Evaluation of Online Valve Diagnostics and Postgre SQL

Henry Boger

Published on: 2019-04-30

Abstract

The two-fold goal of this study is to evaluate the ability of OVD (Online Valve Diagnostics) to identify KPI’s of the individual assets of Plant A, and secondly to determine a method useful to a plant engineer to identify those valves most deserving of attention prior to a shutdown.

Keywords

Online Valve Diagnostics; Online Valve Diagnostics

Objective

The two-fold goal of this study is to evaluate the ability of OVD (Online Valve Diagnostics) to identify KPI’s of the individual assets of Plant A, and secondly to determine a method useful to a plant engineer to identify those valves most deserving of attention prior to a shutdown.

Overall Impression of OVD

The first view of the dashboard is useful because of the overall scope. The ‘In Alarm” feature is a plus for a user looking for only valves out of spec. The “diagnose” feature is easy to use. The screen of individual tests is very informative with trend, friction, step and summary on one page. Report generation seems easily available. However, in my opinion there are too many valves “in alarm”, 48 out of a total of 150 (32%). I think the number should be narrowed to 10% maximum. This is the reason for the second goal described above, to identify those valves most deserving of attention prior to a shutdown.

Methodology

The search for a method to achieve the second goal involved using tools available in Postgres SQL. The first step used is to open the window under “pgAdim III”.A screenshot of the window is shown as (Figure 1).The next step is to expand the “Tables” icon. This shows a listing of 27 tables. The most useful is the table labeled “test”. Clicking on a “view data” tool in the task bar results in a spreadsheet of 9231 rows shows recorded data as well as calculated results. The task bar also allows filtering of data. One most useful filter is shown in Figure 2.The use of this filter reduces the number of tests (rows) to 147. The next step is to transfer the data to an Excel spreadsheet. The text is semicolon delimited and must be delimited in the resulting spreadsheet. The method then sorts by “valve id”, and then eliminates single occurrences resulting in 144 tests (rows). The next step is to reduce the rows by eliminating all but one test per valve id. This reduces the number of valves to nine (6.0%of the total number of valves). Screenshots of the trend screens of these nine tests are shown as Figure [3].

Second Search

The method includes a second search of test data using a second filter as shown in Figure 4.
Implementing this filter results in 78 tests (rows). Again, this filtered table is copied to an Excel spreadsheet. The tests are reduced to one per valve id as before. The result is shown in Figure 5.

Figures

Figure 1: Screenshot of pgAdmin III.

Figure 2: Screenshot of Filter A.

(A-B-PV221) Again, stick-slip operation

(A-B-PV205) Again, stick-slip operation

(A-B-FV233) This test shows stick-slip operation.

(A-C-FV109) This test shows control loop cycling

(A-E-LV049B) Here is stick-slip operation.

(A-K-FV081) Here, a perfect example of control loop cycling.

 

(A-G-LV001) This test shows a form of loop cycling.

(A-GLV101) Here again is a form of loop cycling

 

(A-GLV101) Here again is a form of loop cycling.

(A-H-LV013) This test shows a form of loop cycling.
Figure 3: Copies of trend screens for valves identified in OVD Plant A filtered data 0325.xlsx

Figure 4: Screenshot of Filter B.

(A-B-PV119) A clear example of shutoff cycling.

(A-D-PV047A) Again, shutoff cycling.

(A-D-FV227) Again, shutoff cycling.

(A-D-FV221B) Another clear example of shutoff cycling.

(A-A-PV011) And again, shutoff cycling.

A-C-LV021 Lastly, another example of shutoff cycling.
Figure 5: Copies of screenshots of filtered data from shutoffcycling0326.xlsx.