MCPInstall

energy-simulation

Building energy simulation and analysis for construction. Calculate heating/cooling loads, evaluate envelope performance, optimize HVAC sizing, and ensure energy code compliance.

Installs
115
Last 24h
—
Trending rank
—
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction --skill energy-simulation
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads energy-simulation automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

Energy Simulation

Overview

This skill implements building energy simulation and analysis. Calculate thermal loads, evaluate building envelope performance, and optimize systems for energy efficiency and code compliance.

Capabilities:

  • Heating/cooling load calculations
  • Envelope thermal analysis
  • HVAC system sizing
  • Energy code compliance
  • Renewable energy integration
  • Life cycle cost analysis

Quick Start

from dataclasses import dataclass, field
from typing import List, Dict, Optional, Tuple
from enum import Enum
import numpy as np

class WallType(Enum):
    CONCRETE = "concrete"
    BRICK = "brick"
    WOOD_FRAME = "wood_frame"
    STEEL_FRAME = "steel_frame"
    CURTAIN_WALL = "curtain_wall"

@dataclass
class BuildingEnvelope:
    wall_area_m2: float
    wall_u_value: float  # W/m²K
    roof_area_m2: float
    roof_u_value: float
    floor_area_m2: float
    floor_u_value: float
    window_area_m2: float
    window_u_value: float
    window_shgc: float  # Solar Heat Gain Coefficient

@dataclass
class ClimateData:
    location: str
    heating_degree_days: float  # HDD base 18°C
    cooling_degree_days: float  # CDD base 18°C
    design_temp_winter: float
    design_temp_summer: float

def calculate_heat_loss(envelope: BuildingEnvelope, climate: ClimateData,
                       indoor_temp: float = 21) -> float:
    """Calculate design heat loss (W)"""
    delta_t = indoor_temp - climate.design_temp_winter

    # Transmission losses
    wall_loss = envelope.wall_area_m2 * envelope.wall_u_value * delta_t
    roof_loss = envelope.roof_area_m2 * envelope.roof_u_value * delta_t
    floor_loss = envelope.floor_area_m2 * envelope.floor_u_value * delta_t * 0.5  # Ground factor
    window_loss = envelope.window_area_m2 * envelope.window_u_value * delta_t

    total_loss = wall_loss + roof_loss + floor_loss + window_loss

    # Add infiltration estimate (simplified)
    volume = envelope.floor_area_m2 * 3  # Assume 3m height
    infiltration = volume * 0.5 * 0.33 * delta_t  # 0.5 ACH, 0.33 Wh/m³K

    return total_loss + infiltration

# Example
envelope = BuildingEnvelope(
    wall_area_m2=500, wall_u_value=0.35,
    roof_area_m2=200, roof_u_value=0.25,
    floor_area_m2=200, floor_u_value=0.30,
    window_area_m2=100, window_u_value=1.4, window_shgc=0.4
)

climate = ClimateData(
    location="Moscow",
    heating_degree_days=5000,
    cooling_degree_days=300,
    design_temp_winter=-25,
    design_temp_summer=30
)

heat_loss = calculate_heat_loss(envelope, climate)
print(f"Design heat loss: {heat_loss/1000:.1f} kW")

Comprehensive Energy Analysis

Building Thermal Model

from dataclasses import dataclass, field
from typing import List, Dict, Optional, Tuple
from enum import Enum
import numpy as np
from datetime import datetime

@dataclass
class MaterialLayer:
    name: str
    thickness_m: float
    conductivity: float  # W/mK
    density: float  # kg/m³
    specific_heat: float  # J/kgK

    @property
    def resistance(self) -> float:
        """Thermal resistance R (m²K/W)"""
        return self.thickness_m / self.conductivity if self.conductivity > 0 else 0

@dataclass
class WallAssembly:
    name: str
    layers: List[MaterialLayer]
    inside_surface_resistance: float = 0.13  # m²K/W
    outside_surface_resistance: float = 0.04

    @property
    def total_resistance(self) -> float:
        return (self.inside_surface_resistance +
                sum(layer.resistance for layer in self.layers) +
                self.outside_surface_resistance)

    @property
    def u_value(self) -> float:
        return 1 / self.total_resistance if self.total_resistance > 0 else 0

@dataclass
class Window:
    name: str
    u_value: float
    shgc: float
    visible_transmittance: float = 0.6
    frame_fraction: float = 0.2

@dataclass
class Zone:
    zone_id: str
    name: str
    floor_area_m2: float
    volume_m3: float
    occupancy: int
    lighting_power_density: float  # W/m²
    equipment_power_density: float  # W/m²
    ventilation_rate: float  # L/s per person
    setpoint_heating: float = 21
    setpoint_cooling: float = 24

@dataclass
class BuildingGeometry:
    zones: List[Zone]
    walls: List[Dict]  # {zone, orientation, area, assembly}
    windows: List[Dict]  # {zone, orientation, area, window_type}
    roofs: List[Dict]  # {zone, area, assembly}
    floors: List[Dict]  # {zone, area, assembly, is_ground}

class ThermalCalculator:
    """Calculate building thermal loads"""

    # Standard climate data (simplified)
    CLIMATE_DB = {
        'moscow': {
            'hdd': 5000, 'cdd': 300,
            'design_winter': -25, 'design_summer': 30,
            'latitude': 55.75
        },
        'new_york': {
            'hdd': 2500, 'cdd': 800,
            'design_winter': -12, 'design_summer': 33,
            'latitude': 40.71
        },
        'dubai': {
            'hdd': 50, 'cdd': 3000,
            'design_winter': 15, 'design_summer': 45,
            'latitude': 25.20
        }
    }

    def __init__(self, building: BuildingGeometry, location: str):
        self.building = building
        self.location = location.lower()
        self.climate = self.CLIMATE_DB.get(self.location, self.CLIMATE_DB['moscow'])

    def calculate_design_heating_load(self) -> Dict:
        """Calculate design heating load for each zone"""
        delta_t = 21 - self.climate['design_winter']
        results = {}

        for zone in self.building.zones:
            # Transmission losses
            wall_loss = 0
            window_loss = 0
            roof_loss = 0
            floor_loss = 0

            for wall in self.building.walls:
                if wall['zone'] == zone.zone_id:
                    u_value = wall['assembly'].u_value
                    wall_loss += wall['area'] * u_value * delta_t

            for window in self.building.windows:
                if window['zone'] == zone.zone_id:
                    window_loss += window['area'] * window['window_type'].u_value * delta_t

            for roof in self.building.roofs:
                if roof['zone'] == zone.zone_id:
                    u_value = roof['assembly'].u_value
                    roof_loss += roof['area'] * u_value * delta_t

            for floor in self.building.floors:
                if floor['zone'] == zone.zone_id:
                    u_value = floor['assembly'].u_value
                    factor = 0.5 if floor.get('is_ground', False) else 1.0
                    floor_loss += floor['area'] * u_value * delta_t * factor

            # Infiltration
            infiltration_loss = zone.volume_m3 * 0.5 * 0.33 * delta_t

            # Ventilation (if mechanical)
            ventilation_loss = zone.occupancy * zone.ventilation_rate * 1.2 * delta_t

            total = wall_loss + window_loss + roof_loss + floor_loss + infiltration_loss + ventilation_loss

            results[zone.zone_id] = {
                'zone_name': zone.name,
                'wall_loss_w': wall_loss,
                'window_loss_w': window_loss,
                'roof_loss_w': roof_loss,
                'floor_loss_w': floor_loss,
                'infiltration_w': infiltration_loss,
                'ventilation_w': ventilation_loss,
                'total_w': total,
                'total_kw': total / 1000,
                'w_per_m2': total / zone.floor_area_m2
            }

        return results

    def calculate_design_cooling_load(self) -> Dict:
        """Calculate design cooling load for each zone"""
        delta_t = self.climate['design_summer'] - 24
        results = {}

        for zone in self.building.zones:
            # Transmission gains
            transmission_gain = 0
            for wall in self.building.walls:
                if wall['zone'] == zone.zone_id:
                    u_value = wall['assembly'].u_value
                    # Apply sol-air temperature correction for orientation
                    sol_air_delta = delta_t + self._get_sol_air_correction(wall['orientation'])
                    transmission_gain += wall['area'] * u_value * sol_air_delta

            # Window solar gains
            solar_gain = 0
            for window in self.building.windows:
                if window['zone'] == zone.zone_id:
                    shgc = window['window_type'].shgc
                    irradiance = self._get_solar_irradiance(window['orientation'])
                    solar_gain += window['area'] * shgc * irradiance

            # Window conduction
            window_conduction = 0
            for window in self.building.windows:
                if window['zone'] == zone.zone_id:
                    window_conduction += window['area'] * window['window_type'].u_value * delta_t

            # Internal gains
            lighting_gain = zone.floor_area_m2 * zone.lighting_power_density
            equipment_gain = zone.floor_area_m2 * zone.equipment_power_density
            people_gain = zone.occupancy * 75  # W per person sensible

