---> GITA 2000 ---> User Perspectives

Development of transformer load management at the Modesto irrigation district

Scott Simpson & Bill Woods

The Modesto Irrigation District believes the economic paybacks in advanced informational technology are of critical importance, especially towards making operational assessments and decisions within the electric utility industry. This paper and the corresponding presentation will emphasis the application of electrical engineering principles and custom GIS (geographic information systems) development required for development of a practicable transformer load management application.

Modesto Irrigation District Background
The irrigation district lies in Californiaís Central Valley, south of Sacramento and east of San Francisco. The district is a public utility providing electric and irrigation service to the city of Modesto and surrounding communities. The bulk of the districtís 90,000 electric power customer accounts are residential, followed by light commercial, irrigation, and large commercial power consumers. Population of these customer sets is similar in some ways to other electric utilities of similar size, but in other ways differs significantly. Agricultural pumping for irrigation customers reflects a broad range of electric demand requirements, while large commercial operations process agricultural products on a sharply defined seasonal basis. These commercial operations include caning operations for vegetable and fruit products, as well as large wine bottling facilities.

Project Background
The Modesto Irrigation Districtís current GIS project is an implementation of Smallworld GIS utilizing a custom electric and irrigation data model, along with a broad range of applications serving those sectors. The system is currently operational in a production environment, though new data modeling and application development continues through the present. One of the new applications in development is the transformer load management system.

Electric transformers in the district are modeled as both underground and overhead equipment arrays. Data fields on the transformer objects in the system capture a wide range of information, including:
  • Phase configuration,
  • Kva load rating
  • Location and equipment identifier numbers,
  • Part numbers,
  • Ownership
  • Map-grid-sequence number identifiers.
Transformer load management concerns itself with assessing, based on the customer types, typical load profiles for customer sets, number of customers and electric power demand history, the relative load demands placed on current and possible transformers in the system. An intelligent transformer load management application needs to accomplish three central goals.
  • First it allow for field validation of the transformer to customer link.
  • Secondly it must provide loading analysis that takes into account environmental and equipment factors, as well as statistically determined load profiles.
  • Finally, it must allow for an efficient yet complete reporting of the resulting load analysis.
The Modesto Irrigation District (MID) chose to pursue completion of these goals by developing the necessary data model and code within their electric GIS. Additionally, because the required customer and meter read data reside in a separate Oracle database at MID, the application had to be built with the ability to capture and process customer data stored externally from the GIS. The relationship between customer (and therefore load) data and the transformer object data stored in the Modesto GIS is implied rather than explicit; no data join relationships exists between transformers and customer objects within the data model. Retrieval of the correct customer load read records from the Oracle database is achieved by referencing map-grid transformer identification numbers, a unique point of commonality that exists in both the GIS and customer information system data stores.

Once retrieved the customer information system data is pre-processed and stored in the GIS database. This duplication of data may seem inefficient on first examination. However, there are a number of good reasons for the approach. On a pragmatic level, by replicating data into the GIS, the transformer load management application can now be downloaded to a laptop computer and be used in the field to verify customer-transformer ties and also by a trouble-shooter when investigating power quality issues and transformer failures. Additionally, it was our experience that performance of the transformer load management analysis was significantly enhanced by translating the data and storing it internally on the GIS. Finally, given the large amount of processing required to use the customer information system data for meaningful analysis, it made sense to store the results of that analysis, thus speeding repeated transformer load management on already processed transformers.

Benefits of Transformer Load Management

Typical MID Load Patterns

The industrial customers differ from those found in many utilities in that they are nontypically seasonal in their loading; most of these customers are agricultural-industrial in nature, including processing, canning, bottling facilities and the like. TLM analysis in the district takes this into account with modification factors specific to these customers. Normal climactic seasonal variation occurs as one would expect. Peak loading occurs late summer with high residential air conditioning use.

One requirement for the development of an effective GIS based transformer load management application was the determination of typical load profiles and factors for typical customer sets within the irrigation district. Load research was undertaken to provide diurnal (hourly) load shape information. This load research effort performed extensive sampling of the four classes of customer represented within the district, including residential, small commercial, large commercial, industrial and irrigation. Statistical sampling was necessary because it was cost prohibitive to do an actual load analysis of all customers in the district. The sampling of these customer sets was performed using statistically valid stratified subsets across the district. Data types collected by the study varied by customer type. The energy consumption for residential and small commercial customers was sampled at 15-minute intervals over 31day intervals. Irrigation customers were sampled for kWh and rated horsepower of water pumps. Large commercial customers were sampled for kW. The final load factors determined thus take into account these typical, seasonal load shapes.

Why Transformer Load Management
By developing the transformer load management application, the irrigation district seeks to use their GIS for realizing direct equipment, labor and materials efficiency. The cost of electric transformers can be viewed as a function of load capacity over time, and the resulting effect on equipment life span. An under loaded transformer is a capital loss; the same transformer could be used to serve higher customer concentrations and therefore make for better utilization of that particular piece of equipment. Alternatively, an overloaded transformer would be expected to have a shorter usable equipment life. By providing transformer load management within the context of the GIS an operator will be able to make more intelligent decisions regarding the continuing use of certain transformers in certain situations. Decisions might then be made to switch out a particular transformer for a higher or lower load capacity unit.

