xpertsystems/enr005-sample

---
license: cc-by-nc-4.0
task_categories:
- tabular-classification
- tabular-regression
- time-series-forecasting
tags:
- synthetic-data
- smart-grid
- ami
- advanced-metering-infrastructure
- smart-meter
- der
- distributed-energy-resources
- solar-pv
- bess
- battery-energy-storage
- ev-charging
- wind-micro
- fuel-cell
- chp
- demand-response
- openadr
- ieee-2030-5
- ieee-1547
- ieee-519
- ansi-c84-1
- ansi-c12-19
- iec-61968
- cim
- two-way-power-flow
- reverse-power-flow
- voltage-regulation
- volt-var
- volt-watt
- cvr
- conservation-voltage-reduction
- microgrid
- islanding
- grid-edge
- grid-edge-analytics
- power-quality
- thd
- harmonics
- power-factor
- net-metering
- prosumer
- hosting-capacity
- short-circuit-ratio
- recloser
- sectionalizer
- statcom
- protection-relay
- fault-detection-isolation
- flisr
- dsm
- demand-side-management
- tou
- cpp
- rtp
- dlc
- vpp
- virtual-power-plant
pretty_name: ENR005 — Synthetic Smart Grid Dataset (Sample)
size_categories:
- 10K<n<100K
configs:
- config_name: ami_telemetry
data_files: enr005_ami_telemetry.parquet
- config_name: der_timeseries
data_files: enr005_der_timeseries.parquet
- config_name: power_flow
data_files: enr005_power_flow.parquet
- config_name: grid_edge
data_files: enr005_grid_edge.parquet
- config_name: dr_events
data_files: enr005_dr_events.parquet
- config_name: meter_registry
data_files: enr005_meter_registry.parquet
- config_name: der_registry
data_files: enr005_der_registry.parquet
---

# ENR005 — Synthetic Smart Grid Dataset (Sample Preview)

**XpertSystems.ai | Synthetic Data Factory | Energy & Climate Vertical**

A **six-table smart grid AMI/DER dataset** spanning the distribution-edge
data stack: smart meter (AMI) telemetry, DER asset registry and timeseries
(Solar PV, Battery ESS, EV chargers, Wind Micro, Fuel Cell, CHP),
OpenADR-compatible demand response event logs, feeder-level two-way
power flow, and grid edge device analytics (smart inverters, Volt-VAR,
protection relays, microgrid islanding, power quality disturbances).
Calibrated benchmark-first against **IEEE 2030.5**, **IEEE 1547**,
**IEEE 519**, **ANSI C84.1**, **ANSI C12.19**, **OpenADR 2.0b**,
**IEC 61968 CIM**, and **EPRI demand response benchmarks**.

This is the **sample preview** — 30 meters × 5 feeders × 25 DER assets ×
2 days × 15-min cadence (~14K records across 7 tables, ~2.6 MB). The
full product covers 5,000 meters × 20 feeders × 1,500 DER assets × full
annual cycle (~1.4B records) with N-1 grid stress scenarios, evening
EV charging surge, high-solar reverse flow events, and microgrid
islanding scenarios pre-built.

---

## Dataset summary

| Table | Rows (sample) | What it contains |
|---|---:|---|
| `ami_telemetry` | 5,760 | Per-meter 15-min telemetry: interval kWh, active/reactive/apparent power, voltage, current, THD, power factor, net metering, outage/tamper flags, data quality |
| `der_timeseries` | 4,800 | Per-DER 15-min operational data: output kW, capacity, SOC (BESS/EV), charge/discharge rates, solar irradiance, inverter status, availability, grid-forming flag |
| `dr_events` | ~174 | OpenADR event participation: program type, signal type, baseline, target/actual reduction, performance %, rebound, opt-out, comfort score, incentive paid |
| `power_flow` | 960 | Feeder-level two-way flow: direction (FORWARD/REVERSE/BIDIRECTIONAL), gross/net/reverse load, voltage regulation, voltage rise from DER, feeder loading %, reactive kVAR, distribution losses |
| `grid_edge` | 2,400 | Edge device ops: Smart Inverter / STATCOM / Volt-VAR / Recloser / Sectionalizer / Cap Bank / DSTATCOM, microgrid mode, islanding detection, protection relay events, PQ disturbance type, hosting capacity, SCR |
| `meter_registry` | 30 | Static meter metadata: feeder assignment, customer class, DER ownership flags, vintage year, tariff code, service voltage, contract demand |
| `der_registry` | 25 | Static DER metadata: type, capacity, install year, manufacturer, interconnection voltage, IEEE 1547 compliance, smart inverter capability |

All seven tables are provided in both **CSV** and **Parquet**. They join on
`meter_id`, `feeder_id`, `der_id`, and `timestamp_utc`.

