Building Technology & Urban Systems Division

Energy Technologies Area

Lawrence Berkeley National Laboratory

Open Building Operating System: an Open-Source Grid

Responsive Control Platform for Buildings

Lazlo Paul, Flavia deAndrade Pereira, Sang Woo Ham, Marco Pritoni, Rich
Brown, Jingjuan Dove Feng

Energy Technologies Area

June 2023

This work was supported by the Assistant Secretary for Energy Efficiency and Renewable Energy,
Building Technologies Office, of the U.S. Department of Energy
under Contract No. DE-AC02-05CH11231.
Disclaimer:

This document was prepared as an account of work sponsored by the United States Government.
While this document is believed to contain correct information, neither the United States
Government nor any agency thereof, nor the Regents of the University of California, nor any of their
employees, makes any warranty, express or implied, or assumes any legal responsibility for the
accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed,
or represents that its use would not infringe privately owned rights. Reference herein to any specific
commercial product, process, or service by its trade name, trademark, manufacturer, or otherwise,
does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United
States Government or any agency thereof, or the Regents of the University of California. The views
and opinions of authors expressed herein do not necessarily state or reflect those of the United States
Government or any agency thereof or the Regents of the University of California.
Open Building Operating System: an
Open-Source Grid Responsive
Control Platform for Buildings

Lazlo Paul Flavia de Andrade Pereira Sang Woo Ham, PhD

Affiliate ASHRAE Member ASHRAE

Marco Pritoni, PhD Rich Brown Jingjuan Dove Feng, PhD, PE

Member ASHRAE Associate Member ASHRAE

ABSTRACT
Grid-interactive efficient buildings (GEBs) with flexible loads are a promising method to decarbonize buildings, shift loads during peak hours, and lower
energy use and electricity costs. Despite the promising benefits of GEBs, automation systems that manage flexible loads in response to energy prices or
other grid signals are still uncommon in small and medium commercial buildings. Recent literature demonstrates such control solutions, but they often rely
on custom integrations lacking the tools and drivers needed for scalability. To address these gaps, our team has created a fully open-source software stack
capable of integrating heterogeneous flexible building loads and implementing integrated portable control applications called the Open Building Operating
System (OpenBOS). The software can be deployed over existing control architecture with a small capital cost. OpenBOS leverages semantic models, which
have been the subject of recent investigations to facilitate application portability. The use of semantic data reduces the labor and expense required to deploy
and update smart control applications, increasing scalability. In this paper, the semantic modeling schema “Brick” was used, but the proposed approach
can also be applied to ASHRAE standard 223P, when released. This paper describes the methodology and software components of OpenBOS and
demonstrates its functionality with a rule-based demand flexibility control application configured using a semantic model. This application was tested at a
real building in NY that uses a dual-fuel heating system made up of five ductless heat pump mini-splits and a central furnace serving a single zone. The
demonstration reduced electricity costs at the site by 27%, demand during a shed event by 49%, and furnace usage by 35%.

INTRODUCTION

Grid-Interactive Efficient Buildings (GEBs) are buildings that combine smart systems and communication
technologies to optimize energy efficiency and actively use demand flexibility while offering a comfortable and
productive environment for occupants. At scale, GEBs could reduce CO2 emissions by 80 million tons per year by
2030 (Satchwell et al. 2021). An essential facet of GEBs is demand flexibility (DF), the ability to shift when loads are
used within buildings. DF can be enabled by control applications that respond to event or price signals that incentivize
a site to change its loads. Small- and Medium- Commercial Buildings (SMCBs) account for half of the floor area of all
commercial buildings, offering significant potential for DF applications (EIA, 2021).

