Combating Condensation at the National Air and Space Museum
Richard Renaud and Brian S. Rose
October 23, 2019
Computational analysis of new mechanical and enclosure systems at the National Air and Space Museum (NASM) helped a design team for the building modernization project better understand the risk for condensation on replacement glazing systems.
Fig. 1. NASM, Washington, D.C., 2014. The north elevation, which overlooks the National Mall, is segmented into four stoneclad pavilions separated by glass-enclosed atria, 2014. All images by the authors’ firms unless otherwise noted.
This article explores the impacts of reintroducing humidification into monumental glazed atriums and examines the analysis techniques for assessing condensation risk.
The experience from this project may help others to better recognize the circumstances that warrant further condensation analysis on other rehabilitation projects. Although most directly related to mid-century modern and newer construction types, the fundamental lessons learned during the NASM project can be applied to any building preservation project with numerous features that may cause an elevated risk of condensation.
The foundation of the National Air and Space Museum on the National Mall lies in a small collection, displayed during the late nineteenth century in the Smithsonian Institution’s Arts and Industries Building, that included various kites and balloons. In 1920 a storage hangar, affectionately labeled the “Tin Shed,” was erected outside the Smithsonian Castle, and the collection grew to include artifacts from World War II. In 1946 President Harry S. Truman signed a bill officially inaugurating the National Air Museum. In 1966, with the space race underway, President Lyndon B. Johnson changed its moniker to include “Space” and expanded the collections, which required a corresponding upgrade to its facilities. In 1971 Congress appropriated $40 million for the construction of a new museum on the National Mall to be called “The National Air and Space Museum” (NASM).
Fig. 2. NASM, rendering of north elevation, 2018. The revitalization project includes the addition of a vestibule inspired by winged flight.
The Smithsonian Institution engaged the architectural firm of Hellmuth, Obata + Kassabaum (HOK) to design the building. The Gilbane Building Company started construction in 1972 and completed the structure in time for the 1976 United States Bicentennial celebrations. Since its inauguration, the NASM has been one of the most visited museums in the world and over recent years has averaged approximately seven million visitors annually.
Gyo Obata, the lead architect, designed the museum’s enclosure with monumental pavilions clad with Tennessee Pink Marble that are separated by monumental atriums enclosed with glazed curtain walls and skylights (Fig. 1). Spanning between the pavilions, exposed structural steel (AESS) spaceframe trusses support the 60-foottall curtain walls and a ceiling of skylights measuring 125 feet by 115 feet.
To reinforce the contrasting design intent for strong solid units separated by light-filled atriums, the marble also covers the interior atrium side walls. Visitors can thus view the aviation artifacts in their “natural habitat,” with the sky as a backdrop. Soon after its opening, Smithsonian staff noticed wintertime fogging on the interior surfaces of the glazing and frame. In 1980 the Smithsonian hired Werner Gumpertz, a cofounder of Simpson Gumpertz & Heger Inc., to assess the performance of the building envelope. The investigation determined that mechanically humidified air condensed on the interior face of the insulating glass units (IGUs) and the thermally improved frames.3 Though the interior and exterior frame components were separated by a low-conductance material, the thermal performance fell well short of today’s assemblies. The project stakeholders elected to reduce the mechanical humidification in order to reduce the condensation. Though not the focus of this article, the history of previous condensation issues informed the recent modernization project.
In 2013 the Smithsonian engaged a team led by Quinn Evans to perform a comprehensive condition assessment of the entire building in preparation for a modernization campaign. The team explored several options: repair-in-place, extensive replacement, and full rebuild. Following a rigorous feasibility study, the Smithsonian elected to pursue recladding of the entire building in order to address the issues with the stone cladding. This work also facilitated replacement of the mechanical systems; provided an enclosure that met the performance requirements (many of which fall outside the scope of this article) of a modern Smithsonian Museum; and allowed for new state-of-the-art exhibit spaces and new security vestibules (Fig. 2).
Between 2013 and 2018, the project team developed design documents for a holistic modernization project. The design phase that led to the analysis discussed in this article was guided by the following principles:
• advance the Smithsonian’s leadership in the museum field
• improve the visitor experience
• safeguard the intent of the original architectural design
• become a leader in museum sustainability
• maintain the safety of occupants and the collections.
