Page 222 - Mechanical Engineers' Handbook (Volume 2)

3 Process Data Acquisition 211

represented in the computer as a digital value but still able to be manipulated as an analog
number; and how a continuously changing value can be stored without exceeding the capacity
of storage or computation of the acquisition system. Discrete manufacturing still has a num-
ber of analog data sources, but a larger proportion involves discrete data, such as motor
stops, starts, and pulses. These have their own issues of acquisition and storage and are often
related to attributes of the process. The next section deals primarily with process data, ad-
ditionally covering some discrete data and issues around data collection, representation, and
storage.
As businesses begin to broaden the scope of optimization and understand their global
processes, the context of the data in terms of product, plant conditions, market conditions,
and other environmental aspects has increasingly added discrete data to the set of data to be
obtained. The interaction of the process with factors such as which crew is managing the
process, which customer’s needs are highest priority, legal controls such as environmental
limits impacting allowable process rates, operator decisions, which product is being manu-
factured, grade achieved, and a large number of other factors become important when a
company is competing with other companies that have already achieved excellent local con-
trol of processes. These data involve less understanding how to deal with continuous data
and more with the coordination of data within and between processes. These can be termed
manufacturing attributes to emphasize their importance in providing an environment around
process data. Later sections deal with manufacturing attribute data and issues around their
collection and coordination with process data.

3 PROCESS DATA ACQUISITION
Most modern data acquisition is via digital systems that may have a lower level analog
collection mechanism but is now so removed from the engineer that the engineer is only
concerned with the digital portion of the system. The ability to use digital microprocessors
as building blocks for data collection, the prevalence of computer tools, and the creation of
widely available operating systems that operate on small footprints have virtually eliminated
the need for analog instrumentation. While the data may be analog in nature, the technology
has been developed to such a degree that the engineer decreasingly needs to pay attention
to the analog aspects of the data.
A digital-to-analog (D/A) and analog-to-digital (A/D) converter performs the actual
processing required to bring analog information from or to the process. While the resulting
signal may be digital, it is a representation of a continuous number that has characteristics
that, if not understood, can result in erroneous conclusions from data, including missing data,
misinterpreting trends, or improperly weighting certain values.
The engineer should be aware of several features of analog data to ensure that the data
are used properly. An understanding of sampling interval, scaling, and linearization will
facilitate the use of data once collected.

3.1 Sampling Interval
One of the important steps with any data collection process includes the proper choice of
sampling interval. As an example of the impact of selection of sampling interval, or fre-
quency, a sine wave with a period of 1-s (Fig. 1a) is measured with several sampling inter-
vals. 0.5-s (Fig. 1b) and 0.1-s (Fig. 1c) intervals both provide different impressions of what
is actually happening. The 1-s sampling rate being in phase with the sine wave yields the
impression that we are measuring a nonvarying level. The 0.5-s sampling rate yields several