Smart Data Models’ Candidates repository now includes a new candidate standard for buildings: project-haystack, with 8 candidate models translated from Project Haystack 4.0.0, the open tagging ontology for building and IoT data. Project Haystack is published under the Academic Free License 3.0. Like every candidate, each model comes with a JSON Schema, an NGSI-LD @context and all four example serializations (NGSI-v2 and NGSI-LD, key-values and normalized), each validated against its schema.
About Project Haystack
Project Haystack is an open-source community effort that standardizes how building automation data is described: sites, spaces, equipment such as air handlers, chillers and meters, and the sensor, command and setpoint points they expose. Entities carry marker tags (discharge air temp sensor point), value tags and references (siteRef, equipRef). Haystack 4 defines these tags as “defs” (haystack-defs, release 4.0.0, 719 defs) and publishes them as JSON, Turtle and JSON-LD.
The models
| Model | Haystack def | Attributes | Example |
|---|---|---|---|
Site |
site |
23 | Bravo hospital, Fort Wayne IN (real) |
Space |
space |
12 | Bravo space @b-0135 (real) |
Equip |
equip |
47 | Rooftop unit RTU-1: ahu rtu, chilled water cooling, hot water heating (real) |
Point |
point |
36 | RTU-1 supply (discharge) air temperature sensor (real) |
WeatherStation |
weatherStation |
19 | Fort Wayne, IN station (real) |
System |
system |
10 | Chilled water system (synthetic) |
Device |
device |
10 | RTU-1 controller (synthetic) |
Network |
network |
8 | BACnet network (synthetic) |
Grounded in a real building
Project Haystack publishes anonymized example buildings under the same licence. Five of the examples come straight from “Bravo”, a hospital with steam, hot water and chilled water plants: the site, a space, rooftop unit RTU-1 (an air handler with chilled water cooling, hot water heating and VAV zones), its supply air temperature sensor, and the Fort Wayne weather station. Bravo has no system, device or network records, so those three examples are synthetic, linked to Bravo, and say so.
Translation notes
- Haystack names, verbatim. Every attribute is a Haystack tag name, and every
@contextmaps it to the def IRI of the official Haystack RDF export. Model names follow the Haystack 5 (Xeto) specs: Site, Space, Equip, Point… - Markers as Haystack itself exports them. Marker tags go in one
hasTaglist (["ahu", "rtu", "hvac", "equip", …]), the same way Haystack’s own RDF export carries them (ph:hasTag). In JSON-LD each marker expands to its Haystack IRI. - Choices as enums. Haystack choices such as
pointFunction(sensor, cmd, sp),pointSubject,pointQuantity,coolingProcessorheatingProcessare also enum properties, so they are easy to query. - References become NGSI-LD Relationships: site, space, equipment, system and weather station references, sub-metering, and the flow references between equipment (hot water, chilled water, electricity…).
- Ready for geo-queries. Sites and weather stations get an NGSI-LD
locationbuilt from Haystack’sgeoCoord. It is the only addition to the Haystack content. Areas are in m².
Try it, and tell us what’s missing
These are Candidates: first, carefully sourced translations that are not yet official Smart Data Models subjects. The main open question is overlap with the Brick Schema and RealEstateCore candidates and with the official dataModel.Building and dataModel.Device subjects, since all of them describe buildings and their equipment. Natural next steps are the remaining Haystack entities (EV charging sessions, point groups, machine-learning variables) and the Haystack 5 Xeto specs.
Browse the models, open an issue, or send a PR:
👉 https://github.com/smart-data-models/Candidates/tree/master/standards/project-haystack

