ARTICLE pubs.acs.


Economic Analysis of Integrated Continuous and Batch Pharmaceutical Manufacturing: A Case Study
Spencer D. Schaber,† Dimitrios I. Gerogiorgis,† Rohit Ramachandran,†,§ James M. B. Evans,‡ Paul I. Barton,† and Bernhardt L. Trout*,‡

Process Systems Engineering Laboratory, Department of Chemical Engineering and ‡Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States ABSTRACT: The capital, operating, and overall costs of a dedicated continuous manufacturing process to synthesize an active pharmaceutical ingredient (API) and formulate it into tablets are estimated for a production scale of 2000 t of tablets per year, with raw material cost, production yield, and API loading varied over broad ranges. Costs are compared to batch production in a dedicated facility. Synthesis begins with a key organic intermediate three synthetic steps before the final API; results are given for key intermediate (KI) costs of $100 to $3000/kg, with drug loadings in the tablet of 10 and 50 wt %. The novel continuous process described here is being developed by an interdisciplinary team of 20 researchers. Since yields are not yet well-known, and continuous processes typically have better yields than batch ones, the overall yields of the continuous processes with recycling were set equal to that of the batch process. Without recycling, yields are 10% lower, but less equipment is required. The continuous process has not been built at large scale, so Wroth factors and other assumptions were used to estimate costs. Capital expenditures for continuous production were estimated to be 20 to 76% lower, depending on the drug loading, KI cost, and process chosen; operating expenditures were estimated to be between 40% lower and 9% higher. The novel continuous process with recycling coupled to a novel direct tablet formation process yields the best overall cost savings in each drug loading/KI price scenario: estimated savings range from 9 to 40%. Overall cost savings are also given assuming the yield in the continuous case is 10% above and 10% below that of the batch process. Even when yields in the continuous case are lower than in the batch case, savings can still be achieved because the labor, materials handling, CapEx, and other savings compensate.

’ INTRODUCTION Continuous manufacturing (CM) is attracting increasing attention within the pharmaceutical industry today because it could lead to significant decreases in production costs while improving product quality.1,2 Historically, production costs were seen as a small enough part of the overall industry expenses that major cost reductions were not needed. Regulations also drove production toward the batch mode, since processes were required to be run in exactly the same way for the lifetime of the therapy. Also, batch production allows verification of quality of each batch from each process before further processing, whereas a “batch” in a continuous process is not contained in the same way.3,4 Today, however, it is becoming more difficult for pharmaceutical companies to meet profit expectations, due to increasing research and development (R&D) costs and competition from generics manufacturers.5 At the same time, regulatory bodies are shifting the emphasis toward process understanding and giving more freedom when such understanding is demonstrated.6 For sufficiently large production scales, continuous processes tend to have lower production costs; CM would also allow manufacturers to use the increased process understanding for online process control, yielding consistently high-quality product and less material wasted as off-spec product.7,8 A review of the fine and commodity chemical industries demonstrates that CM could offer both operating expenditure (OpEx) and capital expenditure (CapEx) savings for the pharmaceutical industry. Labor for transporting material between batch units, labor for quality assurance/quality control (QA/QC), and in-process inventory (working capital) can all be significantly reduced in continuous
r XXXX American Chemical Society

processing.1,9 Processing equipment for fine chemical synthesis can be made much smaller by moving to continuous processing, as well has having larger surface area to volume ratios, which implies a safer plant (a smaller holdup of solvents in reactors and enhanced heat transfer for safe handling of highly exothermic reactions), a smaller investment in reactors, and faster change over in multipurpose plants.7,8,10 More rapid mixing, reaction, and quenching are possible in continuous flow,11 enabling reactions that would produce significantly more impurities if run in batch mode, such as in the first reaction in the novel continuous process presented in this work. Plant footprint can also be reduced due to smaller processing equipment, with commensurate energy savings for heating, ventilation, and air conditioning.12 Pharmaceutical processes often contain continuous or semicontinuous processing steps, such as milling and tablet compression, but the processes are started and stopped to mirror the batch processing in other steps. These steps can be more naturally run in a continuous manner, potentially yielding more consistent product quality.1,4 Scaleup of batch granulation can be difficult, and is sometimes easier in continuous mode, so development of a needed granulation process could begin on continuous equipment, easing scaleup for production.3 Recently, the lack of continuous tablet coating equipment was a bottleneck for
Received: Accepted: Revised: April 1, 2011 July 27, 2011 June 28, 2011

A | Ind. Eng. Chem. Res. XXXX, XXX, 000–000

OpEx. as well as the ability to bring therapies to ailing people more quickly. Raw Materials Requirements for All Processes at 50 wt % API Loadinga. continuous pharmaceutical production. which is on par with the production scale of a very high-volume “blockbuster” drug. and produces a final drug product: tablets. XXXX. and present cost of a dedicated batch process and four continuous processes that are enabled by new technologies developed for continuous production.5 Despite studies on the individual differences between batch and continuous processing. whereas the continuous process is being developed in the Center for Continuous Manufacturing.Industrial & Engineering Chemistry Research ARTICLE Figure 1.12 to date. Eng. ’ PROCESS DESCRIPTION For both the batch and continuous processes.4 but now it is available.doi. easing scale-up and leading to additional time during which the product can be sold exclusively by the patent holder. RC. as long as it proves cost-effective. The production scale is 2000 t of tablets per year. We are not permitted to disclose further details of the process. The batch process has been extensively developed by Novartis. can enhance process understanding early in the patent life of a product. XXX. with excellent time stability. Res. Process flow diagram for batch (Bx) manufacturing route.1021/ie2006752 |Ind. In this work we estimate CapEx. While many B pharmaceutical production processes use multipurpose equipment to manufacture several drugs in partial-year campaigns. API loadings in the tablet of 10 and 50 wt % were used to account for variations in API potency.13 In addition to cost savings.1 Recent developments in process analytical technology (PAT) will allow manufacturers to complete the shift to continuous manufacturing. using microreactors for instance. dx. Both processes produce the same drug product. very highvolume drugs are sometimes produced on dedicated equipment. The Novartis-MIT Center for Continuous Manufacturing (CCM) is focused on a holistic approach where we consider manufacture of the final drug product from starting materials available as fine chemicals. direct tablet formation. Table 1. three synthetic steps before the final active pharmaceutical ingredient (API). The sequence of unit operations for the batch process is given in Figure 1.1.8. developing continuous processes early DTF. the assessment starts with a late-stage organic key intermediate (KI) molecule. The raw materials requirements and costs for one scenario are given in Table 1 and Table 2. an integrated analysis of the continuous manufacture of a final drug product from a late-stage organic KI has not been published. roller compaction. Continuous powder mixing has been shown to perform as needed. 000–000 .9. Batch Process. Chem. materials organic reagents inorganic reagents organic solvents water excipients and coatings total a Bx 1 955 000 5 508 000 34 090 000 22 907 000 1 004 000 65 464 000 CM1R/DTF 1 597 000 3 659 000 24 659 000 7 803 000 1 000 000 38 718 000 CM1/DTF 3 112 000 3 659 000 29 497 000 9 965 000 1 000 000 47 233 000 CM1R/RC 1 597 000 3 659 000 24 659 000 7 803 000 1 000 000 38 718 000 CM1/RC 3 112 000 3 659 000 29 497 000 9 965 000 1 000 000 47 233 000 All values in kg/year.

