Enart Energy Selects Wind Turbine Blade Materials with Data Using EDMAP
Enart Energy was selecting blade materials intuitively across wind farms with 300+ MW of installed capacity. The EDMAP model developed by Corius predicts mechanical properties with R²=0.9997 accuracy, reducing material selection time from 2–4 weeks to 8 minutes and extending blade lifespan by 30%.
Blade material selection was made intuitively across varying temperature conditions; incorrect choices shortened blade lifespan and led to high maintenance costs.
EDMAP (EpoxyDMA-Predictor) — an MLP-based machine learning model that predicts the mechanical properties of epoxy composites with R²=0.9997 accuracy using temperature and material identity inputs.
Blade lifespan extended by 30%, annual maintenance cost reduced by 22%, material selection time dropped from 2–4 weeks to 8 minutes, full ROI in 6 months.
What Operational Challenges Was Enart Energy Facing?
Enart Energy is a major renewable energy company operating 15+ wind farms across the Aegean and Marmara regions with 300+ MW of installed capacity. Turbine blades are manufactured from glass fiber-reinforced epoxy composites that operate under high stress. However, temperature ranges across different regions of Turkey (−10 °C to +45 °C) dramatically affect the mechanical behavior of these materials. Without a systematic material selection process accounting for this variable, the team either defaulted to costly premium epoxies or faced temperature-induced fatigue failures.
- The temperature-mechanical property relationship was determined through laboratory testing; the process took 2–4 weeks per material.
- Premature cracking and delamination damage from incorrect material selection caused an average of 3–5 blade replacements per year.
- Storage modulus (E′) and energy dissipation (tanδ) values were measured via manual DMA tests and tracked in spreadsheets.
- Existing tests were only conducted at the prototype stage; there was no way to retroactively predict outcomes for temperature scenarios observed in the field.
- There was no standard numerical criterion for comparing epoxy formulations from different suppliers.
How Did Corius Solve These Challenges?
Corius developed the EDMAP (EpoxyDMA-Predictor) machine learning model together with Enart Energy's R&D team. The model predicts Storage Modulus (E′), Loss Modulus (E″), and Tan Delta (tanδ) values in milliseconds using temperature and material identity inputs. Trained on an MLP architecture and exported to ONNX format, the model was integrated into both engineering workstations and field tablet applications.
Enart's historical DMA (Dynamic Mechanical Analysis) test data was compiled. Data was cleaned and normalized by epoxy formulation and temperature profile, and a model training set was created.
- Digitization and consolidation of 3 years of laboratory DMA test records
- Extraction of temperature-property curves for 12 different epoxy formulations
- Completion of missing values via interpolation and outlier detection
- Train / validation / test split (70/15/15)
A multilayer perceptron (MLP) architecture was designed and hyperparameter optimization was performed. R²=0.9997 accuracy was achieved for Storage Modulus and R²=0.9980 for Tan Delta. The model was exported to ONNX format.
- MLP architecture: 3 hidden layers, ReLU activation, dropout regularization
- Hyperparameter optimization: 200 trials with Optuna
- Separate output heads for E′, E″, and tanδ
- ONNX Runtime export for production deployment
- Cross-validation with 30 engineer-approved scenarios
EDMAP was integrated into Enart's material selection workflow. Engineers can now enter the expected operational temperature range for an installation site and compare the most suitable epoxy formulations within seconds.
- EDMAP service layer with REST API wrapper
- Integration with internal materials database
- Lightweight ONNX Runtime setup for field tablet (Android)
- Engineer training and user guide preparation
How Was EDMAP Used in the Field?
Region-Based Material Selection
Engineers enter the annual min/max temperature range of an installation region into EDMAP. The model presents comparative E′, E″, and tanδ curves for each epoxy candidate and recommends the best balance in terms of mechanical performance.
Temperature Scenario Simulation
Extreme temperature events experienced by existing blades (heat waves or frost events) were retroactively simulated to check whether the material had exceeded critical thresholds.
Supplier Benchmarking
New epoxy formulations from different suppliers were pre-screened with EDMAP before undergoing standard DMA testing; only the most promising candidates were directed to full testing.
Blade Lifespan Prediction and Maintenance Planning
Operational temperature data and EDMAP predictions were combined to model cumulative fatigue on a per-blade basis, generating periodic maintenance scheduling recommendations.
What Results Did the EDMAP Project Produce?
The model completely changed our energy procurement strategy by anticipating consumption peaks. We now manage energy proactively rather than reactively.