> ## Documentation Index
> Fetch the complete documentation index at: https://tact.akleao.com/llms.txt
> Use this file to discover all available pages before exploring further.
# DNV Validation
> Validate your results against DNV RP-0661 industry standards
Validate your turbulence intensity adjustments against industry-standard DNV RP-0661 criteria.
## Quick Validation
```python theme={null}
from tact.validation import validate_dnv_rp0661
# After running adjustment
validation = validate_dnv_rp0661(
adjusted_data=results["adjusted_data"],
reference_col="ref_ti",
adjusted_col="adjTI_RSD_TI",
wind_speed_col="ref_ws",
bin_col="bins",
criteria_type="LV" # Load Verification
)
# Check results
print(f"MRBE: {validation['overall']['MRBE_%'].iloc[0]:.2f}%")
print(f"RRMSE: {validation['overall']['RRMSE_%'].iloc[0]:.2f}%")
print(f"Pass: {validation['overall']['pass'].iloc[0]}")
```
## Understanding DNV Criteria
**Load Verification Criteria**
Used for turbine load calculations and site assessments.
**Acceptance Criteria:**
* MRBE (Mean Relative Bias Error) ≤ 5%
* RRMSE (Relative Root Mean Square Error) ≤ 15%
```python theme={null}
validation = validate_dnv_rp0661(
adjusted_data=results["adjusted_data"],
reference_col="ref_ti",
adjusted_col="adjTI_RSD_TI",
wind_speed_col="ref_ws",
bin_col="bins",
criteria_type="LV"
)
```
**When to use:**
* Site assessment studies
* Turbulence load calculations
* Due diligence evaluations
**Power Curve Criteria**
Stricter requirements for power performance testing.
**Acceptance Criteria:**
* More stringent than LV
* Per IEC 61400-12-1 standards
```python theme={null}
validation = validate_dnv_rp0661(
adjusted_data=results["adjusted_data"],
reference_col="ref_ti",
adjusted_col="adjTI_RSD_TI",
wind_speed_col="ref_ws",
bin_col="bins",
criteria_type="PC"
)
```
**When to use:**
* Power performance testing
* Energy yield assessments
* Warranty verification
## Validation Workflow
First, complete your turbulence adjustment:
```python theme={null}
from tact import TACT
from tact.utils.load_data import load_data
from tact.utils.setup_processors import setup_processors
data = load_data("data.csv")
bp, tp, sp = setup_processors("config.json")
data = bp.process(tp.process(data))
tact = TACT()
results = tact.adjust(data, "ss-sf", {"split": True, "config_path": "config.json"})
```
Validate the adjusted results:
```python theme={null}
from tact.validation import validate_dnv_rp0661
validation = validate_dnv_rp0661(
adjusted_data=results["adjusted_data"],
reference_col="ref_ti",
adjusted_col="adjTI_RSD_TI",
wind_speed_col="ref_ws",
bin_col="bins",
criteria_type="LV"
)
```
Review overall and per-bin validation:
```python theme={null}
# Overall metrics
print("Overall Validation:")
print(validation['overall'])
# Per-bin details
print("\nPer-Bin Validation:")
print(validation['by_bin'])
```
Create professional validation visualizations:
```python theme={null}
from tact.visualization import plot_dnv_validation
plot_dnv_validation(
validation_results=validation,
adjusted_data=results["adjusted_data"],
reference_col="ref_ti",
adjusted_col="adjTI_RSD_TI",
unadjusted_col="rsd_ti",
wind_speed_col="ref_ws",
bin_col="bins",
method_name="SS-SF",
output_dir="output/plots"
)
```
## Validation Output
### Overall Metrics
The `validation['overall']` DataFrame contains:
| Metric | Description | Target (LV) |
| --------- | ------------------------------- | ----------- |
| `MRBE_%` | Mean Relative Bias Error | ≤ 5% |
| `RRMSE_%` | Relative Root Mean Square Error | ≤ 15% |
| `pass` | Whether criteria are met | True |
| `N` | Number of observations | - |
### Per-Bin Metrics
The `validation['by_bin']` DataFrame shows results for each wind speed bin:
```python theme={null}
# Example output
bin N MRBE_% RRMSE_% pass
0 4-5 120 3.2 12.1 True
1 5-6 245 4.1 13.5 True
2 6-7 312 2.8 11.2 True
...