and the overall yield of the continuous process without recycling (CM1) is 10% below that of the batch process.000 15. Continuous reactions scale up very predictably and in an economically favorable way. Also. Process CM1R is identical to CM1.000 3.507.4% to 98.948.000 3. which can be observed as an increasing pressure drop across the reactor. RC is a wellestablished pharmaceutical technology.000 15.899. roller compaction. it is believed that a mature continuous process will have yields equal to or better than the batch yields.263. Novel CM Route (CM1).824. Raw Materials Costs for All Processes at 50 wt % API Loading and $3000/kg KIa materials organic reagents inorganic reagents organic solvents water excipients and coatings total a ARTICLE Bx 3.000 2.000 CM1/RC 3. Process flow diagram for continuous manufacturing route CM1.000 27.1021/ie2006752 |Ind.356.898. Figure 2) that utilizes pathways that are not feasible in a batch and 3 are plug-flow reactors. The API synthesis is coupled with two downstream process options: roller compaction (RC) and a novel direct tablet formation (DTF) process.936.000 15.893.Industrial & Engineering Chemistry Research Table 2.784. In each case. Chem.394.000 CM1R/RC 3.893.784.000 996.000 4. processing in a continuous-flow reactor enables a much more rapid deprotection reaction than batch reaction.000 22. Figure 2.219. Eng.8 Novel CM Route with Recycle (CM1R). For example.898.674.000 27.784. several workarounds to the plugging and fouling issue are possible. Reactors 1. since the continuous process already has competitive yields despite being developed for fewer than half as many years.784.420.948.263.293.7 Microreactors have been successfully used to produce hundreds of kilograms of product in a few weeks.8 The methods for using microreactors with heterogeneous catalysts or severe precipitation are not mature. showing both options for forming tablets.000 15.000 3.375.893. 2. Res.000 4. translating the CM1 route into a standard batch process would result in significant degradation of the product because the required rate of reagent addition cannot be achieved in batch mode.888. at a much smaller scale. RC.219. based on including strategic solvent choice.7 However. except that a single recycle loop and appropriate separation equipment are added to increase the effective yield (from 86.420.000 All values in $/year.000 CM1R/DTF 3. 000–000 .145.000 780.000 92. direct tablet formation.893.888. The CCM team developed a new synthetic route (CM1. Since the yields for the final continuous processes are not known precisely.000 3.000 2.7.doi. The crystallizer and combined reactor/crystallizer are agitated tanks. XXX.000 22. XXXX.000 4. the continuous processes save an average of 61% of the annual water usage and 21 wt % of the annual solvent usage compared to batch.182.000 15.000 CM1/DTF 3. and device geometry. temperature.8 One issue currently limiting the savings is microreactor plugging or fouling. and by fewer people.864.000 3. flow velocity.000 780.899.824. The actual yields that have been demonstrated in bench-scale continuous reactions are bracketed by these scenarios. a patent application is being prepared for the novel direct tablet formation process. yields have been set such that the overall yield for the continuous process with recycling for the first reaction (CM1R) is equal to that of the batch process.375.000 996. DTF. so it is not described in detail here.864. the overall yield for process CM1 is 10% below the corresponding overall yield for process CM1R.5%) in the first step (Reactor 1) of the dx.000 4. Overall cost savings are also given for the case where overall yield C for process CM1R is 10% below and 10% above that of the batch case.

