[Updated below: Feb. 27, 2019] I have just received written confirmation from a Senior Staff Architect at the International Code Council (ICC) that the unenclosed stairs in Cornell’s Fine Arts Library atrium are noncompliant with the International Building Code (IBC), and that the roof-top gallery above the atrium is also noncompliant as designed. The 2015 New York State Building Code is derived from the IBC and contains the same Code language cited in the Code opinion by ICC. Of course, this Code opinion has no legal force, but it does indicate that the fire safety problems I have identified are real and need to be addressed. Here is the written opinion in full:
Date: Thursday, February 14, 2019 at 11:17 AM Re: 2015 IBC Sections 404.6, 1005.3.1, 1017.3.1, 1019, 1023.1, 1023.2,
Question 1: Can the path of egress travel to an exit pass through more than one adjacent story in an atrium?
Answer 1: There is nothing saying someone could not follow the stairway in the atrium as a path of egress travel for as many stories as they wanted. However, if the stairway in the atrium is a considered a required exit access stairway, the egress path for the required means of egress (i.e. number of exits off of a floor) can only go down one story till occupants could choose to move to the required exits (1006.3.1, 1017.3.1, 1019). If the stairway in the atrium is considered a required exit stairway, while Section 1023.2 Exception 2 does allow for the atrium to meet the construction requirements for the exit stairway, the atrium must meet all the remaining provisions for an exit stairway in Section 1023, including termination at the exterior, not be used for any other purpose other than exit (e.g. no uses on the ground floor of the atrium) and no paths that go through the atrium to get to the 2nd exit.
Question 2: Can an unenclosed interior exit stairway, as permitted in Section 1023.2 Exception 2, terminate in the middle of an atrium floor that is above the level of exit discharge.
Answer 2: No. Exit stairways must discharge directly to the exterior (1023.1) and cannot stop at an upper floor. There are the options for the stairway to discharge through a lobby or vestibule (1028), however, this lobby cannot be within the stairway/atrium enclosure and it cannot be at other than the level of exit discharge.
Question 3: Does an occupied roof (Group A-2 assembly) above an atrium need to be separated from the atrium with a 1-hour horizontal assembly?
Answer 3: Yes, an atrium must be separated from occupied spaces. While a roof is not a story, it is an occupied space, so it must be separated from the atrium where the floor of the occupied roof is over the atrium. Where an occupied roof floor is around the atrium, a separation would not be required.
Code opinions issued by ICC staff are based on ICC-published codes and do not include local, state or federal codes, policies or amendments. This opinion is based on the information which you have provided. We have made no independent effort to verify the accuracy of this information nor have we conducted a review beyond the scope of your question. This opinion does not imply approval of an equivalency, specific product, specific design, or specific installation and cannot be published in any form implying such approval by the International Code Council. As this opinion is only advisory, the final decision is the responsibility of the designated authority charged with the administration and enforcement of this code.
I hope that this answers your question in full. Please feel free to contact me again if you have any additional questions on this issue.
Kimberly Paarlberg, RA
International Code Council
Codes and Standards, Senior Staff Architect
5332 Woodfield Drive, Carmel, IN
888-422-7233, Ext. 4306
[February 27, 2019 update: I just received a code interpretation from the New York State Division of Building Standards and Codes that agrees in full with the ICC answers shown above (“We reached the same conclusions as did the ICC representative, Kimberly Paarlberg”). I have forwarded their response to the City of Ithaca Director of Code Enforcement and other parties involved in this fiasco. Depending on how they react, I may well need to file a formal complaint.]
Links to all my writings on the Fine Arts Library at Cornell can be found here.
I’ve invited Christine Sheppard, Ph.D. to give a talk at Cornell on Friday, Nov. 2, 2018 at 4:45 pm in 101 West Sibley Hall on the subject of bird-friendly design. The talk is open to the public and free. Continuing education credit (one LU/HSW credit) is available for architects and LEED APs or Green Associates.
Rand Hall at Cornell University (photo and PhotoShopped window and bird by Jonathan Ochshorn, Oct. 2018)
Date, time, and place: Friday, November 2, 2018, 4:45 pm in 101 West Sibley Hall, Cornell University
Abstract: Birds are potent cultural symbols. They play fundamental roles in ecosystems and habitat regeneration and are important natural controls for insects. Hundreds of millions are killed yearly by colliding with glass in the US alone. Birds cannot see glass, striking it as they fly towards reflections of clouds, sky and vegetation or as they approach real habitat seen through glass. Birds collide with glass on structures of every size, from shacks to skyscrapers, in urban, suburban and rural area. Advances in technology are increasing use of glass curtain walls and other large glass features, increasing the rate of mortality.
Until recently, this problem has been almost unrecognized as an issue of sustainability. However, the Green Building Council has responded by adding a Pilot Credit, Reducing Bird Mortality, to the LEED rating system. Toronto, San Francisco, Oakland and the state of Minnesota now mandate bird-friendly construction in some cases and more legislation and voluntary guidelines are pending. Moving into the future it will be increasingly necessary to design structures with impact on birds in mind.
This class explains how to recognize hazards to birds in the built environment. Case studies and a slide show illustrate many currently available strategies for reducing bird mortality and how bird-friendly design can add value to strategies often deployed to control heat and light or promote security. We review use of the LEED credit and important features of legislation. Techniques now in use for evaluating the relative threat level to birds of different materials are described, along with typical results.
Continuing education credits: Available for registered architects/engineers as well as LEED APs and Green Associates.
Short bio: Christine Sheppard earned her B.A. and Ph.D. in Ecology and Evolutionary Biology at Cornell University. Working with Dr. Tom Cade, who used captive breeding to restore the Peregrine Falcon to the eastern US, developed her interest in captive propagation as a tool to save endangered species. This led her to the Wildlife Conservation Society’s Bronx Zoo, where she started as curatorial intern, in 1978, and ended as Curator and Chair of the Ornithology Department. Zoos deal not only with issues of their buildings causing mortality of wild birds. Glass exhibit walls, windows and handrails bring bird collision problems inside and curators have a vested interest in finding ways to make glass safe for birds. Interest in the issue led to Dr. Sheppard to join the board of the Bird-safe Glass Foundation as science advisor, in 2007; she became President in 2017. She is also conducting basic research into quantifying the effectiveness of different materials and patterns in preventing bird collisions. In 2009, she moved to the American Bird Conservancy as Collisions Program Director. She authored both editions of ABC’s publication, Bird-friendly Building Design. She has also created AIA/LEED continuing education classes on Bird-friendly Design. She helped create San Francisco’s Standards for Bird-safe Buildings and has subsequently been involved in creating code and legislation in many different jurisdictions. She led the team that developed USGBC LEED Pilot Credit 55: Reducing Bird Mortality. She was named an Engineering News-Record Top 25 Newsmaker for 2014 because of her work on glass testing and has worked with most major glass manufacturers on design and evaluation of bird-friendly materials.
