Database Technology for Real-Time Analysis and Control

Large-scale manufacturing, sewage treatment and power generation facilities are required to leverage increased process visibility which provides better and faster decisions, increased productivity and reduced costs for a sustainable competitive advantage. The key factors and criteria that organizations consider important towards managing, measuring and controlling production processes and costs can be logged into a database for analysis. Process historian database technology, vast protocol connectivity, and intelligent analysis application are all important to complex industrial decision making. Most large systems have a need to store massive amounts of process and commercial data. Different databases often need to work together. Learning differences between databases helps in selecting the best database for your situation.

Limited computer storage capacity became a bottleneck in the IT system. The information revolution has led to far more data now than in the past. Large database systems are generating a flood of new data all the time. With the growth of computer storage capacity, there is a tend to permanently save all sorts of data. More information can be acquired and more information can be stored. Early in the information revolution, security trading systems often only stored recent transaction details. They discarded old information and wrote over the allotted memory space. Now most enterprises tend to save everything that can be saved like every transaction, every phone call, each click on a web site and each switch in communications. Due to this trend, massive amounts of computer storage is getting used. In enterprise-level applications, the cost of saving massive data is often shocking.

Relational databases are commonly used in commercial applications like customer relationship management systems. Commercial applications usually require many fields to be stored like: customer name, company name, address, phone number, and email address. Industrial applications are usually simpler and require fields like a tag name, a measurement value and a time stamp to be stored. Production data is relatively much simpler, yet the point count is usually very large. Real-time and historical data processing is greater than the processing ability of a relational database. A great advantage of the process historian database is simple massive production data and historical data generation.

A database comparison study was done by Wellintech, Inc. on an environmental protection management information system. They converted the system’s data into a process historian database from an Oracle based relational database. The database took up 90 percent of the systems hard disk. The system had been operational for three years. The database held a great amount of data and was based on a GIS system which stored GPS information, maps, locations, time stamps, spacial map information, and locations. The database also held a lot of information on the management of the environmental monitoring system. The database was substituted with process historian database and through compression the process historian database cut down the storage space by 25 percent. The space the database occupied was reduced and the querying was much faster with the process historian database.

Process Historian Databases compresses data through a multiple compression algorithm. The changes of industrial production process field data often has waveform laws. Only a small portion tags or variables change in value frequently. The values of the other tags change ery slowly, and users can allow accuracy of data loss within a certain range. Data compression in process real-time/historical databases is a very important technology since it can save massive amounts of space and can aid in query speed.

The CHANGE (0) Compression Algorithm is available for any type of variable compression. It only detects the time-out of compression and verifes the same value detection. It stores the value when a variable has changed. It does not store a variale if nothing has changed. For any kind of compression algorithm, the first step is to check the time and the quality stamp.

The principle of dead banding compression algorithm is very simple. It stores the data when the change in value change reaches a certain threshold. For many variables that change slowly in the actual production process, it can dramatically reduce the amount of data stored.

The swinging door compression algorithm was first proposed by OSI Soft PI. The algorithm s relatively simple and PI opened the entire algorithm to the public. Now, almost all process historian databases have adopted this compression algorithm, and have added new technologies in with it to make it more optimal. The principle of the swinging door algorithm judges whether a data point needs to be saved. It draws a straight line between data from a previous saved point P to the next data point N. It checks the absolute bias of those data points (including A) between two points (P, N) and the data points on that line corresponding to the time stamp. If there is bias of the point that exceeds the compression bias, that point A should be saved.

These compression technologies can help you to save storage space, enhance data querying speed. A vast amount of data is being collected in industrial databases from measurement instruments and control hardware. Many industrial communication protocols are being used in many different industries. BACnet and LonWorks are commonly used in HVAC systems, 102 Protocol in power plants, and Modbus in process control. A large amount of connectivity of data collection is very important in intelligent information systems.

Intelligent analysis for efficiency in all industries requires collection of data from hardware, storage of data to a database, and conversion of the data into usable information for decision-making. Data can be converted into information through process historian database analysis tools which can help you compute things like how many tons of water gets treated in a sewage treatment plant in one week. The historical data analysis tool can predict the future events, or estimate uncertain past events.

Relational databases are great for commercial or smaller systems. Process istorian databases are great for industrial applications where measurement data does not dramatically change over time or requires fast querying speed on vast amounts of data.