Thus. One production line per plant was assumed. and piping costs. The continuous plant has not been built at large scale. Capital Expenditures (CapEx). The total cost of processing equipment excluding any ancillary equipment.98) by a power law in the following cases: plug-flow reactor. These assumptions are optimistic for batch production. with good processes having 74% and “best-in-class” production lines reaching 92%. For example.0 4. dryer.14 Wroth factors allow quick estimation of installation and other necessary equipment costs. Res. For Table 2. This is the percentage of time when the process equipment is actually processing material. XXXX. Wroth factors (Table 3) were used to calculate delivered installed equipment cost.20. Scaling of cost could be approximated well (R2 g 0. which includes ancillary equipment. cleaning. whereas costs for other raw materials are from vendor quotes. Material Balances. as compared to batch processes which are typically operated in an open-loop manner.33. exponent 0. and QA/QC savings represent our best estimates at this time.5 Wroth factor batch pharmaceutical production of 30%. and utilities. ðdelivered installed equipment costÞ ¼ ðWroth factorÞ Â ðdelivered equipment costÞ ð1Þ Additional CapEx heuristics used in the present analysis are summarized in Table 4. engineering. Material requirements and costs for all processes are given in Table 1. emptying. so the values for labor. ARTICLE ’ COST ANALYSIS METHODS Green-field construction of a new. assumed to be treated at a nominal cost indicated in Table 5 35% of annual sales14 w used 35% of annual materials costs for batch. piping. working capital.05 Â FOB cost). Operating Expenditures (OpEx). and engineering (4) battery-limits installed cost (BLIC) (5) buildings and structures (6) contingency (7) offsite capital (for a grass-roots plant) (8) service facilities (9) waste disposal (10) working capital (11) total CapEx sum of items (1) to (3)14 20% of BLIC14 20% of BLIC14 150% of BLIC14 20% of BLIC14 not included in CapEx. the cost of the KI (one of the organic reagents) is $3000 kg/year. a related metric. Summary of CapEx Heuristics Used item (1) FOB cost (2) delivery (3) installation: ancillary equipment. or piping expenses is termed the FOB (free on board) cost. Operating expenditures were calculated for KI prices of $100. OpEx savings are expected in QA/QC since some manual sampling and analysis can be replaced by online analysis. engineering. and must adhere to stricter hygiene regulations. delivery. and startup/shutdown. and the smallest unit of sufficiently large size was selected. exponent 0. respectively. When price quotations were only available for batch equipment. to account for the increased process engineering (CapEx) required to operate a process continuously with feedback control. significant savings overall are estimated. Batch process effective utilization time was assumed to be 85% for upstream processes and 55% for downstream processes. XXX. Eng. Separation equipment and recycle are essential in order to reduce formation of the primary impurity. electrical. 3.Industrial & Engineering Chemistry Research API synthesis from the KI. exponent 0. that of the batch process is 79%. and cleaning the batch processing unit.4 the overall equipment effectiveness (OEE).5 3.1 4. 95% was assumed for all continuous processes. Equipment Size and Cost Estimation.5% for continuous. dedicated plant was considered in all cases. takes a typical value in Table 3. labor. Selected Wroth Factors14 unit distillation tower and internals instrument process tank reactor storage tank all other equipment 4. quality assurance and the primary product of the reaction can undergo a subsequent reaction to form an impurity. Since pharmaceutical production scales are smaller than typical commodity chemical production scales. or simply waiting for material to be processed.42. due to savings in CapEx. representing lean batch operations in dedicated production: According to Vervaet and Remon. Table 4. Without the separation step. materials handling. electrical. the values used were the upper bounds of the ranges given by Couper et al. The remaining time is spent filling. and are typically much less than $3000/kg.doi. materials handling. $500. delivery. and are commonly used at this stage of an economic analysis. The heuristics used are summarized in Table 5. automation. filtration equipment.1 (factor into appropriate process tanks and other equipment) 3. From delivered equipment cost (1. Chem.1021/ie2006752 |Ind. Even without recycling.21. Calculation of Overall CapEx from Individual Process Equipment Costs. 000–000 cost sum of processing equipment units14 5% of FOB cost14 [(Wroth factor) À 1] Â (delivered equipment cost)14 . the additional expenses are expected to comprise a larger fraction of the CapEx. electrical. waste handling. The overall yields (mol drug substance/mol KI) of processes CM1 and CM1R are 69% and 79%. exponent 0. a 10% price premium was assumed for continuous units relative to a batch unit of the same size. and $3000/kg. since throughput times are expected to be 10x lower in continuous processing sum of items (4) to (10) D dx. A 335-day working year was considered. Vendor price quotations for all process equipment were obtained for both batch and continuous equipment over a wide range of sizes. agitated vessel/CSTR/crystallizer. with 30 days left for maintenance.

the following quantity was defined: ðcontribution to present cost differenceÞ À ¼ ð present cost of contributor for Bx processÞ À ðpresent cost of contributor for CM processÞÞ =ðpresent cost of Bx processÞ To clarify the above definition.000/year per operator. so savings are similar for E dx.Industrial & Engineering Chemistry Research Table 5.1021/ie2006752 |Ind. especially for the KI. Present cost of the project is similar to net present value (NPV. dominates CapEx. To quantify the contributions of different expenses to the cost differences for . present cost of the project (eq 2) was calculated for each processing option. CapEx dollar amounts are provided in square brackets for the base case of a batch process. Summary of OpEx Heuristics Used item (1) labor and supervision (2) materials handling and storage (3) off-spec product (4) quality assurance and control (QA/QC) (5) utilities (6) waste disposal (7) total OpEx 15 ARTICLE cost $160. construction period was 1 year. Process CM1 with direct tablet formation has the largest CapEx savings (31À76% savings vs batch processing). This is the discounted total cost of the project. To quantify overall cost differences accounting for both CapEx and OpEx. upstream CapEx. CM relative to batch.00/gallon for all other material17 sum of items (1) to (6) plus raw material costs Table 6.50/gallon for water and organic solvents.16 twice as many operators required for batch processes as for continuous continuous is estimated at 40% of batch 0% for batch and continuous continuous is estimated at 50% of batch $1. downstream CapEx. not NPV. Contributors to Overall Cost Savings. the working capital. excluding any revenue. D. but does not include revenue. At the highest KI price.doi. eq 3). CapEx. note that ðcontributorsÞ ð4Þ ðpresent costÞ ¼ ðCapExÞ þ ∑ i i ¼ 1 ð1 þ rd Þ τ ðOpExÞ ð2Þ ∑ ðcontribution to present cost differenceÞ ð5Þ ðNPVÞ ¼ À ðCapExÞ ( ) τ ÀðOpExÞ ðrevenueÞ þ i þ ð1 þ rd Þi i ¼ 1 ð1 þ rd Þ ¼ ðpercentage present cost difference vs batchÞ ð3Þ ∑ ’ RESULTS Capital Expenditures (CapEx). Overall Cost of Production. and project lifetime (τ) was 15 years. U. This figure of merit was chosen because we are comparing costs. Eng. $15.50/kg material input $2. number of operators estimated as in Ulrich. Res. CapEx (Including Working Capital) Differences for All Process Options. Chem. Relative to Batch Case. 000–000 Discount rate (rd) was 7%. For Upstream and Downstreama API mass loading = 10% cost of KI : Batch (basis for differences) $100/kg Tot: [$315M] U: [$73M] D: [$242M] CM1R with direct tablet formation Tot: À28% U: À31% D: À27% CM1 with direct tablet formation Tot: À31% U: À43% D: À27% CM1R with roller compaction Tot: À20% U: À31% CM1 with roller compaction D: À17% Tot: À23% U: À43% D: À17% a API mass loading = 50% $3000/kg Tot: [$542M] U: [$300M] D: [$242M] Tot: À54% U: À76% D: À27% Tot: À55% U: À78% D: À27% Tot: À49% U: À76% D: À17% Tot: À50% U: À78% D: À17% $100/kg Tot: [$429M] U: [$173M] D: [$256M] Tot: À39% U: À52% D: À31% Tot: À42% U: À59% D: À31% Tot: À34% U: À52% D: À21% Tot: À36% U: À59% D: À21% $500/kg Tot: [$585M] U: [$329M] D: [$256M] Tot: À53% U: À70% D: À31% Tot: À55% U: À73% D: À31% Tot: À49% U: À70% D: À21% Tot: À50% U: À73% D: À21% $3000/kg Tot: [$1565M] U: [$1308M] D: [$256M] Tot: À76% U: À85% D: À31% Tot: À76% U: À85% D: À31% Tot: À75% U: À85% D: À21% Tot: À74% U: À85% D: À21% $500/kg Tot: [$346M] U: [$105M] D: [$242M] Tot: À33% U: À49% D: À27% Tot: À36% U: À57% D: À27% Tot: À26% U: À49% D: À17% Tot: À29% U: À57% D: À17% Abbreviations: Tot.