[Updated Oct. 8, 2018 and Oct. 15, 2018]: [October 8, 2018 update: I just made a video based on the blog post below. Also, I’ve removed two of the six mezzanine “issues” I had initially written about: first, the 200-foot limit for exit access travel within the atrium doesn’t prevent a different path of exit access travel from being compliant, since only one of the two egress paths (the shortest) defines the exit access travel distance; second, while I still think that the area of the so-called mezzanine at the roof level is greater than half of the area of the atrium below, it’s too hard for me to definitively measure such floor area to be absolutely sure. There is the additional issue that the open-air roof gallery is not exactly “sprinklered,” so it’s unclear if it’s area should be 1/3 or 1/2 of the room it is in. In any case, there are still four problems with the roof-top mezzanine assumption being argued by Cornell and the City of Ithaca, any one of which would render the Rand Hall Fine Arts Library scheme noncompliant. These four remaining problems are explained below, and in the video.]
I received an email yesterday from the Director of Code Enforcement for the City of Ithaca Building Division claiming that Cornell’s proposal for a fifth-floor roof gallery—above the Fine Arts Library now being constructed in Rand Hall—was actually a mezzanine within the atrium below. This is important since a 5-story building would be absolutely noncompliant, whereas a 4-story building (with mezzanine)—while still egregious—is supported by numerous Code variances that Cornell has obtained over the years. The Building Division Director wrote: “The roof top gallery must meet the code requirements for a mezzanine otherwise the building is non-compliant. I have discussed this issue with the design team several times and have been assured that it will be sized so that it does meet the code requirements for a mezzanine. The sizing of the roof top area must include the permanently enclosed space at that level and all the open area that can be occupied. I do not believe that the code requires the mezzanine to be open to the atrium because the mezzanine has two exits.”
He then quoted a portion of the 2015 NYS Building Code (based on the 2015 International Building Code, or IBC) which seems to support his argument: “Exception 2. A mezzanine having two or more exits or access to exits is not required to be open to the room in which the mezzanine is located.”
In response, I replied as follows (where all references to the Code can be found here):
You have quoted only part of the mezzanine requirements, which state that a mezzanine need not be open to the room in which the mezzanine is located if it meets certain criteria (having at least 2 exits). However, there are six other criteria for qualifying as a mezzanine that the roof gallery does not meet. If even one of the following six criteria is not met, then the building, as designed, is noncompliant.
1. The definition of a mezzanine in Chapter 2 of the 2015 IBC states that a mezzanine is “an intermediate level or levels between the floor and ceiling of any story and in accordance with Section 505.” Clearly, the open roof gallery is not between a floor and a ceiling of any space, since there is no ceiling above it. Therefore, it is not a mezzanine.
2. The definition of an atrium in Chapter 2 of the 2015 IBC states that an atrium must be “closed at the top.” Therefore, it is impossible to include an open-air roof gallery as part of an atrium, since the atrium, in that case, would not be “closed at the top.”
3. Section 505 states that a mezzanine must be “in” the room for which it is so designated, not “on top of” the room, as is the case with the roof gallery. Specifically, Section 505.2.1 states: “The aggregate area of a mezzanine or mezzanines within a room shall be not greater than one-third [can be increased to one-half with sprinklers] of the floor area of that room or space in which they are located.” [emphasis added]. This section twice reiterates that the mezzanine must be “within” or “in” the room or space. This is not just a figure of speech, but is deliberate and important. Mezzanines are allowed to have certain fire safety benefits, unlike ordinary stories, precisely because they are “in” another room, where occupants of the mezzanine are more aware of any fires that may originate in that room, since the mezzanine has what used to be called a “common atmosphere” with the room in which it is located. Being “in” the room is therefore integral to the very essence of a mezzanine, whether or not the “opening” between the mezzanine and the room it is in has been closed off per the exception; the advantages of being “in” the room still apply. It is improper to ignore that part of the definition, and just look at the requirements for maximum area. If one used only the logic of maximum area, then any space, say a third-floor office, could be considered a mezzanine of a larger second-floor space that happened to be below it, even if there were no logical connections between the two spaces, but only a vertical adjacency.
4. It is also problematic to even consider that an atrium can have a mezzanine within it, since atriums are just “openings connecting two or more stories … which are closed at the top and not defined as a mall. Stories, as used in this definition, do not include balconies within assembly groups or mezzanines that comply with Section 505.” What this definition refers to when it states that mezzanines do not count as stories is not a mezzanine within the atrium (which would make no sense), but rather a mezzanine adjacent to the atrium, i.e., within a story surrounding the atrium opening. Only a conventional room or space can have a mezzanine. An atrium is not a conventional room or space, but rather is defined as an “opening” adjacent to such rooms or spaces. Once “mezzanine” floor area is inserted into an atrium, that floor area is, by definition, no longer an opening between stories, and so cannot be part of the atrium. Rather, it would be part of the adjacent stories that surround the atrium.
[Oct. 8, 2018 update] Issues #5 and #6 are no longer applicable, for the reasons explained in the update note at the top of the blog post. 5. Even if the roof gallery were considered as a mezzanine within the atrium, it still would be noncompliant, since Section 404.9.3 of the 2015 IBC limits the “portion of the total permitted exit access travel distance that occurs within the atrium” to no more than 200 feet. If the gallery is considered as a mezzanine within the atrium, then the exit access travel distance within the atrium, measured from the most remote point of the roof gallery, and following the atrium exit access stair “B” all the way down to the 2nd-floor exit, would be far greater than the 200-foot limit for exit access travel distance “that occurs within the atrium.”
6. The total area of the enclosed and open occupied spaces at the roof level (what is being called a mezzanine) is approximately 2,275 square feet; whereas the floor area of the atrium (measured at the 2nd floor) is approximately 4,335 square feet. Therefore, the so-called mezzanine exceeds the allowable area, based on the atrium floor area.
