is that change is required at all stages of the drug-to-market process to realise the economic benefits of the technology.
There are four scenarios on the Manufacturing Chemist website (www.manufacturing-chemist.info) that examine in detail the particulars of small scale continuous processing. However, as a more general outline, the designer of a continuous pharma plant needs to combine three sets of design principles:
•continuous plant design principles: process design; HAZOP of continuous plant; run-standby/redundancy of equipment and controls; FMEA; process buffering; integrating analysers (PAT) into the process control system, etc.
•API facility design: GMP; material handling; containment; validation; GAMP; PAT etc.
•small scale processing and equipment: transferring knowledge from other industries, such as secondary pharmaceuticals; healthcare products; semiconductor; nuclear, etc.
Very little of the technology required for small scale continuous plants is completely new, and significant benefit can be achieved by 'free lessons' from other industries. For example, the requirements of GAMP and 21 CFR Part 11 do not make the control system for a continuous pharma plant so different from a conventional petrochemical plant, where integrating PAT systems has been done for many years.
The only disadvantage of continuous processing, with its different skill and knowledge requirements, is that companies will need to change their ways of working to take full advantage of its benefits. With, typically, 90% of manufacturing costs locked in at the design stage, and the cost to change a process increasing exponentially the further down the development route it is implemented, good management of early process design and development are essential.
In contrast to batch manufacturing, where the process is sequentially developed and handed 'over the wall' to the next group in the development chain, all of the stages in continuous processing must be developed as a whole process. In addition, equipment development must be carried out alongside chemistry development. This interlinking of the design of all the process stages and the equipment requires an integrated, concurrent approach, and I am therefore suggesting that the optimum development team will combine development chemists with process engineers with continuous operating experience.
Extensive study of the automotive industry concluded that by using concurrent design practices a product can be developed 25% faster; is three time less likely to be delayed, and eight times faster to achieve steady state quality. (This can be broken down into three main project areas, namely project cost, schedule and risk).
The technology of continuous processing enables every aspect of a process to change, from opening up new pathways for a development chemist to exploit, to enabling the business to radically change the way that products are developed, manufactured and delivered to the customer.
