UniversitatdeLleida/omie-blockchain-study

---
license: apache-2.0
language:
- en
pretty_name: OMIE Intraday Auction Subsets for Blockchain Governance Study
size_categories:
- 10K<n<100K
task_categories:
- tabular-classification
- time-series-forecasting
tags:
- electricity-market
- energy
- blockchain
- market-design
- intraday
- omie
- tabular
dataset_info:
features:
- name: submission_ts
dtype: timestamp[us]
- name: auction_session
dtype: int64
- name: type
dtype: string
- name: period
dtype: int64
- name: order_id
dtype: int64
- name: num_blocks
dtype: int64
- name: total_amount_mw
dtype: float64
- name: min_price_eur_mwh
dtype: float64
- name: max_price_eur_mwh
dtype: float64
- name: blocks_json
dtype: string
- name: participant_id
dtype: string
configs:
- config_name: default
data_files:
- split: train
path: omie_intraday_2024_10_01_full.csv
- config_name: full
data_files:
- split: train
path: omie_intraday_2024_10_01_full.csv
- config_name: p10
data_files:
- split: train
path: omie_intraday_2024_10_01_010pct.csv
- config_name: p20
data_files:
- split: train
path: omie_intraday_2024_10_01_020pct.csv
- config_name: p40
data_files:
- split: train
path: omie_intraday_2024_10_01_040pct.csv
- config_name: p60
data_files:
- split: train
path: omie_intraday_2024_10_01_060pct.csv
- config_name: p80
data_files:
- split: train
path: omie_intraday_2024_10_01_080pct.csv
---

# OMIE Intraday Auction Subsets for Blockchain Governance Study

This repository contains cleaned and size-scaled tabular datasets derived from one representative day of the OMIE European intraday auction market. The data were prepared for experiments on blockchain-based verifiable governance of electricity flexibility and auction markets.

The dataset is designed for three uses:
1. replaying a realistic intraday auction day end-to-end,
2. benchmarking a blockchain-governed architecture against a centralized baseline, and
3. evaluating how performance changes as the number of participating market agents increases.

## Source context

OMIE operates the European intraday auction market for the Iberian electricity price zones. The market is cleared in discrete auction sessions, and bids are associated with bidding units, delivery periods, and block-based price–quantity structures. OMIE documentation describes the intraday market as a sequence of auction sessions with marginal-price clearing and period-specific bid submission. :contentReference[oaicite:0]{index=0}

The files in this repository are not raw OMIE exports. They are processed research datasets derived from OMIE intraday auction records for **2024-10-01**, using the final effective version of each order before gate closure.

## Dataset contents

The repository contains six CSV files:

- `omie_intraday_2024_10_01_010pct.csv`
- `omie_intraday_2024_10_01_020pct.csv`
- `omie_intraday_2024_10_01_040pct.csv`
- `omie_intraday_2024_10_01_060pct.csv`
- `omie_intraday_2024_10_01_080pct.csv`
- `omie_intraday_2024_10_01_full.csv`

The `010pct` to `080pct` files contain progressively larger subsets of participants, intended for scaling experiments. The `full` file contains the complete cleaned replay trace used as the reference case.

## Row semantics

Each row is a canonical bid object with the following fields:

- `submission_ts`: bid submission timestamp
- `auction_session`: intraday auction session identifier
- `type`: bid side (`buy` or `sell`)
- `period`: hourly delivery period
- `order_id`: row-unique identifier generated during preprocessing
- `num_blocks`: number of price–quantity blocks in the bid
- `total_amount_mw`: total quantity across all blocks
- `min_price_eur_mwh`: minimum block price in the bid
- `max_price_eur_mwh`: maximum block price in the bid
- `blocks_json`: ordered JSON list of blocks, each containing:
- `block`
- `price_eur_mwh`
- `amount_mw`
- `participant_id`: anonymized or normalized participant/bidding-unit identifier

## Preprocessing summary

The preprocessing pipeline applied the following steps:

1. parse OMIE intraday header and detail records,
2. retain only the latest version of each order,
3. build one canonical bid per `{session, participant, type, period, order}`,
4. renumber bid blocks so they always start at 1 and increase consecutively,
5. merge duplicated identifier columns into a single `auction_session` and `participant_id`,
6. regenerate `order_id` so each row has a unique identifier, and
7. sort the final dataset by `submission_ts`.

## Intended use

This dataset is intended for:
- replay-based validation of auditability and bid-integrity mechanisms,
- systems benchmarking,
- data engineering experiments on block-structured electricity bids,
- market-governance research.

It is not intended to support claims about price forecasting, dispatch optimization, or power-system operation beyond the specific replay and benchmarking scenarios of the associated study.

## Loading with `datasets`

You can load any file directly:

```python
from datasets import load_dataset

ds = load_dataset(
"UniversitatdeLleida/omie-blockchain-study",
data_files="omie_intraday_2024_10_01_full.csv"
)
print(ds["train"][0])
````

Or load one named config:

```python
from datasets import load_dataset

ds = load_dataset("UniversitatdeLleida/omie-blockchain-study", "p40")
print(ds["train"][0])
```

Hugging Face dataset repositories support CSV files directly and can auto-configure loading through repository structure or explicit `configs` metadata in the dataset card. ([Hugging Face][1])

## Citation

If you use this dataset, please cite the associated paper once available.

### Suggested citation

```bibtex
@dataset{universitatdelleida_omie_blockchain_study,
author = {Universitat de Lleida},
title = {OMIE Intraday Auction Subsets for Blockchain Governance Study},
year = {2026},
publisher = {Hugging Face},
url = {https://huggingface.co/datasets/UniversitatdeLleida/omie-blockchain-study}
}
```

## Acknowledgements

The original market context and raw intraday auction structure are based on OMIE market documentation and files. OMIE remains the authoritative source for market rules and raw operational data. ([Hugging Face][2])
This research was supported by the Research Council of Norway (Grant No. 350468). Additional support was provided through the Industrial Doctorates Plan of the Department of Research and Universities of the Government of Catalonia (Grant Nos. 2024 DI 00046 and 2024 DI 00061).