Lazlo Paul is an associate data scientist at Lawrence Berkeley National Lab, Berkeley, California with researchers Sang Woo Ham, Marco Pritoni, Rich Brown,
and Flavia de Andrade Pereira. Flavia is also a PhD candidate at University College Dublin. Jingjuan Dove Feng is an associate director at TRC Engineers Inc.
 In SMCBs, Internet of Things (IoT) devices such as WiFi-connected thermostats have rapidly become popular
(Ford et al. 2017). They are low-cost, discrete, and can be installed with minimal disruption, which is particularly
useful in SMCBs that have limited space and budget constraints. These IoT devices are often connected to vendor
cloud platforms and controlled individually. Some of their potential for DF applications has been recognized and
explored through aggregators or utility demand response programs that directly control WiFi-connected thermostats
(Rajabi et al. 2017). However, without coordinating the IoT devices installed in a building, the full potential of DF in
SMCBs cannot be realized.
Current control platform technologies, such as VOLTTRON (Katipamula et al. 2016), allow the integration of
heterogeneous IoT devices within a building, however, the deployment of control applications on top of such
platforms is still labor-intensive and error-prone. While a control platform may be able to handle various APIs, the
metadata (semantic description) for their related data points (e.g., measurement and control points from IoT devices
associated with building systems such as HVAC) is often only contained in unstructured and non-standard sources
such as drawings, English language documents, and staff knowledge (Bergmann et al. 2020). This limits the discovery
and interpretability of these data points, resulting in costly manual and ad-hoc point mapping processes when
configuring new applications (Pritoni et al. 2021). To deploy a control application at a building, technology providers
need to integrate required devices, map their data, and reprogram or reconfigure applications to use that data.
Additionally, if a building owner wants to upgrade a control application or add an additional application from a
different vendor, this process has to be completed again at the same cost. Since HVAC systems in SMCBs have
relatively lower total energy costs, and thus less opportunity for savings, such manual and costly processes of
developing and deploying control applications do not make sense economically.
OpenBOS is a new software solution that innovates upon current building control technologies to provide a
scalable method for deploying control applications in SMCBs. It accomplishes this through the use of semantic
models. Semantic models enable the development of applications that can be portable across various buildings and
over the whole life cycle of an individual building (Fierro et al. 2020). This paper first describes the capabilities of
semantic modeling. Second, it introduces OpenBOS, describing the methodology behind it and each of its software
components. Finally, it demonstrates an example application, a rule-based DF control application that can be easily
configured using a semantic model, and shows the results from deploying OpenBOS in a real building in New York.

SEMANTIC MODELING

Semantic models offer a structured approach to defining information, allowing buildings to be represented
digitally in a consistent and standardized way (Fierro et al., 2022). These models describe information about building
spaces (e.g., thermal zones), equipment (e.g., rooftop units), including their components (e.g., fans), measurement and
control points (e.g., sensors and setpoints), and the relationship between them. The models are represented by graphs
and can be queried to extract knowledge in a machine-readable format. By querying semantic models, software tools,
such as OpenBOS, can determine the meaning of available measurement or control points and their association with
spaces and equipment using common descriptions. This allows for the seamless integration of data from various
vendors and the definition of replicable, semi-automated application configuration processes (Roth et al. 2022). The
latter helps applications to be portable across different buildings without extensive reprogramming. This is beneficial
in SMCBs, where heterogeneous IoT devices with diverse naming conventions and data acquisition procedures are
commonly deployed, and the deployment of control applications needs to be low cost.
The potential of portable applications using semantic models has been demonstrated by MORTAR, an
open-source platform containing semantic models for over 90 buildings that allows analytics applications to query
semantic models for configuration over multiple buildings (Fierro et al. 2018). However, real-world demonstrations of
semantics-driven control applications are scarce in the literature. As such, this paper aims to introduce and
demonstrate the use of semantic models with OpenBOS to enable semi-automated configuration of portable control
applications. Brick is the semantic modeling schema currently used by OpenBOS. Brick is an open-source semantic
modeling schema that uses semantic web technologies to describe the physical, logical, and virtual assets in a building
and their relationships in a machine-readable, uniform, and interoperable way (Roa et al. 2022). Brick semantic
models contain a digital representation of each data source in a building, such as sensors or setpoints, as well as their
relationships to the equipment and locations (e.g., what zone a sensor is measuring, or which equipment a setpoint
belongs to). A Brick model is thus able to provide a consistent representation of building systems and their data,
which can be relevant to a wide array of control applications. Brick models usually do not include telemetry data about
real-time or historical building system operations, but they point to other data sources (e.g., time series database,
BMS/IoT gateway, or other digital storage) that applications can use to retrieve data or write new commands.
Although OpenBOS employs Brick, other similar semantic schemas can be included in the future, such as ASHRAE
Standard 223P (ASHRAE 2018), which aims to model building systems in greater detail but is currently still under
development.

