Predictive Maintenance Maximizes Machinery Health

As the name implies, predictive maintenance anticipates equipment outages rather than reacting to them. Benefits include fuller useful life for equipment and lower backup inventory. It goes beyond preventive maintenance, which at least strives to shift plant downtime to noncritical periods. Initial cost of predictive monitoring systems tends to be high, but much less than forced reactive ...

By Frank J. Bartos, Control Engineering July 1, 1999

KEY WORDS

Machine control

Predictive maintenance

Motors & motor control

Asset management

Sidebars: Share the data through MIMOSA

As the name implies, predictive maintenance anticipates equipment outages rather than reacting to them. Benefits include fuller useful life for equipment and lower backup inventory. It goes beyond preventive maintenance, which at least strives to shift plant downtime to noncritical periods. Initial cost of predictive monitoring systems tends to be high, but much less than forced reactive maintenance due to just one serious outage of a manufacturing line. In an era of intensive competition, where asset usage and plant operating efficiency must be maximized, it’s predictive maintenance that gets a closer look.

Various nondestructive techniques, such as vibration analysis, oil quality (or wear debris) analysis, infrared thermography, and several electrical motor analyses/monitoring methods go to work for preventive maintenance. Combined with new sensing, monitoring, and information management systems, they help advance predictive capabilities.

Business risk of equipment

As predictive maintenance methods become more cost-effective, their usage will spread to smaller and more varied equipment. However, size of machinery is not the main criterion. “Criticality of the machine is important,” says Brad Law, strategic market manager for Control & Automation Systems at Bently Nevada Corp. (Minden, Nev.). “Continuous on-line methods focus on the business risk of the equipment.” Protecting processes vital to an enterprise gets first attention.

At the same time, available products continue to grow in variety and number. This calls for integration tools to help users manage maintenance tasks. Bently Nevada’s answer is System 1, a complete machinery asset management system that unifies diverse data sources for maintenance management. The modular platform integrates 45 applications, according to Mr. Law, including Bently’s various on-line/off-line data collection hardware systems and software applications; System 1 also connects to third-party data sources. For example, it integrates Bently’s Machine Condition Manager 2000 (MCM2000) that works with Data Manager 2000 software, using embedded knowledge about a plant’s maintenance procedures, to identify problems and pinpoint locations, timing, severity, and root causes. MCM2000 can be customized for different malfunction levels, how to correct problems, and which personnel should do it.

System 1 (available 3Q99) promises openness, via use of industry standards, including ODBC, OLE, OPC, Internet, and MIMOSA (see sidebar). Visit Online Extra at www.controleng.com for more about related products.

National Instruments (NI, Austin, Tex.) notes the widening use of predictive maintenance methods, aided by the power and connectivity of PCs. “Historically, automating predictive maintenance systems required substantial integration costs and proprietary interfaces,” explains Ryan McDonald, NI’s automation software marketing manager. “High costs have kept many users in fixed equipment maintenance schedules without insight into actual machine health.” Today’s PC hardware and software are changing that picture.

Various PC form-factors add to available solutions. For example, portable units with data acquisition hardware and software can combine measurement functions of several service instruments, while offering storage, customized interfaces, and simple connections to other factory systems. Mr. McDonald sees chassis-based PCs, with industry-standard CompactPCI/PXI technology, delivering the ruggedness needed in critical high-performance applications. “Microsoft Windows CE promises to bring even greater compactness to PC-based measurement tools,” he adds.

Spincraft Automation (San Diego, Calif.)—a National Instruments Alliance Program member—uses PC-based data acquisition products and BridgeView software from NI for its predictive maintenance application to reduce unacceptable downtime in machine tools. In the shaping of complex aluminum extrusions at Boeing Corp.’s Huntington Beach, Calif. plant, skin mill operators receive abnormal condition status via flashing lights, audible signals, and on-screen warnings. Remote operators can view vibration information on an internal web page.

Gensym Corp. (Cambridge, Mass.), a notable developer of knowledge-based systems, offers its main G2 graphical, object-oriented software as a development and deployment platform for intelligent real-time applications. In a cooperative product relationship, G2 forms the core of Bently Nevada’s MCM2000 system. With G2, the user’s plant-specific machinery knowledge can be embedded into a condition management system to customize preventive maintenance processes.

