Skip to contents

aemo

CRAN status CRAN downloads Total Downloads R-CMD-check Lifecycle: experimental License: MIT

An R package for accessing public market data from the Australian Energy Market Operator. Wholesale electricity prices, regional demand, dispatch, interconnector flows, rooftop PV, generator bids, predispatch forecasts, FCAS markets, and gas markets across the National Electricity Market.

What is AEMO?

The Australian Energy Market Operator is the independent system operator and market operator for the National Electricity Market (NEM), which serves the eastern and southern states of Australia, and the reformed Wholesale Electricity Market (WEM) in Western Australia (SCED and essential system services markets, go-live 1 October 2023). AEMO also operates the Victorian Declared Transmission System and the Short Term Trading Market for gas (Adelaide, Brisbane, Sydney hubs), the Wallumbilla Gas Supply Hub, and the East Coast Gas System functions. The Australian Energy Regulator (AER) is the economic regulator; the Australian Energy Market Commission (AEMC) is the rule maker.

The NEM is one of the longest interconnected transmission systems in the world and generates around 200 TWh of electricity a year (2024 calendar year was 217 TWh including rooftop PV) across five regions (Queensland, New South Wales plus the ACT, Victoria, South Australia, and Tasmania). Annual NEM wholesale turnover typically falls in the AUD 15 to 25 billion range, with 2023-24 at AUD 18.6 billion (AER Wholesale Electricity Market Performance Report, December 2024), plus tens to a few hundred million in frequency control ancillary services (FCAS) depending on volatility. Since 1 October 2021 the NEM has run on a 5-minute dispatch and settlement cycle (previously 5-minute dispatch with 30-minute settlement); every five minutes AEMO publishes prices, demand, generator output, interconnector flows, bids, and forecasts.

All of this is free and public. AEMO publishes the raw dispatch files to NEMweb (http://nemweb.com.au) within minutes of each dispatch interval, and retains historical archives in the Market Management System Data Model (MMSDM) going back to the start of the NEM in December 1998. Flagship planning publications (the Integrated System Plan (ISP), Electricity Statement of Opportunities, and Gas Statement of Opportunities) are released on aemo.com.au.

Why does this package exist?

NEMweb is generous but unfriendly. There are over 100 report folders. Files are zipped CSVs with a proprietary row-dispatch format (C, comments, I, column headers, D, data rows, F, footers) that many R-native CSV readers can’t parse. A single dispatch file might contain four different tables (prices, demand, interconnector flows, unit output) all interleaved in one CSV. Monthly archives for some tables are multi-gigabyte. Schema versions change between MMSDM releases (the v4-to-v5 transition in April 2021, coincident with the 5-minute settlement program, added columns to DISPATCHLOAD). AEMO is decommissioning the NEMweb HTTP endpoint on 7 April 2026 and migrating the base URL on 30 April 2026.

# Without this package
listing <- rvest::read_html("http://nemweb.com.au/Reports/Current/DispatchIS_Reports/")
hrefs   <- rvest::html_attr(rvest::html_elements(listing, "a"), "href")
zip     <- paste0("http://nemweb.com.au/Reports/Current/DispatchIS_Reports/",
                  tail(grep("DISPATCHIS", hrefs, value = TRUE), 1))
path    <- tempfile(fileext = ".zip")
download.file(zip, path)
tmp     <- tempfile(); utils::unzip(path, exdir = tmp)
csv     <- list.files(tmp, pattern = "\\.CSV$", full.names = TRUE)[1]
# ... read the file line by line, split by "I,"/"D,"/"F," row type,
#     match I headers to D rows, build separate data frames per table,
#     validate that schema matches the MMSDM version, ...

# With this package
library(aemo)
aemo_price("NSW1", "2024-06-01", "2024-06-01 01:00:00")

NEMweb directory listings are parsed into tidy data frames. The C,/I,/D,/F, row-dispatch format is handled per-file so schema drift is transparent. Base URL is configurable via options(aemo.base_url = ...) to handle the April 2026 migration. Rate-limiting and caching are in-box.

Package history and name reclaim

The aemo package name was previously held by Imanuel Costigan’s package, which was on CRAN from June 2014 to April 2020 (v0.1.0 to v0.3.0) and archived on 29 December 2021. That package implemented three functions covering prices and demand only. This rewrite is an independent implementation that shares only the package name and expands scope to the full NEMweb and MMSDM data surface.

Because the old registry entry still exists, the standard CRAN status and download badges pull from Costigan’s historical data (CRAN 0.3.0, ~5 downloads per month, 40K lifetime) rather than from this rewrite. Those three badges are omitted above until v0.4.0 of the new package lands on CRAN and the registry snaps to it. The version starts at 0.4.0 so it sits strictly above the last archived version (0.3.0), per CRAN policy on reclaimed names.

