See how Willow can help you cut costs, save energy, and operate your buildings more efficiently and sustainably.
by Rick Szcodronski, Chief Product Officer, Willow
Ontology, shapes, and data form the foundation of Willow’s data layer. For many organizations, adopting a shared ontology or traditional tagging model is a logical starting point. However, the lack of consistent data shapes across vendors makes true plug-and-play integration and seamless migration difficult to achieve. Willow takes a proactive approach to identifying and addressing ontology and data shape challenges as they arise, focusing on removing key barriers to scaling across the industry. As new connectors are developed and new use cases emerge, data shapes are created or evolved thoughtfully while maintaining consistency. This tight, iterative approach enables Willow to evolve the ontology rapidly and ultimately enable fast onboarding.
In a previous article, we explored the data layer that enables Willow’s operational AI platform comprising a rich ontology, shapes, and data. Let’s now explore real-world applications that show how ontology, shapes, and data come together to bring solutions to life.
Let’s start with a use case that illustrates the role of ontology and shape, and how modeling a hot water commissioning system in Willow enables reasoning at the system level rather than the point level.
Most energy-efficient heating systems adjust the hot water supply temperature based on outdoor conditions. During milder weather, lower supply temperatures are sufficient to heat the building, reducing energy consumption and unnecessary heat loss. As the outdoor air temperature drops, the system increases the hot water supply temperature to maintain comfort and proper heating performance.
Willow’s Heating Plant HW Supply Setpoint Not Reset per Design OAT Skill continuously monitors whether the hot water supply setpoint is being reset correctly based on the Outdoor Air Temperature (OAT). The Skill evaluates both the mechanical sequence of operation and the implemented control program in the BMS to confirm that the programmed logic matches the engineer’s specified reset schedule. It identifies conditions such as fixed setpoints, overrides, or incorrect control logic that may prevent proper resetting, and it verifies the accuracy and calibration of the outdoor air and supply water temperature sensors to ensure the strategy is functioning as intended.
Modeling a hydronic hot water system can be done in many ways, and Willow’s team carefully weighs the trade-offs of each approach to ensure scalability and consistency across portfolios. While there is no single “best” modeling pattern, a decision must be made to support customer outcomes. In this example, the system’s design parameters, specifically the OAT-based reset schedule are stored as properties on the digital twin. By comparing this baseline to live telemetry from the equipment, Willow can detect anomalies, control drift, and performance degradation over time.
Now let’s walk through a use case that illustrates the role of Willow’s data layer, and how ontology for fume hoods in Willow enables modeling cross-systems interactions between pressurization zones and fume hoods.
Critical spaces such as laboratories require monitoring and control beyond temperature and humidity signals. Enabling real-world use cases like sash hood position management for optimizing energy consumption require the system to describe positive or negative pressurization in HVAC Zones and make up airflow to point-of-use exhaust assets modeled as Fume Hoods. Willow’s ontology for fume hoods supports the concept of a pressurization zone and grouping exhaust fans that control an exhaust plenum together.
In Willow, the HVAC Zone model forms the foundation for managing conditioned spaces, providing a structure to link all relevant air supply and exhaust devices, including terminal units, general exhausts, and point exhausts like fume hoods, to the zones they serve. HVAC Pressurization Zone inherits from HVAC Zone and has child models, HVAC Negative Pressurization Zone and HVAC Positive Pressurization Zone that deliver specific intent for spaces that must reliably keep contaminants contained or protected, respectively. Pressure state is computed by connecting to real-time telemetry for supply and exhaust airflow, and differential pressure sensors.
The Fume Hood model in Willow inherits from the Ventilation Hood model, which in turn inherits from HVAC Equipment, and its parent Asset model. The Fume Hood model has deeper specializations, to describe if it is a snorkel exhaust, benchtop or walk-in fume hood.
Properties of the Fume Hood model capture details like required face velocity setpoint to capture how fast air is meant to enter the hood opening at a standardized sash position, e.g. at 18 inches sash height. Bypass type with values of open versus restricted captures how freely air is allowed to enter through the bypass opening when the sash is lowered. Min and Max airflows capture design specs.
Ontology models for Fume Hoods and HVAC Zones define standardized types and relationships as digital twins are created and linked. This example is a great showcase for the ontology’s ability to granularly classify assets such as VAV Airflow Control Valves being used for Supply, Return, or Exhaust. Exhaust Valves are further defined as being used for Fume or General Exhaust. These are just five of the 3,000+ models in Willow’s ontology that make the difference in properly analyzing these complex controls.
Now, as real-time telemetry flows into the graph, Skills reason over these against the design specs to monitor behavior and generate alerts. This includes identifying unsafe or energy inefficient sash positions and their downstream impact on zone airflow and performance.
Willow’s ability to properly model the complete system in the robust data layer enables key use cases such as pressurization tracking, air change compliance, fume hood sash position detection, and energy conservation measures such as exhaust airflow reset and fume hood closure strategies as allowed by ASHRAE.