The preservation of the collections proved to have an extensive impact on the building enclosure systems.
Reintroduction of humidity control.
Dry air conditions can extract moisture from many materials and degrade their mechanical properties. Similarly, rapid changes in air temperature and relative humidity also incrementally deteriorate certain materials. Wood, glue, and ivory are the most vulnerable to humidity related dimensional response, followed by paper, parchment, and textiles. Because of the wide variety of materials (each with its own environmental response) housed in most museums, conservators have established guidelines for humidity in museums, typically between 37 percent and 53 percent, which are intended to protect a range of artifacts.
Conditioning the air within museums is not a new concept. One of the first discussions of relative humidity (RH) levels in museums is found in a paper delivered in 1930 by John McCabe, superintendent at the Cleveland Museum of Art, which describes the use of “air conditioning” and recommends RH levels between 55 percent and 60 percent “independent of temperature or the time of year.” McCabe notes that the first system for humidification in a museum was likely the one installed in the Museum of Fine Arts, Boston, when it was built in 1908.
The original mechanical systems at the NASM were designed to provide a year-round relative humidity, though ultimately terminated, of approximately 50 percent through spray cooling coils (comparable institutional buildings in the mid-Atlantic states without mechanical humidification typically experience wintertime interior RH levels of 15 percent to 25 percent). Early in the recent modernization project, the designers worked closely with the conservators to establish humidification minimums that would preserve the collections and requirements for gradual seasonal transitions. The modernization design includes adiabatic spray injection that provides 40 percent RH and 68°F during the winter and then transitions to 50 percent RH and 75°F in the summer over the course of multiple weeks. Unlike the original systems, the modernization design controls the humidity to within 3 percent of the design target.7 For reference, the exterior winter and summer design temperatures for Washington, D.C., are 17.9°F and 94.7°F, respectively.
A proven strategy to minimize condensation in many humidified spaces is to provide heat and direct airflow to the interior of the enclosure in order to keep of the enclosure in order to keep the surfaces warm. For the NASM project, two fundamental design objectives limited the effectiveness of the air supply within the glazed atriums.
Fig. 3. NASM, America by Air gallery atrium, 2013. The current slot diffusers (at the arrow) on the atrium side walls will be reused for the upcoming modernization, but artifacts and the future shades (not shown in the photograph) will interrupt the air supply.
First, the modernization design incorporated tinted and fritted glazing, fixed shading baffles, and operable shades to increase the general lighting levels while reducing direct (particularly ultraviolet) light, which is harmful to artifacts. However, though not continuous, the fixed and operable shades significantly interrupt the flow of heated air from the HVAC supply. The design team was concerned that, without a reliable heat source, the interior enclosure temperatures could drop below the dew point.
Second, the placement of the new, modernized air-supply and return lines was greatly limited within the atriums because exposed ducts hanging from the skylights would clutter the space and violate HOK’s original design intent. Therefore, the original supply and return air slots on the stone-clad interior partitions enclosing the atriums were reused, and they were supplemented with rows of heaters along the glass curtain wall. The heaters were incorporated into the appearance of the supporting pipe truss to preserve the original open-design intent (Fig. 3).
Due to the massive size of the artifacts and the huge volume of the atriums, the design required warm air from the diffusers to travel more than 60 feet to portions of the curtain wall and skylights. Additionally, this air had to navigate around trusses, mullions, and shades to warm the glazing sufficiently to prevent condensation. Along with the common consequences of uncontrolled interior moisture (such as damaged and stained finishes and the potential for organic growth), water droplets falling from the skylights could damage irreplaceable artifacts. Due to the unknown airflow patterns and untenable consequences, condensation analysis quickly rose to the forefront of the modernization design.
Condensation is water vapor from air that has reverted to the liquid phase, which occurs at and below the dew-point temperature. Per the psychrometric chart, increasing the relative humidity of air increases the amount of water vapor and also increases the dew-point temperature, if the dry-bulb temperature remains constant. Condensation collects on constant. Condensation collects on surfaces that are colder than the ambient dew-point temperature.