due to the lower overall yield without recycling. Contributors to Overall Cost Savings. Overall Cost of Production. downstream raw materials OpEx. If the overall yield is 10% higher for CM1R than for the batch process F (Table 12). Detailed results are given in Table 8. The novel continuous process also has lower solvent usage. Detailed results are given in Table 6. If the overall yield of process CM1R is 10% below that of the batch process (Table 11). Relative to Batch Casea API mass loading = 10% cost of KI : batch (basis for differences) CM1R with direct tablet formation CM1 with direct tablet formation CM1R with roller compaction CM1 with roller compaction a ARTICLE API mass loading = 50% $3000/kg [$542M] À54% À55% À49% À50% $100/kg [$429M] À39% À42% À34% À36% $500/kg [$585M] À53% À55% À49% À50% $3000/kg [$1565M] À76% À76% À75% À74% $100/kg [$315M] À28% À31% À20% À23% $500/kg [$346M] À33% À36% À26% À29% Total CapEx dollar amounts are provided in square brackets for the base case of a batch process. summarized results are given in Table 7. all processes. process CM1 options show slightly less savings. reducing OpEx in the “other” raw materials category. XXXX. 19À35% savings can be achieved versus batch in all low API loading cases by choosing process CM1R with direct tablet formation. Process CM1R with either direct tablet formation or roller compaction has the lowest annual OpEx of any process option (6À40% savings). Eng. URM. Moving toward higher KI price in the table means the KI makes up a higher fraction of the dx. total OpEx.Industrial & Engineering Chemistry Research Table 7.doi. since equally priced excipients are used and the overall yield of drug substance from KI is assumed equal in CM1R and batch.1021/ie2006752 |Ind. See Table 10. Annual OpEx Differences for All Process Options. summarized results are given in Table 9. all other OpEx. CM1R with roller compaction is the next best option. the present cost of the project for the baseline case in which CM1R yield is equal to batch yield is broken down into contributions for each category (cf. 000–000 . Res. the overall costs of continuous processing are between 4% higher and 35% lower than batch processing if the best process is chosen for each API loading/KI price scenario. The cost of the other organic reagents needed in the novel continuous process are lower. eq 4). DRM. In Table 13. The expenditures for the KI and for excipients are the same for CM1R/DTF and batch. all continuous processes are estimated to be more expensive than the batch process. upstream raw materials OpEx. Relative to Batch Casea API mass loading = 10% cost of KI : Batch (basis for differences) $100/kg Tot: [$136M] URM: [$49M] DRM: [$25M] Oth: [$62M] CM1R with direct tablet formation Tot: À33% URM: À36% DRM: 0% CM1 with direct tablet formation Oth: À45% Tot: À19% URM: À6% DRM: 0% Oth: À36% CM1R with roller compaction Tot: À33% URM: À36% DRM: 0% CM1 with roller compaction Oth: À45% Tot: À19% URM: À6% DRM: 0% Oth: À36% a API mass loading = 50% $3000/kg Tot: [$785M] $100/kg Tot: [$531M] URM: [$246M] DRM: [$16M] Oth: [$269M] Tot: À40% URM: À36% DRM: À1% Oth: À47% Tot: À22% URM: À6% DRM: À1% Oth: À38% Tot: À40% URM: À36% DRM: À1% Oth: À47% Tot: À22% URM: À6% DRM: À1% Oth: À38% $500/kg Tot: [$979M] URM: [$693M] DRM: [$16M] Oth: [$269M] Tot: À22% URM: À13% DRM: À1% Oth: À47% Tot: À6% URM: 7% DRM: À1% Oth: À38% Tot: À22% URM: À13% DRM: À1% Oth: À47% Tot: À6% URM: 7% DRM: À1% Oth: À38% $3000/kg Tot: [$3777M] URM: [$3491M] DRM: [$16M] Oth: [$269M] Tot: À6% URM: À2% DRM: À1% Oth: À47% Tot: 9% URM: 13% DRM: À1% Oth: À38% Tot: À6% URM: À2% DRM: À1% Oth: À47% Tot: 9% URM: 13% DRM: À1% Oth: À38% $500/kg Tot: [$226M] URM: [$139M] DRM: [$25M] Oth: [$62M] Tot: À20% URM: À13% DRM: 0% Oth: À45% Tot: À6% URM: 7% DRM: 0% Oth: À36% Tot: À20% URM: À13% DRM: 0% Oth: À45% Tot: À6% URM: 7% DRM: 0% Oth: À36% URM: [$698M] DRM: [$25M] Oth: [$62M] Tot: À6% URM: À2% DRM: 0% Oth: À45% Tot: 8% URM: 13% DRM: 0% Oth: À36% Tot: À6% URM: À2% DRM: 0% Oth: À45% Tot: 8% URM: 13% DRM: 0% Oth: À36% Abbreviations: Tot. Summary of CapEx Differences for All Process Options. The values are for the novel continuous process with recycling with direct tablet formation (CM1R/DTF) and are similar to those for the other novel CM options (not published). Oth. At $3000/kg for the KI. OpEx dollar amounts are provided in square brackets for the base case of a batch process. XXX. Process CM1R with direct tablet formation has the lowest present cost (9À40% savings). with very similar savings. Table 8. Operating Expenditures (OpEx). reducing costs of other raw materials and waste handling.