How does aemo compare to NEMOSIS and nemwebR?

The authoritative tool for NEM data analysis is UNSW-CEEM’s NEMOSIS, written in Python. It is well-maintained, feature-rich, and used across the Australian energy research community. If you’re already in Python, use it.

In R, nemwebR is a single-author GitHub-only utility with no release. This package (aemo) fills the equivalent niche on CRAN.

Package Language Status Coverage
NEMOSIS Python (CEEM) Active, released on PyPI Most MMSDM dynamic and static tables, plus forecast tables
NEMSEER Python (CEEM) Active Forecast tables (P5MIN, PREDISPATCH, PASA)
nempy Python (CEEM) Active NEM dispatch simulator, not a data wrapper
nemwebR R GitHub, no release Prices, dispatch, bids, DUDETAILSUMMARY
aemo (this package) R CRAN Current NEMweb + reference data plus size-guarded bids

Installation 

install.packages("aemo")

# Or install the development version from GitHub
# install.packages("devtools")
devtools::install_github("charlescoverdale/aemo")

Quick start 

library(aemo)

# Reference data (no network required)
aemo_regions()
aemo_interconnectors()
aemo_price_caps()       # MPC / MPF / CPT by financial year

# 5-minute wholesale prices for NSW, last hour
p <- aemo_price("NSW1", Sys.time() - 3600, Sys.time())
head(p)
# Key columns: settlementdate (POSIXct, AEST), regionid, rrp (AUD/MWh)

# Regional demand (operational)
d <- aemo_demand("VIC1", Sys.time() - 3600, Sys.time())

# FCAS prices across all 10 services
f <- aemo_fcas("NSW1", Sys.time() - 3600, Sys.time())

# Binding constraints with shadow prices (price-spike forensics)
c <- aemo_constraints(Sys.time() - 3600, Sys.time())
head(c[order(-c$marginalvalue), c("settlementdate", "constraintid",
                                   "marginalvalue", "rhs")])

Timestamps: AEST not AEDT

All timestamps are AEST (UTC+10, fixed year-round) matching AEMO’s market clock (NER clause 2.2.6). Using "Australia/Sydney" would silently shift every summer interval by one hour. The tzone attribute on every settlementdate column is "Australia/Brisbane".

Timestamps are period-ending: a row stamped 14:05:00 covers the 5-minute interval ending at 14:05.

attr(p$settlementdate, "tzone")  # "Australia/Brisbane"

Intervention runs

DISPATCHPRICE contains two run types in every file: the market pricing run (INTERVENTION = 0, used in settlement) and the physical/intervention run (INTERVENTION = 1, issued during market directions or suspension). All functions default to intervention = FALSE, returning market pricing rows only.

During the June 2022 NEM suspension AEMO issued directions across all regions for several days. Querying that period with the default settings returns only the market pricing run, not the direction-influenced run:

# Default: market pricing run only (correct for settlement analysis)
p_market <- aemo_price("NSW1", "2022-06-13", "2022-06-14")

# Both runs (use for direction-cost analysis)
p_both <- aemo_price("NSW1", "2022-06-13", "2022-06-14",
                     intervention = TRUE)
table(p_both$intervention)  # "0": market run rows, "1": intervention rows

Worked example: price spike forensics 

library(aemo)

# Step 1: find a high-price interval in SA
p <- aemo_price("SA1", "2024-03-01", "2024-03-02")
spike <- p[which.max(p$rrp), ]
spike[, c("settlementdate", "rrp")]

# Step 2: which constraint had the highest shadow price at that time?
c <- aemo_constraints(
  start = spike$settlementdate - 300,
  end   = spike$settlementdate
)
head(c[order(-c$marginalvalue), c("constraintid", "marginalvalue", "rhs")])

# Step 3: which generators were providing FCAS at that time?
e <- aemo_fcas_enablement(
  start = spike$settlementdate - 300,
  end   = spike$settlementdate
)
# Aggregate raise6secmw across DUIDs to see market-wide enablement
if ("raise6secmw" %in% names(e)) {
  aggregate(e$raise6secmw, by = list(e$settlementdate), FUN = sum, na.rm = TRUE)
}

# Step 4: what loss factor applies to a generator in SA?
mlf <- aemo_mlf(year = "2024-25")
head(mlf[mlf$regionid == "SA1", c("duid", "mlf")])

Functions

Prices

Function Description Interval
aemo_price() Regional wholesale price (DISPATCHPRICE or TRADINGPRICE). Pre-5MS (before 1 Oct 2021) uses TRADINGIS; post-5MS aggregates six 5-min prices per trading interval. 5-min or 30-min
aemo_fcas() FCAS prices across ten services (eight contingency plus two regulation, including R1/L1 Very Fast from 9 October 2023) 5-min
aemo_price_caps() Market Price Cap, Floor, CPT, and APC by financial year (FY 2015-16 onwards). Static reference table. Annual