You correctly state that the building is noncompliant if the roof-top gallery, including all enclosed and occupied open spaces on the roof, does not qualify as a mezzanine. I think it is clear that—for the 6 reasons given above—the roof-top occupied spaces cannot be considered as a mezzanine. Therefore, the building is noncompliant.
Please reconsider your preliminary Code interpretation and let Cornell and their architects know that their notion of a roof-top mezzanine cannot be logically sustained.
[October 19, 2018 update]
On October 4, 2018, I asked the Director of Facilities for Cornell’s College of Architecture, Art, and Planning (copied to many other relevant parties, including the Director of Code Enforcement for the City of Ithaca Building Division, Mike Niechwiadowicz) to “confirm whether the entire roof structure of Rand Hall—including all columns that support the roof, and all beams that brace those columns, and all floor decks that brace the beams that brace the columns—has a 1-hour fire rating? This is important since an atrium, per 2015 IBC Section 404.6, needs to be separated from all adjacent spaces by a 1-hour horizontal assembly and/or fire barrier; as you know, there is an occupied space above the atrium’s roof that would need to be separated from the atrium by 1-hour fire-rated construction.”
On October 15, 2018, I got the following reply: “All the supporting elements, columns, braces, and floors, have a one hour rating. There was a variance granted by the state to not fireproof the horizontal roof steel. In regards to the roof, Per Tim D. from GHD, it is outside of the building and not a space within the building and therefore does not need a one hour separation. Also, the penthouse is a mezzanine to the atrium below and therefore does not need to be separated.”
On October 16, 2018, I responded as follows: “Thanks for the detailed answers. The occupied space above the atrium remains extremely problematic. Cornell’s 2016 Code variance was granted on the basis of an atrium roof with no occupied spaces above it, and so the variance cannot be applied to the current scheme. Section 404.6 of the 2015 Building Code specifically requires that ‘atrium spaces shall be separated from adjacent spaces by a 1-hour fire barrier … or a horizontal assembly…’ The occupied roof-top gallery is clearly an ‘adjacent space’ and so must be separated from the atrium with a 1-hour horizontal assembly. Cornell is claiming—improperly—that the roof-top occupancy is actually a mezzanine within the atrium below; this is a transparent attempt to get around the prohibition of a fifth story and the prohibition of an occupied space above a non-fire-rated atrium roof (horizontal assembly). But what Cornell is calling a mezzanine violates both the definition of mezzanine and the definition of atrium in Chapter 2 of the Code, not to mention the requirement that a mezzanine be ‘within’ a space or room [see detailed explanation below]. And as Mike has written to me previously, if the roof-top gallery space is not a mezzanine—which it clearly isn’t— the whole scheme is noncompliant.
“Please ask the design team to reconsider their plans for a fifth-floor above the atrium. It is likely that a formal complaint will be filed with the City of Ithaca and, if necessary, with the New York State Division of Code Enforcement and Administration if construction goes forward as currently planned—unless you can explain to me why none of the four reasons I have listed below applies to the current proposal. Please respond to my specific questions about the mezzanine designation, i.e., why you believe that every one of the four reasons I have cited below is incorrect. I’ll email again in a couple of weeks before initiating the complaint process, hoping that common sense will prevail during that time period.”
Then, on October 19, 2018, I wrote a follow-up email to the the Facilities Director (and other relevant parties): “I just re-read your email dated Oct. 15, 2018, and now am totally confused about Cornell’s intentions with respect to the roof gallery. Mike Niechwiadowicz wrote that the building permit for the Ho Fine Arts Library was granted on the basis of the roof gallery being treated as a mezzanine within the atrium. Your quote from Tim D. from GHD, on the other hand, implies that the roof gallery is not only not a mezzanine, but is not even a relevant consideration, being ‘outside’ of the building. Let me quote from Mike’s email to me, dated Oct. 4, 2018: ‘The roof top gallery must meet the code requirements for a mezzanine otherwise the building is non-compliant. I have discussed this issue with the design team several times and have been assured that it will be sized so that it does meet the code requirements for a mezzanine. The sizing of the roof top area must include the permanently enclosed space at that level and all the open area that can be occupied.’ [emphasis added]
“So please clarify how the roof gallery is being considered from a fire safety (Building Code) standpoint. On the one hand, if it is being considered as a mezzanine, then my comments in the prior email stand. On the other hand, if it is being simply ignored, please consider the following explanation as to why that latter standpoint is equally flawed.
“You wrote: ‘In regards to the roof, Per Tim D. from GHD, it is outside of the building and not a space within the building and therefore does not need a one hour separation.’ There are two claims here, amounting to the same thing: (1) The roof gallery is outside the building; and (2) the roof gallery is not a space within the building. From these two claims, the conclusion is drawn that the atrium does not need to be separated from the occupied roof gallery by a 1-hour fire-rated horizontal assembly.
“But the Code, and common sense, do not support this conclusion. First, from the Code: Section 404.6 states that ‘Atrium spaces shall be separated from adjacent spaces by a 1-hour fire barrier… or a horizontal assembly…’ Any adjacent (occupied) space—not only an adjacent ‘story’—needs to be separated from the atrium. But is the roof gallery part of the building? The 2015 Building Code specifically identifies ‘occupied roofs'” as being part of the building since, per Section 1006.3, ‘The means of egress system serving any story or occupied roof shall be provided with the number of exits or access to exits based on the aggregate occupant load served in accordance with this section’; and the means of egress is defined as ‘a continuous and unobstructed path of vertical and horizontal egress travel from any occupied portion of a building or structure to a public way…’ Since means of egress systems are, by definition, in buildings; and since occupied roofs must have means of egress, it follows that occupied roofs are part of buildings. One could put this in the form of a logical syllogism:
Major premise: Means of egress systems are in buildings (Definition, Chapter 2).
Minor premise: Occupied roofs must have means of egress (Section 1006.3)
Conclusion: Occupied roofs are part of buildings.
“From all this, it follows that the occupied roof gallery is not only a space adjacent to the atrium (which is really the only relevant criterion), but is also part of the building, contrary to the unsupported assertion made by Cornell’s Code consultant. The requirement in Section 404.6 is for separation of the atrium from ‘adjacent spaces,’ not only from adjacent stories. And an occupied roof is both a ‘space’ and a space that is part of the building.
