top of page

Cohesive Powder Feeding Modelling

October 7, 2020

Continuous manufacturing is emerging as a reliable, robust, and even preferred means to produce oral solid dosage forms. As dosage unit potency is determined by the relative flowrates of the solid components, precise control of powder feed rates and levels of variability is essential to adequate control of product quality.  Material is typically introduced into a continuous process by loss-in-weight (LIW) feeders. LIW feeders consist of a reservoir hopper filled with powder sitting above horizontal screws that rotate at a controlled speed.  The entire assembly sits atop a scale and change in mass is recorded over time as the screws rotate and pull material from the hopper and dispense at the end of the screw.  Screw design and speed can be altered to achieve a wide range of flow rates for a given material.  However, consistent feed rates can be difficult to achieve due to the cohesive nature of materials in pharmaceutical formulations, most notably the active pharmaceutical ingredient (API). The API is typically in the micron particle size range where interparticle forces are similar to the gravitational force per particle.  High feed rate variability might result in high unit dosage variability, which may pose challenges to continuously manufacture some oral formulations.  In addition, as multiple kilograms of material are necessary to fill the reservoir hopper to relevant fill levels, proper evaluation for feasibility of continuous manufacturing may have to be delayed until later in process development.  At this point switching to continuous processing from typical batch processing (or the reverse if the evaluation is unsuccessful) can be costly and delay development and filing timelines.  
The objective of this proposal is to seek a partner to develop a model that scientists, operators, and engineers can use to predict feeding performance (flow rate and associated variability) of a granular material based on a short list of material properties, which can be measured on a small sample of material.  Such a model can provide appropriate estimates of probabilities of success for continuous manufacturing early on during development.  

.Download the Request for Information and submit your response

RFP ISSUED October 7, 2020



Q&A Received (Updated 11/09/2020 - newest questions on top)

  1. Should the model address the refill stage of feeding. This could greatly complicate model complexity. The inclusion of the refill stage of feeding is desired, but not necessary.  Please describe the inclusion or exclusion of the refill stage in the proposal. 

  2. What range of LIW feeder hopper fill levels should be considered? 80%-20% or something else. The larger the range of hopper fill level needed will add computational burden than can affect the time period needed to develop the model. Applicability of the model across a 0-100% fill range is desired, but accommodation of a narrower operating range, e.g. 80-100%, can be accommodated, provided the proposal includes suggestions for extrapolation of performance and outlines the process to diagnose deviation of model predictions from actual performance outside the specified operating range. 

  3. It is possible to select from a variety of screw designs for commercial low-in-weight feeders, which can have an impact on mass flow rates for cohesive materials. For instance tests by W. E. Engisch and F. J. Muzzio, “Loss-in-Weight Feeding Trials Case Study: Pharmaceutical Formulation,” J. Pharm. Innov., vol. 10, no. 1, pp. 56–75, Mar. 2015, doi: 10.1007/s12247-014-9206-1. Indicate coarse auger screws exhibited adhesion and long residence times for API whereas a coarse concave screw was self-cleaning and eliminated that problem. There appears to be a variety of screw designs who’s selection can affect feed rate performance in different ways.  The RFP indicates that "Commercial feeder equipment with standard screw designs and integrated LIW system.” be used. Can the team provide more detail on what types of screw designs they are contemplating and how many designs are they wanting a model to accommodate? We agree that the feeding performance can be a function of the particular screw and feeder design, in addition to material properties.  With the numerous combinations available, it’s overly prescriptive at this point to specify a given feeder or screw type.  Our preference is for a model that is screw type and feeder agnostic, but at the very least it should have factors to accommodate major equipment features, e.g. coarse vs. fine augers.  Please include the feeders and screw types in the proposal for consideration by the partners. 

  4. Requirement in Section 2.3.1: “Appropriate software with minimal access requirements or license management, for example open source downloadable code in Python, Matlab, or Excel, or a web-based portal for user access control and model execution. The source code for the tool must be easy to work on and maintain on a variety of operating systems and security protocols”  Could you please provide more detail on this requirement? Is the motivation behind having open-source access to a model driven by a desire to customize the model, in which case, licensed software that is customizable may meet that requirement? Or is the desire to have a tool that does not require additional commercial investment beyond software licenses that are already in place at the member companies, such as Excel and Matlab? If all of the eventual ETC project sponsors already have licensed access to a commercial modeling platform and the to be developed models are provided in open source within that platform, will that platform fall into the same category as Excel and Matlab for the purposes of this requirement?  What does the ETC have in terms of requirements to deal with the challenges of open source models, e.g. quality assurance? The partner companies are flexible with the final form of the tool with EXCEL, Python, and Matlab as the preferred platforms that are most readily deployed at the member companies.  However, if the tool is deployed in another tool widely utilized by the member companies, that is also acceptable.  More important is the function and access of the code to 1) be deployable with minimal tinkering from the companies for a standard use case defined in a transfer protocol and 2) be customizable to specific needs of each company, i.e. the members do not want a black box.  Final determination will be made by the participating companies based on the tool and deliverables described in the proposal.