Liberty Technologies (Conshohocken, Pa.), a division of Crane Nuclear Inc., combines advanced diagnostics with software analysis to prevent unplanned equipment downtime, improve asset usage, and increase plant safety. Its proprietary products interpret operating data gathered from valves, turbines, engines, compressors, motors, and other motor-driven equipment to monitor their condition and performance. See Online Extra at www.controleng.com for more motor management products.

Motor as a ‘sensor’

Electric motors power thousands of crucial manufacturing processes and represent a vast industrial investment in their own right. This has spurred the development of various condition monitoring techniques and products. Ultimately, motor intelligence will function as a predictive maintenance “sensor” of the machine or system to which the motor is attached.

MotorStatus, from Computational Systems Inc. (CSI), now part of Emerson Electric (St. Louis, Mo.), is a self-contained unit with onboard sensors. It collects and stores information on motor operational and cumulative effects (temperature, vibration, electrical parameters, loads, starts, etc.) to track the condition of fixed-speed ac motors. MotorStatus mounts to any size motor. Data collection for analysis takes place via an infrared link to MS-Windows CE capable laptop and palmtop computers or CSI’s machinery analyzer interface.

Framatome Technologies Inc. (FTI, Lynchburg, Va.) calls its motor testing product the EMPATH (for Electric Motor Performance Analysis and Trending Hardware) system. EMPATH determines impending machinery problems based on Motor Current Signature Analysis (MCSA)—a method FTI considers “a proven indicator of the health or deterioration” for electric motors and motor-driven machinery. EMPATH can be used periodically or installed permanently for continuous data collection. A system consists of a portable PC, analysis software, and probes.

Perhaps the most sophisticated example of ac motor monitoring is IQ Intelligent motor with PreAlert technology, available for ac induction motors of 2 to 500 hp (1.5-375 kW) from Rockwell Automation/Reliance Electric (Greenville, S.C.). Embedded sensors, logic, and an onboard microprocessor “electronically inspect” conditions of motor windings, bearings, and rotor for early warning of possible failures. Software performs diagnostic analysis online in real time while the motor operates. IQ Intelligent motor communicates via RS-232, DeviceNet, or wireless means.

But more than motor health is involved. “Vibration and current signatures can be evaluated to check incoming power condition and make assessments about driven equipment,” says, Richard Schaefer, customer service manager for Rockwell Automation’s Athens Motor Plant, (Athens, Ga.). IQ Intelligent motor ( CE , Nov. ’97, p. 7; March ’98, p. 96) has not yet taken the market by storm. “This intelligent product still has not realized its potential. Few users know all the things it can do.”

Future condition-monitoring systems will grow even “smarter.” With operational and historic data available in quality as well as quantity to estimate remaining machine life, they will raise predictive maintenance capabilities to new levels.

Share the data through MIMOSA

Scattered, inconsistent information is sometimes worse than lack of information. One organization tackling this sizable problem is the Machinery Information Management Open Systems Alliance (San Diego, Calif.). MIMOSA is a growing nonprofit corporation of over 50 companies and 200 individuals worldwide—suppliers and users of instrumentation and maintenance information management technology. They include 12 of the 16 top suppliers of condition-monitoring products and systems representing about 50% of total worldwide product sales. Among MIMOSA sponsors (top level of voting membership) are Rockwell Automation and Siemens.

MIMOSA champions open exchange of information about equipment condition—in process, production, and manufacturing areas—among condition assessment, process control, and maintenance information systems. Information types handled include mechanical and operating condition; projected lifetime; presence, identification, and severity of problems; operating and maintenance recommendations; design specifications; and history.

MIMOSA participants have developed a comprehensive information model that forms the basis for exchanging equipment-condition-related information among higher level systems. The model consists of five major sources and consumers of equipment condition information:

Enterprise Resource Manager;

Condition Measurements;

Decision Support;

Maintenance Management Systems; and

Distributed Control Systems.

Another major MIMOSA focus has been the development of a vendor-independent file exchange format called Common Relational Information Schema (CRIS) that permits integration of multiple sources of machinery information.

For more on CRIS, the Information Model, and MIMOSA, see Online Extra at