“Second, based on common sense: Hundreds of assembly occupants may gather on the roof gallery above the atrium and their safety is threatened by placing them above an atrium without any fire-rated separation. This is what the 2015 ‘Code Commentary’ says about the need for fire-rated enclosure: ‘One of the basic premises of atrium requirements is that an engineered smoke control system combined with an automatic fire sprinkler system that is properly supervised provide an adequate alternative to the fire-resistance rating of a shaft enclosure. It is also recognized that some form of boundary is required to assist the smoke control system in containing smoke to just the atrium area. The basic requirement, therefore, is that the atrium space be separated from adjacent areas by fire barriers and horizontal assemblies having a fire-resistive rating of at least 1 hour.’ [emphasis added] The current scheme not only violates this basic requirement for separation, but also violates the explicit logic of a smoke control system (i.e., ‘in containing smoke to just the atrium area’) by venting atrium smoke—not harmlessly into the outside air—but rather right at the roof level, immediately adjacent to the occupied roof gallery itself! I’ve noted previously that the exit access stairway will also serve as a smoke vent system when its door is opened, thereby making egress from the roof gallery even more precarious. If your intention was to design a building that puts occupants at maximum risk, you have succeeded admirably.
“Mike Niechwiadowicz has written explicitly that ‘the roof top gallery must meet the code requirements for a mezzanine otherwise the building is non-compliant.’ Since the roof gallery is not a mezzanine, the building is non-compliant. All other Code interpretations seeking to justify this dangerous proposal are equally flawed.
“I’m still eager to hear back from you before I initiate a formal complaint.”
Without the Code variances that have been granted, Cornell’s Rand Hall Fine Arts Library, as a Type V-B building with A-3 occupancy, would be limited to two stories; as it is, the current proposal (including the anticipated roof-top pavilions) is for a five-story building. The proposal, without the variances, would also greatly exceed the allowable per-floor area permitted by the Code, owing to its attachment, without any fire wall separation, to Milstein and Sibley Halls.
Not only does the current proposal exceed the height limits granted by prior variances, but the variances themselves do not even appear to apply to the current proposal at all, since the prior variances were for a 3- or 4-story library, not a 5-story library. This is a classic bait-and-switch move, entirely unjustified by any fire science rationale.
I have already written extensively about the flawed logic underlying these Code variances and also underlying the Fine Arts Library proposals themselves, so I won’t repeat those arguments.
What I will comment upon are the new (for me) Code analysis drawings supplied by the architects that are now available in the Architecture, Art, and Planning Dean’s Office, based upon which a building permit was granted.
There are at least five serious Code violations that I have noticed in these new drawings:
1. The art gallery will become a fifth story as soon as even temporary pavilions are constructed. As such, these pavilions would not be permitted under the Code, even given the already egregious assumptions made about Code compliance for the proposed 4-story building. Because these pavilions are not shown in the drawings, any such construction will require a new building permit, even if the pavilions are only temporary. The pavilions will not be compliant if the Code is followed, since an A-3 assembly space (art gallery with pavilions) is not permitted above the third floor of a sprinklered Type II-B building, and certainly not on the fifth floor.
2. The entire egress strategy for the building, based on having unenclosed interior exit stairs in the atrium, is flawed, to the extent that it is based on 2015 IBC Section 1023.2, exception 2. This section allows interior exit stairways to forgo construction of otherwise required fire-barrier enclosures if they are within an atrium enclosed per 2015 IBC Section 404.6.
However, 2015 IBC Section 404.6 requires the atrium to be separated from adjacent spaces (not only adjacent “stories” but any adjacent “spaces”) by a 1-hour horizontal assembly (and/or fire barrier). There is an occupied space above the atrium (the roof-top art gallery) and yet a 1-hour fire-rated horizontal assembly is not provided that would separate the atrium from the art gallery above. In fact, there are even roof hatches designed to pop open in the event of fire, between the art gallery and the atrium, a strategy which completely violates both the spirit and letter of this Code requirement. Since the requirements of Section 404.6 are not met, having unenclosed interior exit stairways in the atrium is not permitted (and, of course, the atrium itself is noncompliant).
3. Along those lines, the architect’s Code drawings incorrectly state that the second-floor slab provides a non-required 1-hour horizontal fire barrier as per variance 2015-0432. First, there is no such thing as a horizontal fire barrier (it would be called a horizontal assembly). Second, this 1-hour fire rating is not discretionary, but is absolutely required, in order to separate the first floor occupancy from the atrium above. It does not provide “extra” fire safety beyond Code requirements, as is implied.
4. The atrium is improperly labeled; it is drawn as if it consisted of the entire space within the exterior walls of Rand Hall, above the first floor and excluding the “bump” on the southern side. In fact, the atrium consists only of an “opening connecting two or more stories” and not the stories themselves. Therefore, all the stack floor areas are not part of the atrium itself, but are simply stories, adjacent to the atrium, for which atrium smoke calculations must be made. This means that the justification for the roof-top mezzanine is flawed, since it is based on a calculation that its area is no greater than 1/2 the area of the “atrium” it is in. But the mezzanine as drawn is not “in” the atrium; only the stair to the mezzanine is conceivably in the atrium, while the rest of the mezzanine is directly over the stack floors and the mechanical room serving Milstein Hall. It therefore does not qualify as a mezzanine, but would be considered a fifth story.
5. Glazed openings in Stairway A appear to be too close to glazed openings in the southern wall of Rand Hall (unless this glass has a sufficient fire-rating, which is not indicated in the drawings I have examined). Per 2015 IBC Section 1023.7: “Where nonrated walls or unprotected openings enclose the exterior of the stairway or ramps and the walls or openings are exposed by other parts of the building at an angle of less than 180 degrees, the building exterior walls within 10 feet horizontally of a nonrated wall or unprotected opening shall have a fire-resistive rating of not less than 1 hour. Openings within such exterior walls shall be protected by opening protectives having a fire protection rating of not less than 3/4 hour. This construction shall extend vertically from the ground to a point 10 feet above the topmost landing of the stairway or ramp, or to the roof line, whichever is lower.”
[Update Aug 26, 2018: Correction made to the statement in the second paragraph contending that prior Code variances were for a 3-story building. In fact, Cornell’s third, 2016, variance was for a 4-story building, although its logic was entirely flawed. Since the current proposal, purported to be for a 4-story building, is actually for a 5-story building, the argument I am making—that prior variances do not apply to the current proposal—still stands.]
[Update: Oct. 23, 2018] Latest image showing highly-reflective glass—just installed for the Fine Arts Library project in Rand Hall at Cornell University—inviting bird collisions:
Photo by Jonathan Ochshorn, October 23, 2018
According to the Rand Hall Fine Arts Library’s June 18 to June 29, 2018 construction activity highlights and project update: “A full mock up window has been installed on the northern facade. It can be viewed from the sidewalk on University Avenue.”
