Skip to content

Interactive Dashboard

Overview

The Google-Go Interactive Dashboard is a web-based visualization tool for exploring the project results. It provides an intuitive interface to analyze multiple scenarios, compare results across different policy configurations, and explore both aggregated metrics and detailed hourly timeseries data.

The dashboard handles large-scale energy system analysis results with millions of data points, providing fast, interactive visualizations with intelligent caching and data management.

Installation and Setup

Requirements

Python version: 3.8+

Core dependencies:

pip install dash dash-bootstrap-components plotly pandas numpy pyarrow

Installation Steps

1. Navigate to the repository:

cd /path/to/google-go

2. Install dependencies:

pip install dash dash-bootstrap-components plotly pandas numpy pyarrow

3. Verify data structure:

ls -lh results/
# Should show:
# `results.csv`
# `results_frontier.csv`
# `results_time_series.parquet` (or `.csv`)
# `colors.csv`

4. Launch the dashboard:

cd dashboard
python app.py

5. Access in browser:

http://localhost:8050

Production Deployment

For production environments, use Gunicorn:

pip install gunicorn
gunicorn app:server -b 0.0.0.0:8050 --workers 4 --timeout 300

Configuration options:

  • --workers 4: Use 4 worker processes
  • --timeout 300: 5-minute timeout for large data queries
  • -b 0.0.0.0:8050: Bind to all network interfaces

Data Structure

The dashboard uses a consolidated data structure where all scenarios are stored in a single file.

Required Files

results/
├── colors.csv                    # Color mappings for carriers
├── results.csv                   # Consolidated results for all scenarios
├── results-*.csv                 # Country-specific results (optional)
├── results_frontier.csv          # Frontier analysis data
├── results_time_series.parquet   # Timeseries data (parquet format, preferred)
└── results_time_series.csv       # Timeseries data (CSV fallback)

results.csv Format

The consolidated results.csv file has a multi-level header structure:

Column headers:

Level 0: year          | 2025              | 2025              | 2030 ...
Level 1: scenario      | baseline          | energy-match-25   | baseline ...
Level 2: scope         | system            | system            | system ...

Row index:

Level 0: Results       | (a) Energy mix    | (a) Energy mix    | (b) Capacity ...
Level 1: y_label       | Net generation... | Net generation... | Capacity ...
Level 2: carrier       | solar             | onwind            | solar ...

Example data snippet:

year,,,2025,2025,2025,2030,2030,2030
scenario,,,baseline,energy-match-25,hourly-match-25-90,baseline,energy-match-25,hourly-match-25-90
scope,,,system,system,system,system,system,system
Results,y_label,carrier,,,,,,,
(a) Energy mix,Net generation (TWh),solar,245.3,267.8,289.1,312.5,345.7,378.2
(a) Energy mix,Net generation (TWh),onwind,456.2,478.9,501.3,534.8,567.1,599.4
(a) Energy mix,Net generation (TWh),CCGT,151.1,119.3,48.4,104.5,31.7,14.9

Key characteristics:

  • Dimensions: ~145 rows × ~400 columns
  • Years: 2025, 2030, 2035, 2040
  • Metrics: 13+ categories (energy mix, capacity, costs, emissions, etc.)
  • Carriers: 15-20 energy carriers (solar, wind, gas, hydrogen, etc.)

colors.csv Format

Maps energy carriers to consistent colors across all visualizations:

Results,carrier,color
(a) Energy mix,solar,#f9d002
(a) Energy mix,onwind,#235ebc
(a) Energy mix,offwind-ac,#6895dd
(a) Energy mix,offwind-dc,#74c6f2
(a) Energy mix,CCGT,#a85522
(a) Energy mix,hydrogen,#ea048a
(a) Energy mix,battery_discharge,#b474de

User Guide

The dashboard provides five main analysis tabs for exploring energy system results.

Tab 1: Single Scenario Analysis

Purpose: Deep dive into a single scenario's results to understand detailed energy system characteristics.

