CASE STUDY: Continuous improvement in efficiency of antibiotic production
In this case study, BHS-Sonthofen looks at how production process redesign, implementing a rotary pressure filter, can improve yield.
The company
BHS-Sonthofen is an owner-operated group of companies in the field of machinery and plant engineering based in Sonthofen, Germany. The company offers technical solutions for mechanical process technology, concentrating primarily on mixing, crushing, recycling and filtration. BHS-Sonthofen has a global presence, with more than 350 employees and several subsidiaries.
The production scenario
A global pharmaceutical company, producing a highly efficient new drug based on the traditional batch process method, encounters rapid success that exceeds internal sales forecasts after market introduction. The company launches a project to increase production and the drug rapidly develops into a blockbuster, calling for a production increase that more than doubles previous capacity.
An analysis of the entire production process is performed to streamline and increase efficiencies in production as quickly as possible.
Previous process
The previous production process adheres to the traditional ‘reaction — separation & washing — crystallisation — drying — packaging’ process chain. Each of these steps involves an autonomous batch process, so the overall production batch is produced in independent steps and in sequence.
The daily active ingredient (solid) production yield of approx. 120 kg is manufactured in a 4,000 L reactor over the course of 8–10 hours. The yield is then transferred to a single-plate Nutsche filter that can accommodate the entire batch.
This filter separates the reaction residue from the mother liquor that contains the active ingredient and rinses it with a solvent to increase the final yield. The solvent is then flushed out with water to allow for proper disposal of the residue. This process step takes between 10–14 hours.
Filtrate samples taken throughout the process ensure the required product quality. However, this measure also introduces some uncertainty regarding the duration of the individual processing steps.
The time to complete each process step can vary depending on the cake structure, washing agent distribution, crack formation etc. Also, changing from turbid to clear flow at the beginning of the process, as well as mixing in washing agents, can lead to a significant loss of the active ingredient. In the following process step, there must be no solids in the active ingredient solution. This reduces the final utility of the original reaction batch to the so-called ‘mean production’, which constitutes roughly 75% of the active ingredient generated in the reactor.
Finally, the active ingredient is precipitated from the cleansed solution in a crystallisation step and then dried in a spray dryer.
The review
The project team detected several critical process steps after analysing the existing production process:
- The reaction batch is always prepared at the start of the late shift, as the reaction can take place without the need for major manual intervention overnight. However, the subsequent separation and cleaning in the Nutsche filter requires several manual actions that must take place during daytime. This means that despite an actual processing time of approx. 10 hours, a single batch alone can be produced in a 24-hour period unless major additional investments are made to upgrade the existing infrastructure.
- Also, the amount of active ingredient produced in the reactor exceeds the actual yield after drying. Most of the loss occurs in the Nutsche filter. The turbid flow stage at the beginning of filtration causes the loss of a large amount of the active ingredient. Further loss takes place due to limited carry-over of solvent into the crystallisation process. An insufficient load alters the crystallisation behaviour, thus resulting in a grain distribution outside the specified range.
- The crystallisation and drying facility is used for just a few hours each day. Yet, due to the long batch filtration cycle time, multiple runs cannot be performed with certainty. Valuable potential is thus wasted. A second reactor for alternating production could theoretically double the output if the downstream process allowed for such a setup.
Looking for a way to significantly increase the production capacity, the project team determines that greater output can be achieved either by installing a second production line or by leveraging the unused potential, ideally in combination with improving the yield.
Based on these insights, the separation stage is found to be the production bottleneck. The project team consequently analyses what causes problems at this stage and works out possible solutions. Extensive lab and pilot tests are executed to gauge various versions and configurations to modify the existing Nutsche filter. This includes ‘disruptive solutions’ that challenge the entire batch-based production concept.
In the end, the advantages of continuous production convince the project team to change the underlying concept of the existing line. Therefore, project management recommends modifying the introduction of a continuously operating BHS Rotary Pressure Filter (RPF) and addition of new reactors.
The solution
A continuous solution increases production to the order of 150%, while incurring only a fraction of the investment costs of the previous manufacturing process. Additionally, changing to continuous filtration increases the yield by nearly a third. The key to achieving these benefits is the ability to monitor the continuous filtration and cleaning process online.
The first challenge relating to this concept is to generate a continuous production flow from the batch reactor. As it is not possible to change the reaction itself, either buffer containers need to be introduced to stretch batch production for downstream processes, or a second reactor needs to be installed to enable alternating operation.
A fully continuous separator such as the BHS RPF can process the resulting 8,000 L of solution produced daily to filter out 240 kg of the active ingredient (in a continuous product flow of approx. 340 L/h). A filter surface of just 0.5 m² is fully sufficient for production. The previous 4,000-L batch system formed a 30- to 50-cm-high filter cake from all solids contained in the batch. In contrast, the BHS Rotary Pressure Filter of type RPF P02 reduces the cake thickness to about 10 mm, which results in lower cake resistance.
With a thinner filter cake, the filtration pressure can be lower. This enables the use of a dense filter medium, which results in clear flow, removing the need for the turbid flow stage. By removing this stage, loss of the active ingredient is reduced from approx. 25% to less than 5%.
Moreover, the compact structure of the filter cake facilitates solid washing. Additional online monitoring of the washing filtrates allows the operator to quickly optimise the quantity of the washing agent. Therefore, the pharmaceutical company is consequently able to reduce the amount of solvents by roughly 20%.
Optimised piston flow cake washing also means that now the entire washing filtrate is usable. Here, too, continuous processing in connection with process-analytical technologies (PAT) leads to a more efficient use of consumables and thus higher yields.
The conclusion
Batch-based operation with a complex process flow, high consumable requirements and mediocre yield was improved by putting a continuous process in place. The success of this production redesign is in large part owing to the rotary pressure filter of type RPF P02 with an active filter surface of 0.5 m².