            # Ventilation
            ventilation_gain = zone.occupancy * zone.ventilation_rate * 1.2 * delta_t

            total = (transmission_gain + solar_gain + window_conduction +
                    lighting_gain + equipment_gain + people_gain + ventilation_gain)

            results[zone.zone_id] = {
                'zone_name': zone.name,
                'transmission_gain_w': transmission_gain,
                'solar_gain_w': solar_gain,
                'window_conduction_w': window_conduction,
                'lighting_gain_w': lighting_gain,
                'equipment_gain_w': equipment_gain,
                'people_gain_w': people_gain,
                'ventilation_gain_w': ventilation_gain,
                'total_w': total,
                'total_kw': total / 1000,
                'w_per_m2': total / zone.floor_area_m2
            }

        return results

    def _get_sol_air_correction(self, orientation: str) -> float:
        """Get sol-air temperature correction by orientation"""
        corrections = {
            'north': 0, 'south': 8, 'east': 4, 'west': 6,
            'northeast': 2, 'northwest': 3, 'southeast': 6, 'southwest': 7
        }
        return corrections.get(orientation.lower(), 3)

    def _get_solar_irradiance(self, orientation: str) -> float:
        """Get design solar irradiance W/m² by orientation"""
        # Simplified peak values
        irradiance = {
            'north': 150, 'south': 450, 'east': 350, 'west': 350,
            'northeast': 200, 'northwest': 200, 'southeast': 400, 'southwest': 400
        }
        return irradiance.get(orientation.lower(), 300)

HVAC System Sizing

class HVACSizer:
    """Size HVAC systems based on loads"""

    def __init__(self, calculator: ThermalCalculator):
        self.calculator = calculator

    def size_heating_system(self, safety_factor: float = 1.15) -> Dict:
        """Size heating system"""
        heating_loads = self.calculator.calculate_design_heating_load()

        total_load = sum(z['total_kw'] for z in heating_loads.values())
        sized_capacity = total_load * safety_factor

        # Recommend system type
        if sized_capacity < 15:
            system_type = "Split system heat pump"
        elif sized_capacity < 50:
            system_type = "Packaged rooftop unit"
        elif sized_capacity < 200:
            system_type = "Central boiler with radiators"
        else:
            system_type = "Central plant with multiple boilers"

        return {
            'total_load_kw': total_load,
            'sized_capacity_kw': sized_capacity,
            'safety_factor': safety_factor,
            'recommended_system': system_type,
            'zone_loads': heating_loads
        }

    def size_cooling_system(self, safety_factor: float = 1.1) -> Dict:
        """Size cooling system"""
        cooling_loads = self.calculator.calculate_design_cooling_load()

        total_load = sum(z['total_kw'] for z in cooling_loads.values())
        sized_capacity = total_load * safety_factor

        # Convert to tons
        capacity_tons = sized_capacity / 3.517

        # Recommend system type
        if capacity_tons < 5:
            system_type = "Split system DX"
        elif capacity_tons < 20:
            system_type = "VRF system"
        elif capacity_tons < 100:
            system_type = "Chilled water with AHUs"
        else:
            system_type = "Central chiller plant"

        return {
            'total_load_kw': total_load,
            'total_load_tons': capacity_tons,
            'sized_capacity_kw': sized_capacity,
            'sized_capacity_tons': capacity_tons * safety_factor,
            'safety_factor': safety_factor,
            'recommended_system': system_type,
            'zone_loads': cooling_loads
        }

    def estimate_annual_energy(self) -> Dict:
        """Estimate annual energy consumption"""
        climate = self.calculator.climate

        heating_loads = self.calculator.calculate_design_heating_load()
        cooling_loads = self.calculator.calculate_design_cooling_load()

        total_heating_load = sum(z['total_kw'] for z in heating_loads.values())
        total_cooling_load = sum(z['total_kw'] for z in cooling_loads.values())

        # Simplified degree-day calculation
        # Heating energy = load * HDD * 24 / delta_t_design
        delta_t_heating = 21 - climate['design_winter']
        heating_kwh = total_heating_load * climate['hdd'] * 24 / delta_t_heating / 1000

        delta_t_cooling = climate['design_summer'] - 24
        cooling_kwh = total_cooling_load * climate['cdd'] * 24 / delta_t_cooling / 1000 if delta_t_cooling > 0 else 0

        # Apply efficiency factors
        heating_fuel_efficiency = 0.9  # Gas boiler
        cooling_cop = 3.5  # Chiller COP

        heating_consumption = heating_kwh / heating_fuel_efficiency
        cooling_consumption = cooling_kwh / cooling_cop

        return {
            'heating_load_kw': total_heating_load,
            'cooling_load_kw': total_cooling_load,
            'annual_heating_kwh': heating_kwh,
            'annual_cooling_kwh': cooling_kwh,
            'heating_fuel_kwh': heating_consumption,
            'cooling_electricity_kwh': cooling_consumption,
            'total_hvac_energy_kwh': heating_consumption + cooling_consumption
        }

