12 Method Development
289
is built over the range of intensity and several lamps (as lamp-to-lamp variability
can be expected), and if necessary, limits on lamp intensity should be set. If the
instrument is in a space where the ambient temperature will vary, variability in the
spectra can be observed. Finally, if the instrument is deployed in-line or on-line,
reproducibility of the positioning of the probe or probe fouling can have a significant
effect on the resulting spectra. This is also the case for sample holders for at-line or
off-line measurements. Any change in how a sample is presented to the sampling
system will increase the error of the measurement.
12.3.6 Method Testing and Validation
As discussed in the introduction, a model is only a part of an analytical method.
Method testing and potential validation does not solely mean applying independent
spectra to the model but also ensuring the spectrometer is deployed as intended
and that the model outputs are provided to the reporting system. Testing should
be performed on relevant samples (independent from calibration and if possible,
from different lots of materials) to evaluate the impact of the risks identified in the
risk assessment on the method performance. If new factors affecting the method
are identified during testing or if the experimental work proved to not adequately
address some of the risks previously identified, the method and model may need
to be updated. Evaluation of any new update to a model should be performed with
independent samples. If the method performs as expected, the method development
effort can be concluded.
12.3.7 Referencing, System Suitability, and Performance
Monitoring
During routine use, the method must be used within the conditions it was built for, the
suitability of the instrument must be confirmed, and the performance of the method
must be monitored. Ensuring that the instrument, sampling, data collection, etc.,
are consistent between method development and method use is critical. While the
samples will change, procedures should be in place to ensure that all the variables that
could affect the method performance are controlled. If there is a significant difference
between the conditions where the method was built and the routine manufacturing
conditions, the method may need to be updated. That scenario will be discussed in a
next section.
289
is built over the range of intensity and several lamps (as lamp-to-lamp variability
can be expected), and if necessary, limits on lamp intensity should be set. If the
instrument is in a space where the ambient temperature will vary, variability in the
spectra can be observed. Finally, if the instrument is deployed in-line or on-line,
reproducibility of the positioning of the probe or probe fouling can have a significant
effect on the resulting spectra. This is also the case for sample holders for at-line or
off-line measurements. Any change in how a sample is presented to the sampling
system will increase the error of the measurement.
12.3.6 Method Testing and Validation
As discussed in the introduction, a model is only a part of an analytical method.
Method testing and potential validation does not solely mean applying independent
spectra to the model but also ensuring the spectrometer is deployed as intended
and that the model outputs are provided to the reporting system. Testing should
be performed on relevant samples (independent from calibration and if possible,
from different lots of materials) to evaluate the impact of the risks identified in the
risk assessment on the method performance. If new factors affecting the method
are identified during testing or if the experimental work proved to not adequately
address some of the risks previously identified, the method and model may need
to be updated. Evaluation of any new update to a model should be performed with
independent samples. If the method performs as expected, the method development
effort can be concluded.
12.3.7 Referencing, System Suitability, and Performance
Monitoring
During routine use, the method must be used within the conditions it was built for, the
suitability of the instrument must be confirmed, and the performance of the method
must be monitored. Ensuring that the instrument, sampling, data collection, etc.,
are consistent between method development and method use is critical. While the
samples will change, procedures should be in place to ensure that all the variables that
could affect the method performance are controlled. If there is a significant difference
between the conditions where the method was built and the routine manufacturing
conditions, the method may need to be updated. That scenario will be discussed in a
next section.