---

## Calibration sources

All ten validation metrics target named industry sources, not generator
self-metrics:

- **IEEE 2030.5** — smart energy profile (DER/EV/meter communications,
microgrid normal operations)
- **IEEE 1547** — interconnection of distributed energy resources (service
voltage band, IEEE 1547-2018 anti-islanding)
- **IEEE 519-2014** — harmonic control in electrical power systems (THD
limits)
- **ANSI C84.1** — Electric Power Systems and Equipment — Voltage Ratings
(Range A 114-126V residential, ±5% from 120V nominal)
- **ANSI C12.19** — utility industry end device data tables (AMI metering
data quality KPIs)
- **OpenADR 2.0b** — demand response signaling protocol (event lifecycle,
enrollment, performance reporting)
- **IEC 61968 CIM** — common information model for distribution management
- **EPRI Demand Response Program Performance Benchmarks** — typical realized
reduction ratio 60-90% of targeted reduction
- **Physics** — DER nameplate capacity envelope, BESS SOC bounds [0, 100]%,
mass balance (net = gross − DER generation)

---

## Validation scorecard (seed = 42)

10/10 PASS · **Grade A+ (100%)** across all six canonical seeds (42, 7, 123, 2024, 99, 1).

| # | Metric | Observed | Target | Tol | Type | Source |
|---|---|---:|---:|---:|---|---|
| 1 | `residential_voltage_ansi_C84_range_A_rate` | 1.000 | 0.98 | ±0.03 | FLOOR | ANSI C84.1 |
| 2 | `power_factor_floor_0_80_rate` | 1.000 | 0.99 | ±0.01 | FLOOR | IEEE 519 / tariff |
| 3 | `thd_ieee_519_8pct_compliance_rate` | 0.918 | 0.90 | ±0.05 | FLOOR | IEEE 519-2014 |
| 4 | `bess_soc_in_valid_range_rate` | 1.000 | 0.99 | ±0.01 | FLOOR | Physics |
| 5 | `der_within_capacity_envelope_rate` | 1.000 | 0.99 | ±0.01 | FLOOR | Physics |
| 6 | `feeder_voltage_pu_ansi_band_rate` | 1.000 | 0.95 | ±0.05 | FLOOR | ANSI C84.1 / IEEE 1547 |
| 7 | `power_flow_net_equals_gross_minus_der_rate` | 1.000 | 0.99 | ±0.01 | FLOOR | Mass balance |
| 8 | `microgrid_grid_connected_share` | 0.987 | 0.98 | ±0.03 | two-sided | IEEE 2030.5 / IEEE 1547-2018 |
| 9 | `dr_performance_mean_pct` | 73.96 | 75.0 | ±10.0 | two-sided | EPRI DR Performance |
| 10 | `ami_data_quality_valid_rate` | 0.968 | 0.95 | ±0.05 | FLOOR | ANSI C12.19 / AMI KPI |

---

## Schema highlights

### `ami_telemetry` (5,760 rows × 16 cols)
`meter_id`, `feeder_id`, `customer_class` (Residential / Commercial_SMB /
Commercial_Large / Industrial), `timestamp_utc`, `interval_kwh`,
`net_metering_kwh`, `active_power_kw`, `reactive_power_kvar`,
`apparent_power_kva`, `power_factor`, `voltage_v`, `current_a`, `thd_pct`,
`outage_flag`, `tamper_flag`, `data_quality_flag` (VALID / ESTIMATED /
MISSING / SUBSTITUTED).

### `der_timeseries` (4,800 rows × 11 cols)
`der_id`, `der_type` (Solar_PV / Battery_ESS / EV_Charger / Wind_Micro /
Fuel_Cell / CHP), `timestamp_utc`, `der_output_kw` (signed: + export, −
import for EVs), `der_capacity_kw`, `state_of_charge_pct` (BESS/EV only),
`charge_rate_kw`, `discharge_rate_kw`, `solar_irradiance_w_m2`,
`inverter_status` (ONLINE / OFFLINE / CURTAILED / FAULT / MPPT),
`der_availability_flag`, `grid_forming_flag`.