A material’s thermal resistance influences the rate at which energy transfers from warm surfaces to cold surfaces. During the winter, the interior surfaces of enclosure systems are colder than the ambient interior conditions of the air because the warmth is conducted to the exterior. Condensation may occur on exterior surfaces of air-conditioned spaces (a common occurrence on muggy summer mornings), but exterior condensation most commonly forms outboard of the enclosure’s water barrier and does not pose a risk to damaging artifacts or building materials, as is the case at the NASM. Exterior condensation may, however, collect dust and lead to unsightly staining.
A lower thermal resistance (R-value) between enclosure systems will cause colder interior surface temperatures and increased condensation risk. Similarly, keeping the enclosure system the same and humidifying the interior air (raising the interior dew point) will elevate the risk for condensation. Fenestration elements are more susceptible to condensation than opaque walls due to the use of thin and/or highly conductive materials, such as aluminum and glass.
Spectrum of risk.
Due to the complexity of the interrelated elements that contribute to condensation, rules of thumb and previous project experience are useful starting points but are rarely useful starting points but are rarely sufficient to illustrate the whole scope regarding when and where condensation may form.
Condensation resistance is a spectrum of performance, with each project falling somewhere between total success and total failure. On the success end of the spectrum, all interior surfaces remain dry no matter how cold it gets outside (an enclosure that owners forget about is a success). At the other end of the spectrum are interior glazing surfaces that fog up and/or have frames encrusted with frost several times throughout the winter. However, most instances occur in the gray area of the spectrum. The ambiguity in the middle of the spectrum gives rise to difficult questions, such as:
• How cold does the outside need to be before condensation is inevitable and accepted?
• How much surface area of condensation constitutes failure?
• How much water needs to form before it drips?
• How sensitive is the system to variations from the ideal design scenario?
• How does the condensation itself affect the system performance?
Some of these questions are easier to answer with some sense of accuracy than others. One aspect that can be estimated is what conditions are necessary to predict the onset of condensation.
Often the most difficult part of the process is simply defining the criteria for success because eliminating all for success because eliminating all condensation under all possible conditions usually conflicts with other project criteria (cost being chief among those). Using industry standards, such as the 2017 ASHRAE Handbook—Fundamentals as Handbook—Fundamentals as benchmarks, owners and project teams define uniquely acceptable results based on the project-specific conditions. Due to concerns about the serious consequences of uncontrolled moistures, the NASM team fell on the conservative end of the spectrum.
Digital design simulations are used to help to establish the bounds of the performance the proposed approach is likely to yield. Project-specific judgement is required to determine which, if any, elements of the design (window frames, interior air conditions, etc.) warrant further analysis. A wide range of calculations, many of which are discussed below, are available to designers. Additional calculations, with increasing levels of effort and accuracy, help to narrow the bounds of the likely performance outcomes. If preliminary analysis with conservative performance values yields acceptable predictions, further analysis is likely not needed. For the NASM project, preliminary analysis showed a wide range of possible results, many of which were unacceptable (Fig. 4). Therefore, the project team elected to conduct more intense calculations that conduct more intense calculations that may exceed what is necessary for many other projects.
Condensation analysis is primarily a question of heat transfer. The analysis fundamentally strives to determine the following inputs (Fig. 5):
• exterior conditions, which set the demand on the enclosure
• enclosure systems, which conduct the heat
• interior conditions, which serve as a heat source.
ASHRAE sets forth the weather conditions and serves as a starting point for the exterior temperature. It is discussed in the section on bounding results below.
The enclosure systems are selected by the design team based on project constraints and previous experience. When value-engineering inevitability constraints and previous experience. When value-engineering inevitability arises, however, comparing the performance of systems becomes a necessity. In addition to the solid materials that govern conduction, every surface is surrounded by a thin film of air that provides considerable insulation and transfers heat from a HVAC supply to the enclosure. “Air film coefficient” is a term that quantifies the energy transfer through this layer; it results from convection, largely governed by the air speed, and by radiation, largely governed by surface material.
