Production monitoring utilises a series of tools to collect, record and analyse the plant information. This information includes process variables, production counts, rejects, abnormal conditions.
Data is collected via a distrbuted data acquisition system consisting of analog-to-digital converters and is recorded in archives, on a continuous basis. These archives may be replayed on demand, in user defined x-y graphs. You may select the start time and span of the data to be shown, which variables will be grouped and how the values will be scaled.
You may for instance chose to see on the screen a line pressure, the motor current of the pump and the vibration of that pump, starting June 6, 2001 at 8:00am, for a period of 4 hours. Since these variables are in very different numerical ranges, each will be scaled as a percentage of bottom and top of scale defined for each one. This method uses the best resolution possible for all the variables : a pressure in the 1000 to 2000 kpa range as well as the vibration in -0.5 to +0.5 inches. You can also enlarge - zoom in - a specific area of the graph.
An event is when some specific situation - that you define - occurs : a pump starts or stops, a circuit breaker trips, the pressure reaches a certain limit, etc. The RTES based production monitoring system allows the capture of these events and their replay on the screen. Typically, the values of the selected variables are shown from a few minutes before the event to a few minutes after the event, with the time of the event in the middle. This is an excellent tool to identify problems that lead to abnormal conditions and how the system reacts. How is that possible? How can a system anticipate that an event will occur. In fact, the recording takes place continuously at a faster rate than the historical recording but the data overwrites itself so that only the last few minutes remain on record at any time. When the event is detected, the data recorded thus far is set aside and the recording continues for the preset amount of time. At the end of that time, the whole data is arranged in chronolical order and placed in a disk file with an appropriate name that identifies the date/time and location. You may then view the contents of the file, using the same method as for the historical data.
The data collected for any single variable may also be displayed in Statistical Process Control (SPC) format. You choose the number of samples and the sampling method (random or periodic), pver the desired period of time. The SPC charts include the median, standard deviation, upper and lower control limits and control capability (cpk). You may switch between S and R type graphs.
The occurence of abnormal situations or failures may be totalized by type. The data is then displayed on demand as a set of bar graphs using the "Pareto" method. Each bar is proportional to the percentage of the total. It is a very powerful visual method to highlight which problem must be the first priority. Say you produced 3000 parts and 50 were defective: 25 because they were broken, 10 because the color was wrong, 8 because a dimension was wrong and 7 because they had a blister. The "Pareto" will show a bar 50 high (25/50), another 20 high (10/50), etc..This will clearly identify 'broken' as the major problem and its correction will recduce the percent rejects by half. In other words, although the broken parts represent only 0.83% of the production (25/3000), they represent 50% of the loss and the pareto charts draws attention to that fact.
We would welcome the opportinity to discuss your specific application. Just drop us a line at fai@rt-sys.com