What You Can Do:

  • Visualize individual scenario results in detail
  • View multiple plot types: bar charts, stacked bars, area charts, pie charts
  • Track year-over-year evolution
  • Filter by specific energy carriers

Controls:

  • Year: Select 2025, 2030, 2035, 2040, or "All"
  • Scenario: Choose from available policy scenarios
  • Metric: Select analysis category (energy mix, capacity, costs, etc.)
  • Plot Type: Choose visualization style

Available Plot Types:

1. Bar Chart: Vertical bars showing carrier breakdown for a specific year

  • Best for: Quick overview of a single year
  • Shows: Individual carrier contributions

2. Stacked Bar (All Years): Multi-year comparison in one chart

  • Best for: Seeing how the mix evolves over time
  • Shows: All years side-by-side with stacked carriers

3. Stacked Area: Cumulative area chart over years

  • Best for: Visualizing cumulative contributions
  • Shows: Smooth evolution of carrier mix

4. Pie Chart: Proportional breakdown

  • Best for: Understanding relative proportions
  • Shows: Percentage contribution of each carrier

5. Year Comparison: Side-by-side bars for all years

  • Best for: Comparing specific carriers across years
  • Shows: How each carrier changes year-to-year

6. Year on Year Evolution: Line chart showing trends

  • Best for: Identifying growth/decline trends
  • Shows: Line trajectories for each carrier

Example Workflow:

Question: How does the energy mix change from 2025 to 2040 under hourly-match-50-90?

  1. Navigate to Single Scenario Analysis tab
  2. Select Scenario: hourly-match-50-90
  3. Select Metric: (a) Energy mix
  4. Select Plot Type: Year on Year Evolution
  5. View the line chart showing how each carrier evolves

Tips:

  • Use "Year on Year Evolution" to identify trends
  • Use "Stacked Bar" to see proportional changes
  • Click legend items to show/hide specific carriers
  • Hover over bars for exact values

Tab 2: Cross-Scenario Comparison

Purpose: Compare up to 4 scenarios side-by-side to understand how different policies affect the energy system.

What You Can Do:

  • Compare multiple scenarios simultaneously
  • View side-by-side bar charts with scenario grouping
  • Analyze year-over-year evolution across scenarios
  • Review summary statistics and differences

Controls:

  • Year: Select specific year or "All"
  • Metric: Choose analysis category
  • Plot Type: Choose comparison style
  • Group By: Group by year or scenario
  • Scenario 1-4: Select up to 4 scenarios to compare

Available Plot Types:

1. Side-by-Side: Grouped bar chart comparing scenarios

  • Best for: Direct scenario comparisons
  • Shows: Scenarios grouped for each carrier

2. Stacked Bar (All Years): Multi-year stacked comparison

  • Best for: Seeing evolution across scenarios and years
  • Shows: Stacked carriers for each scenario/year combination

3. Stacked Bar + Total Line: Stacked bars with total overlay

  • Best for: Comparing totals while seeing composition
  • Shows: Stacked bars with line showing total

4. Year Comparison: Compare specific years across scenarios

  • Best for: Year-specific analysis
  • Shows: How scenarios differ in specific years

5. Year on Year Evolution: Evolution lines for multiple scenarios

  • Best for: Trend comparison
  • Shows: Multiple scenario trajectories on one plot

Example Workflow:

Question: How much more solar capacity is needed at 95% vs 90% hourly matching?

  1. Navigate to Cross-Scenario Comparison tab
  2. Select Year: 2040
  3. Select Metric: (c) Capacity mix
  4. Select Scenario 1: hourly-match-50-90
  5. Select Scenario 2: hourly-match-50-95
  6. Select Plot Type: Side-by-Side
  7. Compare the solar bars between scenarios

Tips:

  • Compare baseline against policy scenarios to measure impact
  • Use "Group By: Year" to see temporal evolution
  • Use "Group By: Scenario" to see scenario differences at each year
  • Summary statistics show total changes and percentages

Tab 3: Energy Procurement Frontier

Purpose: Visualize the energy procurement frontier showing the trade-off between energy matching and hourly matching.

What You Can Do:

  • View frontier curves (clean hourly matching vs. energy matching)
  • Identify trade-offs between matching types
  • Compare frontiers across scenarios and countries
  • Analyze regional sensitivities to policy constraints

Controls:

  • Year: Select specific year or "All"
  • Scenarios: Select up to 5 scenarios (or "All")
  • Countries: Select up to 5 countries (or "All")

What is the Energy Procurement Frontier?

The Energy Procurement Frontier includes the maximum achievable hourly matching level (percentage of hours where clean/renewable generation meets or exceeds demand) for a given energy matching level (percentage of annual clean energy produced over the demand). The region within the Energy Procurement Frontier includes the shares of hourly and energy matching being possible.

How to Read Frontier Curves:

  • X-axis: Clean energy matching percentage (0-100%)
  • Y-axis: Clean hourly matching percentage (0-100%)
  • Procurement frontier: Boundary of the dead zone, representing the maximum achievable hourly matching level for a given energy matching level.
  • Tipping points: Where curve suddenly steepens

Example Workflow:

Question: To which extent policies on renewable targets and CO2 price can affect the dead zone?

