Gerber Products Company
Gerber's data mart brings
In the manufacture of baby food,
quality is key. To keep to its high standards, Gerber Products Company,
Fremont, Mich., left no stone unturned in its effort to build a process
control system that gives management a significantly improved ability
to an-alyze information stored in a new quality data mart.
"The old program was first built
in the 1970s, so we were looking to replace that and build an enterprise
quality solution," said Ann Baker, director of quality assurance at Gerber,
a unit of Basel, Switzerland-based Novartis International AG. "We wanted
to link consumer data to the shop floor data. We [also] wanted to link
all of the shop floor data, and provide access to management and to the
people in research and development."
The older system was updated regularly
but could not meet new corporate goals due to a lack of data management
capabilities, an inability to adequately link to other systems and a need
for basic data capture functions, Baker said. The older system also could
not support advanced analysis techniques and suffered performance degradation
as data files grew. In addition, the older system could not integrate
with any of Gerber's quality information management systems, had limited
statistical analysis capabilities, and was said by operators to be laborious
and difficult to use.
The QA team found that available
packaged solutions could not meet all of the requirements for the needed
system; a decision was therefore made to build it in-house, Baker said.
She noted that in addition to the features planned for the new system,
it required tight integration with the corporate data warehouse. On the
advice of its parent company, a longtime user of tools from SAS Institute
Inc., Cary, N.C., Gerber developers decided to use a variety of SAS offerings
in the project after a three-day evaluation and testing process, Baker
Process Control System (IPCS)
perform detailed analysis on supplier quality performance.
management with a previously unavailable level of data analysis;
immediate availability of information during processing.
Unix, NT, mainframe
The new system, called the Integrated
Process Control System (IPCS), gives Gerber's quality management operation
100% electronic capture capabilities compared to 40% in the older system.
Through last fall, Gerber ascribed more than $450,000 in savings for the
firm to the system's improved accuracy, precision and control of product
Baker said the development team
worked closely with several specific user groups, including representatives
from the three food manufacturing plants, as well as the purchasing, engineering,
research and quality assurance units. At the start of the process, the
user team listed inadequacies they saw with the older system and created
a "wish list" for the new IPCS system.
The group created a strategy for
the new system that first and foremost called for a plant floor statistical
control system that could provide middle and upper management with electronic
reporting tools for assessing and monitoring quality performance. This
would provide so-called high-end statistical users and corporate-level
users with statistical analysis and reporting capabilities. Baker said
the system provides such capabilities using color-coded process flow diagrams
that allow users to spot trouble or potential troubles immediately.
SAS/ADX Interface, SAS/Warehouse Administrator,
SAS/QC, SAS/Share, SAS/Analyst, SAS/Insight, Base SAS, SAS/Access,
Prism Warehouse Manager (now part of the Informix Data Stage Suite)
Requirements also included an
extremely aggressive defect rate goal of two parts per billion. Baker
noted that such an aggressive goal is typical of food manufacturers, "that
can't have intolerances in the thousands. We're not the auto industry."
Baker said the new
IPCS system allows plant managers to view the entire production floor
quality performance of their plant via its new process-flow reporting
capabilities. Simple tree diagram interfaces let managers drill down to
gain further performance, product and comparative statistics.
— Michael W. Bucken