Understanding loss model calculations and differences from an EIA
What is a loss model?
To calculate the losses for a given issue, WindESCo leverages loss model calculations specific to the issue of interest to ensure that an appropriate estimation is made. For example, static yaw misalignment has a loss model specific to it using a cos2 relationship:
In contrast, the issue “Pitch curve inconsistent over time” uses a fractional loss model by applying a loss fraction to the filtered power data:
The loss model appropriate for the issue of interest is selected by our subject matter experts (SME) who have analyzed these issues for many years and leverage their experience, first principles, and the latest industry thinking.
How to understand a loss model:
Each loss model also includes a description of the approach at a high level followed by the time period of analysis and the total energy produced during the period. This is graphically displayed on the right as well. Our goal is to provide enough information that another individual understands the assumptions made and can validate if they agree, or disagree, with the approach.
For a full list of the loss models that WindESCo uses, please refer to our Glossary.
How is a loss model different than an Energy Improvement Assessment (EIA)?
When an issue is corrected, the gains are tracked through an inverse of what the losses would have been if the issue were not corrected. To ensure that gains are not tracked when a Fix is incorrectly applied, WindESCo continues to review for all issues on a continuous basis. Any issue that is re-identified will then have their gains appropriately adjusted.
The loss model approach allows customers to implement correct several issues in parallel and continuously track gains rather than being restricted to allow for a specific measurement time period. This means loss models are intended to be illustrative (within an order of magnitude of accuracy) to support decision-making in site improvements rather than a precise measurement technique.
In comparison, WindESCo’s EIA approach is a DNV-approved power comparison tool using machine learning to calculate changes in static parameters, sensors, or mechanical features of individual turbines. The EIA is a robust tool that requires three months of data to develop a baseline, as well as three months of data after an implemented change to calculate the gains. (To learn more, please refer to the EIA Methodology knowledge page.) While very powerful, the EIA has several drawbacks including:
- Requiring six months of high-frequency data
- Vulnerability to the time of year a change is implemented due to seasonal variations on the wind site (this can cause control turbines to have a low similarity score between the control and test periods meaning the results cannot be considered reliable)
- Sensitivity to other changes on the site – if, during the course of normal operations, other changes are introduced on the wind farm that does not leave sufficient control turbines then results can be invalidated..
Given the practicalities of running a wind plant it is difficult to ensure the criteria for an EIA can be fulfilled. Therefore, WindESCo developed the loss models as alternatives to provide customers decision-making information to improve performance and reliability on their wind.
Why didn’t WindESCo use X model?
WindESCo is continuously reviewing leading literature on energy performance calculations and the pros / cons that come with each approach. As further information is released, we will continue to refine our calculations to stay consistent with leading industry practices.