Table 10. However. for the 15-year project lifetime. This may initially seem counterintuitive since the high-loading processes require five times more KI. Table 11. 000–000 . expenses. excluding any revenue. that expenditure is proportional to KI price.1021/ie2006752 |Ind. with the same or slightly more savings than CM1 with roller compaction. for the 15-year project lifetime. Summary of Present Cost Differences for All Process Options. respectively. (second) utilities. the working capital (in-process inventory) savings measured in dollars are increased moving toward higher KI price. so percentage savings are reduced. the percentage savings due to working capital is similar for high and low loadings. Relative to Batch Casea API mass loading = 10% cost of KI : batch (basis for differences) CM1R with direct tablet formation CM1 with direct tablet formation CM1R with roller compaction CM1 with roller compaction a API mass loading = 50% $3000/kg [$7472M] À9% 4% À9% 4% $100/kg [$5117M] À40% À24% À40% À23% $500/kg [$9225M] À24% À9% À23% À8% $3000/kg [$34902M] À9% 5% À9% 5% $100/kg [$1515M] À32% À21% À30% À20% $500/kg [$2337M] À22% À10% À21% À9% Present cost is the total discounted cost of the project. fifth. the largest savings consistently come from (first) other raw materials. and all other expenses dx. This is one of the first market segments in which continuous pharmaceutical manufacturing may be implemented. (fifth) labor and materials handling. (sixth) other organic reagents. respectively. Percentage savings are greatest when KI prices are lower. Overall cost savings of 9 to 40% are predicted if the appropriate process is selected for the API loading/KI price scenario at hand. Present cost dollar amount is provided in square brackets for the base case of a batch process. (fourth) CapEx excluding working capital. or eighth contributor to cost savings for KI prices of $3000/kg. XXXX. ’ DISCUSSION Batch and continuous production of a very large-scale pharmaceutical product produced in dedicated batch or continuous plants was analyzed. (second) utilities. excluding any revenue. (seventh) CapEx excluding working capital.Industrial & Engineering Chemistry Research Table 9. For a given KI the high-loading cases also have about four times greater overall cost than the corresponding low-loading cases. Summary of Annual OpEx Differences for All Process Options. The novel process with recycling (CM1R) with direct tablet formation is consistently one of the most favorable processes. sixth. G (third) labor and materials handling. so the percentage savings are similar. $500/kg. $500/kg. Summary of Present Cost Differences if CM1R Yield is 10% below Batch Yielda API mass loading = 10% cost of KI : batch (basis for differences) CM1R with direct tablet formation CM1 with direct tablet formation CM1R with roller compaction CM1 with roller compaction a API mass loading = 50% $3000/kg [$7472M] 3% 18% 3% 18% $100/kg [$5117M] À35% À16% À34% À15% $500/kg [$9225M] À14% 3% À14% 3% $3000/kg [$34902M] 4% 20% 4% 20% $100/kg [$1515M] À28% À16% À27% À14% $500/kg [$2337M] À15% À1% À13% 0% Present cost is the total discounted cost of the project. (third) waste handling. In the high API loading scenarios. and $100/kg. Present cost dollar amount is provided in square brackets for the base case of a batch process. Relative to Batch Casea API mass loading = 10% cost of KI : batch (basis for differences) CM1R with direct tablet formation CM1 with direct tablet formation CM1R with roller compaction CM1 with roller compaction a ARTICLE API mass loading = 50% $3000/kg [$785M] À6% 8% À6% 8% $100/kg [$531M] À40% À22% À40% À22% $500/kg [$979M] À22% À6% À22% À6% $3000/kg [$3777M] À6% 9% À6% 9% $100/kg [$136M] À33% À19% À33% À19% $500/kg [$226M] À20% À6% À20% À6% Annual OpEx dollar amounts are provided in square brackets for the base case of a batch process. The significance of working capital varies depending on the KI price. This is because the expenditure for the KI is the same in batch and CM1R. and (seventh) other organic reagents. The significance of working capital again depends on the KI price. Eng. the largest savings consistently come from (first) other raw materials.doi. it is the first. Chem. it is the first. (fifth) QA/QC. Res. In the low API loading scenarios. (fourth) waste handling. However. and $100/kg. or seventh most significant contributor to cost savings for KI prices of $3000/kg. the most expensive raw material. since the batch process has 10 times more in-process inventory. (sixth) QA/QC. XXX.

