Brick Schema or Project Haystack: Choosing a data model

Understanding semantic data models for energy management

In modern Smart Buildings, collecting data from numerous IoT devices, sensors, and automation systems is just the first step. The true value of this data is unlocked through its interpretation and analysis, especially in the context of energy management. However, without a standardized semantic data model, integrating and understanding information from disparate systems becomes technically complex, time-consuming, and economically unfeasible. Semantic interoperability allows systems to communicate and understand each other, interpreting not just numerical values but their meaning: what they represent, where they originate, how they relate to other systems, and how they should be used. This is critical for optimizing performance, automated fault detection and diagnostics, and providing energy system services.

Semantic data models, such as Brick Schema and Project Haystack, create a unified description of building systems, their components, properties, and interrelationships in a standardized, queryable manner. This simplifies the installation and configuration of analytics and building management software, transforming disparate operational data into a stable 'contract' that software can utilize.

Brick Schema: A detailed ontology for buildings

Brick Schema is an open-source project aimed at creating a unified schema for representing metadata in buildings. It is based on ontology and uses Semantic Web technologies like Resource Description Framework (RDF) and SPARQL for querying. Key components of Brick include an RDF class hierarchy describing building subsystems, entities, and equipment, a minimal set of relationships to link them in a directed graph, and an encapsulation method for composing complex elements. Brick describes buildings in a machine-readable format, allowing software to programmatically explore various operational, structural, and functional aspects of a building.

Advantages of Brick Schema:

• High detail and structure: Brick is a formal ontology that defines explicit classes (e.g., Air_Handling_Unit, Zone_Air_Temperature_Sensor) and typed relationships between them (feeds, hasPoint, is_part_of). This ensures high accuracy and unambiguous modeling.
• Powerful querying and validation capabilities: Thanks to RDF and SPARQL, Brick models are easily queryable. Tools like SHACL allow programmatic validation of models against defined constraints, for example, that a certain VAV (Variable Air Volume) type must contain specific points.
• Simplified analytics development: Brick reduces the cost of deploying analytics, energy efficiency measures, and intelligent control by simplifying the development of such applications.
• Integration with other standards: Brick actively collaborates with other standards, such as RealEstateCore for describing building spaces and the upcoming ASHRAE 223 standard.

Challenges of implementing Brick Schema:

• Implementation complexity: Building a Brick model requires a deep understanding of ontologies and Semantic Web technologies, which can be more challenging for automation engineers accustomed to traditional approaches.
• Higher initial costs: Due to its formality and detail, initial Brick implementation can be more labor-intensive and require greater investment in expertise.

The Brick Schema community is actively developing, releasing minor versions approximately every 6 months, which include backward-compatible extensions and new ontology features. Documentation is available on the official website brickschema.org and GitHub.

Project Haystack: A flexible approach to data tagging

Project Haystack is an open initiative that provides a set of technologies for modeling IoT data, particularly in buildings and embedded environments. Unlike Brick, Haystack traditionally uses a tagging system where an entity is described by attaching tags to it (e.g., ahu, discharge, air, temp, sensor), and the meaning is derived from the combination of these tags. This is a lightweight and common approach that naturally maps to BACnet point lists.

Advantages of Project Haystack:

• Flexibility and ease of implementation: The Haystack tagging system is quick to apply, allowing automation engineers to rapidly tag existing points. This makes it attractive for quick deployment in large facilities.
• Broad industry support: Haystack is widely adopted in the building automation world, and many BAS (Building Automation System) vendors and analytical solutions already support it.
• Active community: Project Haystack has an active developer community that continuously expands the scope of definitions for additional data types.
• Evolution towards formal modeling: While Haystack started as a tagging system without strict rules, current documentation defines three semantic levels: dictionary, taxonomy, and ontology. The development of Haystack 5, particularly through Xeto, is moving towards specifications, model validation, querying, and improved interoperability, with announced support for RDF.

Challenges of implementing Project Haystack:

• Potential inconsistency: The lack of formal rules for tag usage can lead to highly customized and inconsistent modeling practices across different facilities. This complicates automated data understanding and processing.
• Limited capabilities for complex queries: Haystack's traditional approach may be less effective for complex analytical queries that require understanding structural relationships between components, such as “find all VAVs downstream of this AHU.”

Official Project Haystack documentation is available on their website, where information about the community and usage examples can also be found.

Comparing architectural approaches: Brick Schema vs. Project Haystack

The choice between Brick Schema and Project Haystack often comes down to balancing detail, flexibility, and implementation complexity. While both standards aim to solve the problem of semantic interoperability in buildings, their architectural approaches differ significantly.

It is important to note that Brick metadata is a superset of Haystack – everything that can be modeled in Haystack can be modeled in Brick. Moreover, tools exist to convert Haystack models to Brick.

Impact on analytics and scalability of energy management

The choice of a semantic data model directly impacts the capabilities of energy consumption analysis and the scalability of the energy management system. Data structure determines the complexity of queries and the efficiency of analytical algorithms.

Brick Schema, thanks to its ontological structure and explicit relationships, provides a high level of semantic depth. This allows for complex queries that can reveal interdependencies between different building components and their impact on energy consumption. For example, it's easy to identify all temperature sensors associated with a specific HVAC system in a particular zone. Such detail is extremely valuable for advanced analytics, Fault Detection and Diagnostics (FDD), creating Digital Twins, and integrating with Machine Learning (ML) and Artificial Intelligence (AI) models for energy consumption optimization. The ability to validate models using SHACL ensures data consistency and reliability, which is critical for automated systems and AI agents.

Project Haystack, with its flexible tagging approach, allows for rapid organization of large volumes of data. This is ideal for quick implementation and data visualization on dashboards. However, for complex analytical tasks requiring an understanding of hierarchical or causal relationships, additional processing or interpretation may be needed. While Haystack is evolving towards more formal semantic modeling, its traditional flexibility can create challenges in ensuring data consistency across large facilities, which, in turn, can complicate the scaling of analytical solutions. Nevertheless, Haystack is actively integrating with AI tools, especially in the context of Large Language Models (LLM) and Retrieval-Augmented Generation (RAG) applications, allowing its use for advanced natural language processing and search scenarios.

Long-term maintenance and development costs also depend on the chosen model. While initial Brick implementation costs may be higher due to the need for expertise, its structured nature can reduce operational costs for integration and development of new analytical applications in the future. For Haystack, a low entry barrier might translate into increased costs for maintenance and ensuring data consistency as system complexity grows.

How AZIOT implements this

The AZIOT platform supports flexible integration with both data models, allowing Smart Building architects to choose the optimal approach for semantic interpretation of energy consumption data coming from various devices and systems, such as MQTT, Modbus, BACnet, KNX, Zigbee, Z-Wave, LoRaWAN, Matter, SCADA, BMS, and ERP. This ensures efficient analytics and automation based on the chosen model, utilizing edge processing, Unity Base, rules/scenarios, dashboards, auditing, and access control.

The choice between Brick Schema and Project Haystack is not always dichotomous. Often, a hybrid approach is optimal, where Haystack tags are used for rapid data collection and initial organization, and then mapped into a more formalized Brick model for deep analytics and Digital Twins. This approach allows retaining the advantages of both standards, minimizing their drawbacks, and providing a reliable foundation for future energy management. Intecracy solutions and inbase.com.ua solutions.