OPENBOS SOFTWARE

OpenBOS is a software solution that utilizes semantic models to enable the deployment of portable control
applications (Figure 1). All of OpenBOS’s components are open-source and completely free to use and can be easily
replicated at other buildings. OpenBOS architecture contains three layers. The top layer includes portable applications,
as well as a data dashboard for monitoring and analysis. The middle layer of the architecture includes the middleware.
The middleware platform used is VOLTTRON, which securely connects the applications to the building and external
data sources, handling the required read and write commands for control application logic. Drivers are parts of the
middleware which manage communication to specific pieces of hardware and external data sources. The semantic
model and database have also been included in this layer. The bottom layer contains the building measurement and
control points (hardware) that are being controlled, and external data sources that are used by the applications. This
section describes the proposed methodology for the deployment of portable applications within OpenBOS and each
component of its architecture.

Figure 1 Components of OpenBOS architecture

 The proposed methodology for enabling OpenBOS to deploy portable DF control applications through
semantic models consists of four steps (Figure 2). First, a portable application is created. The application must be
generalizable across heterogeneous buildings, which means that it cannot be hard-coded for a specific building.
Second, a Brick semantic model must be created for a given building. The model should describe the metadata about
the building systems and measurement/control points, as well as pointers to their respective sources, from where data
can be read and written (e.g., time series databases and control points via the middleware drivers). Third, the portable
application is semi-automatically configured by extracting required pointers and other metadata through automated
semantic queries bundled with the application. Certain applications may require a limited amount of information that
is not in the Brick model, such as occupant comfort constraints or price signals. Finally, the application is executed,
utilizing the middleware for data reading and writing.

Figure 2 The proposed methodology for enabling OpenBOS to deploy portable DF applications leveraging semantic
models and middleware.

While the first step is relevant from the perspective of application developers, who can more easily reuse
developed control applications across many buildings, the last three are important from the perspective of building
owners/operators, who can reuse the same semantic model for a given building to facilitate and reduce costs for the
deployment of several applications.

Portable Applications

OpenBOS portable applications contain supervisory control logic that can run on many buildings without
extensive reprogramming. Along with the semantic model used for automated configuration, the OpenBOS
middleware abstracts the specific protocols and data structures used to communicate with the diverse building devices
and external data sources. This allows portable applications to exclusively contain logic, with no communication
details, enabling them to run on various applicable systems. It is important to note that although portable applications
are intended to run across various heterogeneous buildings, certain applications may be more suitable for given
building systems.

Semantic Model

The OpenBOS software solution requires the creation of Brick semantic models for specific buildings. For that,
OpenBOS leverages BuildingMOTIF (Fierro et al. 2022), a toolset that helps to create Brick models following a
well-defined workflow process that ensures they are constructed correctly and are able to run applications. The
created Brick model includes building thermal zones, HVAC systems and their measurement and control points, as
well as the relationships between all of them. While the created model does not include telemetry (real-time or
historical) data related to the operation of the building systems, it does provide their meaning and context, as well as
their addressing information (i.e., pointers to where to find and query them via the OpenBOS middleware drivers and
database). The model is then used by OpenBOS to query these pointers in order to configure the application. While
the semantic model acts as the informational interface between applications and hardware, the middleware acts as the
communication interface.

