Distribution utility enterprise integration - A new paradigm
The GPU energy experience
GPU Energy is made up three companies that operated under the names of JCP&L - Jersey
Central Power & Light Co., Met-Ed - Metropolitan Edison Co., and Penelec - Pennsylvania
Electric Co. These companies combined serve 2.1 million customers and 24,000 square miles of
service territory.
Where Are We Today
Because of operating under three different companies we currently have three different IBM
GFIS systems, and three different Graphic update environments. The systems are all basically
the same, but each has their own uniqueness that causes each to be different. We also maintain
multiple different legacy databases for each company such as transformer records, property
records, joint use, and street light. Each company also has their own operating procedures,
which is often even harder to change into one system. Most important of where we are today is
that the systems are NOT INTEGRATED.
Where Are We Going
We are currently in the midst of moving to a completely different platform and are planning to
be implemented by the GITA conference. Where we are going is listed below:
Smallworld GIS - Convert from IBM GFIS to Smallworld
ERP- SAP R/3 - Enterprise wide solution for all applications
Disconnect - Tie to SAP with our GIS objects
CH Expert Design - On-Line Design system
Smallworld PowerOn - New outage management system
MDSI - Computer Assisted Dispatch
INTEGRATED
In the following sections we will relate the experiences we have gained by implementing each of
these products and the impact it has when trying to do it with an ERP such as SAP.
The first hurdle for GPU in moving to the new GIS was to make a common model. This
included first understanding the three different models and then combining them into one model.
This was complicated by the fact that we were implementing an ERP at the same time and had to
insure that both systems were kept in sync. The next problem encountered was moving away
from the point connector model and using a more graphical model. This was not an easy move
for our developers and users of the system. It was very hard early on during conversion to
understand how the model worked and how it was being converted. Early on we did not have
tools to provide us help in determining if our data was being converted correctly.
Enterprise Resource Planning
The implementation of our ERP system was based on process flows more so than data
requirements. This proved to be a problem and a blessing for the GIS team. While it was
fmstrating in that we were not being provide with data requirements when they finally decided it
was time to concentrate on data we provided them with our model. The other aspect of the ERP
team was that they were working on new business procedures while requiring old procedures for
day one implementation. This was mostly a focus problem in that we needed to know what was
required for implementation rather than where we were going in the future. There is a fine line
between spending time looking into the future and insure that what you are planning to
implement will meet those requirements and what is needed day one.
The second issue of ERP implementation that had indirect impacts on the GIS implementation
was the need to load data from the legacy databases and then integrates it with GIS. The three
different operating companies had varying forms of integration from none to some. This caused
concern on how resolve the differences in the databases and who was the driver of the data. A
decision was made early that the GIS data was the driver and it would be utilized as the base to
match against the legacy databases. Now it was a matter of how to match the different systems
with the GIS data and then what data could be utilized.
GISConnect
Disconnect is a solution for ERP/GIS Integration. It is a piece of software that through
configuration allows you to link your GIS object to the technical equipment objects in ERP.
These objects in our case were decided upon to be the capitalized units. The biggest issue we
had with this process is the knowledge base available on the product. The product has been
implemented in other countries, but had not been implemented in the United States and we had a
very limited knowledge base to work with early on. The other issue here was that it was
integrating different data models. We were a facility based electrical model system and ERP was
trying to implement a costing and property records model. A key decision early on here and one
that was hard fought was to not try and model the electrical network within ERP. Allow the GIS
to keep themodel and only provide ERP with the attributes required for their model. This
decision does require that the GIS can provide the required network reporting through the use of
GISConnect.
