Revamping Energy Storage Including Degradation Effects

The Revamping Energy Storage Including Degradation Effects (RESIDE) project, carried out 
at the JRC Smart Grid Interoperability Laboratory (SGILab), focuses on optimizing the opera
tion of Battery Energy Storage Systems (BESS) by improving the modelling of the main under
lying phenomena which characterize and influence their performance. 
As the share of renewable energy in power systems continues to increase, managing the var
iability and uncertainty of these resources becomes critical. BESS plays a pivotal role by offer
ing flexible storage solutions, fast response services, and contributing to Capacity Markets 
(CM), where they ensure security of supply during peak load periods. The main objective of 
the project is to enhance a novel revenue maximization model for BESS operation, which is 
designed to optimize its functioning in both Wholesale Energy Markets (WEM) and CM while 
accounting for short-term degradation and long-term optimal cell replacement. 
The primary motivation for the project stems from the need to better understand and accurately 
model the BESS efficiency during charging and discharging operations. Existing models often 
assume fixed efficiency, which does not reflect real-world behavior where efficiency can vary 
depending on factors like charging/discharging power and the state of charge. This discrep
ancy can lead to suboptimal dispatch strategies, increased operational costs, and inaccurate 
energy availability estimates. The RESIDE project addresses these issues by developing a 
dynamic efficiency model that adapts to changing operational conditions, making it more suit
able for effective integration into optimization frameworks. 
The experimental component of the project involved performing a series of tests on a Li-ion 
BESS at SGILAB, where the system was subjected to both full and partial charging and dis
charging cycles under varying power levels. These tests were designed to capture the impact 
of different operational conditions on the system’s efficiency. Key electrical parameters—such 
as voltage, current, and power—were recorded, and the state of charge was monitored to 
assess the system’s efficiency across different stages of operation. The data collected during 
the tests was processed and divided into two subsets: one for training the model and another 
for testing its accuracy.
Initially, machine learning techniques like Regression Trees and Gaussian Process Regres
sion were explored to predict BESS efficiency. However, due to the complexity and computa
tional demands of these methods, the focus shifted to more straightforward techniques, such 
as polynomial fitting and linear regression. These models provided a good balance between 
prediction accuracy and computational efficiency, allowing for better integration into the 
broader optimization framework. 
To evaluate the accuracy of the model, key statistical metrics such as R-squared, Root Mean 
Squared Error (RMSE), and Minimum Absolute Error (MAE) were calculated for both the train
ing and testing datasets. The Power Conversion System (PCS) efficiency estimation model 
demonstrated excellent accuracy, with an R-squared value of 98.00% for the charging process 
and 91.96% for discharging. Similarly, the battery efficiency model achieved R-squared values 
of 63.00% for charging and 61.48% for discharging. These results confirm that the model is 
effective in predicting BESS efficiency under different operational scenarios and can be reliably 
used for optimization purposes. 
The project successfully generated analytical expressions that quantify BESS efficiency as a 
function of operational power and state of charge, enhancing the understanding of how these 
factors influence system behavior. By treating the charging and discharging processes sepa
rately, the model was able to capture the particular trends in the system’s operational dynam
ics, further improving the accuracy of the efficiency estimations. 
Finally, a comparative financial assessment was conducted over a 10-year investment horizon, 
incorporating the improved efficiency model into BESS revenue optimization. The results show 
that accounting for dynamic efficiency lead to a +47.44% increase in net revenues, significantly 
overcoming conventional models.  
The findings of this study offer valuable insights into the performance of BESS under varying 
operational conditions and demonstrate the feasibility of integrating a dynamic efficiency model 
into the broader optimization process. This enhanced model not only improves short-term op
erational decisions but also provides the foundation for long-term planning, including consid
erations for capacity degradation and cell replacement. By addressing both efficiency and deg
radation, the model enables a more comprehensive characterization of BESS operations, en
suring that the systems are better equipped to meet the demands of renewable energy inte
gration and contribute more effectively to Capacity Markets. 
In conclusion, the RESIDE project makes significant strides in advancing the modeling and 
optimization of BESS operations. The development of a dynamic efficiency model, coupled 
with the integration of degradation effects, provides a more accurate and realistic representa
tion of system performance. This will ultimately contribute to better decision-making, lower op
erational costs, and enhanced long-term financial viability for BESS deployments, further sup
porting the transition to a sustainable and reliable energy system.