Middleware and Drivers

VOLTTRON (Katipamula et al. 2016) is the middleware platform used in OpenBOS. VOLTTRON handles
data capture and retrieval, and provides secure management and communication between the other components of
OpenBOS. It uses an agent-based approach that allows one to easily create new drivers to integrate diverse types of
devices and external data sources.
Drivers connect the middleware to devices and provide a uniform interface for scalable device integration and
control. VOLTTRON has existing drivers for common protocols, such as BACnet and Modbus. IoT devices that
communicate via HTTP API frequently include software development kits that can be adapted into new device
drivers. The communication protocols supported and the ability to develop new drivers for IoT devices allows for the
integration of systems typical of SMCBs. Semantic models augment the capabilities of this approach, seamlessly
integrating applications with diverse new devices with minimal manual effort.
To handle cybersecurity concerns driven by the open-source aspect of VOLTTRON, detailed threat profiles
have been published for its current version, describing possible vulnerabilities and mitigation measures for
deployments (Himes et al. 2021). Many threat types are considered, including denial of service, elevation of privilege,
spoofing, and tampering. The mitigation strategies follow standard security procedures released by the National
Institute of Standards and Technology (NIST) in SP 800-53 Rev. 5, Security and Privacy Controls for Information
Systems and Organizations (Joint Task Force 2020). Some of these mitigation strategies are implemented within the
middleware code, such as input validation to make sure that messages received by the middleware are not a threat.
Others must be undertaken when deploying the platform, such as setting up a firewall to restrict access to the
platform and database.

Database and Monitoring Dashboard

OpenBOS currently uses a Timescale database to handle the time-series data storage. Timescale is an
open-source extension to PostgreSQL that is a relational database built to handle large amounts of time-series data
(Timescale 2023). As for data monitoring and analysis, OpenBOS leverages the open-source dashboard Grafana
(Chakraborty and Kundan 2021). Grafana has flexible alerting capabilities and can be used to create interactive and
dynamic dashboards by querying and transforming data from an attached database and then displaying it in a matrix
of panels.

DEMONSTRATION

This section details the deployment of a DF portable control application with OpenBOS in a real building in
New York. First, it describes the building, its systems, and baseline controls. Then, it presents the proposed rule-based
DF control for load shifting and shedding and discusses the results.

Building Description

The demonstration of the proposed semantics-driven control application within OpenBOS took place at a
real small office building in New York over a four-day period during March 2023. The demonstration site consists of a
single zone of 3,780 ft2 (351.2m2) of office space served by a dual-fuel heating system of five ductless heat pumps
(HPs) and one attic-mounted gas furnace (GF). The ductless HPs had inaccessible internal thermostats, and required
infrared (IR) remotes to change their setpoints or operating mode. Thus, WiFi-enabled IR remotes with built-in
temperature sensors were used to monitor the space and change the setpoints of the HPs. The electrical loads of each
HP were measured minutely by a WiFi-enabled electricity meter, and the GF was controlled by a standard
WiFi-enabled thermostat. All of these devices were connected to the internet by a 4G cellular router. Individual
drivers were created for each device to integrate them into OpenBOS, and they were controlled via vendor cloud
APIs.

