GUIDEBOOK · CEA-4 APP TRAINING

Energy Supply System Optimisation

Energy Supply System Optimisation

Supply system features evaluate energy system configurations against cost, emissions, and energy consumption. Building scale (decentralised) compares a fixed set of configurations per building; district scale (centralised) uses a multi-objective genetic algorithm.

Supply System Optimisation: Building-Scale

⚠️ Note: This feature is only available via Command Line Interface (CLI). It is not accessible through the CEA-4 App dashboard.

Overview

Evaluates decentralised energy supply systems for individual buildings. There is no search algorithm: for each building, CEA simulates a fixed set of heating and cooling supply configurations, compares them one-to-one, and flags the best one. Each building is treated independently, as if it were disconnected from any district network.

When to Use

• Designing building-level energy systems
• Comparing retrofit options for existing buildings
• Supporting building-level investment decisions
• When district systems are not feasible

What It Evaluates

Heating (13 configurations per building):

• Natural gas boiler
• Biogas boiler
• Fuel cell
• Ground-source heat pump with natural gas backup boiler, at ten heat pump / boiler capacity splits. Configurations whose heat pump exceeds what the building’s footprint area allows for boreholes are disqualified

Cooling (6 configurations per building):

• Direct expansion / mini-split (not fully built yet)
• Vapour compression chiller with cooling tower
• Flat-plate or evacuated-tube solar collectors with a single-effect absorption chiller, boiler and cooling tower
• Combinations that supply the sensible cooling load separately from the air-handling / recirculation load

How the best configuration is chosen: configurations are ranked separately by total annualised cost (TAC) and by GHG emissions. The configuration with the best combined rank is flagged as Best configuration. If the rank is tied, the configuration with the lowest compounded relative cost and emissions is chosen (a random pick only if those are identical too).

Prerequisites

• Energy Demand Part 2 - Total_demand.csv and building loads
• Solar collector potential (SC1 flat plate and SC2 evacuated tube) for every building with a cooling load - the cooling calculation reads both
• Building supply properties, zone geometry and a weather file (used for the ground-source heat pump)

Key Parameters

No other parameters: technology data come from the scenario’s supply-system databases.

How to Use

Note: This feature must be run from the command line. See CLI Documentation for detailed parameter configuration.

1. Complete prerequisites:

   • ✅ Energy Demand Part 2
   • ✅ Solar collector potential (SC1 and SC2)

2. Run via CLI:

   cea decentralized --scenario /path/to/scenario

3. Configure parameters (optional):

   • Only --multiprocessing and --number-of-cpus-to-keep-free
   • Enable multiprocessing (strongly recommended)

4. Processing time: Depends mainly on the number of buildings and available CPU cores. Each building is simulated independently, so multiprocessing helps directly.

Output Files

Note the folder spelling. Results are written to outputs/data/optimization/decentralized/ with a z, even though the feature is named with an “s” elsewhere in the app.

All files are in {scenario}/outputs/data/optimization/decentralized/, one set per building:

Understanding Results

• Compare configurations within one building using the TAC_USD and GHG_tonCO2 columns; the flagged best configuration is a compromise between the two rankings, not a Pareto front.
• Building-scale results are not Pareto optimisation results, so they are not used by Plot - Pareto Front.

Tips

• Enable multiprocessing: Buildings are simulated independently, so runs scale with available CPU cores
• Keep a few CPUs free so your machine stays responsive during long runs
• Compare all rows, not just the flagged best, if you care about a specific trade-off between cost and emissions

Troubleshooting

Issue: Run fails with a missing file (e.g. solar collector results)

• Solution: Run the solar collector potential for both SC1 and SC2 first, and Energy Demand Part 2

Issue: Optimisation runs very slowly

• Solution: Enable multiprocessing
• Solution: Test on a scenario with fewer buildings first

Issue: A heat pump configuration is missing or marked as not best

• Solution: Heat pump sizes are limited by the building’s footprint area (borehole space); larger heat pump shares are disqualified for small footprints

District Supply System Optimisation

Overview

Optimises centralised energy supply systems for entire districts. This feature finds optimal configurations for central plants, distribution networks, and building substations, considering both individual building requirements and district-level synergies.

When to Use

• Designing district heating/cooling systems
• Planning energy hubs or neighbourhood systems
• Comparing centralised vs decentralised approaches
• Optimising plant locations and capacities
• Supporting district-level energy master planning

What It Optimises

Decision Variables:

• Central plant technology types and capacities
• Network configuration (buildings to connect)
• Plant location
• Thermal storage size and operation
• Renewable energy integration (solar fields, geothermal, etc.)
• Peak vs base load equipment sizing
• Energy import/export strategies

Objectives: Any combination selected under Objective functions:

1. Minimise cost (cost)
2. Minimise GHG emissions (GHG_emissions)
3. Minimise system energy demand (system_energy_demand)
4. Minimise anthropogenic heat (anthropogenic_heat)

Additional Complexity vs Building-Scale

District optimisation must account for:

• Network costs and losses (pipe lengths, diameters, heat losses)
• Load diversity (coincidence of peaks across buildings)
• Economy of scale (larger central equipment often more efficient)
• Technology synergies (CHP, waste heat recovery, etc.)
• Spatial constraints (plant locations, network routing)

Prerequisites

• Energy Demand Part 2 - All building loads
• Streets network - For network routing (if thermal network)
• Thermal Network Part 1 (optional but recommended) - Network layout

Required Input Files

• Total demand summary
• Street network (for district heating/cooling)
• Cost, emission, and technology databases
• Renewable energy potential data (if available)