### `dr_events` (~174 rows × 17 cols)
`dr_event_id`, `meter_id`, `dr_program_type` (TOU / CPP / RTP / DLC /
AutoDR / VPP_Dispatch), `dr_signal_type` (SHED / SHIFT / MODULATE /
PRICE_SIGNAL / EMERGENCY), `dr_event_start_utc`, `dr_event_end_utc`,
`duration_hours`, `baseline_kw`, `dr_target_reduction_kw`,
`actual_reduction_kw`, `dr_performance_pct`, `non_event_baseline_kw`,
`rebound_kw`, `dr_incentive_paid_usd`, `customer_opt_out_flag`,
`comfort_score`, `customer_class`.

### `power_flow` (960 rows × 14 cols)
`feeder_id`, `timestamp_utc`, `power_flow_direction` (FORWARD / REVERSE /
BIDIRECTIONAL), `gross_load_kw`, `der_generation_kw`, `net_load_kw`,
`reverse_flow_kw`, `feeder_loading_pct`, `voltage_regulation_pu`,
`voltage_rise_pu`, `reactive_kvar`, `conservation_voltage_reduction_factor`,
`substation_load_kw`, `distribution_losses_pct`, `thermal_rating_kw`.

### `grid_edge` (2,400 rows × 13 cols)
`feeder_id`, `device_id`, `edge_device_type` (Smart_Inverter / STATCOM /
Volt_VAR / Recloser / Sectionalizer / Cap_Bank / DSTATCOM),
`timestamp_utc`, `volt_var_mode` (VOLT_VAR / VOLT_WATT / CONSTANT_PF /
FIXED_Q), `reactive_power_support_kvar`, `feeder_automation_status`
(MANUAL / AUTOMATIC / FASR / FLISR), `islanding_detection_flag`,
`microgrid_mode` (GRID_CONNECTED / ISLANDED / TRANSITIONING),
`protection_relay_event` (PICKUP / TRIP / RECLOSE / LOCKOUT / NORMAL),
`pq_disturbance_type` (SAG / SWELL / INTERRUPTION / HARMONIC / FLICKER /
NORMAL), `hosting_capacity_kw`, `short_circuit_ratio`.

---

## Suggested use cases

- **AMI load disaggregation** — train NILM (non-intrusive load monitoring)
models to separate base load, HVAC, EV charging, and solar from
`interval_kwh` and `net_metering_kwh`
- **DER forecasting** — short-horizon prediction of `der_output_kw` for
solar PV from irradiance + cloud features; wind micro turbine output
from synthetic wind speeds
- **BESS dispatch optimization** — learn charge/discharge policies from
`bess_soc_pct`, time-of-day, and feeder loading signals
- **EV charging behavior modeling** — predict charging session start time,
duration, and energy from customer class, charge rate, and historical
patterns
- **Demand response performance prediction** — regressor for
`dr_performance_pct` and `actual_reduction_kw` from baseline, signal
type, comfort score, and customer class
- **VPP / aggregator dispatch ML** — train portfolio-level DR fleet
optimizers across heterogeneous customer classes
- **Two-way power flow classification** — predict
`power_flow_direction` from DER generation, gross load, and
feeder topology features
- **Voltage rise modeling on high-PV feeders** — ML to predict
`voltage_rise_pu` from `der_generation_kw`, R/X impedances, and
feeder loading
- **Volt-VAR control optimization** — train smart inverter Volt-VAR
curves from PCC voltage and feeder loading conditions
- **Microgrid islanding detection** — anomaly/classification on
`islanding_detection_flag`, `protection_relay_event`,
`pq_disturbance_type` joint signals
- **Hosting capacity prediction** — regressor for `hosting_capacity_kw`
given short-circuit ratio, R/X, and existing DER penetration
- **Power quality classification** — multi-class on `pq_disturbance_type`
(SAG / SWELL / INTERRUPTION / HARMONIC / FLICKER) from voltage and
THD time-series features
- **AMI data quality / outage detection** — classifier for
`outage_flag`, `tamper_flag`, and `data_quality_flag` transitions
- **Net metering / prosumer billing analytics** — model bill components
from `net_metering_kwh`, TOU tier (joinable from registry), and
retail rate schedules