Baseline Controls

The baseline control scenario was a schedule-based setpoint operation, with the gas furnace operated using
“droop” control. This means that the gas furnace was given a lower heating setpoint than the HPs to prevent it from
operating when the HPs were able to provide sufficient heating. This scenario is the current standard method for
controlling dual-fuel systems made up of units with independent thermostats. During the occupied period (7:00 AM -
8:00 PM), the HPs had a heating setpoint of 70°F (21°C), and the GF had a heating setpoint of 68°F (20°C). During
the unoccupied period, all systems operated with a heating setpoint of 60.8°F (16°C). The HPs were put into heat
mode because this test occurred during winter. Information about the devices and baseline control strategy for the site
is summarized in Table 1.
Table 1. Building Summary
Details
Occupied: Indoor Heating Temperature Setpoint 70°F (21°C) from
7:00 AM - 08:00 PM
Baseline Control Strategy
Unoccupied: Indoor Heating Temperature Setpoint 60°F (16°C) at all
other times
Electricity Cost 7:00 AM - 10:00 PM: On-peak (18.62$/kWh).
(conEdison ToU small business1) Other times: Off-peak (1.38$/kWh)

OpenBOS DF Control Application

The proposed application introduces a rule-based control scheme for DF that allows for individual control of
packaged HVAC units to shift electrical loads to earlier in the day. It was designed to be portable and run effectively
on the packaged units with zone-level thermostats that are common in SMCBs. For this purpose, the deployment of
the application started with an initialization routine to semi-automatically configure the control application with

1
https://www.coned.com/en/accounts-billing/your-bill/time-of-use
building-specific information. In this process, the semantic model created for the building was queried to retrieve the
pointers to access the required data for the application, as illustrated in Figure 3 for zone temperature sensors. The
application was also configured with additional data not available in the model, such as occupant comfort boundaries.
Following this configuration process, the control was executed.

Figure 3 Snippet of the Brick model of the building and example of a query to retrieve the pointers to
temperature sensors needed for the DF control application.

The control flow started by checking runaway conditions through the monitoring of the current temperatures
in the zones against a predefined comfort range. If temperatures were outside of the predefined comfort range, the
control application would revert to the baseline control strategy. This was followed by shift and shed event checks
triggered by occupancy, electricity price, or DF event signals indicating potential grid stress conditions. The shift
strategy involved preheating the building to its maximum comfort bound by a configurable amount of time before the
shed event. This increases electrical loads and stores thermal energy in the building. The shed strategy involved
reducing zone temperature setpoints when the building was occupied and the price exceeded a threshold, or there was
a shed event signal. This decreases electrical loads. Additionally, logic was used to rotate the operation of different
units to avoid simultaneous demand peaks. This logic was based on priority levels assigned to each unit depending on
the difference between their monitored temperature and setpoint, and their inactive period.
In this demonstration, the DF control application managed all five HP units and the GF installed at this site.
During the preheating strategy, the HP units were used in order to reduce furnace runtime. In order to maintain
occupant comfort while reducing energy consumption, the control used conservative maximum and minimum
comfort bounds of 71.6°F (22°C) and 66.2 °F (19°C), and a 2-hour period to preheat before a shed event. The site
was sent a simulated shed event from 7:00 - 9:00 AM, when winter heating loads in NY caused a demand peak
(National Grid, 2023). The existing occupancy schedule (7:00 AM - 8:00 PM) was used, and when a shift or shed event
was not identified, the baseline setpoints were used.

Results

The performance of the DF control application is evaluated based on the daily amount of load it was able to
shift from all the HP units, electricity cost savings, change in peak power, and furnace run-time reduction. Load
shifting is represented by two metrics, the percent change of load during the preheating period and the percent change
of load during the shed period. The percent change in peak power refers to the difference in the total peak from all
HPs at any time during the simulated morning peak period. As shown in Table 2, the results indicate an increase in
flexibility, with a load increase of 117% during the shift (preheating) period, and a load decrease of 49% during the
shed period. While executing demand flexibility, the energy costs were reduced by 27%, and furnace runtime was
reduced by 35%, however, there was an increase of 3% in the maximum power during the simulated morning peak
period. This may have been due to the HPs not staggering operation as expected, which is a result of the use of
one-way IoT IR remotes for control that did not allow the real state of the HPs to be accurately determined and
controlled.
Table 2. Performance comparison between baseline and the semantics-driven DF
control application
Performance metrics Daily baseline Daily DF Percent Change
Load during preheat (kWh) (05:00 - 07:00) 0.432 0.940 117%
Load during shed (kWh) (07:00 - 09:00) 1.193 0.610 -49%
Electricity consumption cost ($/day) 65.541 47.582 -27%
Peak power during shed (kW) (07:00 - 09:00) 6.29 6.471 3%
Furnace Runtime (Hrs/day) 0.53 0.34 -35%