Demand and generation

Function Description Interval
aemo_demand() Regional demand: operational, operational-less-SNSG, or native (includes estimated rooftop PV) 5-min
aemo_dispatch_units() Per-DUID generator output (SCADA actual or dispatch target MW) 5-min
aemo_rooftop_pv() Region-level rooftop PV actual or forecast 30-min

Dispatch and constraints

Function Description Interval
aemo_constraints() Binding transmission and system constraints with shadow prices (marginal values). Use this to answer “why was the RRP AUD 15,000?” 5-min
aemo_fcas_enablement() FCAS enablement volumes (MW) per DUID across ten services (from DISPATCHLOAD). Quantity side of the FCAS market. 5-min
aemo_interconnector() MW flow, losses, and limits on the seven NEM interconnectors 5-min
aemo_bids() Generator bid stack: daily (BIDDAYOFFER_D) or per-interval (BIDPEROFFER_D, with size guard) Day or 5-min

Forecasts

Function Description Interval
aemo_predispatch() Price and demand forecasts: 5-minute-ahead (P5MIN, 12 intervals) or 40-hour-ahead predispatch (30-min resolution) 5-min or 30-min
aemo_pasa() Projected Assessment of System Adequacy: short-term (1-7 day) or medium-term (2-year) Hourly or daily

Reference data

Function Description
aemo_units() DUID registry downloaded from MMSDM DUDETAILSUMMARY (500+ DUIDs with region, schedule type, fuel class). Fallback to a 12-row sample if MMSDM is unreachable.
aemo_mlf() Marginal Loss Factors by DUID and financial year, from MMSDM MARGINALLOSSFACTOR. Used in PPA revenue reconstruction and settlement.
aemo_dlf() Distribution Loss Factors by connection point and financial year, from MMSDM LOSSFACTORMODEL.
aemo_regions() NEM region metadata (5 regions, static)
aemo_interconnectors() NEM interconnector topology and energisation dates (7 links, static)

Low-level access and configuration

Function Description
aemo_nemweb_ls() List files in any NEMweb directory
aemo_nemweb_download() Raw zipped-CSV download from NEMweb
aemo_cache_info(), aemo_clear_cache() Cache management
aemo_throttle() Request-rate configuration

Gas

Function Description Interval
aemo_gas() STTM (Adelaide, Brisbane, Sydney) and DWGM (Victoria) gas markets Daily

Size guards

AEMO’s bid tables are large. BIDPEROFFER_D (per-interval bid price and volume bands) has monthly archives on the order of 0.5 to 1.5 GB zipped (much larger uncompressed). A naive full-year pull is several gigabytes of compressed data and tens of millions of rows.

aemo_bids() enforces a 30-day span guard by default and refuses longer spans without explicit consent:

# Works (1-day span)
aemo_bids(duid = "BW01", start = "2024-06-01", end = "2024-06-02")

# Errors with a clear message
aemo_bids(duid = "BW01", start = "2024-01-01", end = "2024-06-01")
# Error: Requested span is 152 days.
# Bids data is large (BIDPEROFFER_D monthly archives are multi-GB zipped).
# Pass allow_large = TRUE to proceed.

# Works with explicit opt-in
aemo_bids(duid = "BW01", start = "2024-01-01", end = "2024-06-01",
          allow_large = TRUE)

Examples

Regional spot price by hour of day 

library(aemo)

p <- aemo_price("NSW1",
                start = "2024-06-01",
                end = "2024-06-02")

p$hour <- as.integer(format(as.POSIXct(p$settlementdate), "%H"))
avg_by_hour <- aggregate(as.numeric(p$rrp) ~ p$hour, FUN = mean, na.rm = TRUE)
names(avg_by_hour) <- c("hour", "rrp")
plot(avg_by_hour, type = "l", xlab = "Hour", ylab = "AUD/MWh")

Interconnector flows 

# Heywood Interconnector (VIC-SA) flows for the past day
i <- aemo_interconnector(flow = "V-SA",
                          start = "2024-06-01",
                          end   = "2024-06-02")
head(i)

# Project EnergyConnect (V-S-N) Stage 1 went live in April 2025
pec <- aemo_interconnector(flow = "V-S-N",
                           start = "2025-06-01",
                           end   = "2025-06-02")

Compare regions 

# Query each region once
nsw <- aemo_price("NSW1", "2024-06-01", "2024-06-01 01:00:00")
vic <- aemo_price("VIC1", "2024-06-01", "2024-06-01 01:00:00")

# Average price
mean(as.numeric(nsw$rrp), na.rm = TRUE)
mean(as.numeric(vic$rrp), na.rm = TRUE)

Reference tables 

# No network needed
aemo_regions()
aemo_interconnectors()

# Small generator registry (fallback shipped in package)
aemo_units()

Explore NEMweb directly 

# List files in a NEMweb report directory
f <- aemo_nemweb_ls("/Reports/Current/DispatchIS_Reports/")
head(f)

# Download a specific file
path <- aemo_nemweb_download(f$url[nrow(f)])
path

Data source and licence

Data is published by AEMO at http://nemweb.com.au under the AEMO Copyright Permissions Notice (not CC BY 4.0; the notice is similar but carries AEMO-specific attribution language). See https://www.aemo.com.au/privacy-and-legal-notices/copyright-permissions for the full terms. This package caches downloads to tools::R_user_dir("aemo", "cache").

