Overview

banner

A comprehensive framework for analyzing energy storage systems using OSeMOSYS with time-series clustering and temporal disaggregation methods.

Warning

This library is under active development. For latest release and updates, please check the GitHub repository.Some of the contents are not displayed as a requirement for the publication process of the paper. Complete content will be available after the publication.

Overview#

Storage-in-OSeMOSYS implements four different temporal representation methods for energy system optimization, each addressing the challenge of modeling energy storage systems with varying levels of computational complexity and accuracy:

  • 🔄 Model_Cluster: Time-series clustering approach for representative day selection

  • 📊 Model_Kotzur: Kotzur temporal disaggregation method with daily mapping

  • ⏰ Model_Welsch: Welsch temporal representation with fixed patterns

  • 🔗 Model_Niet: Niet temporal method with direct chronological linking

Key Features#

  • 🔄 Time-series clustering for representative day selection using K-means

  • 📊 Multiple temporal methods comparing different storage modeling approaches

  • ⚡ OSeMOSYS integration with OtoOle for seamless data processing

  • 📈 Automated visualization and comparative results analysis

  • 🛠️ Modular design with utilities for configuration management

  • 🔧 GLPK solver integration for linear programming optimization

  • 🐍 Python 3.13 support with modern pathlib usage

  • 📋 Conda environment management with comprehensive Makefile

Research Context#

This framework addresses the critical challenge of modeling energy storage systems in large-scale energy planning. Storage systems fundamentally differ from conventional generation and demand technologies because they move energy across time, creating complex temporal dependencies that are computationally expensive to model accurately.

The Storage Modeling Challenge#

Energy storage modeling is complex because:

  1. 🔁 Time-shifting behavior: Storage moves energy across time periods rather than just producing or consuming

  2. ⏳ State continuity: The state of charge depends on all previous time periods

  3. 🔗 Temporal coupling: Decisions in one hour affect options in future hours

  4. ⚡ Multi-scale dynamics: From minute-level frequency regulation to seasonal energy shifting

Our Solution#

Storage-in-OSeMOSYS provides a systematic comparison of four established temporal representation methods, allowing researchers to:

  • Evaluate trade-offs between computational efficiency and model accuracy

  • Compare results across different temporal abstraction approaches

  • Select appropriate methods based on specific research questions and constraints

  • Understand limitations of each approach for storage system analysis

About Storage Modelling#

Storage modeling is complex because storage systems behave differently than generation or demand. Storages move energy across time, not just produce or consume it. This means:

Warning

This library is under active development and some of the contents are not displayed as a requirement for the publication process of the paper. Complete content will be available after the publication.

🔁 1. Storage shifts energy in time

Unlike solar panels (generate) or homes (consume), storage stores energy when it’s abundant and releases it when needed. That means:

  • You can't look at one hour in isolation — what happened before matters.

  • The state of charge (SOC) today depends on all previous hours.

    Analogy: It's like a savings account — what you can spend today depends on how much you saved earlier.

⏳ 2. Time resolution matters

Should you model every single hour? Every day? A few representative days? Each choice impacts:

  • Accuracy (high resolution = realistic, but slow to compute)

  • Computational effort (simplified models lose detail)

    If you model storage with only a few sample days, you risk missing important patterns, like a long wind lull or weekend charging habits.

🔗 3. Storage needs continuity

Many simplification techniques (e.g. clustering days) break the continuity of storage:

  • The model might forget how full the battery was yesterday.

  • Artificial resets can distort when charging/discharging actually happens.

This makes it tricky to model seasonal storage or multi-day balancing.

⚡ 4. Storage interacts with everything Storage isn’t just an isolated battery:

  • It interacts with renewables (absorbing variability)

  • It supports grid reliability (fast response)

  • It responds to prices, constraints, and emissions goals

So modeling it well requires integrating with the rest of the energy system.

state of the art energy storage algorithms:#

Welsch (Fixed-Pattern) — “Photocopied Day”#

  • Analogy: This person writes down one typical day of feelings and then photocopies that same page for each day of the week. So Monday, Tuesday, etc., look exactly the same.

  • Key idea: Same daily pattern repeated; no inter-day variation.

  • Use case: Fast and simple but ignores variation between days.

Kotzur (Unverified Cluster) — “Alternating Diaries”#

  • Analogy: This person has two types of days — maybe “busy” and “lazy” days — and alternates between those two. They use pre-written diary templates for each. But they don’t bother connecting one day’s feelings to the next, so it can feel jumpy or inconsistent.

  • Key idea: Captures daily diversity using clusters, but no continuity.

  • Use case: Captures more variety than Welsch but lacks smooth transitions.

Novo (Hourly-Verified) — “Linked Journaling”#

  • Analogy: This person still uses simplified mood templates for different kinds of days, but they always make sure each day starts where the previous one left off. So it feels realistic and connected, even if it's not a full diary.

  • Key idea: Combines typical days with continuity between hours.

  • Use case: A practical middle ground between realism and simplicity.

Documentation Structure#

User Guide:

Technical Reference:

Examples & Tutorials:

Quick Start#

1. Environment Setup#

# Clone repository
git clone https://github.com/DeltaE/storage-in-OSeMOSYS.git
cd storage-in-OSeMOSYS

# Setup conda environment
make setup
conda activate storage_osemosys

# Verify installation
make test-env

2. Basic Usage#

# Run analysis with default configuration
python main.py

# Check results
ls Results/

3. Configuration#

Edit config/config.yaml to customize:

scenario_name: "k4h1WND"
days_in_year: 365
n_clusters: 4
hour_grouping: 1

Project Team#

Research Lead: Bruno Borba Development Support: Md Eliasinul Islam
Affiliation: Delta E+ Research Lab, Simon Fraser University
License: MIT License

Citation#

If you use Storage-in-OSeMOSYS in your research, please cite:

@software{storage_osemosys_2025,
  title={Storage-in-OSeMOSYS: A Framework for Temporal Representation Methods in Energy Storage Modeling},
  author={Islam, Md Eliasinul},
  organization={Delta E+ Research Lab, Simon Fraser University},
  year={2025},
  url={https://github.com/DeltaE/storage-in-OSeMOSYS}
}

Note

For technical support, please open an issue on GitHub. For research collaboration inquiries, contact the Delta E+ Research Lab.

Contents