Interfaces: Smallworld-based engineering studies with customer information loads

Interfaces: Smallworld-based engineering studies with customer information loads

Fred Boland and Larry Schanteli
UtiliCorp United
20 West 9th Kansas City, Missouri 64105
Email: fboland@utiiicorp.com and lschantel@utilicorp. com
816-467-3822, 816-467-3818

Why it is important
Since the inception of the UtiliCorp FAME system, the ability to model and analyze customer load and revenue data integrated with facility data has been a significant source of benefits.

UtiliCorp engineers analyze the distribution network in order to understand the performance of the network, test alternate configurations, and make recommendation to improve the systems utilization. Analyses currently performed are load flow analysis, voltage drop analysis, and fault current analysis.

Interfacing the facilities data with an analysis application will generate benefits in two general areas:

Labor: Reducing the labor required to analyze the distribution networks allows cost savings to be achieved.

Capital Investment: Increasing the accuracy of facility and load models also increases the validity of the analysis. This ultimately leads to cost avoidance by deferring or eliminating system improvement projects.

Requirements gathering
The first step of the intetface design was to examine the current practice of performing an analysis including processes, applications, and data.

Processes
The original process was labor intensive; it involved manual data entry of the distribution facilities and estimated loads. The following steps are required to perform an analysis.

Gather paper maps of the area in the analysis
Manually extract conductor and equipment information from the map using a scale
Record the conductor and equipment data onto data sheets
Enter the data sheets into the analysis application
Perform checks on the loaded data and correct errors in the analysis application
Enter estimated loads at significant points in the network
Run the analysis
Adjust loads using trial and error until a feasible solution was reached
Manually update the analysis model as new construction adds facilities to the system

Not only is this a manually intensive process, but in some cases, the actual network changed faster than an engineer can build the model.

This process was also error prone. The amount of data entry required allows introduction of keying mistakes. Some mistakes were simple to find and correct, but it was not unusual to spend several hours debugging an analysis because of a keying mistake.

The process also required redundant data entry. Both distribution maps and analysis models were updated manually and independently.

Another major disadvantage to this process was the engineer would correct data errors in the analysis application, but these corrections did not make their way back to the maps.

Requirements
The analysis of the manual process revealed that significant savings could be obtained by implementing two high level requirements.

Automate the input of conductors and equipment from the GIS into the analysis Application
Use actual vs. estimated loads

To accomplish this, the application would need an interface between the GIS and the analysis application in order to transfer network information to the analysis application.

Additionally, the system would need to tie the actual Kilowatt Hour (KWh) load data from the Customer Information System (CIS) with the distribution facilities in order to increase the accuracy of the load model.

During the detail design phase the following requirements were defined:

The interfaces would not be real-time
1. The distribution analysis methodology at UtiliCorp is to take a snapshot of the distribution network during peak load conditions. Therefore, real-time data is not needed.

The process would utilize the ASCII loader provided by the analysis application.
1. The ASCII loader was already in use at UtiliCorp and engineers are familiar with it .
2. The ASCII loader checks for data errors during the load process, and provides reasonable error logs.
3. The ACSII loader was previously purchased, therefore no additional cost.

Use a standard file format designed to import customer information into the Smallworld datastore.
1. UtiliCorp has two legacy Customer Information Systems and a new CIS is being installed.
2. The data from the various legacy systems would be formatted to fit the single file format.

The data extracted from the CIS will include a 12 month history of billed usage, demand, and revenue.
1. There are several fields in the CIS that contain usage and revenue amounts. There are raw reads that are multiplied by a factor based on the meter type, plus there are usage and billing adjustments and other data anomalies.
2. Due to the nature of load allocation algorithms, these adjustments and anomalies would have minimal effect on the results of an analysis.
3. Using the billed amounts simplifies the extraction of the CIS data.

The link between the CIS data and the GIS data is facilitated by the tag number of the transformer that serves the customer.
1. UtiliCorp has an ongoing process of physically tagging a number on the transformer, recording the tag number in each CIS customer record, and recording the tag number on the distribution map.
2. Eliminates the need to convert or add secondaries and services.
3. Eliminates the need to graphically represent and connect each customer.

Data modeling considerations
The second step of the interface design was to identify data model changes.

Analysis Interface
The initial data model was designed with interfaces to analysis applications in mind. However, during the course of the project, a new version of the analysis application became available. The new version has enhanced functionality and increased data requirements. Minor additions to several objects were implemented to account for the discrepancies.

CIS Interface
To accommodate the data extracted from the CIS systems and relate it to existing facilities, three objects were modified and several new objects were added to the data model, see Figure 1. The ultimate requirement of the design is to allow connection of the usage data to a transformer without requiring that customers be graphically placed in the system. However, graphic placement of customers is allowed, if desired.

Electric Load Group
The Electric Load Group is a new object that simply maintains the relationship between a transformer installation and service points. Relating service points to transformers was considered but would limit flexibility. Implementing a load group provides a bucket for service points that can be related to any object. One immediate use of the load group is for CIS records that do not have a matching transformer. These records are related to a load group that is associated to the zip code of the customer.