---

## Loading examples

```python
from datasets import load_dataset

ami = load_dataset("xpertsystems/enr005-sample", "ami_telemetry", split="train")
der = load_dataset("xpertsystems/enr005-sample", "der_timeseries", split="train")
flow = load_dataset("xpertsystems/enr005-sample", "power_flow", split="train")
print(ami.shape, der.shape, flow.shape)
```

```python
import pandas as pd
from huggingface_hub import hf_hub_download

# Load AMI + DR events; join on meter_id
ami = pd.read_parquet(hf_hub_download(
"xpertsystems/enr005-sample", "enr005_ami_telemetry.parquet",
repo_type="dataset",
))
dr = pd.read_parquet(hf_hub_download(
"xpertsystems/enr005-sample", "enr005_dr_events.parquet",
repo_type="dataset",
))

# Per-class DR performance
print(dr.groupby("customer_class")["dr_performance_pct"].agg(["mean", "std", "count"]))
```

```python
# Compute net feeder demand profile
import pandas as pd
from huggingface_hub import hf_hub_download

flow = pd.read_parquet(hf_hub_download(
"xpertsystems/enr005-sample", "enr005_power_flow.parquet",
repo_type="dataset",
))
flow["t"] = pd.to_datetime(flow["timestamp_utc"])
flow["hour"] = flow["t"].dt.hour
hourly = flow.groupby("hour").agg(
gross=("gross_load_kw", "mean"),
der=("der_generation_kw", "mean"),
net=("net_load_kw", "mean"),
).round(2)
print(hourly) # see the duck-curve shape
```

```python
# Check power flow direction transitions per feeder
import pandas as pd
from huggingface_hub import hf_hub_download

flow = pd.read_parquet(hf_hub_download(
"xpertsystems/enr005-sample", "enr005_power_flow.parquet",
repo_type="dataset",
))
for fid, g in flow.groupby("feeder_id"):
g = g.sort_values("timestamp_utc")
transitions = (g["power_flow_direction"] != g["power_flow_direction"].shift()).sum() - 1
print(f"Feeder {fid[:8]}: {transitions} direction transitions, "
f"reverse rate = {(g['reverse_flow_kw'] > 1).mean():.3f}")
```

---

## Limitations and honest disclosures

This sample is calibrated for **structural fidelity, not bit-exact reproduction
of any specific utility's AMI archive.** Specifically:

- **AMI/DER/power_flow tables auto-truncate to 2 days** when `hours < 8760`
in the generator's `main()` (lines 791-794, 808-810, 829-830); `grid_edge`
truncates to 1 day (line 839). At this sample's hours=48 setting,
AMI/DER/flow all cover the full 48h while grid_edge covers 24h. **Time
ranges across tables don't align 1:1** — be aware when joining grid_edge
with other tables.
- **Power factor distribution clips at 0.80** — generator uses
Beta(8,2) clipped to [0.80, 1.00], so values below 0.80 are pushed up
to exactly 0.80 (heavy mass at the lower bound). This matches utility
residential tariff minimums, but real Commercial_Large and Industrial
loads can dip to 0.70-0.85 without correction. Don't use this dataset
to study deep low-PF events.
- **THD distribution has fat upper tails.** The generator's
`thd_base + exponential(1.5)` shifts ~60% of rows above the IEEE 519 5%
residential limit, with ~8% above the 8% short-duration bound. **For PQ
research, treat these as bus-level coupled THD, not point-of-common-
coupling compliance measurements.**
- **Distribution losses average 0.46%** (Beta(2,20)×5), well below the
US industry typical 4-7% range. The generator's loss distribution is
intentionally narrow for compact statistical learning; **do not use
`distribution_losses_pct` for absolute loss benchmarking.**
- **Reverse power flow at sample scale fires rarely.** At ~6 meters per
feeder × default 35% solar penetration, DER generation seldom exceeds
gross load. The full product activates reverse flow via the
`High_Solar_Day` scenario config (`solar_penetration=0.65`).
Wrapper validates the STRUCTURAL identity (`reverse_flow_kw =
max(0, -net_load_kw)` and `net_load_kw = gross_load_kw -
der_generation_kw`) rather than an aggregate rate.
- **`hosting_capacity_kw` and `short_circuit_ratio` are sampled per-row**
(not per-device-or-feeder properties). Use as advisory features in ML
pipelines, not as static topology attributes.
- **`interconnect_status` (CONNECTED / CURTAILED / TRIPPED / ISLANDED)
on grid_edge is sampled independently of generation flow** at the
feeder level. Treat as device-state telemetry, not a causal label
for a flow event.
- **`solar_irradiance_w_m2` is a simplified mid-US-latitude clear-sky
model** (latitude hardcoded to 37.5°). All meters share the same
solar noon and seasonal cycle. **Do not use for geographically-
varying PV studies** — use the full product or join with ENR-002 for
per-site irradiance.
- **DR event participation uses an O(N) pandas lookup per participant**
in the generator's `generate_dr_events` (line 426-427). At sample
scale (30 meters × 12 events × ~50% enrollment ≈ 180 participants)
it's fast (< 0.1s); the full product (5000 meters × 48 events × ~50%
enrollment ≈ 120,000 participants) runs slower. Not a data-quality
issue, just a perf note.
- **Solar PV `der_output_kw` is non-negative; EV charger `der_output_kw`
is non-positive (load convention).** When aggregating "DER export"
use `der_output_kw.where(der_output_kw > 0).sum()` to avoid loads
canceling generation.
- **`dr_events.start_utc` references the full 12-month timestamp range**
via the generator's `months=12` hardcode (line 401), but **the
timestamp draw is bounded** by `len(timestamps) - 16`, so at sample
scale all events fall within the AMI window. This is by design but
is something to be aware of when adapting the generator.

