When comparing the kilovolt-amp demand loading on southern company substations with the actual connected transformer nameplate kVA, it was apparent that, on average, the distribution transformers were not being heavily loaded. This finding was supported by additional historical evidence that a very small number of transformers fail due to overloads. This information led Southern Company to the hypothesis that the utility could realize significant savings with an improved transformer size selection process for new facilities. The utility found it could save in investment and operations by more closely sizing transformers to actual load, while still operating within acceptable risk and safety limits.

At the heart of this problem is a basic business issue that a decision must be made to allocate an investment based on balancing risk, value and performance in an environment of uncertainty. Before the building to be served is built and the actual load metered, the transformer size and design must be selected and installed. To minimize risk and investment, each step of the process needed to be investigated and improved. The first and most crucial step in correctly sizing a transformer is to estimate the future customer's kVA demand. This includes the initial gathering of information and then applying the information to forecast the demand.

Southern Company supplies energy to approximately 4.2 million customers over a 120,000-square-mile (310,799-square-km) service territory spanning most of Georgia and Alabama, southeastern Mississippi and the panhandle region of Florida. Southern Company is comprised of four regulated retail electric utilities: Alabama Power, Georgia Power, Gulf Power and Mississippi Power. Annually, Southern Company purchases approximately 70,000 distribution transformers.

ESTIMATING COMMERCIAL LOADS

The initial phase of the study focused exclusively on commercial loads. For demand rate customers, actual metered demands were available from billing data. The maximum winter and summer demands were determined from two years of history and were compared with the original estimated demand forecasted for the facility. These findings supported the initial concerns. Frequently, the customer's estimated demand, which is used to size the transformer, was significantly greater than the actual maximum demand. It was obvious that if we wanted to do a better job of sizing and loading our distribution transformers, a better tool was needed to estimate the customer's demand. Even a small improvement in transformer sizing could result in very significant annual dollar savings. In order to get better results, representatives in the field needed better information and a way to use that information.

The easiest and lowest hanging fruit for solving the load-estimating problems was to look at our national accounts such as Wal-Mart, The Home Depot and McDonald's, for which we have very good demand records. These types of businesses generally build new facilities that closely match existing facilities. A simple listing of the existing facility demands provided an easy and accurate estimate of what the new kVA demand would be. These facility demands provided an initial starting point for evaluations.

Unfortunately, not all new commercial customers are national accounts. For customers with no comparative examples, a better estimating tool was needed. Working in conjunction with Z Solutions Inc. (Atlanta, Georgia, U.S.), a consulting company specializing in statistical analysis and load research, a project was initiated to develop software for estimating commercial loads. Working with Z Solutions the project team, with members from each of the four operating companies, developed a process to minimize the uncertainty of forecasting the building's demand.

The field engineer or marketing representative has many factors to consider in the process for estimating the load. These factors include the type of building: office, restaurant, retail and so forth; the size of the building; the appliances in the building; and the diversity for each appliance. The diversity is the percent of the total maximum connected load of the appliance expected to be seen at the time of the building's peak usage. Selection of the correct diversity factors is critical in estimating the new customer's demand and, therefore, the correct transformer size. Previously, the marketing representative depended on past experience or general rules of thumb passed down from generation to generation. Recognizing this is a difficult task and an uncertain science. A software tool was developed to provide a greater degree of guidance to the field-marketing representatives.

A database was produced for existing customers that contained customer type, square footage, electrical end-use equipment, original forecasted maximum demand and actual metered maximum demand. Z Solutions used this database to develop a distribution of actual demands per square foot and end-use equipment diversity factors. The estimated diversities (percent of connected load operating at the time of the building peak demand) are used in calculating the expected building maximum demand. Using actual metered demands and totals of installed appliance connected loads from plans and diversities that were estimated from the data and the square footage from the plans, an initial estimate could be developed. The estimator uses the demand-per-square-foot distributions to check the forecasted demand against actual history. This way, regardless of the size of the building, the estimator would know how the new building compares to other similar uses. The estimator could then check this demand versus the history. If the judgment is that this building is typical, then the building kVA demand per square foot should be close to the average. The majority of buildings are close to the average kVA demand per square foot. Therefore, if the software user is forecasting a deviation from that point there needs to be a solid reason for the deviation.

NOW IT IS CLEAR

The software program, called CLEAR, which stands for Commercial Load Estimating and Referencing, was developed to implement these concepts. CLEAR lets the user select the customer facility type, such as restaurant, office building or church, and square footage of the building. Next, the user enters the connected load of the appliances to be used in the building (the end uses). This information is derived generally from the plans for the facility and contains information such as: the kVA connected load of the HVAC equipment, the total of the lighting (literally counting light bulbs), motor loads and so on.

Based on an analysis from billing and survey data, the software then applies a diversity factor to the end-use loads specified by the user. The user has the option of using the default diversity factors calculated from the database of all buildings or, if there are strong reasons to, modifying them. The sum of the diversified end-use loads is the kVA demand for that building.

The calculated demand is then compared with the kVA per square foot for other customers of that same type as a check to ensure that the value is in the expected range. Another important feature of the software is the calculation of the hours-use number for the customer. This number is defined as the total annual kWh divided by the maximum kW demand, and can also be thought of as the number of hours of annual peak demand. The magnitude of this number is an excellent indicator as to the general shape of the load duration curve for the customer. The higher the hours-use number, the flatter the load shape — closer to a constant load. The lower the number, the more “peaked” the load shape. Knowing the general shape of the load-cycle curve is essential to knowing the loading effect on the distribution transformer. Loads with a very high demand but a very low hours-use number may be served more effectively with a smaller transformer than loads of the same kVA demand with higher hours-use loads.

