Scenarios

Scenarios enable what-if analysis and sensitivity studies by allowing you to create alternative versions of your network without duplicating the entire model. Using a sparse override architecture, scenarios only store attribute values that differ from the base network, making them highly efficient.

Overview

Every Convexity model has a base network that defines the default network configuration. Scenarios let you create variations by overriding specific attribute values while inheriting everything else from the base network.

Key benefits:

• Memory efficient: Only differences are stored, not entire networks
• Easy comparison: All scenarios share the same network topology
• Flexible analysis: Test multiple what-if cases without model duplication
• Stochastic optimization: Assign probabilities to scenarios for uncertainty modeling

Sparse Override Architecture

The sparse override system works as follows:

1. Base network: All attributes with scenario_id = NULL represent the default network
2. Scenario overrides: Attributes with a specific scenario_id override the base value for that scenario only
3. Inheritance: If a scenario doesn't override an attribute, it uses the base network value
4. Efficiency: A scenario with 10 overrides on a 1000-component network only stores 10 values. This makes the setup highly efficient (do not hestitate to add many scenarios)

Example:

Base network: Generator "Coal Plant" has marginal_cost = 50 EUR/MWh
Scenario "High Fuel Costs": Override marginal_cost = 80 EUR/MWh
Scenario "Low Demand": No override → uses base value 50 EUR/MWh

Scenarios Table Properties

Property	Data Type	Default	Required	Description
id	INTEGER	(auto)	-	Primary key (auto-increment)
name	TEXT	-	Yes	Scenario name (unique)
description	TEXT	-	No	Scenario description
probability	REAL	NULL	No	Probability for stochastic optimization (0-1, NULL for deterministic what-if scenarios)
is_system_scenario	BOOLEAN	FALSE	No	True for system-reserved scenarios (cannot be deleted)
system_purpose	TEXT	NULL	No	Purpose of system scenario (e.g., "actual" for measured values)
created_at	DATETIME	(auto)	No	Timestamp when scenario was created

Scenario Types

1. Deterministic What-If Scenarios

Standard scenarios for sensitivity analysis and alternative cases.

Characteristics:

• probability = NULL
• Used for comparing discrete alternatives
• Each scenario solved independently

Examples:

• "High Fuel Costs" - Test impact of 50% higher gas prices
• "Delayed Transmission" - Model postponed infrastructure investments
• "Accelerated Renewables" - Explore faster clean energy deployment

2. Stochastic Scenarios

Scenarios with probabilities for uncertainty modeling.

Characteristics:

• probability set between 0 and 1
• Probabilities across scenarios should sum to 1
• Solved together in stochastic optimization

Examples:

• "Low Demand" (probability = 0.3) - Mild winter scenario
• "Medium Demand" (probability = 0.5) - Normal winter
• "High Demand" (probability = 0.2) - Severe winter

3. System Scenarios

Special reserved scenarios that cannot be deleted.

"Actual" Scenario:

• Automatically created for every network
• Purpose: Store measured/actual values for validation
• is_system_scenario = TRUE, system_purpose = 'actual'
• Not included in optimization solves
• Used for comparing model results against real-world data

Database Storage

Scenarios are stored efficiently in the database:

scenarios table:
  id=1, name="Base Network", (implicit, no row)
  id=2, name="High Fuel Costs", probability=NULL
  id=3, name="Low Demand", probability=0.3

component_attributes table:
  component_id=5, attribute="marginal_cost", scenario_id=NULL, value=50  (base)
  component_id=5, attribute="marginal_cost", scenario_id=2, value=80     (override)
  component_id=7, attribute="p_set", scenario_id=3, value=[...]           (override)

Learn More

• PyConvexity Examples - Programmatic scenario creation
• Component Attributes - How attributes work with scenarios