Figure 1. Additions to FAME Model

Service Point
The Service Point is an existing object containing information about the customer that cannot be found in the CIS, such as design demand. It also contains the premise number of the associated customer, if available. The Service Point has a geometty field allowing the user to geographically place the Service Point. The geometry field is not mandatory due to the large number of Service Points created when loading the CIS data.

Premise
The Premise object contains customer information, such as account number, name, address, premise number, SIC code, etc.

Usage
The Usage object contains monthly bill information (usage and revenue) for the associated premise.

Facilitv Link
The Facility Link object matches an old transformer number and division code to the new unique facility ID.

Issues
During the design and implementation of the interfaces, several issues arose. The most notable issues and their solutions are discussed below.

Matching Premises to Transformers
Issue	CIS records contain transformer tag numbers that are only unique across an operating district.
Solution	The transformer records converted into the GIS contain the same tag number as the CIS. The tag number is only unique across an operating district. Therefore, it was programmatically prefixed with the code of the operating district to which it belonged (example: 1234 became 1901234 where 19 was the code for the operating district).

Issue	Tag numbers should be, but are not guaranteed to be unique.
Solution	The transformer installation has a unique facility ID that is a derivative of the system ID. The Facility Link table was implemented to match the tag number in the CIS to the corresponding unique facility ID in the GIS. This process ensures that a CIS record is associated to a unique transformer.

Issue	Only 75% to 80% of the customer records contain the tag number of the connected transformer.
Solution	The load process searches the GIS for a transformer with a matching tag number. If a match is found, the customer is added to the load group of the transformer. If a match is not found, the customer is added a load group associated to the zip code. By associating the customer to a zip code, the engineer is easily able to review only the discrepancies in the area of interest.

Incomp/ete Data
Issue	Not surprisingly, the facilities data in the GIS is not complete. For example, a conductor span may have been converted without a size or type. Although it does not affect map production, analysis failing due to omissions frustrates engineers.
Solution	Design and implement a user editable file to hold default values. The file lists default values for objects and attributes of facility data. If the extraction process encounters an object that is missing data required for analysis, it references the default values and substitutes the unset value with the specified default. Although the default value is not always the actual value, it allows the engineer to perform an analysis without manually updating missinq data.

GIS Topology vs. Analysis Topology
Issue	In the GIS, transformers, capacitors and other equipment are modeled as point objects connected to conductors. The analysis applications model these connections with a database relationship between the section and the equipment, usually with an attribute defining whether the equipment is at the load or source end of the section.
Solution	The extraction process must obtain the section ID of the conductor span where the equipment is located and the ID of the node that is nearest to the equipment. The section and node IDs are written to the ASCII file with the equipment characteristics.

Issue	In the GIS, it is valid to begin with an ABC phase conductor, split the A phase away from the others for some distance, bring the three phases back together into an ABC phase conductor. This situation happens frequently in the real world, but analysis applications typically interpret this configuration as a loop condition, and the analysis will not complete.
Solution	The extraction process does not allow the phases to merge back together. It creates additional separate sections parallel to each other.

Correcting Data
Issue	Engineers always encounter errors in the data when performing an analysis. Due to the manual nature of building an analysis model, the corrections have only occurred in the analysis model. The corrections are not carried through to the facilities data.
Solution	The technology put in place allows a change to the process. The engineer extracts the data from the GIS, and imports it to the analysis application. If errors are encounter, the engineer has the ability and responsibiiitv to make the corrections in the GIS and re-extract the data.

Processes and procedures
The following figures illustrate the process of loading CIS data into GIS, Figure 2, and the process of extracting data from GIS for analysis applications, Figure 3.

CIS Data Loader Process	Analysis Intetiace Process
Figure 2- CIS Data Loader Process	Figure 3- Analysis Interface Process

Data mainentance
As in all enterprise systems, there must be an effective means data. of maintaining accurate.

Updates from GIS to Analysis
The GIS data will always be the most current data available to the engineer. The tools and processes provided to the engineer, allow for quick extraction of current information. Updates from the GIS to the analysis application take place each time the engineer extracts data for a distribution study. If data errors require correction during the distribution study, the engineer corrects the data in the GIS and quickly re-extracts the area of interest.

CIS updates to GIS
As new customers connect and disconnect, the service person doing the work will record the ID number of the transformer that powers the customer. This ID number is recorded by the CIS.

Each month the customer data is extracted from the CIS systems and loaded into the GIS. During the load process any record that is not associated to a load group will trigger a search for a transformer with a matching ID. When the matching transformer is found the new customer is linked to the associated load group.

Conclusions
The interfaces implemented at UtiliCorp have two major benefits, reduced labor and increased accuracy. The labor reduction is achieved by automating a very tedious process. The increased accuracy is achieved by using actual customer loads rather than estimated.

UtiliCorp has implemented the interfaces in a cost efficient manner. The method of linking customers into the facilities network is automated, customers do not need to be geo-positioned or connected to services and secondaries.