Energy Code Compliance

@dataclass
class EnergyCodeRequirements:
    code_name: str
    climate_zone: str
    wall_u_max: float
    roof_u_max: float
    floor_u_max: float
    window_u_max: float
    window_shgc_max: float
    lighting_lpd_max: float  # W/m²

class ComplianceChecker:
    """Check energy code compliance"""

    CODES = {
        'ASHRAE_90.1_2019_4A': EnergyCodeRequirements(
            code_name="ASHRAE 90.1-2019",
            climate_zone="4A",
            wall_u_max=0.45,
            roof_u_max=0.27,
            floor_u_max=0.32,
            window_u_max=2.0,
            window_shgc_max=0.40,
            lighting_lpd_max=9.0
        ),
        'IECC_2021_5A': EnergyCodeRequirements(
            code_name="IECC 2021",
            climate_zone="5A",
            wall_u_max=0.35,
            roof_u_max=0.20,
            floor_u_max=0.30,
            window_u_max=1.7,
            window_shgc_max=0.40,
            lighting_lpd_max=8.5
        )
    }

    def __init__(self, code_key: str):
        self.requirements = self.CODES.get(code_key)
        if not self.requirements:
            raise ValueError(f"Unknown code: {code_key}")

    def check_envelope(self, building: BuildingGeometry) -> Dict:
        """Check envelope compliance"""
        results = {
            'code': self.requirements.code_name,
            'climate_zone': self.requirements.climate_zone,
            'compliant': True,
            'issues': []
        }

        # Check walls
        for wall in building.walls:
            u_value = wall['assembly'].u_value
            if u_value > self.requirements.wall_u_max:
                results['compliant'] = False
                results['issues'].append({
                    'element': f"Wall {wall['zone']} {wall['orientation']}",
                    'actual': u_value,
                    'required': self.requirements.wall_u_max,
                    'issue': 'Exceeds maximum U-value'
                })

        # Check windows
        for window in building.windows:
            u_value = window['window_type'].u_value
            shgc = window['window_type'].shgc

            if u_value > self.requirements.window_u_max:
                results['compliant'] = False
                results['issues'].append({
                    'element': f"Window {window['zone']} {window['orientation']}",
                    'actual': u_value,
                    'required': self.requirements.window_u_max,
                    'issue': 'Exceeds maximum U-value'
                })

            if shgc > self.requirements.window_shgc_max:
                results['compliant'] = False
                results['issues'].append({
                    'element': f"Window {window['zone']} {window['orientation']}",
                    'actual': shgc,
                    'required': self.requirements.window_shgc_max,
                    'issue': 'Exceeds maximum SHGC'
                })

        # Check roof
        for roof in building.roofs:
            u_value = roof['assembly'].u_value
            if u_value > self.requirements.roof_u_max:
                results['compliant'] = False
                results['issues'].append({
                    'element': f"Roof {roof['zone']}",
                    'actual': u_value,
                    'required': self.requirements.roof_u_max,
                    'issue': 'Exceeds maximum U-value'
                })

        return results

    def check_lighting(self, zones: List[Zone]) -> Dict:
        """Check lighting power density compliance"""
        results = {
            'compliant': True,
            'issues': []
        }

        for zone in zones:
            if zone.lighting_power_density > self.requirements.lighting_lpd_max:
                results['compliant'] = False
                results['issues'].append({
                    'zone': zone.name,
                    'actual_lpd': zone.lighting_power_density,
                    'required_max': self.requirements.lighting_lpd_max
                })

        return results

Quick Reference

ComponentGood U-ValueCode Maximum
Wall< 0.25 W/m²K0.35-0.45
Roof< 0.15 W/m²K0.20-0.27
Floor< 0.20 W/m²K0.25-0.32
Window< 1.2 W/m²K1.7-2.0

Resources

  • ASHRAE 90.1: Energy standard for buildings
  • IECC: International Energy Conservation Code
  • EnergyPlus: DOE building simulation
  • DDC Website: https://datadrivenconstruction.io

Next Steps

  • See co2-estimation for carbon analysis
  • See cost-prediction for energy cost modeling
  • See bim-validation-pipeline for model integration

FAQ

What does the energy-simulation skill do?

Building energy simulation and analysis for construction. Calculate heating/cooling loads, evaluate envelope performance, optimize HVAC sizing, and ensure energy code compliance.

How do I install energy-simulation?