Fig. 4. Summary table of the preliminary sensitivity analysis that roughly approximates the spectrum of likely interior conditions near the curtain wall for various design options. Decreasing the near surface interior temperature below the design set point of 68°F and the air-film coefficient (Hc=1.6), which is related to air speed, results in more related to air speed, results in more condensation risk. Note that even when the near-surface air is above the supply set-point temperature, the edges of standard double-pane IGUs are at risk for condensation when the air is sufficiently stagnant (Hc=0.5).
The ambient interior conditions are governed by the building’s use. Since no building is a simple box, the air conditions are not uniform throughout a space. Therefore, both the temperatures near the enclosure’s interior surface and the air-film components differ from the ideal design conditions.
During the initial stages of the modernization design at the NASM, the project team analyzed the proposed glazing assemblies using thermal modeling (see below) and varied uncertain inputs. Commonly referred to as a “sensitivity analysis,” this type of study identifies how a variable impacts the results and to roughly what extent. The preliminary sensitivity analysis determined that the interior temperature and air velocity had the greatest impact on the results, but the range of possible interior conditions was considerably wide because of the airflow travel wide because of the airflow travel distance, obstructions between the diffusers, and the fenestration surfaces. The results showed a spectrum of both low and high condensation risk. The built reality likely falls somewhere between. Holding true to the artifact preservation guiding principle, the Smithsonian requested that the design team continue the analysis and narrow the bounds of possible results.
Fig. 5. Typical procedure for analyzing the condensation risk for enclosure systems. The exterior conditions to the right of the enclosure are set by historic climate data (step 1). The interior conditions on the left of the enclosure are analyzed by the CFD model (step 2). The CFD results are used to more accurately determine the inputs for the thermal models (step 3). By adjusting variables from the thermal models, the designer can evaluate the risk for condensation (step 4).
Selecting systems with long and proven track records under similar conditions and performance demands is vital to project success. The key is finding exemplary applications that are comparable in terms of critical metrics to the current project without unnecessarily constraining the design.
Unlike appearance or wind-load resistance, which are relatively easy to assess through published literature, condensation is more difficult condensation is more difficult to evaluate due to the numerous interrelated facets of each enclosure product and the project conditions that often yield a unique design scenario for each project. Subtle differences in thermal-break material, size, configuration, corner detailing, etc., make each proprietary assembly different. Similarly, the fenestration size, mullion layout, rough opening detailing, interior air supply, etc., make each project application different.
To help specifiers, the National Fenestration Rating Council established condensation resistance (CR) values based on computational-modeling procedures (NFRC 500), and the American Architectural Manufacturers Association established condensation resistance factor (CRF) values based on laboratory testing procedures (AAMA 1503). Experts in building-enclosure design, however, caution that while the CRF has some merit for comparing the generic (good, not-so-good, etc.) performance of similar products installed in similar constructions and climates, it fails to provide a true, physical measure of condensation resistance. Designers seeking to specify fenestration must become aware of these limitations and begin making use of the actual test data and modern computer simulations to accurately specify and verify the performance of fenestration products.
Similarly, manufacturers’ published CR data may be based on inputs and configurations that differ in key ways from the project in question.
For the recent NASM modernization project, the Roschmann Group was selected as the specialty-glazing subcontractor responsible for developing custom steel glazing systems to address the performance criteria. As the system has not yet been installed at the museum, the system specifics are still being developed. Future articles will discuss the glazing assemblies in more detail.
Computational Fluid Dynamics
Computational fluid dynamics (CFD) is a numerical modeling tool that can simulate fluid flow (typically based on the finite volume method) and allows the discretization (breaking up a geometry into finite elements) of any volume. With respect to building-enclosure condensation analysis, CFD can simulate airflow patterns and velocities and can simulate heat transfer through convection, conduction, and radiation. With these CFD results, designers can determine with more confidence what the air conditions are near surfaces of the enclosure systems, which the initial sensitivity analysis demonstrated were critical to the system success but difficult to control.
