|
|
"All the other
visual data analysis products that we tried tended to choke on the tremendous
volume of data," says Ronald A. Bray, formerly a process control
specialist at the company. "They lacked functionality and didn't
easily line up with our varying data sources." Bray's first use
of PV-WAVE occurred in 1990 during a project that required statistical
analysis and display of process data for control and optimization. He
built several PV-WAVE routines to pull data from a historical archive,
where process control data, descriptions, units and time stamps were stored.
Time periods and tag lists were entered as parameters. Later, as the six
teams met for the massive process-control project, Bray spent some time
reworking his existing PV-WAVE routines to see if they would prove useful.
"During a short demonstration, the other engineers were immediately
impressed with PV-WAVE's ability to organize and display results,"
Bray says. "A previous project of this type required two weeks of
effort just to collect and organize the data. But with PV-WAVE, the entire
data-analysis portion was essentially complete within that amount of time. "Interest in
PV-WAVE grew quickly as the engineers discovered they could become productive
with the application in less than an hour," Bray says. "Other
data-analysis tools were quickly abandoned in favor of PV-WAVE,"
he adds. A New Standard for Visual Data Analysis (VDA) PV-WAVE lets technical
and business computing professionals interactively explore, manipulate
and analyze complex numeric data. Tightly integrated graphics, statistics,
image processing, signal processing and animation let users rapidly interpret
complex data sets, quickly identify important features and trends and
share their results. The emphasis during
the multiteam project was on analyzing process control data to improve
manufacturing productivity throughout the company. Data on temperatures,
pressures, flows and other process control information were collected
and analyzed. "One of the teams collected nine hours of one-second
data on 40 tags [about one million samples] and was able to get the data
from the disk into PV-WAVE in just two or three minutes," Bray says.
"And that was using a VAX®station 3100 computer, which
is considered slow by today's standards." The key to PV-WAVE's speed
is an array-oriented, fourth-generation programming language (4GL) designed
specifically for visual data analysis. "The code is compact and efficient,"
Bray says. "Also, since it is an interpreted language, it is very
easy to develop and debug applications. It has the speed of a compiled
language, where calculations are done with large arrays." Bray's custom VDA
routines were developed with PV-WAVE Foundation, Visual Numerics' flagship
UNIX® product for data visualization. "Other visualization
tools that I had tried were painfully slow and inadequate for these large
data sets," Bray explains. "They usually required data to be
put in their own special format before they could be accessed, which required
tedious programming. "PV-WAVE's 4GL,
by contrast, reduces coding efforts by up to 80 percent compared with
traditional programming languages, such as C and FORTRAN," Bray adds.
"It takes time to get used to the syntax and language constructs.
But once you get used to it, you can become very proficient at data analysis
and visualization. It is my preferred language because you can quickly
create applications that are easy for users to understand, yet have all
the capabilities required by power users." Bray's early work with
PV-WAVE was performed on Digital's VMS-based VAXstations. As interest
in PV-WAVE spread, Bray was able to port his PV-WAVE routines to a variety
of UNIX platforms, as well as Windows® NT and MS-Windows.
"PV-WAVE runs
as an X application on the VAX, so any PC or Macintosh®
computer that does X emulation can be used to access PV-WAVE applications,"
Bray says. "Porting a PV-WAVE application among platforms takes minimal
effort. Any calls specific to a particular operating system must be changed,
but there are very few of these." Today, Bray is employed
at another company, Houston-based Litwin Process Automation. Litwin provides
advanced control and training simulator applications to the refining and
petrochemical industry. PV-WAVE is still playing a critical role in Bray's
work. His team purchased five licenses of PV-WAVE to assist with their
work creating dynamic simulations to help regulate manufacturing processes
for Litwin clients. They work with large data sets, typically 50 to 400
tags, with anywhere from 5,000 to 30,000 points per tag. All in all, Bray
credits himself for creating the actual PV-WAVE routines, but credits
Visual Numerics for making it possible for one person to quickly develop
applications that improve the overall productivity of several process
control project teams. "Without PV-WAVE, it would have been impossible for a single engineer to develop and support such a powerful and useful application as a sideline to the usual work that must be done," Bray concludes. "We have been very impressed with the speed and robustness of the Visual Numerics products, yet I feel we are just beginning to explore their full potential. They integrate well with many types of data and software tools and dramatically increase productivity where visual data analysis is concerned."
| ||||||
