How do PKS database data interact with the station?
Thread Content
A problem arose during production: the company implemented a control system using Aspen at the upper layer of PKS, with PKS serving as the OPC server and Aspen handling data acquisition. More than a year after this system was in operation, strange problems were discovered. The historical curve of a data point collected by ASPEN is different from the historical curve collected by the station! In ASPEN, what is observed is a gradual process: the data decreases steadily over a period of about 5 minutes before there is a sudden change. Such a curve is generally accepted in terms of process dynamics. The curve collected by the station shows no gradual change in the first 5 minutes, only a sudden change at the end! The same problem exists in other steps of the same process. This problem overall seems quite strange. Because it is basically possible to rule out problems with the site or the controller; the only issues are those related to data collection by the database and communication between the database and the station. But when generating a sine wave manually in CB, the data in ASPEN and Station remains constant, or the data is forced to be the same across all readings… It seems difficult to understand, but that’s how it is. When data from the same database results in two different readings under certain conditions, it must mean that there is some deviation in one of the reading processes – that’s how I understand it. But my supervisors are too afraid to say this, for fear that the production team will claim that there are flaws in our control system! I wonder if anyone has encountered this situation before? Are there any means or methods to further investigate where the problem lies? Any suggestions or opinions are welcome; we’ll just be the lab rats, haha... Finally, if there is indeed some flaw in the station, how can it be identified, and how should we communicate with Honeywell? Everyone is welcome to offer suggestions and discuss!
Check the historical sampling period of the station
The historical curve of the station already uses rapid sampling, but due to system limitations, only 5-second sampling is supported; as a result, the gradual changes over time cannot be seen in the entire history, only the final sudden changes are visible. However, these 5 seconds are more than enough to capture changes that occur over those 5 minutes or so. Yet the data in the station remains very stable throughout that time, which is quite strange... Last year, an engineer from Honeywell came and mentioned an issue related to certain versions; many versions of PKS are unstable and contain various bugs, which is why there are continuous updates and patches released... I wonder if this is related to it
Looking at this post, it seems like it’s going to be deleted again… Actually, I came up with an idea: we could query the data for that specific point in the real-time database. But I don’t know how to access it or what methods are available for viewing such data. Do anyone have any suggestions? I hope experts can give me a lot of guidance!
I was busy all morning, and finally things became clear: PKS and ASPEN are both working properly; the difference in the historical curves is simply due to their different data collection methods. The signals collected by the field instruments are like this; it can be said that the location where the instruments are installed is not appropriate – it’s a problem related to the process design……