```
## Complete Example
```python Full Pipeline theme={null}
from tact import TACT
from tact.utils.load_data import load_data
from tact.utils.setup_processors import setup_processors
from tact.validation import validate_dnv_rp0661
from tact.visualization import plot_dnv_validation
# 1. Load and process
data = load_data("data.csv")
bp, tp, sp = setup_processors("config.json")
data = bp.process(tp.process(data))
# 2. Run adjustment
tact = TACT()
results = tact.adjust(data, "ss-sf", {"split": True, "config_path": "config.json"})
# 3. Validate
validation = validate_dnv_rp0661(
adjusted_data=results["adjusted_data"],
reference_col="ref_ti",
adjusted_col="adjTI_RSD_TI",
wind_speed_col="ref_ws",
bin_col="bins",
criteria_type="LV"
)
# 4. Print results
overall = validation['overall']
print(f"\n{'='*50}")
print(f"DNV RP-0661 LV Validation Results")
print(f"{'='*50}")
print(f"MRBE: {overall['MRBE_%'].iloc[0]:6.2f}% (target: ≤5%)")
print(f"RRMSE: {overall['RRMSE_%'].iloc[0]:6.2f}% (target: ≤15%)")
print(f"Pass: {overall['pass'].iloc[0]}")
print(f"N: {overall['N'].iloc[0]}")
print(f"{'='*50}\n")
# 5. Generate plots
plot_dnv_validation(
validation_results=validation,
adjusted_data=results["adjusted_data"],
reference_col="ref_ti",
adjusted_col="adjTI_RSD_TI",
unadjusted_col="rsd_ti",
wind_speed_col="ref_ws",
bin_col="bins",
method_name="SS-SF",
output_dir="output/plots"
)
print("Validation plots saved to output/plots/")
# 6. Save validation results
validation['overall'].to_csv("validation_overall.csv", index=False)
validation['by_bin'].to_csv("validation_per_bin.csv", index=False)
```
```python Compare Methods theme={null}
# Validate multiple methods
methods = ["ss-sf", "ssws", "sswsstd"]
validation_results = {}
for method in methods:
# Run adjustment
results = tact.adjust(data, method, {"split": True, "config_path": "config.json"})
# Validate
validation = validate_dnv_rp0661(
adjusted_data=results["adjusted_data"],
reference_col="ref_ti",
adjusted_col="adjTI_RSD_TI",
wind_speed_col="ref_ws",
bin_col="bins",
criteria_type="LV"
)
validation_results[method] = validation
# Compare
import pandas as pd
comparison = pd.DataFrame({
method: {
'MRBE_%': val['overall']['MRBE_%'].iloc[0],
'RRMSE_%': val['overall']['RRMSE_%'].iloc[0],
'Pass': val['overall']['pass'].iloc[0]
}
for method, val in validation_results.items()
}).T
print(comparison.sort_values('MRBE_%'))
```
## Visualization
The `plot_dnv_validation` function generates:
* Before vs after adjustment comparison
* 1:1 reference line
* Regression fit line
* R² correlation coefficient
* MRBE and RRMSE per wind speed bin
* Acceptance criteria threshold lines
* Pass/fail indication by color
* Residual plots
* Bias distribution
* Error by wind speed bin
## Interpreting Results
**Both MRBE and RRMSE meet criteria**
```
MRBE: 3.21% (target: ≤5%) ✅
RRMSE: 12.45% (target: ≤15%) ✅
Pass: True
```
**What this means:**
* Adjustment is successful
* Results meet DNV standards
* Data suitable for its intended use
**Next steps:**
* Generate final report
* Save validated results
* Proceed with site assessment
**❌ MRBE exceeds 5% threshold**
```
MRBE: 8.5% (target: ≤5%) ❌
RRMSE: 13.2% (target: ≤15%) ✅
Pass: False
```
**What this means:**
* Systematic bias remains
* Adjustment method may not be optimal
**Troubleshooting:**
* Try different adjustment method (e.g., SS-SF instead of SSWS)
* Check data quality and synchronization
* Review per-bin results for specific problem ranges
* Verify configuration column mappings
**❌ RRMSE exceeds 15% threshold**
```
MRBE: 3.1% (target: ≤5%) ✅
RRMSE: 18.9% (target: ≤15%) ❌
Pass: False
```
**What this means:**
* High scatter/random error
* Data may have quality issues
**Troubleshooting:**
* Check data quality filtering
* Verify sensor calibration
* Review RSD-reference distance (should be less than 200m)
* Increase dataset size if possible
* Check for measurement synchronization issues
**❌ Both metrics fail**
```
MRBE: 12.3% (target: ≤5%) ❌
RRMSE: 25.7% (target: ≤15%) ❌
Pass: False
```
**What this means:**
* Significant data or methodology issues
* Adjustment not effective
**Troubleshooting:**
1. Verify data quality and completeness
2. Check time synchronization between sensors
3. Try all available methods (compare results)
4. Review raw data correlations
5. Consider data collection issues
## Next Steps
Create publication-ready plots
Fix validation failures