CONCLUSION

This paper describes the development of OpenBOS and demonstrates its potential to deploy
semantics-driven portable control applications in SMCBs. OpenBOS was deployed in a real small office building in
New York and demonstrated with a rule-based portable demand flexibility control application. In this demonstration,
OpenBOS and the control application were able to reduce electricity costs at the site by 27%, demand during a shed
event by 49%, and furnace usage by 35%. This demonstration serves as a concrete example of the proposed
semantics-driven approach. OpenBOS was able to successfully perform demand flexibility strategies using a control
application with generalizable logic that did not require manual modification to map the application to the specific
data points at the site. Instead, it took advantage of the proposed automated configuration using a Brick model and
semantic queries, resulting in a more streamlined configuration process with minimal manual inputs. This successful
deployment of OpenBOS and the portable control application highlights the potential for significant cost and time
savings in the deployment and operation of demand flexibility strategies across multiple buildings.

ACKNOWLEDGMENTS

This work is supported by the New York State Energy Research & Development Authority (NYSERDA)
through the NextGen HVAC Innovation Challenge program, and by the CBIM-ETN funded by the European Unions
Horizon 2020 research and innovation program under the Marie Skodowska-Curie grant agreement No 860555. The
authors would also like to acknowledge the support and leadership of the TRC team, Jinjuan Dove Feng and Gwelen
Paliaga, and additional support from the LBNL team, Armando Casillas, Anand Krishnan Prakash, Weiping Huang,
and Peter Grant.
REFERENCES
ASHRAE. 2018. “ASHRAE’s BACnet Committee, Project Haystack and Brick Schema Collaborating to Provide Unified
Data Semantic Modeling Solution.” n.d. Accessed March 29, 2023.
https://www.ashrae.org/about/news/2018/ashrae-s-bacnet-committee-project-haystack-and-brick-schema-collaborati
ng-to-provide-unified-data-semantic-modeling-solution
Bergmann, Harry, Cory Mosiman, Avijit Saha, Selam Haile, William Livingood, Steve Bushby, Gabe Fierro, et al. 2020.
“Semantic Interoperability to Enable Smart, Grid-Interactive Efficient Buildings.” Lawrence Berkeley National Lab
(LBNL). https://www.osti.gov/biblio/1735554.
Chakraborty, Mainak, and Ajit Pratap Kundan. 2021. “Grafana.” In Monitoring Cloud-Native Applications, 187–240.
Berkeley, CA: Apress.
Energy Information Administration (EIA). 2021. “U.S. commercial buildings energy consumption survey.” n.d. Accessed
March 29, 2023. https://www.eia.gov/consumption/commercial/data/2018/bc/html/b1.php.
Fierro, Gabe, Anand Krishnan Prakash, Cory Mosiman, Marco Pritoni, Paul Raftery, Michael Wetter, and David E. Culler.
2020. “Shepherding Metadata Through the Building Lifecycle.” In Proceedings of the 7th ACM International
Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation, 70–79. BuildSys ’20. New York, NY,
USA: Association for Computing Machinery.
Fierro, Gabe, Marco Pritoni, Moustafa AbdelBaky, Paul Raftery, Therese Peffer, Greg Thomson, and David E. Culler. 2018.
