TL;DR
This paper proposes a data-driven framework for power grid operators to decide whether to repair or replace aging main transformers. By combining Grey Theory (GM(1,1)) to predict equipment failure rates with a Whole Life Cycle Cost (LCC) model, the researchers provide a method to move away from rigid, schedule-based maintenance toward more economical, status-based decision-making.
What Problem Does This Paper Address?
Managing a modern power grid is a balancing act between reliability and cost. Traditionally, China's power grid enterprises have relied on "regular planned maintenance." This "one-size-fits-all" approach often leads to two extremes: over-maintenance (wasting money on healthy equipment) or under-maintenance (leading to unexpected failures).
Furthermore, existing methods for forecasting when to replace equipment (technical transformation) often rely on the "asset wall" theory, which focuses primarily on the age of the equipment. The authors argue that age alone is a poor predictor of health; two transformers of the same age might have very different "wear and tear" based on their specific operating conditions. There is a need for a model that integrates real-time operational health with long-term economic forecasting.
Methods
The authors employ a two-pronged analytical approach:
-
Operation Status Assessment (Grey Theory): To predict future failure rates from limited historical data, the paper uses a GM(1,1) dynamic prediction model. Grey Theory is particularly useful here because it excels at analyzing systems with "small samples and poor information." The model takes original sequences of equipment defects, smooths them through accumulation, and solves differential equations to forecast future health trends.
-
Whole Life Cycle Cost (LCC) Modeling: To compare different maintenance strategies (e.g., a major repair vs. a full replacement), the authors calculate the LCC using five dimensions:
-
Initial Investment (CI): Purchase, installation, and demolition costs.
- Operating Cost (CO): Daily inspections and energy consumption.
- Maintenance Cost (CM): Planned and unplanned repairs.
- Failure Loss Cost (CF): Costs associated with troubleshooting and power outages.
- Retirement Disposal Cost (CD): Labor, transport, and environmental costs minus residual scrap value.
Because a repaired transformer and a new transformer have different expected lifespans, the authors use the Net Annual Value (Equal Annuity) to normalize the costs, allowing for a fair "apples-to-apples" comparison.
Datasets
The study utilizes operational data from a power grid company, specifically focusing on:
- Oil-immersed transformers: A dataset covering equipment with operational lives ranging from 1 to 17 years.
- Case Study Subject: A 110kV main transformer from the Guangdong Foshan Transformer Factory, commissioned in 1996 (18 years of service at the time of study) with a rated capacity of 31,500 kVA.
Key Insights and Ideas
- Status Over Age: The core "aha moment" is that equipment health (predicted via Grey Theory) should be the primary driver for investment, not just the commissioning date.
- The "Equal Annuity" Equalizer: Comparing a repair that extends life by 12 years to a replacement that lasts 30 years is difficult. By calculating the "Equal Annuity," the researchers show that even if a new transformer is more efficient, the massive upfront capital cost often makes the repair scheme more economically viable in the long run.
- Predictive Accuracy: The GM(1,1) model showed high alignment between predicted and actual defect rates (most results falling within the 0.905 to 1.105 range), validating Grey Theory as a viable tool for transformer health forecasting.
Key Results
The researchers conducted an empirical analysis comparing a Repair Project (extending life by 12 years) vs. a Technical Transformation Project (replacement with a 30-year life):
- Total LCC: The repair project cost approximately 1.41 million RMB, while the full replacement cost 6.87 million RMB.
- Equal Annuity (Average Annual Cost): The repair scheme cost 147,776 RMB/year, whereas the replacement scheme cost 283,823 RMB/year.
- Decision: Despite the replacement scheme having lower annual operating and maintenance costs (CO and CM), the sheer magnitude of the initial investment (CI) made the repair scheme the recommended choice from a life-cycle perspective.
Limitations and Future Work
- Data Granularity: The paper relies on general "defect rates." Future models could benefit from more granular sensor data (dissolved gas analysis, winding temperature, etc.) to refine the Grey Theory predictions.
- Scope of Costs: While the LCC model is comprehensive, it may not fully account for the indirect costs of carbon footprints or the systemic risks of keeping older equipment on the grid as it becomes less compatible with modern smart-grid technologies.
- Single Case Study: The results are based on one specific 110kV transformer; broader validation across different voltage levels and manufacturers would strengthen the findings.
Summary updated