Data, Data Everywhere...

Predictive analysis using data from continuous automated diagnostic inputs will make unscheduled mechanical and process failures almost unheard of

 

Are you responsible for helping to identify and provide data collection and data analysis tools for your machines? Then, you may feel like a shipwreck victim adrift on the ocean. You are drowning in a sea of machine data, and what you really need is a way to make sense of it all.

On the data-overload front, there is good news and bad news. The bad news is new sensors and control systems promise to provide exponentially more data. The good news is these components also will be equipped with embedded tools that will analyze and make sense out of the sea of data.

The sheer amount of data that can be collected by current software products is staggering. OSIsoft (

www.osisoft.com) sells an embeddable data historian that can archive 900,000-plus variables per second. Further technology advances will soon allow individual sensors to use embedded software and hardware to store these massive amounts of data.

OSIsoft foresees a future in which each piece of machinery will become both a control and a data collection point, maintaining a record of its performance and its status for its entire life.

As an example, a turbine could contain accelerometers combined with an embedded time-series data historian and vibration-analysis software. This would allow capture of every single vibration of each turbine blade. At each start-up, an operator could easily perform a full-range dynamic test.

Real-time comparison of this profile against a "standard" profile for that specific turbine, generated over many thousands of hours of operation, would enable operators to learn immediately if there are any potential causes of failure.

OSIsoft believes future generations of data gathering and analysis tools will fundamentally change machine operation and maintenance. "Maintenance technicians will come to rely more and more on continuous automated diagnostic inputs," says Matt Miller, OEM product manager at OSIsoft. "Predictive analysis using these data will make unscheduled mechanical and process failures almost unheard of."

One of the ways to make sense out of data is to use a global database with expert rules. Such a database looks at data inputs, compares this input to embedded rules, and automatically makes decisions. "Global experience databases will build rules and dynamically improve them over time," speculates Bill Lydon, product manager of electronic products at Wago (

www.wago.com). "These databases will determine optimum maintenance and efficient operating parameters, as well as predict performance."

Some observers see data collection rates advancing at a much faster rate than communication technologies, necessitating local data storage for buffering. "The speed of memory and processors will advance to such an extent that almost every device on the plant floor will have its own built-in data recorder," observes Ken Hall, vice president of architecture advanced technology at Rockwell Automation

(www.rockwellautomation.com).

According to Hall, these recorders have the potential to significantly expand the scope of today's plant-wide data collection capabilities since they can record samples faster than the communications systems can transmit them. The recorders will use local data buffering to play back the last few seconds of an event with very high resolution.

Relatively slow communication speeds can cause problems when the transducers employed in gathering statistical process control (SPC) data are different from those actively involved in controlling the process. "In fast-moving or very precise processes, this has the effect of allowing the SPC data readings to be out of synchronization with critical events happening in the process," says Rick Meyerhoefer, regional applications specialist for Delta Computer Systems (www.deltacompsys.com).

Delta is addressing this problem by enabling its controllers to act as data acquisitions units. "Our motion controller does double-duty, as a control system element and as a remote data acquisition system," he adds. "In addition to improving the quality of process data sampling, this architecture cuts system costs because a separate data acquisition system is not needed."

Another way to synchronize control and data is offered by Robert Burridge, senior R&D engineer with DVT (

www.dvtsensors.com). "Currently, data collection systems feed SPC and SQC tools that produce output for databases and ultimately for human analysis," he says. "In the near future, data analysis will get pushed down toward the embedded systems/sensors to reduce the volume of information."

Burridge also thinks the software analyzing the data at the collection nodes will get smarter--able to make real-time global control adjustments autonomously.

E-mail Dan at dhebert@putman.net.

Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.

Comments

No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments