CVE is a unique process visualization tool that allows engineers to display hundreds of thousands of measurements across hundreds of variables, allowing them to interrogate historical process and laboratory data. CVE allows users to easily explore operating windows and operating envelopes based on their domain knowledge and without the need for deep understanding of higher mathematics.
Predictive Analytics is a time-consuming method which requires a data scientist to interpret the results, and which provides generalized answers through simplifications which can destroy much of the richness of the data. By contrast, Geometric Process Control (GPC) – a technology developed at PPCL – avoids these pitfalls.
By providing engineers with graphical tools to work with datasets spanning their entire plant, it allows them to create low-cost predictive models, without equations, to develop new process understanding quickly and easily.
Times
24 October webinar registration is open now. Choose from four time slots which are suitable for users in all parts of the world: 06.00 (British time), 09.00, 14.00 or 21.00. |
GPC allows engineers to fully explore their data and make discoveries that they can’t now. It covers the entire process from incoming analysis through processing conditions to final quality variables, KPIs and performance variables.
Engineers have been using the basic features of CVE for many years to work more efficiently with their plant data. It allows faster and easier management of key areas including alarm rationalization, process optimization and plant troubleshooting as well as building and evaluating real-time operating envelope models from C Process Modeller (CPM), PPCL’s online real-time process monitoring and event prediction product.
The new features introduced with CVE 2.8 represent advances in three key areas: improved user experience working with large datasets; across the board display improvements; and new options for alarm-based variables. These will be demonstrated in real time during the October webinars, starting with analysis of a process with 750 variables over a year at 10-minute intervals.
For improved working with large datasets, network versions now use a 64-bit executable, effectively allowing datasets limited only by the computer’s memory. Datasets in excess of 200 million values (product of variables and time points) have been routinely used on typical desktop machines during the development of version 2.8. To help users work with these large datasets, many common tasks have been simplified or automated, including bulk file join and append and bulk creation of function variables. Variable names can be colored, allowing classes of variables to be defined, for instance-controlled variables, valve positions, or even definition by unit.
Display improvements include normalization options on distribution plots, which allow much easier comparison of populations within the operation. More query colors are also now included, along with the ability to visually remove background data without using focus, as well as many new plot options, including cumulative distributions.
New options for alarm-based variables (alarm count and annunciation rate) now allow direct comparison of alarm performance across different modes of operation, in addition to the existing capability of comparing different alarm sets. For the user the benefit is much quicker evaluation of existing and proposed alarm limits across the range of plant operation. Summary statistics, such as total alarms and clean board rate, are always displayed.
A pre-processing filter has also been added to read the output from the online monitoring tool, C Process Modeller (CPM). This makes it much quicker and easier to analyze replay results and evaluate model performance.
New licencing options make CVE more easily affordable for geographically dispersed organizations.
@ppcl_news #PAuto #Training
No comments:
Post a Comment