Attribution on derivative work (AEMO’s requirement in full): Source: AEMO. AEMO makes no representation as to the accuracy or completeness of this data.

Python equivalent: UNSW-CEEM’s NEMOSIS is the established Python library for NEM data. Use it if you’re in Python.

R ecosystem:

Package Description
cer Australian Clean Energy Regulator (ACCUs, Safeguard, NGER, LRET, SRES)
carbondata Global carbon markets (EU ETS, UK ETS, RGGI, California, Verra, Gold Standard)
climatekit 35 climate indices (temperature, precipitation, drought)
readnoaa NOAA climate and weather data
readaec Australian Electoral Commission
readabs Australian Bureau of Statistics
readrba Reserve Bank of Australia

A note on GST

All NEM wholesale prices in AEMO’s published feeds, and hence in this package, are exclusive of GST. See National Electricity Rules, Chapter 10 (Definitions: “spot price”). When computing revenue figures for comparison against published financial statements or retail prices, gross up by 1.10 as appropriate.

References

Methodology choices in this package follow the design conventions of the established NEM literature. Where a design decision is non-obvious, the docs cite the original:

  • Gorman, N., Haghdadi, N., Bruce, A., & MacGill, I. (2018). NEMOSIS: NEM Open Source Information Service: open-source access to Australian National Electricity Market Data. Asia-Pacific Solar Research Conference. The authoritative reference for NEM data-model conventions.
  • Gorman, N., Bruce, A., & MacGill, I. (2022). Nempy: A Python package for modelling the Australian National Electricity Market dispatch procedure. Journal of Open Source Software 7(70), 3596. doi:10.21105/joss.03596.
  • Prakash, A., Bruce, A., & MacGill, I. (2023). NEMSEER: A Python package for downloading and handling historical National Electricity Market forecast data. Journal of Open Source Software 8(92), 5883. doi:10.21105/joss.05883. The aemo_predispatch(run_datetime =) vintage pattern follows NEMSEER.

Academic uses of NEM data this package is designed to support:

  • Gonçalves, R. & Menezes, F. (2022). The price impacts of the exit of the Hazelwood coal power plant. Energy Economics 113, 106398. doi:10.1016/j.eneco.2022.106398. Uses DISPATCHLOAD + BIDPEROFFER_D + DISPATCHPRICE at 5-min resolution.
  • Mountain, B. (2025). The exercise of market power in Australia’s National Electricity Market following the closure of the Hazelwood Power Station. Victoria Energy Policy Centre working paper. doi:10.2139/ssrn.5285882.
  • Nelson, T., Easton, L., Wand, M., Gilmore, J. & Nolan, T. (2024). Electricity contract design and wholesale market outcomes in Australia’s National Electricity Market. Australian Journal of Agricultural and Resource Economics 68(4), 784-804. doi:10.1111/1467-8489.12588.
  • Rangarajan, S., Svec, J., Foley, S. & Trück, S. (2025). Revisiting the crisis: An empirical analysis of the NEM suspension. Energy Economics 141. doi:10.1016/j.eneco.2024.107983. Uses MARKETNOTICEDATA + DISPATCHPRICE with INTERVENTION = 1.
  • Simshauser, P. & Wild, P. (2024). Rooftop Solar PV, Coal Plant Inflexibility and the Minimum Load Problem. The Energy Journal. doi:10.1177/01956574241283732.
  • Simshauser, P. & Gilmore, J. (2025). Demand Shocks From the Gas Turbine Fleet in Australia’s National Electricity Market. Australian Journal of Agricultural and Resource Economics. doi:10.1111/1467-8489.70065.
  • Australian Energy Regulator (2025). State of the Energy Market 2025, Chapter 2: National Electricity Market.

Citation 

citation("aemo")

Issues

Please report bugs or requests at https://github.com/charlescoverdale/aemo/issues.

Keywords

AEMO, Australian Energy Market Operator, NEM, National Electricity Market, electricity prices, dispatch, demand, FCAS, MMSDM, NEMweb, gas market, STTM, DWGM, Wallumbilla, AEMC, AER, energy data, electricity market, R package, Australian energy data