  1. Navigate to Dead Zone Analysis tab
  2. Select Year: All
  3. Select Scenarios: baseline-rps, baseline-co2-price25, baseline (for comparison)
  4. Select Countries: EU
  5. Observe the dead zone extension

Tips: - Compare dead zones between multple scenarios - Use "All" years to see how dead zones shift over time - Country-level analysis reveals regional sensitivities - Look for "tipping points" where curves suddenly steepen

Tab 4: Timeseries Exploration

Purpose: Explore hourly operational data (8,760 hours/year) to understand detailed energy system operations.

What You Can Do:

  • Explore hourly timeseries data for the full year
  • Select specific time ranges (week, month, season)
  • Overlay multiple scenarios for comparison
  • Filter by specific energy carriers

Controls:

  • Year: Select year to analyze
  • Scenarios: Multi-select scenarios to overlay
  • Timeseries Type: Choose data type (Electricity Balance, Storage Levels, etc.)
  • Country: Select geographic scope
  • Carrier: Optional filter for specific carriers
  • Time Range: Select temporal zoom level

Time Range Options:

  • Full Year: All 8,760 hours
  • Week 1: First week of January
  • Winter Week: Representative winter week (Jan 15-21)
  • Summer Week: Representative summer week (July 1-7)
  • Specific Month: 2013-01, 2013-02, etc.

Example Workflow:

Question: How does battery storage cycle during a typical winter week under hourly-match-50-90?

  1. Navigate to Timeseries Exploration tab
  2. Select Year: 2035
  3. Select Scenarios: hourly-match-50-90
  4. Select Timeseries Type: Storage Levels
  5. Select Country: EU
  6. Select Carrier: battery
  7. Select Time Range: Winter Week
  8. Observe daily charging/discharging cycles

Tips: - Start with weekly views to identify patterns - Expand to full year to see seasonal trends - Compare multiple scenarios to see policy impacts on operations - Look for correlation between demand, solar/wind, and storage - Winter weeks show system stress, summer weeks show surplus

Tab 5: Key Insights

Purpose: View comprehensive statistical analysis and strategic recommendations derived from 3,080 scenario runs.

What You'll Find:

Executive Summary: 12 critical findings including:

  • Universal 10% tipping point analysis
  • LDES (Long-Duration Energy Storage) criticality
  • EU frontier anomaly
  • Regional extremes (25x variation)
  • Non-linear cost patterns

Key Sections:

1. Critical Tipping Points

  • The universal 10% barrier
  • Cost acceleration patterns
  • Technology-specific thresholds

2. LDES Criticality

  • Statistical significance (p<0.001)
  • Cost impact (+4.24% without LDES)
  • System feasibility analysis

3. Frontier Curve Analysis

  • Cost elasticity rankings
  • Maximum acceleration points
  • Scenario-specific patterns

4. Robustness Paradox

  • Stricter policies → more predictable outcomes
  • Variability analysis across scenarios
  • Planning uncertainty implications

5. Regional Analysis

  • Country-specific sensitivities
  • 25x variation (Luxembourg +25%, Czechia -7.3%)
  • Recommended focus countries

6. Strategic Recommendations

  • Data-driven procurement strategies
  • Optimal matching percentages (40-50%)
  • Technology investment priorities
  • Policy effectiveness rankings

Key Findings Highlight:

  1. 10% Tipping Point: All scenarios show dramatic cost acceleration at 10% hourly matching threshold
  2. LDES Non-Negotiable: +4.24% cost without it, statistically significant (p<0.001)
  3. EU Frontier Anomaly: EU scenarios accelerate at 2% vs 97-117% for national scenarios
  4. Increasing Returns: 46% cheaper per percentage point at 25-50% matching vs 0-25%
  5. Regional Extremes: 25x variation - Luxembourg +25%, Czechia -7.3%

Tips:

  • Read this tab first to understand key findings
  • Reference specific insights when exploring other tabs
  • Use findings to guide scenario comparisons
  • Share strategic recommendations with stakeholders

Scenario Configuration

Format: [type]-[additionality]-[matching]

Examples:

  • energy-match-25 = Energy matching with 25% additionality
  • hourly-match-50-99 = Hourly matching with 50% additionality at 99% matching level
  • hourly-match-noadd-50-99 = Hourly matching with NO additionality, 50% target, 99% matching

Key Parameters:

Additionality Level:

  • 25 = 25% of clean energy must be new/additional
  • 50 = 50% of clean energy must be new/additional
  • noadd = No additionality requirement (can use existing clean energy)

Matching Level:

  • 90 = 90% hourly matching requirement
  • 95 = 95% hourly matching requirement
  • 98 = 98% hourly matching requirement
  • 99 = 99% hourly matching requirement

Customization

Changing Analyzed Scenarios

The dashboard automatically detects all scenarios in results.csv. To add or remove scenarios:

1. Modify results.csv:

  • Add new columns with format: year | scenario | scope
  • Ensure new scenarios follow existing data structure

2. Restart dashboard:

bash python app.py

3. Verify: New scenarios appear in dropdown menus

No code changes required - the dashboard dynamically loads all available scenarios.