The full ENR005 product addresses these by per-site latitude-aware
irradiance, broader PF / THD distributions, reverse-flow and EV-surge
scenarios pre-built, multi-month time spans, and full meter/DER fleet
scale — contact us for the licensed commercial release.

---

## Companion datasets in the Energy & Climate vertical

- **ENR-001** — Synthetic Power Grid Operations Dataset (transmission-side
bus telemetry, line flows, generation dispatch, frequency, contingency)
- **ENR-002** — Synthetic Renewable Energy Generation Dataset (utility-
scale solar/wind/hybrid SCADA, weather, forecast, PCC, BESS)
- **ENR-003** — Synthetic Electricity Demand & Load Forecasting Dataset
(zone-level demand, multi-horizon forecasts, peak events, EV/DER, TOU,
LMP)
- **ENR-004** — Synthetic Upstream Oil & Gas Production Dataset (well-
level production, decline curves, PVT, commodity prices, Subpart W
methane)
- **ENR-005** — Synthetic Smart Grid Dataset (you are here) — **the
distribution-edge complement to ENR-001's transmission focus**: meter-
level AMI, behind-the-meter DER, OpenADR demand response, feeder
power flow, and grid edge analytics.

Use **ENR-001 + ENR-005** together for full transmission + distribution
grid ML workflows; combine with **ENR-002 + ENR-003** to add renewables
and demand forecasting in the same modeling stack.

For subsurface companion data (seismic, well logs, reservoir simulation,
geological formations), see the **OIL series** (OIL-001 through OIL-004)
in our [Oil & Gas vertical](https://huggingface.co/xpertsystems).

For the broader catalog:
- [Materials & Energy](https://huggingface.co/xpertsystems) — MAT-001
- [Insurance & Risk](https://huggingface.co/xpertsystems) — 10 SKUs
- [Cybersecurity](https://huggingface.co/xpertsystems) — 11 SKUs

---

## Citation

```bibtex
@dataset{xpertsystems_enr005_sample_2026,
author = {XpertSystems.ai},
title = {ENR005 Synthetic Smart Grid Dataset (Sample Preview)},
year = 2026,
publisher = {Hugging Face},
url = {https://huggingface.co/datasets/xpertsystems/enr005-sample}
}
```

---

## Contact

- **Web:** https://xpertsystems.ai
- **Email:** [e-mail]
- **Full product catalog:** Cybersecurity, Insurance & Risk, Materials & Energy,
Oil & Gas, Energy & Climate, and more

**Sample License:** CC-BY-NC-4.0 (Creative Commons Attribution-NonCommercial 4.0)
**Full product License:** Commercial — please contact for pricing.