Key Parameters

How to Use

1. Complete prerequisites:

   • ✅ Energy Demand Part 2 for all buildings
   • ✅ Streets network
   • ✅ (Optional) Thermal Network Part 1

2. Configure optimisation:

   • Navigate to Energy Supply System Optimisation
   • Select District Supply System Optimisation
   • Select the network layout and network type (DH or DC)
   • Choose the heating, cooling and heat rejection components, energy sources and potentials
   • Choose the objective functions
   • Set GA population size and number of generations

3. Run optimisation:

   • Click Run
   • Processing time: 2-24+ hours depending on:
     • District size (number of buildings)
     • Complexity (network + technologies)
     • Population × generations
     • Available CPU cores
   • Consider running overnight or on high-performance computer

Output Files

All district results live under {scenario}/outputs/data/optimization/centralized/{DES_id}/, where {DES_id} identifies one near-Pareto-optimal district energy system (current_DES for the base case). Each candidate system gets its own folder, so the Pareto set is a set of folders rather than one file.

Note the folder spelling: optimization with a z.

Supply systems summary: .../{DES_id}/Supply_systems/Supply_systems_summary.csv

• Technology mix, equipment types and capacities for that district energy system
• Annual performance metrics

Individual supply system structure: .../{DES_id}/Supply_systems/{system_id}_supply_system_structure.csv

• One file per network (N0000) or standalone building (B0000)

Network layout: .../{DES_id}/networks/{network_id}_layout.geojson

• The optimised network geometry

Network performance: .../{DES_id}/Supply_system_operation_details/network_performance.csv

• Detailed thermal and hydraulic performance of the networks

Hourly operation: .../{DES_id}/Supply_system_operation_details/{system_id}_operation.csv

• Plant dispatch and storage operation, hour by hour

Annual breakdown: .../{DES_id}/Supply_system_operation_details/{system_id}_annual_breakdown.csv

• Annual energy demand, generation and losses per system

Understanding Results

District vs Decentralised Trade-offs

District systems advantages:

• Economy of scale (lower cost per kW)
• Load diversity (lower peak capacity needed)
• Enable waste heat recovery
• Centralised maintenance
• Higher efficiency central equipment

District systems disadvantages:

• Network capital cost and losses (10-20%)
• Requires suitable density
• Less flexibility for individual buildings

Decision threshold:

• High-density areas (>0.5 MW/hectare): District often optimal
• Low-density areas (<0.2 MW/hectare): Decentralised often better
• Medium density: Depends on specifics

Typical Results

Min cost solutions might include:

• Gas CHP for baseload
• Gas boiler for peaks
• Moderate thermal storage
• Connect ~60-80% of buildings

Min emission solutions might include:

• Heat pumps with renewable electricity
• Large thermal storage
• Maximum building connections
• Solar thermal fields

Tips

• Very computationally intensive: Plan for long run times
• Use high-performance computing if available
• Start with coarse optimisation (smaller population/generations) to understand solution space
• Consider running overnight or over weekend
• Review network assumptions: Network costs heavily influence results
• Compare to decentralised: Run building-scale optimisation too for comparison

Troubleshooting

Issue: Extremely long computation time (>24 hours)

• Solution: Reduce population size and generations significantly
• Solution: Simplify technology options
• Solution: Consider smaller sub-district
• Solution: Use high-performance computing cluster

Issue: Network costs dominate results

• Solution: Verify network layout is reasonable (use Thermal Network Part 1)
• Solution: Check pipe cost database values
• Solution: Consider if district system is appropriate for this density

Issue: No centralised solutions in Pareto frontier

• Solution: This may indicate decentralised is truly better for this case
• Solution: Check that district options are enabled
• Solution: Verify building density is sufficient

Comparing Building-Scale vs District-Scale

When to Use Each

Use Building-Scale Optimisation when:

• Low building density
• Buildings have very different profiles
• No existing district infrastructure
• Individual building owners want autonomy
• Testing building-level options quickly

Use District-Scale Optimisation when:

• High building density
• Master planning for new developments
• Existing or planned district infrastructure
• Centralised ownership/management
• Access to waste heat or renewable sources

Running Both for Comparison

Best practice workflow:

1. Run building-scale optimisation first (faster)
2. Review building-level optimal solutions
3. Run district-scale optimisation
4. Compare costs and emissions:
   • Sum of building-scale solutions = fully decentralised
   • District-scale results = centralised options
   • Choose based on cost/emission/practical considerations

Optimisation Best Practices

Parameter Selection (district-scale)

• Testing: 50 population × 25 generations (~1,000 evaluations)
• Production: 200 population × 100 generations (~20,000 evaluations)
• Publication: 300 population × 150 generations (~45,000 evaluations)

Validation

• Check solutions are technically feasible
• Verify capacities are reasonable
• Compare to rule-of-thumb sizing
• Test sensitivity to key assumptions

Interpretation

• No single “best” solution: Pareto frontier shows trade-offs
• Select solution based on priorities (cost, emissions, energy)
• Consider practical constraints (space, permits, expertise)
• Account for uncertainties (future energy prices, regulations)

Related Features

• Energy Demand Forecasting - Provides loads (prerequisite)
• Renewable Energy Assessment - Provides renewable potential
• Thermal Network Design - Network layout for district systems
• Life Cycle Analysis - Detailed cost and emission calculations
• Visualisation - Plot Pareto frontiers

← Back: Life Cycle Analysis | Back to Index | Next: Data Management →

Source: view raw on GitHub ↗