In 2002, Orange & Rockland Utilities Embarked on a Project to interface circuit analysis capability to its geographic information system (GIS). The intent was to have GIS provide component information for the analytical software to automatically build the integrated electrical system model. The model would automatically attach loads at their specific geographic location. Then circuit changes, device status and other updates captured by the GIS system would automatically update the system model. All of this would interface through to the customer information system.

THE SOFTWARE SELECTION

O&R (Pearl River, New York, U.S.) selected the Distribution Engineering Workstation (DEW) from Electrical Distribution Design (Blacksburg, Virginia, U.S.) as its software of choice. It is an open-architecture analytical tool with sophisticated per-phase circuit-analysis capability. Once the model is built, any analysis application can be attached to the model. Isolated models are no longer needed to solve specific problems.

Using a common model to solve all problems reverses the paradigm where 90% of the time was spent gathering data and 10% of the time was spent on analysis. For instance, the DEW power-flow application uses a patented solution technique that runs extremely fast, can be parallel-processed and is much less susceptible to numerical instability than matrix methods. This allows a detailed solution of a very large system in near real time. For example, the entire O&R integrated distribution system model, which contains more than 365,000 components, is solved by the DEW power flow in less than 30 seconds on a standard laptop computer. Distributed processing can be employed to achieve faster solution rates. A test on a blade server is planned for 2009, and it is anticipated that solution time will be reduced to less than 2 seconds.

SWITCH ORDERS

A Web services interface was developed that allows GIS updates to be sent to DEW on a daily basis through a batch process. Working with EDD, load-research statistics such as kilowatt-hour to kilowatt conversion factors, diversity curves and load profiles by customer class were developed from data supplied by field-installed interval recorders. Additional interval recorders were added to the load-research program to obtain more accurate customer load-research statistic models. Understanding the time varying load behavior of different classes of customers is essential for accurate load estimates.

Additional functionality exploiting both Web services and the fast analytical capability were developed. At O&R, distribution switch orders are developed in the GIS system. Once a switching sequence is planned, the operator can request that the switching moves be checked to insure that system parameters, such as loads and voltage levels, do not exceed preset limits. Once the GIS system submits the planned switching steps to DEW via the Web interface, DEW changes the switch states within the model and performs a power flow on the circuits affected and provides a report that displays on the GIS map viewer. The report details loads and voltages before and after the switching moves. Other statistics — such as the number of customers transferred, dropped or restored — are also displayed. Locations with low voltage and overloads are identified and reported as alarms. Areas left without power as a result of the proposed switching also are identified and alarmed.

DATA SOURCES

The DEW model of the O&R distribution system is automatically built from several external data sources, including:

• Customer load measurements, including hourly kilowatt-hour and demand measurements

• GIS data

• Load research data

• SCADA data.

O&R provided additional information during database table setup, which was used in model creation: data for equipment impedances, and ratings and settings for regulating transformers and capacitor controllers.

The major issue encountered in building the model was data accuracy from the above sources. Power-flow simulations were used extensively to discover inconsistent and/or inaccurate data. When data errors were discovered, the error was fixed in the data source and the model rebuilt. The data-cleansing effort took more than two years. Processes are now in place, which include continual use of the power-flow simulation, to discover any data errors or inconsistencies that might be introduced into the system.

The DEW implementation diagram shows three servers along with planning workstations.