“Mortar: An Open Testbed for Portable Building Analytics.” In Proceedings of the 5th Conference on Systems for
Built Environments, 172–81. BuildSys ’18. New York, NY, USA: Association for Computing Machinery.
Fierro, Gabe, Avijit Saha, Tobias Shapinsky, Matthew Steen, and Hannah Eslinger. 2022. “Application-Driven Creation of
Building Metadata Models with Semantic Sufficiency.” In Proceedings of the 9th ACM International Conference on
Systems for Energy-Efficient Buildings, Cities, and Transportation, 228–37. BuildSys ’22. New York, NY, USA:
Association for Computing Machinery.
Himes, Catie, Patrick O’Connell, Garret Seppala, Torri Simmons, Angela Steinmetz, Chance Younkin. 2021. “VOLTTRON
Threat Profile for VOLTTRON Version 8.0” Pacific Northwest National Laboratory (PNNL).
Jensen, Søren Østergaard, Anna Marszal-Pomianowska, Roberto Lollini, Wilmer Pasut, Armin Knotzer, Peter Engelmann,
Anne Stafford, and Glenn Reynders. 2017. “IEA EBC Annex 67 Energy Flexible Buildings.” Energy and Buildings 155
(November): 25–34.
Joint Task Force. 2020. “Security and Privacy Controls for Information Systems and Organizations” National Institute of
Standards and Technology (NIST). https://doi.org/10.6028/nist.sp.800-53r5.
Katipamula, Srinivas, Jereme Haack, George Hernandez, Bora Akyol, and Joseph Hagerman. 2016. “VOLTTRON: An
Open-Source Software Platform of the Future.” IEEE Electrification Magazine 4 (4): 15–22.
NationalGrid. “Demand Response Program.” n.d. Accessed March 29, 2023.
https://www.nationalgridus.com/NY-Business/Energy-Saving-Programs/Demand-Response.
Pritoni, Marco, Drew Paine, Gabriel Fierro, Cory Mosiman, Michael Poplawski, Avijit Saha, Joel Bender, and Jessica
Granderson. 2021. “Metadata Schemas and Ontologies for Building Energy Applications: A Critical Review and Use
Case Analysis.” Energies 14 (7): 2024.
Rajabi, Amin, Li Li, Jianfeng Zhang, and Jianguo Zhu. 2017. "Aggregation of small loads for demand response programs
— Implementation and challenges: A review," 2017 IEEE International Conference on Environment and Electrical
Engineering and 2017 IEEE Industrial and Commercial Power Systems Europe (EEEIC / I&CPS Europe), Milan,
Italy, 2017, pp. 1-6, doi: 10.1109/EEEIC.2017.7977631.
Roa, Carlos Duerte, Paul Raftery, Ruiji Sun, Lazlo Paul, Anand Prakash, Marco Pritoni, Gabe Fierro, Therese Peffer, 2022.
“Towards a Stronger Foundation: Digitizing Commercial Buildings with Brick to Enable Portable Advanced
Applications.” Lawrence Berkeley National Laboratory. http://dx.doi.org/10.20357/B7ZG6R Retrieved from
https://escholarship.org/uc/item/0dj216nq
Roth, Amir, Michael Wetter, Kyle Benne, David Blum, Yan Chen, Gabriel Fierro, Marco Pritoni, Avijit Saha, and Draguna
Vrabie. 2022. “Towards Digital and Performance-Based Supervisory Hvac Control Delivery.” Lawrence Berkeley
National Laboratory. http://dx.doi.org/10.20357/B70G62 Retrieved from
https://escholarship.org/uc/item/59z6d46m
Satchwell, Andrew, Mary Piette, Aditya Khandekar, Jessica Granderson, Natalie Frick, Ryan Hledik, Ahmad Faruqui, et al.
2021. “A National Roadmap for Grid-Interactive Efficient Buildings.” Lawrence Berkeley National Laboratory.
https://doi.org/10.2172/1784302.
Timescale. “Timescale Documentation.” n.d. Accessed March 29, 2023.
https://docs.timescale.com/timescaledb/latest/overview/.