ESTIMATING LOAD-CARRYING CAPACITY

Southern Company is a member of the Distribution System Testing Application and Research (DSTAR) group and was heavily involved with the project to develop a transformer total owning cost software (TOCS) program. The DSTAR program uses annual 8760-hour load data and weather temperature data to predict transformer life based on the ANSI/IEEE transformer loading formula. Using the DSTAR software as a basis, transformer design parameters from Southern's transformer suppliers and the 8760-hour load research data, we could accurately calculate the size transformer needed for each account. Z Solutions took the calculations of the same ANSI/IEEE transformer loading procedures, which DSTAR had developed in spreadsheet form, and implemented them into a statistical package. This greatly improved the efficiency of data input and the output of results.

Before, an analysis would consist of only a few load models and transformer designs. Now, an analysis of multiple load shape properties and all possible transformer designs may be performed. Literally thousands of results may be organized and reported in order to reveal differences in transformers and how they may be used. These results often lead to unexpected findings.

New sizing guidelines were developed based on the load-carrying analysis reflecting both energy and demand impacts. In general, the newly sized transformers were one to two transformer sizes smaller than what was generally specified in the past. This reflected both the new analyses and greater confidence in the improved forecasting methods.

NEXT STEPS

After completion of the commercial analysis, the project team turned their attention to the residential market. This task again involved the analysis of specific loads gathered from residential cost-of-service load research data. In the residential market, one transformer usually serves multiple homes, including high-density multi-family developments, where as many as 200 units may be served. The key factor in estimating the kVA demand of these applications is not only the size of the homes and the types of appliances used, but also the coincidence of the maximum demand of the homes. Again, multiple analyses were performed to estimate the impact of all these factors on sizing requirements. The work is being finalized and reviewed for implementation at this time.

The last step in this process will be to close the loop on the entire process. This will include the consideration of new transformer designs and sizes based on the findings from the previous work and the frequency of occurrence of the kVA demands on the system.


G. Bruce Shattuck began his career with Alabama Power as a student engineer in 1969 and graduated from the University of Alabama with a BSEE degree in 1971. He progressed through various positions of increasing responsibility to his present position as principal engineer in the Power Delivery — Distribution Engineering Services group. He is past chairman of the Southeastern Electric Exchange UD Committee and the DSTAR Committee. He is a member of the Association of Edison Illuminating Companies (AEIC) Cable Engineering Committee and serves as the Southern Company lead product engineer for underground cables. GBSHATTU@southernco.com

R. Gary Huff is the president of Flamingo Solutions, LLC. He is an expert on analysis of consumer and business demand behavior and was a subcontractor with Z Solutions when the work of this article was conducted. Huff was previously manager of the load research at Southern Company, where he directed load research sample design and analysis for its four operating companies. garyhuff02@earthlink.net

System 12.47 kV/208 V Three-Phase Pad Sizing Table - All Transformers

Maximum summer kVA range
Lower kVA 0 61 96 101 106 111 116 161 171 181 189 196 206 216 226 282 301 376 401 426 451 576
Upper kVA 60 95 100 105 110 115 160 170 180 188 195 205 215 225 281 300 375 400 425 450 575 600
Hours-use                                            
0 - 999 45 75 75 75 75 75 112.5 150 150 150 150 150 150 150 225 225 300 300 300 300 500 500
1000 - 1499 45 75 75 75 75 75 112.5 150 150 150 150 150 150 225 225 225 300 300 300 300 500 500
1500 - 1999 45 75 75 75 75 75 112.5 150 150 150 150 150 150 225 225 225 300 300 300 300 500 500
2000 - 2499 45 75 75 75 75 75 112.5 150 150 150 150 225 225 225 225 225 300 300 300 500 500 500
2500 - 2999 45 75 75 75 75 75 112.5 150 150 150 150 225 225 225 225 225 300 300 300 500 500 500
3000 - 3499 45 75 75 75 112.5 112.5 112.5 150 150 150 225 225 225 225 225 225 300 300 500 500 500 500
3500 - 3999 45 75 75 75 112.5 112.5 112.5 150 150 150 225 225 225 225 225 225 300 300 500 500 500 500
4000 - 4499 45 75 75 75 112.5 112.5 112.5 150 150 150 225 225 225 225 225 225 300 300 500 500 500 500
4500 - 3999 45 75 75 75 112.5 112.5 112.5 150 150 150 225 225 225 225 225 225 300 300 500 500 500 500
5000 - 5499 45 75 75 75 112.5 112.5 112.5 150 150 150 225 225 225 225 225 225 300 300 500 500 500 500
5500 - 5999 45 75 75 112.5 112.5 112.5 112.5 150 150 225 225 225 225 225 225 225 300 500 500 500 500 500
6000 - 6499 45 75 75 112.5 112.5 112.5 112.5 150 150 225 225 225 225 225 225 225 300 500 500 500 500 500
6500 - 6999 45 75 112.5 112.5 112.5 112.5 112.5 150 150 225 225 225 225 225 225 225 300 500 500 500 500 500
7000 - 7499 45 75 112.5 112.5 112.5 112.5 112.5 150 225 225 225 225 225 225 225 300 300 500 500 500 500 750

Transformer size selection based on the customer's kVA demand range, hours of use and transformer design