for the 15-year project lifetime.0% À0. PAT). Since the continuous yield is identical to the batch yield. we are developing another novel tablet formation process (not yet published) that should have significantly better yields than roller compaction. present cost savings approach the working capital savings from the reduced in-process inventory.6% À32% =[À$485M] $500/kg À1.3% À1. but the standard value for “other equipment” of 3.2% 0. Contributions to Present Cost Difference Relative to Batch for Novel Continuous Process with Recycling (CM1R) with Direct Tablet Formationa API mass loading = 10% cost of KI : organic reagents KI other raw materials excipients labor and materials handling waste handling utilities QA/QC CapEx excluding working capital working capital/in-process inventory total: $100/kg À2.0% À4.9% À4.6% À0.9% À4.0% À2.9. Eng.0% À11.g. That is.8.1% À9% =[À$689M] $100/kg À3.0% À8. the equipment cost is subject to more uncertainty than most of the other costs.5% 0. making scaleup easier and less expensive over time.8% À2. XXX. Res. because the pharmaceutical industry has less experience with continuous processing and the absence of conventional batches in the highly regulated industry demands more process understanding and online instrumentation (i.6% 0.0% À2.4% 0.0% À0. This was accounted for as a 10% price premium for continuous processing equipment at the same scale as an equivalent batch process.9% À0.12 no articles have been published comparing the overall economics of batch and continuous pharmaceutical processes producing drug product (tablets) from an organic key intermediate.18 claims that continuous granulation processes can be developed more quickly.2% 0.9% À0..0% À6. 000–000 . The novel direct tablet formation process should also be more broadly applicable than roller compaction and eliminate other costs related to powder handling that are not considered in this study.2% À1.4% À0. Summary of Present Cost Differences if CM1R Yield is 10% above Batch Yielda API mass loading = 10% cost of KI : batch (basis for differences) CM1R with direct tablet formation CM1 with direct tablet formation CM1R with roller compaction CM1 with roller compaction a ARTICLE API mass loading = 50% $3000/kg [$7472M] À19% À7% À18% À7% $100/kg [$5117M] À44% À30% À44% À29% $500/kg [$9225M] À31% À18% À31% À17% $3000/kg [$34902M] À19% À6% À19% À6% $100/kg [$1515M] À35% À25% À33% À24% $500/kg [$2337M] À28% À17% À27% À16% Present cost is the total discounted cost of the project.4% À4. Table 3).3% À5. This assumption allows for cost increases specific to the pharmaceutical industry and may be unnecessarily high. Although different aspects of continuous pharmaceutical production have been analyzed. With the current trend toward smaller continuous processes (e. the contribution of KI cost to overall savings is identically zero in all cases.8% À3. In addition to the direct tablet formation approach reported here. Some unit operations are easier to characterize in continuous mode: Dhenge et al.0% À5.2% À1. meaning that the realized cost savings for the direct tablet formation process may be greater than those estimated here. A presentation at the American Association of Pharmaceutical Scientists meeting on Drug Product Manufacturing claimed 58% CapEx savings and 67% annual OpEx savings for a continuous pharmaceutical processing facility versus a batch processing facility. Once the process is operational.7% À1.7% À2.2% À3.2% À9% =[À$3066M] a Difference in present cost relative to batch case is provided in square brackets.19 A study on production of ethanol dx. XXXX..9% À2.6% À2.8% À0.5% À24% =[À$2188M] $3000/kg À0.8% À0. The capital expenditure for the novel direct tablet formation process was based on a vendor quotation.5 has been assumed (cf. Contribution to present cost difference relative to batch is defined by eq 4. but since it has not H been used in the pharmaceutical industry.1. more process understanding will be obtained at early stages of the process development.9% À2. Furthermore.Industrial & Engineering Chemistry Research Table 12.0% À0.7% 0.6% 0.0% À3. Chem.6% À1. excluding any revenue.0% 0.0% À1.e. The process development costs will tend to be greater for continuous manufacturing processes as opposed to batch processes.2% À22% =[À$513M] $3000/kg À0. labor costs are typically lower as well. however.0% À1. Present cost dollar amount is provided in square brackets for the base case of a batch process.4% 0.1021/ie2006752 |Ind.1% À4.0% À6. the smaller scale required for each unit in a continuous process (due to greater effective utilization time) usually offsets the additional process understanding and control required. Table 13.8% À3.0% À5. microreactors).9% 0. except working capital are insensitive to KI price. A typical Wroth factor for the process under consideration in other industries is about 2. as KI price approaches infinity.8% 0.6% À0.0.8% À3.7% À2. with associated savings in API material during development.doi.7% À40% =[À$2048M] API mass loading = 50% $500/kg À1.

the savings are 9 to 40% and 19 to 44%. as well as having a lower residence time: the continuous process took 12 min as opposed to 9 h and 40 min using the batch process. a 10% yield difference can shift a continuous process from providing cost savings to providing cost increases (see Tables 11 and 12). The beststudied method for continuous granulation is extrusion. US$ PAT = process analytical technology QA/QC = quality assurance/quality control RC = roller compaction ’ REFERENCES (1) Plumb. Res. continuous manufacturing of pharmaceuticals is a viable way for the pharmaceutical industry to achieve substantial cost savings. alternative methods for doing so may prove even more efficient.7À9. three continuous API synthesis processes. XXXX. US$ Bx = batch manufacturing process CapEx = capital expenditures.25À30 On the contrary. 2005. Continuous Processing in the Pharmaceutical Industry: Changing the Mind Set. Eng. and two continuous drug product formation processes. on which the first papers for pharmaceutical applications were published in 1986. 000–000 ’ CONCLUSION An integrated cost estimation of the production of a final drug product from a key organic intermediate was performed. XXX. To make the analysis applicable to a wider range of products. The overall cost of production can be reduced most by changing to the continuous .26À30 Commercial equipment for creating the final dosage form such as continuous tableting and coating is already available. Plantwide dynamic models are essential for this task. whereas others require further study. whose comments led to significant improvements to the paper. United States. Microreactors and other continuous-flow reactors have received quite a lot of study. respectively. the analysis was performed with two API loading levels in the final drug product.25 and much subsequent work has been completed. US$ KI = key intermediate LLE = liquidÀliquid extraction NPV = net present value OEE = overall equipment effectiveness OpEx = operating expenditures. § ’ ACKNOWLEDGMENT This work was funded by Novartis Pharmaceuticals as part of the Novartis-MIT Center for Continuous Manufacturing. Again. Des. efficient chemistry for the particular product being produced is absolutely crucial. 83. analysis of smaller-scale production and the considerations of a multipurpose continuous production line rather than a line dedicated to a single product. Continuous granulation processes have also been widely studied. If the CM process with recycling has 10% lower yield than the batch plant.doi. more economic analysis and system-level research is also required. Apart from considerations of continuous pharmaceutical manufacturing unit operations. Other effective separations technologies may become more promising under continuous mode as well. For example. and plantwide modeling are all expected to lead to more economical processes. Some areas of continuous processing in pharmaceuticals are well understood. I dx. appropriate solvents and conditions must be chosen for each specific purification step in each particular manufacturing process. ’ AUTHOR INFORMATION Corresponding Author *E-mail: trout@mit. Particularly in the case of high KI price. Present Addresses Department of Chemical and Biochemical Engineering.31 The recent research is promising: one study found that the purification of an API using a continuous oscillatory baffled crystallizer is much more predictable to scale up than batch crystallization.31À43 Despite recent promising results on continuous crystallization of pharmaceuticals. using a batch process and four continuous processes. three prices for the most expensive KI organic feedstock. Res. K. as well as the economic implications thereof.1021/ie2006752 |Ind.44À47 Vervaet and Remon4 wrote a review article on six different methods of continuous granulation. studies on continuous crystallization for pharmaceuticals has been lacking until recently. When combining the economic advantage with more consistent product quality and greater regulatory freedom. ’ ABBREVIATIONS API = active pharmaceutical ingredient BLIC = battery-limits installed cost. The authors are grateful to the anonymous reviewers. savings can be achieved for all scenarios except the highest KI price. in those two yield scenarios. Many opportunities for further study exist: developing more efficient chemical routes.Industrial & Engineering Chemistry Research estimated 57% CapEx savings by shifting the process from batch to continuous Rutgers University. The authors thank Kristopher Wilburn (MIT) for vendor price quotations and insight into the batch process and the full Novartis-MIT Center for Continuous Manufacturing team.21 A study on production of fine chemicals on dedicated batch versus continuous equipment found that continuous production is economically favorable at all production levels studied (as low as 200 t/year). final dosage form production. Eng. Piscataway.4. US$ CCM = Novartis-MIT Center for Continuous Manufacturing CM1 = novel continuous manufacturing process CM1R = novel continuous manufacturing process with recycle DP = drug product (final dosage form) DTF = direct tablet formation FOB = free on board (cost of equipment before delivery).23. Chem.20 A study on production of cell culture media estimated overall cost savings at 34% for switching from batch to continuous production at 100 000 L/year capacity.31 The easy scaleup in this type of continuous crystallizer compares favorably with the 10 different schemes for scaling-up batch crystallization enumerated by Lawton et al. how much time is needed to change between products. when the production line has not reached steady state. this case study of a specific pharmaceutical product found results similar to those found in other industries.18.18.48 ARTICLE process with recycling (CM1R) with the novel direct tablet formation process in all scenarios tested if overall yields for the continuous process meet or exceed those of the batch process.22 separations technologies. especially Berthold Schenkel. New Jersey. and how much waste material is generated during startup and shutdown. the maximal savings can be achieved by choosing process CM1R with the novel direct tablet formation process.24 however. specifically. The break-even KI price is $1700/kg. 730–738. Chem.