The NASM project team developed the CFD model in the program Star-CCM+ by CD-Adapco (recently acquired by Siemens). The model was evaluated using a 3D steady-state analysis and assumes nighttime conditions and an exterior design temperature of 17.9°F. The team modeled one of the three NASM atrium galleries that extends the full north-to-south width of the museum and is clad primarily with the glazed curtain walls and skylights (Fig. 6). The atrium model also includes an open mezzanine, but the gallery enclosed by partitions on the second floor was excluded. The continuous air-volume model measured approximately 192 feet by 115 feet in plan and was 80 feet tall. This model was exceptionally large because most buildings are typically partitioned into smaller volumes with contained air-distribution systems.
CFD models must include the enclosure systems, partitions, and any other interior elements that significantly impact the airflow. This included massive rockets, aircraft, and satellites scheduled to remain in place. The NASM CFD model also included the interior mullion profiles, ridge skylight shape, pipe trusses, and fixed sun shades because the accuracy of these components most closely impacts the conditions of the air near the enclosure.
Thermal performance values were calculated using THERM (see below), based on the basis-of-design systems, and applied to the surface elements to incorporate heat loss to the exterior. Through close coordination with Mueller Associates, the mechanical, electrical, and plumbing engineer of record, the air-supply and heating systems were incorporated into the model.
Fig. 6. Surface elements included in the CFD model. Large artifacts are shown in brown, and diagonal sunshades are shown in light gray below the skylight.
Fig. 7. Airflow streams and temperatures were calculated using the CFD model. Average near-surface temperatures were determined for IGU areas at controlling (cold) zones of the enclosure and used for subsequent analysis.
Once the model geometry and energy inputs were defined, the air volume was subdivided into cells to create the was subdivided into cells to create the “mesh” of data points, where the air properties would be calculated. The size of each cell varied, with the smallest cells positioned near the enclosure for improved accuracy at the location of interest. The NASM model contained approximately 57 million cells. Offsite super computers were required to discretize and analyze such a substantial discretize and analyze such a substantial model. Data transfer, analysis time, and other “big data” issues quickly impacted the schedule and mandated countless late nights of troubleshooting.
Next, the model was run, and near-surface air speed, temperature, and other data points of interest were processed (Fig. 7). The 3D model also allowed the project team to review variations throughout the enclosure and to identify vulnerable areas where the enclosure was not receiving beneficial heat and/or airflow. Some results met expectations, such as the cold curtain-wall head located 50 feet from a diffuser. Other results were surprising, such as the checkerboard pattern of heating on the skylight (likely a result of the shades or similar blockages). Through iterations in the CFD analysis, the team was also able to digitally commission, or adjust the settings of, the curtain-wall heaters within their operating parameters to supply enough heat to the glazing while not overheating the space for occupants. Like any mechanical system, the heaters must still be commissioned in the real world, but the set-point temperature in the mechanical design is based on the CFD results.
Bounding the Results
Each thermal model set the exterior boundary layer to the ASHRAE 99.6 percent dry-bulb heating-design temperature (17.9°F for Washington, D.C.) that is used, in part, for sizing HVAC elements in accordance with ASHRAE 90.1. This low exterior temperature threshold is statistically derived from 30 years of weather data for each project location, but the exterior conditions are predicted to be lower than this design temperature for 35 nonconsecutive hours (0.4 percent of the 8,760-hour year) on average.
The selection of the exterior design temperature is partly arbitrary and may be influenced by the project’s risk profile, but 99.6 percent is a good starting point as it aligns with the design assumptions used for sizing the HVAC systems. By varying the exterior conditions above and below the 99.6 percent dry-bulb heating-design temperature, the project team was able to evaluate the sensitivity of the enclosure systems to outdoor conditions.
The team was also able to compare proposed design alterations by iterating between the CFD and THERM models. Due to diminishing returns, eliminating all condensation for 100 percent of the possible outdoor temperatures is almost always impossible for humidified buildings, but the team used the analysis as a tool for the cost-benefit analysis between options. One such iteration demonstrated that stainless-steel IGU spacers would likely result in localized condensation at a greater number of glazing lights compared to IGUs with “warm-edge” spacers. The output—skylight plans annotated with the risky glazing lights—helped the team visualize the extent of risk (Fig. 9).
Fig. 9. NASM, partial skylight plan. The highlighted lights of glass are at risk for condensation near the perimeter of the IGU for 35 nonconsecutive hours per year. Note that the atrium is bisected with only the western half of one atrium shown, and the curtain wall is shown to the right.
