Building energy modeling accuracy isn’t just important—it’s critical. That’s why we put EP3, our SketchUp extension for EnergyPlus, through exhaustive validation testing against hundreds of verified models and thousands of data points.
Our Validation Methodology
EP3 is unique in the EnergyPlus ecosystem—in that it can import complete EnergyPlus IDF files, including complex HVAC systems. This capability has unlocked entirely new workflows:
- Reusable libraries: create libraries for internal loads, schedules, constructions, materials, and HVAC systems. Libraries can be integrated into a project at any time.
- AI integration (in development): Leverage AI’s training on open-source EnergyPlus data by importing AI-generated model components directly, and modifying components based on AI-suggestions.
- Cross-platform inspection: Examine and learn from models created in other software tools
- Accelerated testing: For our team, this means we can test specific components or model functionality without building complex systems from scratch
This import functionality has dramatically expanded our ability to test EP3 itself—we can validate against hundreds of existing models rather than creating test cases manually. But with such powerful capability comes the need for rigorous validation: we must prove that imported models maintain their accuracy.
Our standard validation workflow follows these steps:
- Source Validation: Start with validated IDF files from trusted sources (EnergyPlus example files, prototype building models, OpenStudio generated BESTest models)
- Baseline Simulation: Run the original file in EnergyPlus to establish reference results
- Import Process: Import the complete IDF file into EP3
- EP3 Simulation: Simulate the imported model
- Results Comparison: Compare outputs to verify accuracy
Tailored Validation Approaches
Standard Building Models: Focus on Energy Performance
For most building energy models, we focus our validation on monthly energy consumption data broken down by end-use. Why? Monthly energy validation is both practical and revealing: it’s straightforward to automate, and any issues with model translation are quickly identified when monthly values fall out of alignment. More importantly, when monthly energy use numbers match to two decimal places as a percentage (calculated as 1 – original/EP3), this indicates that the fundamental building physics and system operations are correctly preserved.
This approach is both practical and scientifically sound—monthly energy consumption represents the integration of all building systems over time, capturing the complex interactions between envelope performance, HVAC operations, and internal loads.
Comparison for EnergyPlus Example File: 5ZoneAirCooled
BESTEST Models: Comprehensive Variable-by-Variable Analysis
For our recent BESTEST validation, we took a different approach. BESTEST models are specifically designed as validation tools—they’re simpler than real buildings but include detailed output variables that measure specific physical phenomena that the test is designed to validate.
This made them perfect candidates for our most comprehensive validation to date: comparing every single detailed output variable between the original OpenStudio-generated models and their EP3 equivalents.
BESTEST Validation Results: The Numbers
We tested EP3 against the complete BESTEST suite: 97 validated building models with 3,860 individual output variables compared.
Understanding the R-squared Analysis:
Before diving into results, it’s important to understand what these R-squared values represent. Each comparison involves either 8,760 hourly values (full year) or 2,160 values (three months of hourly data). When two data series are identical or nearly identical, we assign an R² value of 1.0. Technically R² becomes undefined when there’s no variance in the data, though perfect matches deserve recognition as perfect correlations.
Achieving R² values above 0.95 across such large datasets means virtually identical performance. Even our few “outliers” below R² = 0.95 represent remarkably close correlation across thousands of hourly calculations.
Correlation Analysis (R-squared values):
Key Findings:
- 95.67% of variables achieved R² ≥ 0.99
- 98.89% achieved R² ≥ 0.96
These results demonstrate that EP3 doesn’t just import EnergyPlus models—it preserves their precision across thousands of individual calculations.
Why This Validation Matters
Complete IDF Import Capability
Unlike other EnergyPlus interfaces that require users to build HVAC systems from scratch, EP3’s import capabilites allow users to take advantage of HVAC systems created elsewhere – no matter where the file is generated. EP3’s IDF import feature maintains the integrity of the original IDF file, keeping their complexity intact. This validation proves that capability works reliably.
Trust in Complex Models
When you’re working with sophisticated building models—district energy systems, advanced HVAC controls, detailed daylighting—you need confidence that your interface isn’t introducing errors. Our validation methodology provides that confidence.
Continuous Improvement Process
Every time we add support for new EnergyPlus features or HVAC equipment types, we run this validation process on relevant model sets. This ensures that EP3’s accuracy improves alongside its capabilities.
The Bottom Line
Energy modeling is only as good as its accuracy. Through systematic validation against hundreds of verified models, we’ve demonstrated that EP3 maintains the precision of EnergyPlus while providing the intuitive interface of SketchUp.
Whether you’re importing a simple prototype building or a complex district energy model, you can trust that your EP3 simulation will deliver results that match the original EnergyPlus calculation. This level of accuracy extends beyond just the import process: it reflects EP3’s reliability as a complete modeling platform, from model creation and modification through final simulation results.
Leave a Reply