The length of a design setup depends on several factors such as the number of numeric and categorical study factors, the investigated ranges and the stage of the development phase. Therefore, a general answer cannot be given.

However, as Fusion calculates specific equations and data models for every analytical response from scratch more experiments (i.e. number of different method conditions) are required for screening and optimization experiments compared to an approach where the equations (i.e. a chemical theory) already exist. Still, with the higher number of experiments compensation is already given for the experimental random error. This can only be accounted for by replicate injections using other technologies.

You should not mix basic and acidic pH values during a study, as these are two different chemistry systems. In which of these regions to perform the study should be investigated by a couple of pilot experiments.

You could either select one acidic and one basic pH value and define these buffers as a buffer type variable or you could run several pH values preferably using the OLP mode to investigate, which pH region would work best for your compounds. During screening you should then use pH as a variable with up to 6 pH values and a range of about 1 pH unit.

This questions probably results from indication such as 70% or 99% of your results will be correct.

Still, this cannot be a way to characterize accuracy in Fusion QbD, as every prediction will have an associated predicition error.

Fusion QbD will indicate the expected mean performance of your results together with a prediction error. The goal should be to keep the predicition error as low as possible. The last is the responsibility of the operator and we provide assistance of how to detect sources for bad predictive value and how to fix this.

Along with the prediction error, there always will be a random error, that is also characterized in Fusion QbD.

From this perspective, you can expect the the reported outcome will always be inside the reported error ranges.

Yes, from the calculated models simulation chromatograms can be displayed taking into account elution time, peak shape and peak resolution. In combination with the peak tracking feature peak names can clearly be assigned to the simulated chromatograms.

Fusion hasn't implemented a feature to scale a method to different column dimensions and operating conditions that give equivalent chromatographic performance.

However, there are plenty of  such tools on the market offered for free by vendors for chromatographic systems and analytical separation columns.

E.g. the Column Calculator from Waters is always shipped together with Empower 3 as a separate app.

The number of compounds actually is not a category Fusion QbD would deal with. We can demonstrate examples with more than 90 compounds.

Still, the more compounds, the more difficult it will be to identify method conditions with satisfactory overall resolution. The multifactorial approach with Fusion QbD will provide a much broader range of possible method conditions compared to any 2D or 3D approach and will let you know, what is possible inside the specified study ranges.

In addition, by extrapolation of the determined response models, promising method conditions outside the actual study space are indicated. Such extrapolation can therefore be considered as proposals that, of course, require verification by a follow-up experiment.

Yes, up to 3 gradient steps can be defined.

However, we recommend trying to avoid a segmented gradient, as with each step a further robustness issue is implemented into the method and a risk for method transfer. Fusion QbD has already proven to develop gradient methods without segmentation and with adequate baseline separation and acceptable run times in multiple customer studies.

Experienced Fusion QbD users know, that the introduction of gradient steps will be a well considered expection, as we can provide many more optimization parameters in a multifactorial approach.

Please also check out our article on this topic:

Segmented Pump Gradient

Fusion QbD 9.9 provides a patented peak tracking technology, that performs hyperaccurate peak tracking by the click on a button. Fusion QbD uses 3D PDA spectral data augmented with standard UV peak results data to automatically identify each peak in each experiment chromatogram. Fusion will also fully utilize 3D mass spectral data for experiments run on LC systems configured with a Waters Acquity single quad mass detector.

As any peak response can be analyzed also not baseline separated peaks can be evaluated using responses such as e.g. retention time difference or p/v ratios.

By defining the permitted and expected maximum variation of all relevant study factors an intended method transfer can already be respected during method development.

We recommend to include an initial isocratic hold time in the pump gradient and study this as a variable method parameter for simulating the dwell volume of different HPLC systems.

Please check the following articles:

Robustness Simulation

Considering Dwell Volume with Fusion QbD

Fusion QbD determines a multidimensional and multivariate design space, that includes robustness responses. This means robustness is not investigated and characterized after the establishing of an MODR only for single method condtions, but already proven for the entire study region from the obtained data models.

This can be reached by running a Monte Carlo Simulation using the determined models and applying process capability metrics (Cp, Cpk, Cpm, …) before taking the decision for single method conditions or an entire MODR.

2302_FQbD - Robustness Simulation