D. J. Biotechnol. H. T. 14.. Continuous Twin Screw Extrusion for the Wet Granulation of Lactose. 2000. 273. 3949–3957. A. The Reality of Continuous Processing. Sk.. M. Continuous Mixing of Powder Mixtures with Pharmaceutical Process Constraints. G. Continuous. F. Fyles. F. B.. N. Edwards. Bioprocess Eng. M. Int. 10. (45) Wind. H. 231–244. Perry. J. K. Vrabel. In Proceedings of 3rd International Conference on Bioreactor and Bioprocess Fluid Dynamics. 19–29. 37. M. J. 905–910. Manuf. G. The Role of Process Simulation in Pharmaceutical Process Development and Product Commercialization. Sci. J. Chemical Reactor Design. Kawasaki. 47.. J. Glavic. M. Process 2008.. A 2000. Baker Perkins MPF50D. A.. L... S. P. R. 12.. 8th ed. Chem. Pharm. A. (46) Yoon. M.. Olbjer.. 2010. 283–288. G. D.. Eng. T. H. E. P. C. Eur. 2002. M. P. Needham. Steffens. J. Kluwer Academic Publishers: Dordrecht. K. Pharm. J.. (32) Nienow.. (22) Gorsek. M. F. 2010. C.. D. 81–90. T. Sci. Johannsen. 64. 000–000 . M. K. E.. 322–329. Boon. 1978. Nienow. 118–128.. J. B. Boston.. Eng. Smith. Harsh Reaction Conditions in Continuous-Flow Microreactors for Pharmaceutical Production. MD. M. 60.. Chem. R. M. Kim. Nienow. 1997. J. R. (36) Kraume. C. Zhao.—Eur. F. (41) Kim. van der Schot.... W. N.. Wagner. J. Roberts. Katz. Microreactor Technology: A Revolution for the Fine Chemical and Pharmaceutical Industries? Chem. R. 1357–1363. New Developments in Chemical Engineering for the Production of Drug Substances. K. Eng... K. D.. R. Eng. P. 895. Technol.. G. Des. R. N. XXXX. Pandit... 843–853. Eng. (7) Kockmann. 3089–3097. McGraw-Hill Professional: New York... D. (35) Smith. Chemical Reactor Development: From Laboratory Synthesis to Industrial Production. Commun. Hoffmann. Parasu Veera. (14) Couper. P. 2004. (42) Evangelista. J. W. Zehner. J. Chem. Bieler. Wang. van der Lans. (13) Berthiaux.. 1990. T. UK. A. (23) Glasnov. 1986. (43) Li. (24) Brocklehurst.. M. 3585–3598. 2009. 2010. (19) Crosby. J.. 2nd ed. (16) Ulrich.. 22. Gr€n. 30. B. McGrawHill: New York. Design and Process Aspects of Laboratory Scale SCF Particle Formation Systems.. Chung. In Perry’s Chemical Engineers’ Handbook. Biochem. Twin Screw Granulation as a Simple and Efficient Tool for Continuous Wet Granulation. Chiral Capillary Electrophoresis As Predictor for Separation of Drug Enantiomers in Continuous Flow Zone Electrophoresis. D.. 9. Shering. Chem. (47) Peper.. K€ppers. L.. Manelius.. M. C. B. Ind. Twin screw wet granulation: Granule properties. I.. D. Dev.. Laird. R.Industrial & Engineering Chemistry Research (2) McKenzie.. E.. Kockmann. L. R. Kiang. 36. Chem. Boston. 1994. Process Design and Economic Studies of Alternative Fermentation Methods for the Production of Ethanol. 15. Technol.. Ducry. CRC Press: New York. Continuous Extrusion Using a Baker Perkins MP50 (Multipurpose) Extruder. 28. W. Penney. 1682–1694. Biotechnol. Eardley. Shinnar. S. J. C. Microreactor Technology and Continuous Processes in the Fine Chemical and Pharmaceutical Industry: Is the Revolution Underway? Org. W. Eng. J. Design of Batch Versus Continuous Processes Part III: Extended Analysis of Cost Parameters. M. Engineering. pp 9À1.. W. 20. Process Engineering Economics. 2005. Microbiol. J. Eng. (5) Behr. Ewers. Cartwright.. (44) Vemavarapu. L€bbert. Wilke. Sep. G. J. Process Res. Continuous Granulation in the Pharmaceutical Industry. Tufvesson. H. 1791–1798. Catal. Int. Baltimore. 9À56. Chem. 318–323. Ind. 285–291. pp 776À779. Kimmel. Synth. Drug Discovery Dev. Mixing in the Process Industries. Continuous Crystallization of Pharmaceuticals Using a Continuous Oscillatory Baffled Crystallizer.. J.. 2004. Sci. Drug Dev. Dev. 99. Concept for Scale-up of Solids Suspension in Stirred Tanks. 52. J. P. Eng. Mollan. T. Technol. 