Page 349 - Intro Predictive Maintenance

340 An Introduction to Predictive Maintenance

The second method uses dynamic limits that monitor the rate of change in the mea-
surement parameters. This type of monitoring can detect minor deviations in the rate
that a machine or system is degrading and anticipate when an alarm will be reached.
The use of dynamic limits will greatly enhance the automatic diagnostic capabilities
of a predictive maintenance system and reduce the manual effort required to gain
maximum benefits.

Data Storage. The microprocessor must be able to acquire and store large amounts
of data. The memory capacities of the various predictive maintenance systems vary.
At a minimum, the system must be able to store a full eight hours of data before trans-
ferring it to the host computer. The actual memory requirements will depend on the
type of data acquired. For example, a system used to acquire vibration data would
need enough memory to store about 1,000 overall readings or 400 full signatures.
Process monitoring would require a minimum of 1,000 readings to meet the minimum
requirements.

Data Transfer. The data acquisition unit will not be used for long-term data storage.
Therefore, it must be able to reliably transfer data into the host computer. The actual
time required to transfer the microprocessor’s data into the host computer is the only
nonproductive time of the data acquisition unit. It cannot be used to acquire additional
data during the data transfer operation. Therefore, the transfer time should be kept to
a minimum. Most of the available systems use an RS 232 communication protocol
that allows data transfer at rates of up to 19,200 baud. The time required to dump the
full memory of a typical microprocessor can be 30 minutes or more.

Some of the systems have incorporated an independent method of transferring data
that eliminates the dead time altogether. These systems transfer stored data from the
data logger into a battery-backed memory, bypassing the RS 232 link. Using this tech-
nique, data can be transferred at more than 350,000 baud and will reduce the non-
productive time to a few minutes.

The microprocessor should also be able to support modem communication with
remote computers. This feature will enable multiple plant operation and direct access
to third-party diagnostic and analysis support. Data can be transferred anywhere in
the world using this technique. Not all predictive maintenance systems use a true RS
232 communications protocol or support modem communications. These systems can
severely limit the capabilities of your program. The various predictive maintenance
techniques will add other specifications for an acceptable data acquisition unit.

The Host Computer
The host computer provides all of the data management, storage, report generation,
and analysis capabilities of the predictive maintenance program. Therefore, care
should be exercised during the selection process. This is especially true if multiple
technologies will be used within the predictive maintenance program. Each predictive
maintenance system will have a unique host computer specification that will include