  5. This RFP suggests a project with two distinct components: (1) collection of experimental data of various material properties and the performance of those materials on a variety of commercial feeders, (2) development and implementation of mechanistic and/or empirical models that describe the relationship between material properties and feed factor in a software tool, including the material database generated in part 1. We are best suited to perform part 2. Would you consider awarding this as two separate projects? If not, we can hook up with one of its partners to provide a joint proposal covering the entire scope.  The proposal can either be joint or from a single vendor.  We will not be selecting two proposals, but may consider suggesting the pairing of proposals where appropriate.  Industrial partners are willing to share data, but it is preferred to have the vendor generate a “fresh” data set for this work.  Pending agreement of the participating partners, it also may be acceptable to use publicly available datasets. 

  6. A requirement is stated as, “A model supplementing that available in the existing literature is to be developed and equations shared publicly with the scientific community for peer review and comment.”  Will only those proposals that introduce new science be considered, or will proposals that implement recent literature also be considered? Will purely empirical correlations (statistical models linking material characteristics to a feed factor) be considered?  The member companies will consider all proposals, but there is a preference for new science to be developed, as there is a gap around adequate prediction of the flow of cohesive powders from twin screw feeders in today’s literature.  Purely empirical models may be considered, but the accompanying dataset will need to be shown to be adequately large. 

  7. The requirements state: “Member companies request access to a prototype tool during development, with full access to the source code at the completion of the project period.”  Will it be acceptable to make the prototype tool available at predefined milestones during the development process? Testing by member companies at pre-defined milestones is appropriate.

  8. A series of round robin testing by the ETC Members may be executed to confirm the applicability of the methodology and model approach utilized on ETC member equipment with ETC member materials.  Will this be used to determine acceptance criteria for the project? If a particular material or equipment does not produce results that are consistent with the model’s predictions, how will it be decided whether that is due to experimental or model error? What is the tolerance level for inaccuracy in the predictions? A protocol will be developed and aligned between the member companies and the vendor to test that the model works as intended.  Quantitative ranges will be established after prototype evaluation.  Evidence will be generated for the performance of the components of a relevant, yet challenging, pharmaceutical formulation to be defined during the course of the agreement.  The solution will be tested according to the protocol, but then will be challenged for additional pharmaceutical formulations – a blind test by the partners.  Success will be determined based upon the outcome of tests on multiple materials from different feeders by multiple partners.  If known, please include any estimates of prediction accuracy in the proposal.

  9. Our experience indicates that particle shape is an important factor in feeder performance as well as cohesion and other factors. Is shape a specific parameter of interest to characterize along with cohesion or to be included as a secondary contributor? For an industrially implementable model, shape is a hard thing to easily assess regularly and quantify or characterize along a continuum.  If shape is a critical parameter for the model, the model must also be able to make predictions assuming a spherical particle shape.  We welcome proposals for a consistent way to define and quantify shape as part of the material characterization element

  10. Is there a length/format restriction to the proposal (section 5.1)?  No there are no length or format restrictions.

  11. Can you provide an annual cost range for typical ETC projects?  This varies greatly and not something we publicize so unfortunately I cannot provide this information.  The project should be priced based upon what is required to deliver what is being proposed.  If ETC is interested and moving forward with a project but the available funding is not sufficient, ETC will work with the 3rd party to look for ways to reduce the cost to match the available budget (e.g., reduce scope, change timeline, external funding, etc.).

  12. In section 5.2, no functional requirements or specifications are listed, but some are discussed in section 2.2. Should the table in section 5.2 be left blank or do we fill this in ourselves using the functional requirements and specifications we identify?  Yes, this is expected to be filled in by the respondent as you outlined.

  13. How involved will the project sponsor(s) be in the project? Are there required monthly and/or quarterly progress reports/meetings?  This varies by project but there is typically some involvement with the project sponsor(s) SMEs to provide feedback and test intermediate deliverables.  The frequency of those meetings can be determined between the team and collaborators but at a minimum need to occur quarterly.

  14. Will all the materials identified by the sponsor(s) be provided to the vendor and include required measured material properties?  In section 2.2 the items that ETC will provide are there.  If additional resources are required, please highlight these in your proposal as requirements.  If these are non-proprietary materials they can be provided through the agreement with ETC… if proprietary materials are required, there would need to be individual agreements executed between those companies and the 3rd party collaborator.

  15. If a vendor does not have all the needed laboratory equipment, is it possible for a sponsor to provide it on loan to the vendor or allow the vendor access to their facility to use the needed equipment?  Section 2.2 has the expectation the selected collaborator will have access to laboratory facilities.  There may be opportunities to come on site but with COVID-19 this may not be possible and would require a separate agreement.  Again, if you need access to equipment please highlight these in your proposal as requirements.

  16. We are interested in submitting a proposal for the cohesive powder feeding model development project. The ETC webpage has provided sufficient information for us to start preparing the proposal. One question that we would like to get some more clarity on is ETC's position on proprietary tools - for instance, would there be interest in a tool that utilizes a commercial (proprietary code) particle solver with a flexible API.  The GUI built using the API can be made available as an open source application. As stated in the RFP, the intent of the project is to deliver a solution that is freely available to the community via an “open-source mechanism” that could allow continued development.  However, if you would like to propose a proprietary/commercial solution that can satisfy the requirements we would be interested in learning more about that and it would be considered by the team.  ETC has partnered with several vendors to support development of new commercial products through providing seed funding and feedback so this sort of relationship would be aligned here.

bottom of page