Biomedical Engineering Reference
In-Depth Information
Fig. 3 Dimensionality of
time-series and single-value
parameters
• Online data capture: Online data comprise real-time streaming data coming
from sensors installed on process equipment that captures process parameters
[e.g. pH, DO, temperature, conductivity etc.]. This capture mode has the highest
data volume and is recorded every few seconds or according to a set, periodic
data sampling frequency. Due to their high volume, these data are usually
retrieved from the native SCADA system storage (since there is limited native
data storage capacity), being compressed and stored by plant-wide data histo-
rians. The data from the historians are then decompressed on demand.
1.3.3 Data Types and Dimensionality
Captured process data can be in numeric, alphanumeric or purely text form.
Numeric data comprise parameter values that are either recorded manually or
captured automatically by the automation system of process equipment. Alpha-
numeric data are typically metadata comprising a numbering system used to
encode identification of materials (e.g. raw materials, product intermediate bat-
ches, equipment etc.). Finally, pure text data comprise all events and observations
related to the process execution. Metadata such as parameter names, product
names and equipment names also comprise full-text data.
The data for parameters (also referred to as time-series parameters) that are
captured at regular time intervals while the batch is progressing are three
dimensional in nature. The Batch IDs constitute the first dimension, the process
time the second dimension and the parameter names (e.g. daily viable cell density,
daily lactate concentration, daily temperature etc.) constitute the third dimension,
as shown in Fig. 3 . In contrast, parameters (usually referred to as single value) that
are captured once per process execution (e.g. step yield, harvest titre etc.) are two
dimensional in nature, with Batch IDs as the first dimension and parameters as the
second dimension, as shown in Fig. 3 .
1.3.4 Multi-Unit Batch Processing Involving Splits and Combinations
Bioprocesses are typically batch processes comprising many unit operations,
usually executed in series (Fig. 4 a). Processes that involve multiple cycles of
chromatography operations also involve batches running in parallel for that unit
Search WWH ::




Custom Search