JMP has many wonderful customers so it was a delight to team up with Somaieh Mohammadi and Gwenola Ninon from Fujifilm Diosynth Biotechnologies for a Chemistry World webinar. For more details and to access the recording of this please click here.
We covered very important challenges in scaling processes including:
- Planning small scale runs that either unlock the perfect settings or trial alternative process conditions.
- Getting the full picture by comparing scales and qualifying the smaller scale.
- Convincing your organisation to adopt this sophisticated approach by conveying process understanding and clear benefits.
The following materials show you how to explore this approach. Don't have JMP just yet? Download your free trial here.
Some prerequisites:
- Understand what you want to achieve and the desirable outcomes/potential benefits. Ultimately, scaling a process requires resources, such as time and money so it's important to understand if there will be sufficient payoff for this activity.
- Plan how to scale, will this be a scale up or down experiment? What are your Critical Quality Attributes (CQA) and Critical Process Parameters (CPP)?
Step 1: Input your CQAs and CPPs into a DOE that is appropriate. If in doubt, it's recommendable to use the Custom Design option, more details in a worked example here. With limited resources, this particular scenario called for a Definitive Screening Design.
Step 2: Perform experiments and input results into the data table generated by JMP after the DOE.
Step 3: Explore your hard earned data! One way to do this is with the Fit Y by X platform, have a scroll through the separate univariate analyses below. Can we make actionable recommendations from these analyses? Personally, I don't think so, there are potentially multivariate relationships between the CPPs (Feed Rate, Catalyst, Agitation, Temperature and Time) and the CQA (Yield).
Step 4: Use the Fit Model (click for a great @HydeMiller talk) platform to establish a model. It is now crucial to interrogate that model, imagine the questions that are required to be asked to be reassured of any actionable insights. The Prediction Profiler (@phil_kay presents a deep dive using the PP), is a brillant tool to help you and your colleagues start asking those questions. Remember you want to transfer from a Parameter Space (all potential factor configurations) to a Design Space (all useful factor configurations that meet your desired CQA).
Step 5: With confidence in your Design Space at a certain scale, look to perform the experiments at another scale level. Below, I've performed experiments at four scale levels (AMBR Bioreactor, Pilot Plant 1, Pilot Plant 2 and Full Scale). As a process is scaled up we are to expect changes in CQA values.
Ideally, we would like these changes to be consistent across CPP ranges, this is seen between AMBR Bioreactor and Pilot Plant 1. Use the QbD Dashboard below to explore this relationship.
Sometimes, relationships are attenuated (the differences between scales diminishes with over the CPP range) or magnified (the differences between scales increases with over the CPP range). Can you see evidence of this in the comparisons between AMBR Bioreactor and Pilot Plant 2 or Full Scale? Ultimately, these are both situations where our process model isn't robust across scales...we may need to explore other CPPs or CPP relationships.
In Summary
JMP is a fantastic tool to assist you in scaling your processes. The set of tools available enable you and your organisation to generate and then utilise processes understanding.
During the webinar, the poll question asked "Which one of these challenges that Design of Experiments tackles is most applicable to your organisation?"
- Reducing the number of experiments (28%)
- Setting a time frame and budget for experimentation (12%)
- Understanding new cutting edge processes (6%)
- Improving/innovating current processes (38%)
- Communicating process understanding effectively (17%)
With 38%, it is clear that organisations are interested in improving their current processes. Would your organisation benefit from being able to scale their current processes robustly? If so, do feel free to get in touch; either below in the comments section, filling the form here, or emailing me ben.francis@jmp.com
Recommended Articles:
Sandner, V., Pybus, L.P., McCreath, G. and Glassey, J. (2019), Scale‐Down Model Development in ambr systems: An Industrial Perspective. Biotechnol. J., 14: 1700766. doi:10.1002/biot.201700766
Veronika Debevec, Tijana Stanić Ljubin, Žiga Jeraj, Tanja Rozman Peterka, Borut Bratuž, Dušan Gašperlin, Stanko Srčič & Matej Horvat (2020) Step-wise approach to developing a scale-independent design space for functional tablet coating process, Drug Development and Industrial Pharmacy, 46:4, 566-575, DOI: 10.1080/03639045.2020.1742140
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.