1. Transformer Loading Considerations
Using AMI Loading Data to Calculate Loss of Life
Presented by Steven Dennis May 2-4, 2018
SWEDE 2018 – Tulsa, Oklahoma

2. Why Consider Transformer Loading
Oil/air cooled transformers are typically between 95 and 99 percent efficient at transformation
The inefficiency of a transformer is materialized as an energy loss in the form of heat
Excessive heat accelerates the deterioration of the paper insulation, oil and gaskets
The breakdown of these components will ultimately lead to failure with all or some of the following consequences
Bridged winding (typically an increased output voltage)
Faulted winding
Tank rupture
Oil spill
The ability to predict and prevent an overloading failure limits environmental impact, improves customer experience and reduces operational expenses
C IEEE Guide for Loading Mineral-Oil-Immersed Transformers
Reference:

3. Factors to Consider When Determining Risk
Traditional Approach
Estimate loading from kWh readings
This method is error prone because of all of the assumptions that have to be made
What the coincident peak is in kVA
The duration of excessive load
How much longer the transformer will last before failure
That the connectivity model is accurate
With this method you have to set an arbitrary threshold such as 140% nameplate at an estimated peak

4. Factors to Consider When Determining Risk
AMI Approach
Loading comes from kWh intervals which gives a more accurate picture of the actual load curves and durations
This method mitigates some of the traditional assumptions
Coincident peak in kVA – Much higher resolution (15 minute vs 30 Days)
The duration of excessive load
Calculate the loss of life in percent (C IEEE )
Validation of the connectivity model with voltage statistics (Pearson Correlation)
This approach allows you to accumulate loss of life and rate of change
With this system you can prioritize by the likelihood of a near term failure

5. Effective Load Metering
When the load changes on a transformer the temperature equilibrium lags behind the actual loading
Individual AMI meters can supply enough resolution to bridge the time constant of thermal lag for calculating internal temperatures
Aggregated loading can be used from the connected meters as long as there is confidence in the accuracy of the connectivity model

6. Transformer Characteristic Tracking
We use an inventory model to track the physical characteristics of each transformer
Rating in kVA
Weight
Oil capacity
Purchase date
Install date
This inventory model is linked to a table of manufacture specifications for each transformer design
No load loss
Full load loss
Thermal time constant
Etc.
Tables in our electrical model hold the meter to transformer connectivity data
The transformer details are accurate but the hit rate back to the connectivity model is less than 100%.

7. Risk Ranking Risk is calculated using
Validated connectivity
Loss of life in percentage
Rate of change for the loss of life
Risk is ranked as a projection to the nearest point of 100% loss of life using historical data

8. Optimal Conditions for Load Modeling
Loss of life modeling is a continuous and accumulative process
The rate of change in loss of life is key to forecasting a failure and is not necessarily correlated to seasonality. In many cases it is associated with connected load changes.

9. Questions?