Adding Custom Metrics

To add new metrics:

1. Add data to results.csv:

csv (n) New Metric,Y-axis Label,carrier,value1,value2,value3,...

2. Add color mappings (optional) in colors.csv:

csv (n) New Metric,carrier1,#color1 (n) New Metric,carrier2,#color2

3. Restart dashboard - new metric appears in dropdown

Customizing Colors

Edit results/colors.csv:

Results,carrier,color
(a) Energy mix,solar,#FFD700  # Change to gold
(a) Energy mix,onwind,#00BFFF  # Change to deep sky blue

Color format: Hex codes (#RRGGBB)

Apply changes: Restart dashboard

Troubleshooting

Common Issues

Dashboard won't start

Error: ModuleNotFoundError: No module named 'dash'

Solution:

pip install dash dash-bootstrap-components plotly pandas numpy pyarrow

No data showing

Error: Blank plots or "No data available"

Possible causes:

  • results.csv not found or incorrectly formatted
  • Column headers don't match expected structure
  • Missing year/scenario/metric in data

Solution:

  1. Verify file exists: ls -lh results/results.csv
  2. Check first few lines: head -5 results/results.csv
  3. Verify multi-level headers are correct
  4. Check dashboard logs for specific errors

Timeseries loading very slow

Issue: Timeseries queries take 30+ seconds

Solution:

1. Convert CSV to Parquet:

import pandas as pd
df = pd.read_csv('results/results_time_series.csv')
df.to_parquet('results/results_time_series.parquet', compression='snappy')

2. Verify parquet file is detected:

ls -lh results/results_time_series.parquet

3. Restart dashboard

Colors not matching

Issue: Carriers show default colors instead of custom colors

Possible causes:

  • colors.csv not found
  • Carrier names don't match between results.csv and colors.csv
  • Incorrect CSV format

Solution:

  1. Verify file: cat results/colors.csv | head -10
  2. Check carrier name spelling (case-sensitive)
  3. Ensure format: Results,carrier,color

Port already in use

Error: OSError: [Errno 98] Address already in use

Solution:

# Option 1: Kill existing process
lsof -ti:8050 | xargs kill -9

# Option 2: Use different port
# Edit app.py, change:
app.run_server(debug=True, host='0.0.0.0', port=8051)

Memory errors

Error: MemoryError or dashboard crashes

Cause: Insufficient RAM for large datasets

Solutions:

  1. Use Parquet format for timeseries (better compression)
  2. Use time range filtering instead of full year
  3. Deploy on machine with more RAM (4GB+ recommended)

Best Practices

Data Exploration Workflow

  1. Start with Key Insights tab to understand high-level findings
  2. Use Single Scenario Analysis to explore individual scenarios
  3. Use Cross-Scenario Comparison to measure policy impacts
  4. Use Dead Zone Analysis to identify cost tipping points
  5. Use Timeseries Exploration to understand operational details

Performance Tips

  1. Use Parquet format for timeseries data (10-20x faster than CSV)
  2. Start with smaller time ranges (week/month) before loading full year
  3. Use caching - repeated queries are instant
  4. Limit scenarios in multi-scenario comparisons (max 4 recommended)
  5. Close unused browser tabs to free memory

Presentation Tips

  1. Use consistent plot types across related analyses
  2. Include baseline in comparisons for reference
  3. Use "Year on Year Evolution" to show trends clearly
  4. Export plots via browser's screenshot or Plotly's built-in save feature
  5. Cite specific insights from Key Insights tab in presentations

Browser Compatibility

Tested and working on:

  • Chrome 90+
  • Firefox 85+
  • Safari 14+
  • Edge 90+

Support

For questions, issues, or feature requests:

  1. Check this guide's Troubleshooting section
  2. Review dashboard logs for error messages
  3. Verify data structure matches expected format
  4. Contact the development team with specific questions

Dashboard Status: Production-ready, actively maintained