Rand Hall’s highly reflective trial window appears as an opening to sky and trees, inviting bird collisions.
Based on the trial window, shown in the image above, it seems likely that these highly-reflective windows will be problematic for birds: the windows appear (to the birds) as openings to trees and sky, thereby inviting collisions. There are better bird-safe glazing techniques that are widely known, for example using patterns on the glass that mimic the multi-pane steel-framed “factory” windows that were recently removed. For more information, see the American Bird Conservancy website.
[Update: July 9, 2018] Here’s another image of the same highly-reflective window.
Another image of Rand Hall’s highly-reflective glazing.
[Updated Oct. 16, 2018: The image below shows how it looked on Oct. 8, 2018, with a few more windows installed.]
View on Oct. 8, 2018 (photo by Jonathan Ochshorn)
Additional writings and blog posts on the Rand Hall Fine Arts Library project are linked here.
I just read an article in the May 2018 ASHRAE JOURNAL about Discover Elementary School in Arlington, Va. designed by VMDO Architects; it is currently the largest zero energy elementary school built in the US. It’s interesting to compare its size, cost, and “sustainable” qualities with Cornell’s LEED gold-certified Milstein Hall, the architecture building designed by OMA and Rem Koolhaas. Both are two-story academic buildings.
Size (sq. ft.)
Cost ($ millions)
OFF-SITE Energy USED
So, in a nutshell, Milstein Hall is half the size, twice the cost,* and barely meets minimum ASHRAE energy standards,** while the Arlington school actually returns energy to the grid.
Aerial views of the Discovery Elementary School in Arlington, VA (top) and Milstein Hall at Cornell University (bottom)
*To be clear, the Arlington school is half the cost and twice the size; in other words, its cost per square foot is approximately 25% that of Milstein Hall.
**Cornell’s stated goal (see my video) since 2008 is for new buildings “to use 30% less energy than current energy standards and strive towards 50% less energy.” Milstein Hall’s energy model projects a whopping 2% energy reduction compared to this minimum standard.
Milstein Hall, the architecture facility at Cornell designed by OMA and Rem Koolhaas, has a “growing” handrail problem: the vines intended to cover the exterior egress stairway have penetrated the metal grid on which they’re supported, and have rendered the handrail noncompliant.
Vines intended to cover the exterior egress stair for Milstein Hall have penetrated inside the structure and rendered the handrail noncompliant (photo by J. Ochshorn, May 28, 2018)
Handrails must be continuous; the NYS Building Code has this advice: “Handrail gripping surfaces shall be continuous, without interruption by newel posts or other obstructions.”
I took a look at several documents that were emailed to me, at my request, on March 21, 2018: the “Building Envelope and Structural Conditions Assessment” for Sibley Hall at Cornell University (prepared by Ryan-Biggs Associates in May 2009) as well as four pages of structural plans and details for the third-floor renovation of E. Sibley Hall (prepared by Robert Silman Associates in March 2015). It seems to me that the structural analysis and mitigation measures proposed for the E. Sibley roof do not reflect the actual behavior of the roof members and do not explain observed displacements, misalignments, and rotations of columns, attic joists, girders, and roof rafters. I am not a structural engineer, but I think that the discrepancies between the Assessment report and structural analysis, on the one hand, and the observed conditions in E. SIbley Hall, on the other hand, warrant further investigation.
First, there is no mention anywhere, in any of the documents, of second-floor and especially third-floor column misalignment, attic joist displacement and rotation, and girder bowing in E. SIbley Hall. Either the engineers simply missed these important structural conditions in their site investigations, or the misalignments, displacements, rotations, and so on occurred after their reports and drawings were prepared — for example, they might have occurred as a result of construction during the third-floor E. Sibley renovation itself. It is also possible that these conditions occurred, or were exacerbated, as a result of the 2009 excavation for Milstein Hall and the underpinning of E. Sibley foundations, or it is possible that they occurred as a result of some original skylight installation decades ago. It would be important to know whether these structural problems (not mentioned in any of the documents) existed before the E. Sibley renovation and the Milstein Hall excavation took place. I was a student here, spending my entire freshman year on the third floor of E. Sibley Hall in 1970–1971; I don’t recall any column misalignment at that time, but, of course, I may not have noticed. I also can’t find any early photos of the interior of E. Sibley Hall that might allow us to determine when this misalignment happened. The two photos showing the interior of E. Sibley Hall provided in the Assessment report are inconclusive, but perhaps the engineers took other photos (not in their Assessment report) that might show the condition of the columns in 2008.
Second, the Assessment report repeatedly emphasizes and notes an outward movement of the exterior masonry walls in E. Sibley Hall. Such an outward movement does not explain (and in fact appears to contradict) the observed inward movement of the interior wooden columns on the third floor of E. Sibley Hall where the skylight opening occurs. This alleged outward movement of the masonry walls is not visible by casual observation, and no evidence is produced to support the contention that the masonry walls have, in fact, displaced laterally. Nor is it clear, even if there has been some outward movement of the masonry walls, that this movement explains the distress in the roof structure.
Third, there is no analysis in the Assessment report or in Silman’s structural drawings of the actual structural behavior of the wooden roof elements in E. SIbley Hall, behavior that is alleged to have caused this outward “thrust” on the exterior masonry walls. In fact, the behavior of this structure is rather complex, since it contains several internal “hinges,” especially at the intersection of the lower and upper sloping members that create the Mansard form. Such hinge conditions could explain why there may well have been both an outward and an inward thrust on the exterior walls (outward) and the interior columns (inward) respectively — especially when attic joists under the skylights were removed. Silman’s structural drawings do not even represent the lower part of the Mansard structure in their “Existing Attic Framing Configuration” section (Figure 1). There is no way to fully understand the structural behavior of E. Sibley’s roof framing without including these Mansard elements in the sections and in the analysis.
Fourth, this mischaracterization of the behavior of the roof apparently led to the conclusion that the two interior lines of girders above the wooden columns on the third floor of E. Sibley Hall should be connected with cables or rods (in tension) to protect against the alleged outward thrust of the roof elements. These tension cables are clearly useless (they are slack) and, in fact, will do nothing to protect against a mode of failure in which the girders above the two lines of inner columns, already leaning towards each other, move further inward under the roof loads (see Figure 4). Silman’s “Design Intent Narrative” is based entirely on the premise of adding tension rods between the third-floor girders to “resolve horizontal spreading force.” There is no mention of the observed fact that, rather than “spreading,” the third-floor columns, along with the girders they support, are actually leaning inward.