Through multiple conversations with the Smithsonian and a wide range of output summaries, the project team was able to define, with a reasonable level of assurance, where the proposed rehabilitation would fall on the spectrum of risk and adjust the solution until an acceptably conservative result was reached. However, condensation will form on interior surfaces of all humidified buildings at some point, whether from outside temperatures dropping below design assumptions, complications with constructed air supply, temporary spikes in the humidity or occupant loads, or other real-world scenarios that computations cannot fully anticipate. The goal is that under most reasonable operating conditions, there is no condensation (or it is managed) and that condensation is sufficiently infrequent and inconsequential to operations. The NASM analysis established confidence that the selected systems can reasonably achieve this goal, but the final proof will come only when the enclosure is constructed and seasoned.
Applications to Other Projects
The Smithsonian Institution, as steward of the artifact collection, has a near-zero tolerance for condensation. However, a large portion of the collection is situated near the glazing systems and is silhouetted against the sky, a cornerstone of HOK’s original design intent. Other buildings will have different conditions and differing levels of what is acceptable.
Unsurprisingly, the CFD analysis described in this article took considerable engineering effort. Every project is different, and many rehabilitation projects may not warrant this level of analysis. However, many historic buildings and most museums require humidification to preserve sensitive artifacts or original finishes. With a careful look at each project’s circumstances, a design team may be comfortable with estimating the interior boundary condition, or the initial range of possible solutions may be sufficient, not needing further refinement. Each project requires experienced design professionals to judge the extent to which intense analysis, such as CFD, is necessary. The following red flags encountered early in a design phase may prompt concern and warrant additional analysis:
• Potential for damage: skylights above a priceless museum collection.
• Extreme exterior weather conditions: the mid-Atlantic NASM environment, while seemingly milder than most, did not raise this red flag, but it still needed careful analysis. A building in Alaska or a mountaintop chapel would certainly need a careful look, while a building in Florida might not.
• Elevated interior moisture: a humidified museum, laboratory, or natatorium inherently pushes the interior air closer to the dew point and demands more of the enclosure system to limit conduction.
• Confidence in the air supply: air that travels long distances or around obstructions between the diffuser and the fenestration will differ from design conditions at the air supply point.
• Unfamiliar systems: vetting a product through calculations is prudent before implementing a cutting-edge curtain wall that claims advanced performance without a proven track record.
Computational analysis helps to inform the project team regarding the likely performance of the actual construction, but calculations alone rely on assumptions and therefore provide no guarantee. Careful engineering judgment is required to identify what variables are unknown and the confidence in the assumptions. Varying the assumed inputs helps to bound the potential range of results, and that range is narrowed with incrementally more accurate analysis methods. An informed and honest discussion with all stakeholders is necessary for owners to decide when an acceptable solution has been reached.
Next Steps for the NASM
For the NASM project, an acceptable solution was reached after a few iterations of the CFD model, adjustments to the mechanical supply, and enhancements to the proposed enclosure detailing. Quinn Evans Architects and their consultants completed the design phase in January 2018. The team constructed and tested a 20-foot-tall performance mock-up at a testing laboratory to validate the computational analysis. The mock-up performance testing successfully passed the specified condensation-resistance criteria and provided the Smithsonian with further confidence to begin construction on the building.
The contractor, Clark-Smooth-Consigli Joint Venture, began mobilization for construction in the last quarter of 2018 and is preparing for a construction phase that will last several years. The work will be phased such that approximately half of the museum will remain operational at any given time. The project will renew one of America’s favorite museums and preserve its collection for generations to come.
is a senior project manager in the Ann Arbor office of Quinn Evans Architects and has over 20 years of experience managing historic-preservation projects. He also provides oversight for complex and highly technical projects. He can be reached at firstname.lastname@example.org.
Brian S. Rose
is a senior member of the Building Technology group at Simpson Gumpertz & Heger Inc. and is experienced with new-design enclosure consulting, rehabilitation, and investigation projects for a variety of architects, general contractors, and owners, including multiple museums along the National Mall. He can be reached at BSRose@sgh.com.