The transformer load management system also has the ability to allow a designer or operator to test hypothetical load situations. For example, customer load expansion might be planned for an existing transformer location. A designer, using the transformer load management application could begin by examining the current and historical loads occurring on a transformer.

Then, the operator may add any number of "new" customers onto the transformer, basing the hypothetical kWh consumption on that typically seen within the neighborhood, on any other amount determined suitable for the analysis. This additional hypothetical customer loading is added to the transformer load management analysis at run-time in an interactive process using GUIs built specifically for this task. No artificial data is added permanently to the GIS or Oracle customer information system database. The resulting load analysis would allow the designer/operator to make a more informed decision concerning where new load can be reasonably added, and when new transformer capacity must be included in the design.

Conversely, customers may be removed from a transformer for the purpose of analyzing how that change would effect the monthly loading. Again, this change would be achieved at runtime through an interactive design process. No actual changes to either the GIS or customer information system data would be realized.

Another use for the transformer management system is interim batch reporting on all transformers in the system. An automated monthly process will be initiated without user interaction where all the transformers undergo load analysis based on the actual customers they serve. This batch reporting takes advantage of off times when the GIS and customer information systems are not being used in order to undertake the extensive processing required to evaluate all transformers.

This report then provides a human reviewer with a list of potentially overloaded or under-loaded transformers that may require further investigation. This approach will also be used to develop a special drawing application that will draw under loaded and overloaded transformers with distinctive styles. This will allow for a visual geographic interpretation of the batch analysis results.

Development of Transformer Load Management
As briefly stated above, the development of the application required an interaction between the GIS and the irrigation districtís Oracle based customer information system. The customer meter read records extracted from the Oracle database comprise a complex set of information fields that are then partially translated for local storage within the GIS. These fields include billed demand (kWh or kW depending on customer type), consumption amounts, billing period dates, customer type, address information, horsepower in the case of irrigation accounts, load factor, meter number, account number, location number, read types, power factor and kva. Some of the complexity associated with extracting and using this information arises from unusual customer account activity. For example, switching meters on and off by reflecting a change in customer must thereby be taken into account by the transformer load management system. Occasionally situations arise where all meters attached to a particular transformer arenít read on exactly the same dates, and sometimes a faulty read results in a meter being reread days after the first reading. For these situations magik code (the proprietary development language for Smallworld GIS) has been developed to correct and normalize the information as much as possible in the pre-processing leading up to the load analysis. In some situations abnormal meter read data can not be readily resolved by the system in an automatic fashion, and operator intervention is necessary. These anomalous data are disregarded by the system when batch load analysis is carried out, and the transformers with those instances are flagged for operator investigation.

Storage of the customer meter read data from the customer information system into the GIS presents its own set of problems. Several data model additions created a set of objects storing the read data, along with the address information, various account and location numbers, against the transformer object in the GIS. However, because the customer information system and GIS do not link changes in real time the storage of customer read data must be periodically updated. Optionally, a user investigating transformers on an individual basis (as opposed to the batch processing) is able to generate load analysis based on the most current customer information by reconnecting to the Oracle database. This action also causes an update on the stored GIS versions of that data. Therefore, changes in the customer data may be reflected immediately in the GIS. When transformers are changed in the GIS as part of the design process and then those changes are enacted on the ground, standard processes are used to reflect those changes in the customer information system. Thus situations may potentially arise where the state of a transformer is temporarily portrayed inaccurately because the customer data is inaccurate.

Load analysis takes place by accumulating the electric consumption for all customers on a transformer and then applying load-modeling calculations to the aggregate consumption. For example, in the instance of residential customers, the total number of customers on the transformer determines a multiplier and addition factor used in translating the aggregate monthly kWh into monthly kva (demand). These calculations are further modified based on the season in which the monthly demand falls.

Tables exist for residential summer and winter months, and for light commercial and irrigation calculations. Heavy commercial accounts record kva directly and thus require less processing, though there are a set of coincident factors used to conform the results to the districtís load profiles.

Since the kva rating for transformers in the GIS is stored as a field, once the kva load has been determined these monthly results may be compared with the transformer rating in determining the relative load of the transformer. The results are presented in a text file report in the instances of batch processing, or in a GUI based report when individual transformers are being actively evaluated by a user. The report format shows a line by line breakdown of the individual customers loading the transformer, including individual demand ratings, as well as the aggregate result. This information is further broken down by month, with the report indicating the month of maximum demand.

Cost of Ownership Analysis
An additional cost of ownership feature will allow for the analysis of relative costs associated with transformer selection and the decision to change out existing transformers. This feature will take into account the cost difference between different transformers, estimated labor and equipment expenses for changing it out, and the relative cost of load and no-load transformer losses. Electric utilities already assume that a certain number of transformers will have shorted life span and suffer premature failure within their networks. By adding cost analysis capability to the districtís load management system there will be a relative basis for more accurate decision making in this regard.

Transformer load management in the Modesto Irrigation District is a realization of part of the promise of GIS. By combining GIS location information with power consumption data extracted from an outside system, the GIS becomes a more powerful facilities management system. It is hoped that transformer load management will allow the district to allocate resources in a fashion that results in direct savings of equipment and equipment life span, while eliminating under utilization wherever possible.

© All rights reserved.