Fifth, there appears to be a much smaller misalignment of several second-floor columns in E. Sibley Hall, but in the opposite direction of the third floor columns. In other words, some of the second-floor columns appear to be leaning outward, while the third-floor columns appear to be leaning inward. This could be the result of some outward spreading of the exterior masonry walls. However, such spreading, to the extent that it has occurred, seems to be relatively insignificant in terms of explaining the behavior of the wooden roof structure. There is no obvious evidence of outward movement of the exterior walls from inside E. Sibley Hall. On the third floor, the top edge of the masonry wall appears perfectly straight, whereas the girders over the interior column lines are remarkably out of alignment (see Figure 2 and Figure 3).
Figure 2. These images show the misalignment of third-floor columns (left) and the bowing of the steel girders that they support (right). Photos by J. Ochshorn, April 2018.
Sixth, these column/girder misalignments appear to happen only where the skylights in E. Sibley Hall occur, presumably because either attic joists were removed, or because the structure was not adequately braced during the renovation of the third floor when the older skylight frame was removed. In other words, my assumption is that these former structural elements (i.e., either the attic joists or skylight mullions in the plane of the roof) allowed the inward-acting tendencies (thrusts) of the north and south mansard sections to counteract each other in compression and thereby provide more than 100 years of equilibrium.
Seventh, Silman’s section notes that the “existing rafters [are] pulling from ridge board.” This observation actually contradicts the premise of their structural design — that the roof structure is thrusting outwards. If that premise were true, one would expect that the rafters would be in compression, pushing against the ridge board, rather than in tension, pulling away from the ridge board. This separation of the rafters from the ridge board is better explained by a rotation of the rafter-attic joist-knee wall (considered as a rigid free-body) over the girder on which it is “balanced” — facilitated by its unstable connection (hinge) to the lower inclined Mansard element. This explanation is much more plausible than the “spreading” explanation, since it is consistent with the observed rotation of these wooden roof elements and with the inward displacement of the supporting girder. I measured the distance from the bottom of the attic joists to the finished floor and found that the attic joists were indeed significantly lower where they connect with the inclined Mansard elements, but only at the locations where the columns were most displaced. As would be expected, these locations of maximum column displacement and maximum attic joist rotation occur toward the middle of the north-south bracing walls, as shown schematically in Figure 3: Attic joist “A” (at the bracing wall or frame) is horizontal, with no apparent rotation, whereas attic joist “B” shows significant rotation, with its northern end 1.5 inches lower than its southern end. This rotation, made possible by the “hinged” connection to the inclined Mansard supports, is easier to visualize by examining the section in Figure 4.
Figure 3. This is an annotated, revised, and schematic version of Silman’s “Attic Floor Existing Framing” plan, showing not only how columns and girders are displaced between the north-south bracing walls, but how the attic joist-rafter elements, considered as rigid free-bodies, have rotated and displaced laterally. Locations of maximum column displacement and maximum attic joist rotation occur toward the middle of the north-south bracing walls: Attic joist “A” (at the bracing wall or frame) is horizontal, with no apparent rotation, whereas attic joist “B” shows significant rotation, with its northern end 1.5 inches lower than its southern end. This rotation, made possible by the “hinged” connection to the inclined Mansard supports, is easier to visualize by examining the section in Figure 4. (Annotations and revisions by J. Ochshorn, April 2018)
Eighth, the inward bowing of the steel girders over the misaligned columns on the third floor of E. SIbley (Figures 2 and 3) indicates that these girders — designed to transfer the gravity loads of the roof structure to the columns — are now also acting as parabolic tension chains, resisting the further inward movement of the columns and attic joists (Figure 3). Clearly, these girders were not intended to act in this manner, and there has been no mention, in either the Assessment report or in Silman’s structural drawings, of this potentially dangerous condition. If these girders fail in tension as a result of the inward-acting lateral forces being applied to them by the rafters and attic joists of the roof structure, it is possible that a total failure of the roof structure could follow. Figure 4 shows the observed and alleged movement of columns and walls in E. Sibley Hall. This movement is consistent with a structural model of the Mansard’s wooden elements that is explicitly unstable (i.e., creating what is called a “mechanism”) because it includes an internal hinge at the top of the inclined Mansard supporting wall, making the structure underconstrained internally. It is possible that the third-floor steel girders, having been transformed into parabolic tension chains, are the only things preventing a structural collapse.
Ninth, if the tension rods were ever tightened to reduce their sag, this would have made the condition even worse.
Figure 4. This is an annotated, revised, and schematic version of the section through E. Sibley Hall provided in the Assessment Report prepared by Ryan-Biggs Associates in May 2009. The black lines show an idealized condition, with no displacement of columns, walls, or girders; the red lines and tones show the current condition (not to scale), with a small outward displacement of the exterior masonry wall (mentioned in the Assessment report, but not visible to the unaided eye), a large inward displacement of the interior wooden columns and their steel girders, and a substantial rotation and displacement of the wooden attic joists and rafters that are supported by the displaced columns/girders on one side, and by the inclined Mansard structure on the other side; and the blue dotted lines show a possible failure mechanism. The roof structure appears to be unstable in its current configuration. (Annotations and revisions by J. Ochshorn, April 2018)
I want to make it clear that I am not suggesting that the E. Sibley Hall roof structure either is, or is not, in danger of collapse. I am not competent to make such a determination. My concern is that neither the Assessment report nor the Silman structural analysis (1) acknowledges the complex nature of this potentially unstable roof structure, (2) notes the actual displaced, misaligned, and rotated condition of the third-floor structural columns, girders, attic joists, and rafters, (3) discusses the structural geometries and forces that would explain these observed conditions, and (4) proposes a remediation strategy that follows logically from an adequate structural analysis.
I strongly believe that Cornell should immediately engage the services of a consulting engineer — preferably an office that was not involved in producing the Assessment report or the structural drawings for the E. Sibley renovation — in order to assess the conditions in E. Sibley Hall and to make recommendations for remediation.
