Content uploaded by Donald Friedman
Author content
All content in this area was uploaded by Donald Friedman on Sep 21, 2019
Content may be subject to copyright.
SAHC2014 – 9th International Conference on
Structural Analysis of Historical Constructions
F. Peña & M. Chávez (eds.)
Mexico City, Mexico, 14–17 October 2014
NEW YORK’S TOWER BUILDING:
STRUCTURAL ANALYSIS OF A PROTO-SKYSCRAPER
Donald Friedman1 and Gabriel Pardo Redondo2
1 Old Structures Engineering, PC
111 Broadway, New York, New York, 10006 USA
e-mail: dfriedman@oldstructures.com
2 Old Structures Engineering, PC
111 Broadway, New York, New York, 10006 USA
e-mail: gpardo@oldstructures.com
Keywords: Cast Iron, Wrought Iron, Building Frame, Skyscraper
Abstract. The Tower Building, completed in 1889 in New York, was an 11-story (39m) early
skyscraper with a hybrid frame. It was among the last tall buildings completed in the United
States before the introduction of skeleton framing 1890, and attracted a great deal of atten-
tion for its extreme slenderness and its unique structural system, which is best described as a
five-story bearing-wall building sitting on top of a six-story frame building.
The building occupied a narrow mid-block site, that made the use of masonry bearing walls
impractical, as the wall thicknesses required by code would occupy nearly half of the lot
width. The constructed solution was to use a cast- and wrought-iron braced frame for the bot-
tom seven stories of the building using the legal fiction that it was an extended basement. Tra-
ditional masonry bearing walls were carried on the top of the frame at the seventh floor and
extended up to the roof over the 11th story. It was the first commercial building of such height
and such an extreme height-to-width ratio, which helps explain the difficulties that Bradford
Lee Gilbert, its architect, had with both the Board of Examiners of the New York City Build-
ing Department and with public perception of safety. In the 1890s, there was an extended dis-
cussion in the American engineering community on the appropriate wind loads and methods
of bracing to be used in tall buildings. At the same time there was discussion in the public
press of the effect of the new type of the building, the “skyscraper,” on public safety.
This paper examines the practicality of the hybrid structure: was it adequate using the codes
and state of knowledge at the time of construction? Would it be considered adequate today?
Using old methods for design and current formulas, the paper compares how close the struc-
tural design was to current standards and how methods of design have evolved for the last
century.
Donald Friedman and Gabriel Pardo Redondo
1 INTRODUCTION
The history of skyscraper structure tends to focus on buildings with skeleton frames and
non-load-bearing curtain walls, because that form is used in all modern tall buildings. How -
ever, tall buildings were constructed before that form was used, and some of the hybrid
frames were important steps toward the development of the skeleton frame.
The first fire-resistant skeleton-frame skyscrapers were constructed in 1890: the Manhattan
Building in Chicago and the London & Lancashire Building in New York. Just before these
buildings were constructed, the Tower Building was completed in New York, and attracted na-
tional attention due to its extreme slenderness for that era – its height was greater than 6 times
its width – and its unique hybrid frame. [1,2]. Because the Tower Building did not have a
modern-style skeleton frame, an analysis is necessary to see if its structural system is rational
by modern standards.
1.1 Historical Context
The Tower Building was, for its time,
quite slender, but it was not exceptionally
tall. The tallest American building of
“modern fireproof” construction in 1889
was the (small footprint) tower of the
(large-footprint) Auditorium Building in
Chicago, which was 275 feet (84m) high,
and the (small) tower of the (large) Tri-
bune Building in New York, at 235 feet
(72m) high. There were some 20 extant
buildings that had higher top elevations
than the Tower Building when it was
completed, either with small towers or en-
tire floors. All had bearing-wall or hybrid
frame structure. The first cage-frame
building, a form of hybrid where the inte-
rior floors were carried entirely on a metal
frame, was the 1885 Home Insurance
Building in Chicago.
The unique hybrid structural system
used at Tower was not a modern skeleton
frame, although it approaches that form at
the lower floors. Its fame stems first from
the impression it made on the local real-
estate community and, via the national
press located in New York, the American
design community (by showing that a
small lot could be developed with a tall
building; and later from the efforts made
by its architect, Bradford Lee Gilbert, to
claim for it the title of “first steel-frame
skyscraper.” [3, 4]
2
Figure 1: The Tower Building (at arrow) in 1901, sur-
rounded by taller, newer buildings. (Photo credit: Li-
brary of Congress, Prints & Photographs Division, De-
troit Publishing Company Collection, LC-DIG-det-
4a08587
New York’s Tower Building
Not until 1892, two years after the first skeleton-frame building was completed and three
years after the Tower Building, did the New York City Building Code recognize the existence
of skeleton framing. The rapid proliferation of skeleton-frame skyscrapers in the downtown
business district of New York meant that the Tower Building was completely surrounded by
tall buildings by 1905, and was eventually torn down as economically obsolete in 1914. (See
figure 1.)