Modeling the complete system is only as good as the analytics’ ability to scale to the various types of system configurations. Willow’s graph-based, real-time analytics engine, Activate Technology, is able to scale to various types of scenarios such as multiple Supply VAV Valves and Exhaust VAV Boxes managing the zone with a common Exhaust Fan System with multiple fans managing multiple zones. This translates to faster deployment due to less customization and greater reliability during operations.
Comfort-related service requests are a relatable use case across domains, as occupants find rooms too hot or too cold. Let’s explore how Willow’s data layer enables this with ontology and data shapes.
A typical hot or cold call starts with a reactive work order being submitted. Now the race is on to triage the incident, find the root cause, and fix it so that the space can be comfortable and the occupants can be productive. The best way to find the root cause is to have a full visibility into the upstream equipment serving the space. And the optimum mechanism to resolve the issue is to have a complete understanding of where to find the equipment to service it. Willow’s ability to onboard both the system topology and the locations of each of the assets helps detect the root cause and indicate precisely where and how to resolve the issue.
In this example, we can see the relationships in the Knowledge Graph that indicate where HVAC equipment like a VAV Box named AHU-4-5 is part of an upstream AHU Group. The graph also captures where the VAV Box is physically located and which rooms it serves. In this case, the physical location of the VAV Box is in Corridor 06, and rooms it serves are Consultation Room 06 and Resource Center. Further, we can see the twin for the Thermostat and its physical location.
The information is historically locked away and not readily available, making this use case unachievable. Willow is able to extract this key link from design and handover documents to hydrate the Knowledge Graph.
With the twins and relationships in place, when a comfort issue is reported in Consultation Room 06, it’s easy for the facilities team to see the location of the Thermostat closest to that room. The ‘Serves’ relationship determines the relevant VAV Box. The ‘Located In’ relationship identifies the physical location of the VAV Box where the technician can be sent to perform repairs or servicing.
Occupancy data is one of the most valuable sources of insight in modern buildings and use cases abound. Understanding space utilization and capacity can help facility teams identify underutilized areas and repurpose them. Occupancy-driven HVAC control strategies can respond to real-time signals such as badge swipes, camera-based human detection, or the counts of networked devices, e.g. laptops with unique MAC addresses used as proxy for employee presence. Room booking data can also be used to predict occupancy for spaces like classrooms in universities.
Catering to a range of systems that may be device-based or space-based, vendor approaches vary and as a result, occupancy data is one of the most inconsistent in our industry. Normalizing space occupancy to account for the variety of ways in which the IoT vendors report data, and how that relates to the organization’s space management topology can quickly get complicated. Some vendors report device-level people counts while others aggregate and report at the room level. And yet other vendors offer workstation sensors which report occupancy for the desk spaces within a larger open area.
Ontologies often include basic building blocks such as Occupancy Zone, Occupancy Sensor Equipment, People Count Sensor Equipment, Occupancy Sensor, and Occupancy Count Sensor. However, building solutions at scale across vendor solutions quickly becomes complicated. Shapes provide the structure needed to normalize this data so that it can be used consistently across portfolios.
A key part of normalization is defining how devices relate to spaces. Willow’s modeling patterns in shapes clarify whether a device should be designated a space with the ‘Located In’, ‘Feeds’, or ‘Serves’ relationship. Another aspect is to manage telemetry data that may come in as a Change of Value (COV), or as a data stream polled at a specific interval, like once every 15 minutes.
When multiple sources are leveraged to capture occupancy data in addition to booking information, a unique interplay lights up. Space utilization analytics can compare actual occupancy with expected occupancy as indicated by reservations.
Space reservation is critical to enabling occupancy-driven HVAC use cases. Organizations report higher energy savings because booking information can often cover more rooms than have occupancy or people count sensors installed. Let’s walk through an example.
The building’s purpose often informs how much energy savings are to be had from this use case, but there is no denying that most building types can take advantage of variations in daily, weekly, and even seasonal (i.e. summer break) changes in occupancy. These patterns often reflect a clear opportunity to optimize energy usage, such as dialing down during downtime while maintaining comfort during active use hours. Willow’s Active Control functionality turns quiet periods into powerful gains in sustainability.
Whether optimizing a lab’s exhaust system, continuously commissioning a hot water loop, or normalizing occupancy data across vendors and automating occupancy-driven HVAC control, the value comes from a data layer that understands both the physical world and the intent behind it.
Across these examples, a common theme emerges. Willow’s data layer, comprised of the ontology, shapes, and data, help facility teams achieve outcomes for operational efficiency. By being the leader in each of these three aspects of the data layer, Willow achieves faster deployment times, better coverage of skills across unique system topologies and types beyond HVAC, and ultimately more value realized.
We challenge you to unlock the true potential of your unique facilities. Envision a world where every kilowatt-hour or liter of water is consumed responsibly and every system is running in its optimal state, maximizing its longevity and optimizing the environment for those within.