We give a lot of introductory webinars. Most of these are focused on continuous processes, because the processes and the data are easier to understand, especially as an introduction, but we get a number of questions afterward about extensions to batch processes, making us realize that we don’t normally mention them enough. We do occasionally give a webinar focused on batch, semi-batch, and transient (start-up, shutdown, grade transition) operation, but there’s a need to give it wider exposure.
If you don’t read the rest of this, the best thing you can do right now to improve your batch analysis is making sure that the batch number and phase number are captured in your plant historian off the batch control sequencer. This will make future analysis easier regardless of the tools or techniques you use.
Of our current users, batch processes are common. I’d estimate that about 25% of sites are primarily/exclusively batch, and another 40% have hybrid or batch operations in some major units. When cyclic operation like pressure swing absorption is included, the total probably reaches 90%. And everyone has start-ups, so it has broad applicability.
Batch process analysis still takes all the tools used with continuous processes, so it’s a good basis, but there are a number of specialized approaches and tools to deal with the time-dependent nature. Whereas quality data in continuous processes is usually linked to process operation by timestamp (the time the sample was taken relative to the time that material was processed), in batch processes we normally have quality measurements taken for individual batches, making batch number the linking element.
CVE has built-in tools (Join-by-Exact) to align batch records, including final quality data, with the batch process history. These same tools can be used to bring in data for lots of feed material used in the batch, constructing a wider view of the batch. It is really helpful if the batch number being processed at each point in time is stored in the process historian. This is the first thing you can do to help your future self now. If you don’t already store the batch number, have it added. Another thing that the batch number can be used for is relative indexing. This is just computing the time elapsed in the batch. This is obviously essential for comparing batch trajectories, time of maximum temperature, etc. What is also useful, but few use now, is the time until the end of the batch.
This can be really helpful for improving batch times, comparing batches in a different way. The same tools can also be used to compute time elapsed and remaining in phase. This really benefits from having the phase number in the historian. If this isn’t currently stored in your historian, get it in there now, future you will thank you. Focusing on and comparing performance within individual phases is very easy with CVE, and provides a powerful tool for identifying key batch differences. If phase information hasn’t been captured in the historian, it can often be retroactively added with CVE, based on signatures that change at the start of individual phases. This takes extra effort and care, though, and is more error-prone than pulling it from the historian.