2 BUILDING DESCRIPTION
The site for the Tower Building was as-
sembled from two lots, one with a 21’-3”
(6.5m) frontage on Broadway measuring
110 feet (33.5m) deep, and one with a 38’-
3” (11.7m) frontage on New Street measur-
ing 57’-2” (17.4) deep. The north lot lines
are the same line, making the entire site a
blunt “L” in plan. It should be noted that,
because of lower Manhattan’s irregular
street grid, the lots were not rectangular and
all dimensions vary slightly from side to
side. (See figure 2). The building was 11
stories with an attic on Broadway; because
of the sloped site, that made it 12 stories
with an attic on New Street. There were two
sub-grade levels, officially known as the
cellar and sub-cellar. The typical floor to
floor height was 11’-9" (3.6m), putting the
attic at an elevation of 129 feet (39.3m)
above grade, with the peaked roof extending
up to 147’-5” (44.9m) above grade.
The building consisted of two wings that were, in structural terms, essentially two different
building. Both had wrought-iron beams supporting terra-cotta tile arch floors, but the basic
frame layouts were different. The east wing, on the New Street lot, had bearing walls at the
north and south sides, a row of cast-iron columns and girders running east-west, and two
spans of ordinary filler beams running from the north bearing wall to the column line to the
south bearing wall. This system, mixing bearing walls with iron “frames” that were designed
for gravity load only, had become common in American tall building construction in the late
1870s. The east wing is therefore of little structural interest.
The west wing, on the Broadway lot, is the portion of the building where constraints of ge-
ometry and code requirements forced the designers to look for a new solution to framing a tall
building. The first constraint was simply the width of the west wing, 21’-3” (6.5m) north to
south. In such a narrow building, developed for speculative commercial use, every bit of
width matters. The minimum thickness of brick wall in use for commercial buildings in New
York at that time was 12 inches (0.3m), or three wythes of bricks thick. If it were possible to
use these minimum walls, they would have occupied more than 9 percent of the floor area of
the wing, which is a non-negligible loss. However, the second constraint was the minimum
wall thickness specified by New York City Building Code, which varied from 20 inches
(0.5m) at the top two floors to 36 inches (0.9m) at the base. [5] At the 11th floor, the side
3
Figure 2: Plot plan of Tower Building, based on land-use
map.
Donald Friedman and Gabriel Pardo Redondo
walls would occupy over 15 percent of the lot; at the first floor, more than 28 percent. This
loss of rentable space was enough to mean the difference between a building that was an eco-
nomic success and one that was not.
The solution worked out by Gilbert and William Birkmire, an engineer working for the
Jackson Iron Works, the company that erected the frame, was to take advantage of ambiguous
wording in the building code. First, the thickness of walls was typically calculated by their
height above the curb, but walls that were supported on metal beams had their thickness
counted from the elevation of that support. Second, the code allowed for "light partition
walls” as thin as 8 inches (0.2m) of brick, with a maximum height of 50 feet (15.2m) from
their support. [5] Third, there was no specific limitation put on supporting walls on iron gird-
ers and columns. The structural scheme for the west wing therefore consisted of (1) a frame
consisting of cast-iron columns and wrought-iron beams and diagonal braces that ran from the
lowest level of the foundations (at the sub-cellar) to the top of the sixth floor, with thin ma-
sonry curtain walls supported at every floor, (2) a five-story bearing wall building sitting on
top of the frame, with its walls entirely supported by girders at the seventh-floor framing
level, and (3) a continuation of the diagonal bracing into the upper bearing-wall portion of the
building. [6] (See figure 3.) There is some dispute as to whether the original design included
the bracing diagonals; modern analysis shows that the building would not have performed
well without them.
It should be noted that Gilbert’s public statement of his logic for this building is incorrect.
In 1899, Gilbert claimed that he “hit upon this thought: ‘The City authorities figure the height
of a building from its foundation only; why can’t I run my foundation far up into the air and
then begin my building? In other words, why can’t I build a structure to the eighth floor,
which will be a composite [frame-supported] affair, and then build on top of that the rest of
the edifice of three or four stories, thus using the foundation as part of the building, and ful -
filling the city’s requirements?’” [3] The problem with this statement is that the building code
clearly states that the height is measured from the curb line, not the foundation top, as shown
in the wall-thickness diagrams accompanying the code text. [7] Since he made the quoted
statement more than ten years after the design was completed, he may have misremembered,
or he may have had some motive for claiming this story.