Run npx skills add https://github.com/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction --skill energy-simulation in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/energy-simulation/ (personal) or .claude/skills/energy-simulation/ (project).

Which AI agents can use energy-simulation?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is energy-simulation safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains energy-simulation?

It is published by datadrivenconstruction in datadrivenconstruction/ddc_skills_for_ai_agents_in_construction. It has 115 recorded installs.

More from datadrivenconstruction/ddc_skills_for_ai_agents_in_construction

View all →

warranty-tracker

datadrivenconstruction/ddc_skills_for_ai_agents_in_construction

Track and manage construction warranties. Monitor expiration dates, claims, and manufacturer documentation.

AI & Agents996 installs

cad-to-data

datadrivenconstruction/ddc_skills_for_ai_agents_in_construction

Convert CAD/BIM files to structured data. Extract element data from Revit, IFC, DWG, DGN files.

AI & Agents333 installs

dwg-to-excel

datadrivenconstruction/ddc_skills_for_ai_agents_in_construction

Convert AutoCAD DWG files (1983-2026) to Excel databases using DwgExporter CLI. Extract layers, blocks, attributes, and geometry data without Autodesk licenses.

Data & Databases309 installs

drawing-analyzer

datadrivenconstruction/ddc_skills_for_ai_agents_in_construction

Open the skill to read its SKILL.md instructions.

AI & Agents259 installs

permit-tracking-automation

datadrivenconstruction/ddc_skills_for_ai_agents_in_construction

Automate construction permit tracking and management. Monitor application status, track renewal deadlines, manage document requirements, and integrate with municipal systems.

cost-estimation-resource

datadrivenconstruction/ddc_skills_for_ai_agents_in_construction

Calculate construction costs using resource-based method. Estimate project costs from work items, physical resource norms, and current prices.

AI & Agents193 installs

Popular ai & agents skills

View category →

google-agents-cli-publish

google/agents-cli

This skill should be used when the user wants to "publish an agent", "publish my ADK agent", "register an agent with Gemini", "publish to Gemini" (formerly Gemini Enterprise), or needs guidance on the agents-cli publish gemini command. Also use when the user wants to "manage agents in Agent Registry", "list/update/delete registered agents", or "register an MCP server". Covers ADK vs A2A registration modes, programmatic and interactive usage, flag reference, auto-detection from deployment metadata, Agent Registry fleet management, and troubleshooting. Part of the agents-cli skills suite. Do NOT use for deployment (use google-agents-cli-deploy).

AI & Agents513.9K installs

google-agents-cli-eval

google/agents-cli

This skill should be used when the user wants to "run an evaluation", "evaluate my agent", "evaluate my ADK agent", "write an eval dataset", "analyze eval failures", "compare eval results", "optimize agent", or needs guidance on the Gemini platform eval methodology and the Quality Flywheel. Covers eval metrics, dataset schema, LLM-as-judge scoring, and common failure causes. Applies to any agents-cli project, whatever framework the agent is written in. Do NOT use for agent API code patterns (ADK: use google-agents-cli-adk-code), deployment (use google-agents-cli-deploy), or project scaffolding (use google-agents-cli-scaffold).

AI & Agents513.5K installs

skill-creator

anthropics/skills

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.

OfficialAI & Agents416.4K installs

codex-pet

prime-skills/runcomfy-agent-skills

Codex Pet generator on RunComfy. Build a Codex-compatible Codex Pet spritesheet.webp + pet.json from a single reference image, drop it into ${CODEXHOME:-$HOME/.codex}/pets/ / and Codex picks it up as a custom Codex Pet next to the 8 built-ins. This skill produces the exact Codex Pet atlas Codex expects (1536x1872 PNG/WebP, 8 cols x 9 rows, 192x208 cells, 9 animation states — idle, running-right, running-left, waving, jumping, failed, waiting, running, review). Calls OpenAI GPT Image 2 edit ONCE via the local RunComfy CLI as runcomfy run openai/gpt-image-2/edit to produce a canonical Codex Pet pose, then assembles all 9 animation rows programmatically with ImageMagick micro-transforms — no Codex Pro, no $imagegen, no OPENAIAPIKEY required, only RUNCOMFYTOKEN. Triggers on "codex pet", "create codex pet", "make codex pet", "hatch codex pet", "/hatch image", "desktop pet codex", "codex pets", "spritesheet.webp", or any explicit ask to build a custom pet for OpenAI Codex.

AI & Agents402.1K installs

full-output-enforcement

leonxlnx/taste-skill

Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.

AI & Agents401K installs

executing-plans

obra/superpowers

Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available

AI & Agents245.7K installs