2003. Can. A. L. J. Influence of Intragranular Porosity and Liquid Saturation. (38) Hoeks. American Association of Pharmaceutical Scientists Conference on Drug Product Manufacturing. Rubin. V. A. A. Enabling Continuous-Flow Chemistry in Microstructured Devices for Pharmaceutical and Fine-Chemical Production. W. Chem. Butterworth-Heinemann: Woburn. (6) International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use. Fair. A Plant-Wide Control Strategy for Continuous Pharmaceutical Manufacturing. J. H. 3990–3994... pp 347À354. Eng. P. R. L. Pharm. Kim. Can Pharmaceutical Process Development Become High Tech? AIChE J. Biopharm... Chem. and Scaleup. J. (37) Thoenes. Vermeire. Cady... 2005. 22. Turbulent Flow in Stirred Tanks: Scale-up Computations for Vessel Hydrodynamics. J. Mechanical Engineering Publications Ltd.. Adv. I. Wet Granulation in a Twin-Screw Extruder: Implications of Screw Design. (39) Shukla. 2002. Extrusion of an Effervescent Granulation with Twin Screw Extruder. Blomsten. (40) Whitton.. 14. M. Optimization.. R.. 2005. Pharm. A. Holm. C. Ind. 183–194. R. Battistoni. A Novel Design of Simulated Moving Bed (SMB) Chromatography for Separation of Ketoprofen Enantiomer. 1421–1444. White. J. L. 2005. (3) Leuenberger. (18) Dhenge. M... Chem. R. 2002. Dev. Brehme. B. J. C. 2545–2566. Elsevier: Amsterdam. C. Chem.. P. (20) Cysewski. Pharm. 2008. Rainone.. Braatz. Eds. Life Sci. Marikh. Wolff. Scale up of Stirring As Foam Disruption (SAFD) to Industrial Scale. M. Studer. E. (48) Lakerveld. Eng. J. 12. Vermeire. D. Green. 2003. 1–8. T. 1988... J. Zimmermann. Bioeng. Remon. Toward a Continuous-Flow Synthesis of Boscalid. Sci. A. Scaleup of Biotransformation Process in Stirred Tank Reactor Using Dual Impeller Bioreactor. V. Cretton. 32. R.. Pharm. o u S. M. 2009. Biotechnol.. R. Thormann. Futran. Gottsponer. Ein Neuartiges System f€r die u Kontinuierliche Feuchtgranulierung. Chem. J. (9) Roberge. Vol. Res. 2002. Bujalski. J. Lehmann. P. 2004. 164. Technol. Process Res.. Harnby. Koulouris. 52. C.. 2004. M. A. Lagonikos.. (4) Vervaet.. G. (25) Gamlen... Pharmaceutical Development Q8(R2). 2001. J ARTICLE (26) Lindberg. O. M. Gatumel.. O. Tom. 2006. P. Symietz. M. Vervaet. Int. Kubiak.. J. Pharm. 7470–7. N. Pharm. B. Hertz. 352. A Scale up of Stirred Tank Contactors for the Liquid Membrane Permeation of Hydrocarbons. XXX. U.. D. pp 600À605. Zimmermann. (31) Lawton. (28) Schroeder. Chromatogr. 292. Nitration Chemistry in Continuous Flow Using Fuming Nitric Acid in a Commercially Available Flow Reactor. M. 2004. A. 80.. Kulkarni.... A. New Trends in the Production of Pharmaceutical Granules: Batch versus Continuous Processing. C. 1701–1713. (29) Keleb. Chemical Process Equipment: Selection and Design. (11) Kockmann. (12) Thomas.doi. Ind. F. M. C. 108. 51–65. T.. 1993. 8. Cytotechnology 1996. J. 1–16. 13. Sep. A.. Org. 674–681. N. D.. Eyholzer... 2002. 49–73. MA.. 2005.. W. W..1021/ie2006752 |Ind.. Remon. M. D. (27) Keleb. Roberge. Eng. R.. (8) Roberge. Enhanced Capital Productivity Through Continuous Processing. Curr. 1984. Barton. 2008. Roberge. Org. Drug Dev. Wilkinson. A. 2009. Gottsponer. E. Placek. F. 289–296. Sun. G. Separation u of Ibuprofen Enantiomers by Supercritical Fluid Simulated Moving Bed Chromatography. G. L. 15–24. S. N. M. Eng. Scale-up Criteria for Stirred Tank Reactors.. E. N. 2001. Process Res. A. J. et al..: Cambridge.. Remon. Park. J... Ed. (30) Thompson. 239.. C. 8. (15) Petrides. Eng.. 2010. C. 2002. 2315–2322... H. H. Continuous Processes for the Production of Pharmaceutical Intermediates and Active Pharmaceutical Ingredients. 69–80. W.. Opin. 78. Kappe. 2008. I. 2011In Press. C.. Scale up Study of Retreat Curve Impeller Stirred Tanks Using LDA Measurements and CFD Simulation. J. P. (21) Jayme. 93.. Meet... L... P. 738–745. 2001. 2007. A Guide to Chemical Engineering Process Design. Pharm. AIChE Annu. (10) LaPorte. 2090–2103. (Amsterdam) 2010. dx. Brunner. I. (33) Couper. AIChE J. R.. Wiley: New York. P.. D. F. Chapter 8. Vecchia. (17) Couper. D.. Steele. 4. Res. 255–261. L. Lodaya. Vervaet. Zimmermann. High Capacity Reconstitution of Nutrient Media from Concentrated Intermediates. T. Tavlarides. (34) Nauman. W.

Sign up to vote on this title
UsefulNot useful