Do you ever find that your process historian is excellent at capturing valuable data but then you have little time to truly understand what the data is trying to tell you? It’s another thing to add to the list of tasks you probably won’t get done today. Eventually, you’ll sit down minutes before the monthly meeting and use your highlighter pen to go over some discrepancies in the monthly data overview. There’s information there – but you don’t have time to play detective with your spreadsheet.
The average process plant generates thousands of rows of process history data in a day’s operation. But unless you have a sadistic desire to pour over this with a fine-tooth comb then this data is being captured but not used. If you could simply use the valuable data you are capturing daily then you would validate the purchase of the process historian, but also give yourself a unique insight into how your process works best. But how?
There are several solutions on the market:
– Spreadsheets You can spend your days plotting the variables you believe are important against time. It helps to give you a picture of how these variables are tracking, but fails to take into consideration their effects upon other variables. Also, the question still remains: are the variables you think are important actually the most important variables?
– Statistical package Evaluates data using equations. It does well in representing information as 2d graphs and assessing process outliers. But it lacks the ability to visualise full sets of data and unfortunately assumes linearity in your data, which can confound results.
– Visualisation and data exploration tools provided by the process historian supplier which have merits in a simple user interface particular to that historian, but are often intended to drive trend-displays requiring only a few variables rather than the larger numbers of variables and rows that you really wanted to retrieve.
– C Visual Explorer (CVE): A different kind of graph allowing you to see hundreds of variables and tens of thousands of data rows in one graph and with all the easy-to-use tools necessary to pick out different types of operation and activity that have occurred in your plant. It is an easy to use system which does not require any mathematics on the part of the user. Unlike other methods, there is no fitting of pre-defined functions to the data, so you are able to see your process and all its undisguised non-linear behaviour across many variables at once. You will learn things you never knew, prove long-held suspicions and dispel myths about how the process whose origins are lost in the mists of time operates. And then you can move on to use it for process stewardship, KPI reporting, alarm management, finding which variables are important and which aren’t, comparing operating envelopes, optimisation and much more.
We believe our Geometric Process Control technology provides the best tools for the many jobs process engineers, control engineers and others are called upon to perform because it is the solution which is easy to use and does not require any mathematics on the part of the user. This means that training is fast and anyone in the organisation can use the tool; so adoption rates are high.
CVE provides a simple drag-and-drop set of one and two-dimensional graphic selectors which are visible and manipulable in the parallel plot and in all the associated distribution and scatter plots.
Focus Levels provide a simple means of sub-setting data to investigate, for instance, particular modes or phases of process operation by hiding all the points not selected by a particular query. A wholly new set of queries can be constructed at the new focus level and queries from the previous focus level can be copied in if desired. There can be as many levels of focus as desired.
CVE does not alter the data, and so is able to show non-linear process behaviour exactly as it occurred in the process thus allowing identification of local optima and other significant non-linear features that are often obscured or hidden by equation fitting methods of data analysis such as principal components analysis (PCA), linear regression and Partial Least Squares (PLS).