As I mentioned in an update to my previous blog post, the Rand Hall Fine Arts Library project at Cornell is still under plan review and a building permit has not yet been issued. [Updated Feb. 7, 2018. But you would never know that a building permit has not been issued from this headline in the online AAP News from Jan. 25, 2018: “Construction Begins on Mui Ho Fine Arts Library.” In fact, only demolition and site preparation permits have been issued; the application for an actual building permit is still under plan review by the City of Ithaca Building Division.] One of the points of contention is the status of the unenclosed exit access stairway (Stair “B”) in the atrium that I discussed in detail in that same previous blog post. I have just learned that a third Code section is being considered in relation to Stair “B,” the idea being that an unenclosed exit access stair terminating above the level of exit discharge can actually be designated as an enclosed “interior exit stair”—even though it is not actually enclosed. This latest argument is based on Section 1023.2 (Exception #2) in the 2015 Building Code of NYS, which states that the requirement that “Enclosures for interior exit access stairways and ramps shall be constructed as fire barriers… or horizontal assemblies…, or both…” (with fire-resistive ratings of 1 or 2 hours) need not be applied within atriums.
There are at two reasons that this latest attempt to salvage an unsafe and noncompliant design should be rejected.
First, the definition of “interior exit stairway” in Chapter 2 of the 2015 Building Code of NYS states that such a stair “provides for a protected path of egress travel.” This means it must be enclosed. That it must be enclosed is reiterated in Section 1023.1 which states unambiguously that an “Interior exit stairways shall be enclosed and lead directly to the exterior of the building or shall be extended to the exterior of the building with an exit passageway.” Neither of these two Code passages are affected by Exception 2 in Section 1023.2. That exception only affects the required level of protection, and does not state that such stairs can be unenclosed. In other words, an enclosure for an interior exit stair in an atrium need not be constructed as a fire barrier with a 1- or 2-hour fire-resistance rating, but, per Section 1023.1 (which isn’t affected by this exception), it still needs to be enclosed. This enclosure is particularly important for smoke control in the atrium space, even without a fire-resistance rating.
Second, even if the first argument is not accepted, calling this exit access stairway an interior exit stairway is still problematic. The reason is that, per Section 1023.1, stair “B” does not lead “directly to the exterior of the building [nor has it been] extended to the exterior of the building with an exit passageway conforming to the requirements of Section 1024, except as permitted in Section 1028.1.” A proposed designation of Stair “B” as an interior exit stair would not meet the criteria in Section 1028.1 of discharging through areas on the level of exit discharge, since it terminates on the second floor. Thus, an exit passageway connecting the base of this proposed (unenclosed) interior exit stair to the enclosed interior exit stair on the second floor must be provided, and that exit passageway must be enclosed. According to its Chapter 2 definition, such an exit passageway must be “separated from other interior spaces of a building or structure by fire-resistance-rated construction and opening protectives, and [it must provide] for a protected path of egress travel in a horizontal direction to an exit or to the exit discharge.” The Code does not permit an interior exit stair to discharge on a second floor (rather it must terminate on the level of exit discharge), a prohibition that is reiterated in Atrium Section 404.10, which only permits “50 percent of interior exit stairways … to egress through an atrium on the level of exit discharge in accordance with Section 1028.”
So, we’re left with Sections 1006.3 and 1019.3 (see previous blog post), and I’ve already commented on the likely outcome based on the consideration of those two remaining code sections.
See this web page for links to, and summaries of, all my writings on the Fine Arts Library proposal.
Cornell’s Fine Arts Library in Rand Hall, currently under construction, is unsafe and seriously noncompliant. Rather than address fire safety concerns and design a safe library, Cornell has repeatedly gone to the New York State Board of Review to request waivers of building code requirements. Yet in spite of being granted code variances on these several occasions, Cornell’s final proposal—designed by Wolfgang Tschapeller in collaboration with Architect STV—is still deficient with respect to fire safety provisions that were not addressed in these prior variance requests. I had a chance to quickly look through the working drawings for the library that were made available in the Dean’s Office on January 19, 2018 (“Conformed set” dated 12/15/17), and I have outlined several serious fire safety violations below. Of course, it’s possible that I missed some exculpatory material, but I thought it would be important to bring these issues to the attention of Cornell and the Ithaca Building Department in order to determine if any remedial action needs to take place. Given the seriousness of these potential violations, it is my view that all construction work on the project should stop immediately until the plans are reconciled with fire safety provisions in the building code. [Updated Jan, 31, 2018: I’ve just learned that only interior demolition and site preparation have been approved by the City of Ithaca; the project itself is still under plan review and a building permit has not yet been issued.]
1. Noncompliant exit access stairway
The open exit access stairway serving the various library stack levels and roof pavilion (Stair “B” per add-alt. #6, as shown schematically in Figure 1) violates Section 1006.3 of the 2015 New York State Building Code. This code section, which appears for the first time in the 2015 IBC, requires that the “path of egress travel to an exit shall not pass through more than one adjacent story.” Since exit access stairway “B” passes through more than one story before finding an enclosed exit on the second floor, it is in violation of this section. The 2015 IBC “Commentary” confirms this interpretation.
Fig. 1. An early architect’s rendering of proposed exit access stairway “B” in Rand Hall showing that an occupant on the 5th floor (roof pavilion) must pass through two adjacent stories (i.e., the 4th and 3rd floors) in order to find an enclosed exit on the 2nd floor (image crudely edited to replace the extension of this exit access stairway to the “lantern” level, which no longer exists, with a schematic representation of the open bulkhead leading to the roof pavilion at the 5th floor).
Cornell’s code consultant, in Exhibit A of Cornell’s 2016 Variance request, stated incorrectly that such an open exit access stair is allowed by Sections 1019.3, exception 5, and 404.9.3 in the 2015 NYS Building Code because “Chapter 4 continues to contain provisions which supersedes [sic] other provisions in the Code.” Nothing could be further from the truth. The first section that he cites (1019.3, exception 5) requires that exit access stairways be enclosed if they are not in an atrium. It does not say that exit access stairways in atriums can pass through an unlimited number of stories. The second section that he cites (404.9.3) limits exit access travel distance in an atrium to 200 feet, and is not relevant to the question at hand. On the other hand, Section 1006.3, which Cornell’s Code consultant doesn’t mention in his analysis and which he seems to be unaware of, prohibits such stairs when they “pass through more than one adjacent story.” These code requirements are documented in Figure 2. According to my conversation with Technical Code Experts at the International Code Council (the organization that creates the International Building Code from which New York State Codes derive), this latter Section 1006.3 is specific, clear, and unambiguous, and places limits on the application of Section 1019.3.
