Conference Agenda

This paper presents whole-building simulation results that demonstrate the use of differentiable predictive control (DPC) for supply-air-temperature reset in a reference office building with stochastic occupancy. A physics-informed DPC was trained on one HVAC zone to optimize for occupant comfort and HVAC energy use. When deployed on 5 zones on the same floor and compared with rule-based control strategies, DPC reduced occupied comfort violations by 54.8% and reduced total HVAC energy use by 21.3%. When the same DPC was then zero-shot deployed on all 15 zones in whole building, it delivered a 39% reduction in comfort violations and 14% reduction in total HVAC energy use. These results demonstrate that DPC is a scalable, occupant-centric framework enabling predictive and energy-efficient building control under real-world stochastic occupancy conditions.

Completing a comprehensive HVAC design for building retrofits is traditionally a labor-intensive process, necessitating disjointed stages of data acquisition, geometric modeling, load calculation, and ductwork layout. This complexity increases project costs and impedes the widespread adoption of HVAC upgrades in existing building stock. This presentation introduces a novel, automated workflow designed to streamline the design pipeline for multi-family residential retrofits by leveraging reality capture data.

We demonstrate a methodology for rapidly converting unstructured LIDAR point clouds into simplified architectural geometries through classification and surface fitting suitable for building thermal analysis and HVAC design. Building block loads are calculated by leveraging the OpenStudio-HPXML workflow and simplified geometry data. The workflow utilizes advanced algorithms to automatically optimize the placement of supply terminals, home run ducts and the location of the central Air Handling Unit (AHU) on the floor plan using heuristics, trained computer vision models and human in the loop feedback. By identifying structural constraints and generating a map of viable routing corridors, the workflow employs a modified A* pathfinding algorithm to connect terminals to the AHU. This ensures a duct design that is optimized for both material efficiency and aerodynamic performance. Furthermore, the system automatically generates designs for soffit enclosures to contain the ductwork, facilitating the prefabrication of key components prior to on-site installation. This integrated approach offers a scalable solution to the logistical and economic challenges of upgrading existing buildings with modern, efficient, HVAC systems.

As a young professional starting my career in ‘high performance buildings,’ there was little nuance in my understanding that electricity is the most sustainable source of energy and that we should move towards it and away from gas at all costs. However, the question of what truly is ‘sustainable’ has a different answer depending on who’s asking the question and what questions they’re asking. Through studying the metrics of cost, carbon, energy efficiency, longevity, and envisioning how each will change in the future, we have a much more holistic, nuanced understanding of the implications of using gas versus electricity. For example, on a per unit of energy basis, electricity is anywhere from 2-6x the price of gas. However, when you then factor in that heat pumps can be 2x more efficient than gas boilers, heat pumps ultimately emit less carbon, use less energy, and are cheaper than gas equipment. This brief presentation will explore the nuances of gas versus electricity and identify how the answer to the question of what is sustainable changes depending on which lens is utilized.

The presentation contains an award winning graphic from the Project Stasio 2025 annual competition.

Laboratory construction across the U.S. Southeast research corridors is accelerating, fueled by expanding research ecosystems, university–industry partnerships, and state-backed economic initiatives. This growth signals opportunity, it also exposes a growing challenge: laboratories are among the most expensive and energy-intensive building types to deliver—financially, technically, and operationally.

In practice, many lab projects rely on layered funding sources and face schedule pauses, redesigns, or phased delivery. When energy codes advance during these delays, project teams often face a “double whammy”: tightened budgets combined with performance targets that may suddenly leap multiple code cycles. This presentation shares lessons from a recent laboratory project in Louisiana that was forced to jump from roughly IECC 2012-era performance expectations to IECC 2021 requirements midstream—an experience that highlighted the risk of taking the ‘business as usual’ approach and designing around the mid-2010s compliance level.

The presentation will discuss in fact vs myth format onsite solar, envelope, lighting and appendix G pathway. On-site solar can function as a late stage “safety net” for energy code compliance. Under the Energy Cost Budget (ECB) method, on-site renewable contribution is now capped at five percent. Furthermore, laboratories are energy-intensive by nature and shaving even five percent off total energy cost often requires a surprisingly large PV array.

Envelope strategies, while still important, deliver diminishing returns post-2015 code levels are assumed. Similarly, lighting power densities in laboratory baselines are already relatively low—often leaving only incremental savings available before safety requirements and research flexibility become constraints. Together, these realities remove three of the traditional “easy wins” from the compliance toolkit. Appendix G with its more flexible nature remains an allowable compliance pathway, but in practice it offers limited relief for labs when paired with building performance factors (BPFs). In many recent projects, teams have found that meeting Appendix G targets is effectively out of reach.

This presentation argues that meaningful code compliance, project resiliency and operational performance—now squarely depends on HVAC-level strategies that create performance cushion. Owners and facilities teams need to be pursued for incorporating perceived “complicated and expensive” HVAC design options at early project phases. Using simulation-driven comparisons, we will walk through several approaches implemented on recent Southeast lab projects, including heat recovery chillers, modular heat pump and heat recovery chillers, advanced ventilation control strategies including aircuity, ductless fume hoods, glycol-based preheat/reheat systems like Konvekta, and radiant conditioning solutions. Payback trends and compliance impacts are shared graphically, focusing on decision-making backed by simulation data.

The key takeaway is simple: at least one high-impact HVAC strategy must be protected throughout design and construction. These lessons, grounded in real project outcomes, offer a roadmap for de-risking lab design as codes and decarbonization goals continue to evolve.

Improving the performance of dynamic heating, ventilation, and air-conditioning (HVAC) equipment and systems requires efficient integrated modeling and optimization. This study optimizes a dedicated outdoor air system (DOAS) with a desiccant wheel (DW) using a Julia-based toolchain that unifies symbolic modeling, numerical solving, and optimization. Gradient-free methods were employed in this study. Two operational parameters, the preconditioning start time and the desiccant regeneration temperature limit, were optimized for three humid locations. Results show that pre-activating DOAS by 1.5 to 2 hours and decreasing the desiccant regeneration temperature reduced humidity violation time by 80–90% and the system energy use by 2–7%, demonstrating the potential of the toolchain for HVAC modeling and simulation-based optimization.