THE BENEFITS OF A DISTRIBUTED SYSTEM
The benefits of a comprehensive, automated process control and data management system make a long list (Table 2), but they
all lead directly to the main points of greater efficiency and shorter time-to-market. Almost all modern instruments are designed
with the capability to communicate and cooperate, but what has been lacking is the orchestra conductor that manages and controls
all the individual instruments as a unified whole. The software working at the process level bridges the gaps between disparate
subsystems. It is independent of the instrumentation and able to communicate with instruments of all brands. It supports all
types of physical connections: Ethernet, serial, analog & digital I/O, OPC, and other standard protocols.
Another benefit that comes from device-independence is that one method of measurement can be replaced with a new one, with
no interruption to the history of the process value. Here, the phone companies set a good example. A customer can buy a new
phone (instrument), but his phone number doesn't change; his identity stays the same. Likewise, when an old mass spectrometer
is replaced with a new kind of gas analyzer, the nitrogen measurement has a seamless history across the two.
Even with the wide spectrum of open instrument communication, there might still be some external analyses from which data
is collected by hand. This "off-line" data can be entered into the unified data management system and is treated as an equal
citizen. It is available for display, analysis, and feedback control, along with all the data from the various instruments
that were automatically collected. Also, all online data analysis or calculated results are equally available for display,
further analysis, and feedback control. Figure 3 shows a Trends display that graphs process data acquired from online instruments,
off-line measurements, and calculated results all together.
Figure 3. Unified data from instruments, offline measurements, and results are all visible on the computer screen.
In addition to communication with each different data island, there need to be bridges between the islands. This is accomplished
by a distributed architecture, linking all computers, instruments, data servers and archives. This requires a robust network
to be in place. A company's IT group has to be an active player here, and should be involved early in the planning. Figure
4 shows examples of connections in this distributed system.
Figure 4. A networked, distributed system can connect to other parts of the system.
The advantages of a distributed system are many. First, all aspects of the process control and data management can be available
from any site on the network. A technician can monitor many reactors from one station, or respond to alarms remotely. A researcher
can check on and adjust experiments from her office, create reports, and be automatically notified by email or pager when
problems arise. Second, instrument resources can be shared. One expensive gas analyzer can serve many reactors. Third, know-how
and results can be shared. A recipe or control strategy developed by one scientist can be used by others for scale-up or for
other runs. Process data and results can be shared and compared on a live system, even from a meeting room. The Operations
Manager of a San Francisco Bay Area biotechnology firm said, "I have immediate access to all the reactors I manage in three
separate facilities, and this has streamlined our post-experiment data analysis."