Section 1019.3 (which says that exit access stairs must be enclosed if they don’t meet various conditions enumerated in that section) is not in conflict with Section 1006.3 (which says that exit access stairs cannot pass through more than one story before finding an exit). The limit placed on the number of stories that an exit access stair can pass through in Section 1006.3 is perfectly consistent with the allowance for an unenclosed exit access stair in an atrium per Section 1019.3. In other words, Section 1006.3 does not prevent exit access stairways from being unenclosed in atriums, but simply places a limit on the number of adjacent stories that such a stairway can pass through.
Fig. 2. IBC Section 1019.3 (top) allows unenclosed exit access stairways in atriums; but IBC Section 1006.3 (bottom) limits the number of adjacent stories that such a stair can pass through.
It is not unusual to find two sections in the building code where one section establishes limits not found in the other. For example, Section 506 (specifically Table 506.2) allows non-sprinklered (NS) Type IA, IB, IIA, and IIIA Group A-3 occupancies such as libraries to have floor areas in excess of 12,000 sq.ft. However, Section 903.2.1.3 requires all Group A-3 occupancies with fire areas exceeding 12,000 sq.ft. to have sprinklers. So, in this hypothetical example, if you wanted to place a non-sprinklered, Type IIA library in a building with a floor area of 15,000 sq.ft., and only looked at the allowable area factors in Table 506.2, you might (mistakenly) conclude that it was code-compliant because Table 506.2 says nothing about fire area limits. It is only by checking Section 903.2.1.3 that you would realize that this building floor, defined as a fire area, would either need to be sprinklered or subdivided into fire areas no greater than 12,000 sq.ft. This latter section is not in conflict with the former section, but merely places limits on its applicability. The Group A-3 library would need to comply with both sections, and designers (or code consultants) cannot simply ignore one of the sections because they would prefer not to acknowledge the limits it imposes. Code requirements relevant to this hypothetical example are documented in Figure 3.
Figure 3. IBC Table 506.2 (top) limits the allowable area for non-sprinklered Type IIA libraries (Group A-3) to 15,500 sq.ft; but IBC Section 903.2.1.3 (bottom) limits Group A-3 fire areas to 12,000 sq.ft. Taken together, Section 903.2.1.3 places a limit on the application of Table 506.2, but does not contradict it.
The two code sections concerning exit access stairways work in exactly the same way as do the two code sections cited in the hypothetical example just given. In the current case, the first code section allows unenclosed exit access stairways in atriums, but says nothing about limitations that may be imposed because of the number of stories that the stairway passes through. Those limitations appear in the second section, 1006.3. Taken together, the two sections 1) allow unenclosed exit access stairways in atriums, and 2) limit the number of stories that such exit access stairways can pass through to a maximum of one adjacent story.
So there is no conflict. But even if a conflict is alleged between the two exit access sections cited above (i.e., 1006.3 and 1019.3), the building code has a remedy, and the outcome is still the same. Section 102.1 of the 2015 IBC states: “Where there is a conflict between a general requirement and a specific requirement, the specific requirement shall be applicable. Where, in any specific case, different sections of this code specify different materials, methods of construction or other requirements, the most restrictive shall govern.” Section 1019.3 does not specifically permit exit access stairways in atriums to pass through unlimited numbers of stories. It just says that if such a stair is not in an atrium (or any of the other seven conditions listed), then it must be enclosed in a shaft enclosure. It is mute on the question of whether exit access stairways in atriums can pass through more than one adjacent story; whereas Section 1006.3 specifically prohibits such a stair from passing through more than one adjacent story. The open stairs are therefore unambiguously noncompliant under the 2015 NY State Building Code because they pass through more than one adjacent story before finding an exit.
[Updated Jan. 23, 2018: I just had a follow-up phone conversation with James (Doug) Connell, P.E., Team Leader, A & E Services, International Code Council (ICC), Inc., Birmingham District Office. This office within the ICC provides expert analysis of the International Building Code to members of the ICC (I am a member). I had emailed him a schematic representation of the Fine Arts Library exit access stairway in Rand Hall showing an open stair from the 4th floor to the 2nd floor. He emphasized that even that exit access stair (i.e., without its proposed extension to the roof deck per Add-alt #6) was noncompliant with IBC Section 1006.3, since it “passes through” more than one adjacent story. In other words, an open exit access stair originating on the 4th floor must find an (enclosed) exit on the 3rd floor; otherwise, it is passing through more than one adjacent story and is therefore noncompliant. He made it clear that the stair’s location in an atrium (per Section 1019.3) has no bearing on the limitations imposed by Section 1006.3.]
2. Add-alt #6 requires 1-hour horizontal separation from occupied roof
Under Add-Alternate #6, the atrium space needs to be separated from the occupied roof pavilion and occupied roof deck with 1-hour fire-resistance rated horizontal assemblies and fire barriers. In addition, all structural steel supporting the roof deck, including the deck itself, roof girders, and all steel columns supporting the girders, need to have a one-hour fire-resistive rating. I noticed that these structural elements in the atrium were specifically excluded from having any fire-resistance rating on working drawing number G-003, and I’m not sure if Add-Alt #6 included provision for this additional fire-proofing. It is required if the roof above the atrium is occupied, per IBC Sec. 404.6 and Sec. 711.2.3.
3. Smoke control conflict (Add-alt #6)
Exit access stairway “B” is within the atrium and extends up to the bulkhead (pavilion) on the roof in Add-Alt #6. It does not seem consistent with atrium smoke control requirements. Not only is it at the highest point in the atrium, where hot smoke naturally migrates, but it also contains an exit door which, when opened along with the exit door from the pavilion area itself, would draw hot smoke from a fire through the exit door at roof level, precisely in the path of occupants attempting to exit through that stairway.
4. Headroom question on Stair A to occupied roof (Add-alt #6)
Section 8 on Drawing Number 109 (Add-alt #6) does not clearly show a roof over the interior exit stairway, and also shows what looks like a brick obstruction drawn halfway across the door leading to the 4th-floor walkway to the stacks. It is hard to see how there would be sufficient headroom in the portion of interior exit stairway “A” leading from the enclosed part at the 4th floor level to the exterior pit where Stair “A” continues up to the occupied roof. I can’t find any drawings which provide definitive details, but unless the roof over Stair “A” is raised considerably higher, there appears to be a major headroom problem.
See this web page for links to all my writings and blog posts on the Rand Hall library proposal.