Two features mark this building as a hybrid: the bearing walls at the upper floors, and the
use of cast-iron columns as part of a braced frame. The first is obviously an artifact of the
transition from bearing-wall to skeleton framing. The second is evidence that the relationship
between the structural engineering profession and building design was still in its early stages
in the United States. Structural engineers at that time were typically concerned with civil
structures (bridges, roads, railroads, dams, and similar works that were not buildable without
engineering analysis and design) and would not become standard consultants during building
design until the twentieth century. Most hybrid- and skeleton-frame American buildings of the
nineteenth century had their iron and steel elements designed by the iron contractors, as at the
Tower Building.
The floors were the standard used for large buildings in the 1880s, and often referred to as
“fireproof”: flat terra-cotta vaults supported by and providing fire-protection for the wrought-
iron floor beams. (See figure 4.)
4
New York’s Tower Building
5
Figure 3: Typical building section look-
ing east. Figure 4: Typical west wing floor plans.
Donald Friedman and Gabriel Pardo Redondo
3 FRAME ANALYSIS
The availability of only partial information on the actual built structure of the Tower Build-
ing prevents a completely accurate analysis. Instead, we have examined the parameters of
analysis, using what was known of building frame analysis at the time of construction and
what is known for certain about the frame as built.
The layout of the frame is a rational one given the architectural and construction con-
straints. The use of cast-iron columns was still standard in New York in 1889 and would grad-
ually diminish over the next fifteen years. Since it was well known that cast iron could fail
without warning if loaded in tension, the use of moment connections for lateral bracing was
not realistic. In such a narrow building, the presence of the public and private hallways along
the north wall limited the width of the bracing “trusses” to approximately 60 percent of the
building width. The south chords of the trusses were the cast iron columns embedded in the
south wall; the north chord was an intermediate vertical member built-up of wrought-iron
angles; and the north-wall cast-iron columns were not part of the bracing frames.
Since there were no rock anchors or similar methods of tying down columns against uplift
forces, both the south-wall columns and the intermediate vertical had to be tied into the ma-
sonry walls to allow the dead weight of masonry walls and floors to counter wind uplift. Since
the south columns supported the 7th floor girders that carried the upper bearing walls and were
tied into the thinner curtain walls below, they were weighted against uplift without any special
details. The intermediate verticals had no such direct tie to gravity load, which appears to be
the reason that the bracing pattern changed below grade. In the subcellar, the intermediate
verticals’ loads were distributed to the side foundation walls through a change in the bracing
pattern, so that there was no need for foundations in the center, where they would be isolated
from the masonry weight.
3.1 1887 NYC code
The governing regulation for the design of the Tower Building was the 1887 building code
of the city of New York. This code was largely concerned with the construction of low-rise
residential buildings, which were the vast majority of buildings constructed at that time. The
discussion of structural analysis and design is extremely limited with, for example, floor loads
specified only for three broad classes of buildings and wind loads not specified at all. [8] No
engineering design data is provided, but rather the reader is directed to use “Trautwine’s Trea-
tise for Engineers,” a common handbook of the era, or whichever textbook is used in the engi-
neering classes at the U.S. Military Academy at West Point.
The dead load of the building was estimated using the known geometry and materials (for
example, assuming common brick for the masonry walls) and estimates where all data is not
known (such as estimating the flat terra-cotta tile arch floors at 80 psf (3.8kPa)). There was no
stated live load for office use in the 1887 New York Code, but rather a 75 psf (3.6kPa) mini-
mum load for all buildings, 120 psf (5.7kPa) for places of assembly, and 150 psf (7.2kPa) for
“manufacturing or commercial purpose[s]”. The rental office space in the Tower would not be
classified as public assembly, but was probably designed for the commercial loading. [8] The
1892 revision of the New York code included office occupancy at 100 psf (4.8kPa) and roof
snow loading at 50 psf (2.4kPa). [9] There is no wind loading mentioned in the New York
code, but Trautwine provides a design wind load of 40 psf (1.9kPa) on portions of the walls
exposed above neighboring buildings. [10]
6
New York’s Tower Building
Trautwine provides several different formulas for computing the allowable stress in both
cast and wrought iron. For tension in wrought iron, the ultimate stress given vary from 45 ksi
(310 MPa) to 76 ksi (524MPa), but the most clear statement is for allowable load: “...good
iron bar should not be trusted permanently with more than about 5 tons per square inch...
[69MPa]” For compression in wrought iron, the values even more widely because of uncer-
tainty about safety factors: the statement “for safety take [the ratio of allowable to ultimate
stress] from 1/3 to 1/8, according to circumstances” makes pinning down a single allowable
stress impossible. Ultimate compressive stresses range from 30.9 ksi (213MPa) (L/r = 90,
“flat ended”) to 7.2 ksi (50MPa) (L/r = 300). Finally, the ultimate compressive stress in square
cast-iron columns with flat ends with a length of 12 feet (3.7m) and a 20 inch (0.5m) side is
66 ksi (455MPa), leading to an allowable compressive stress of 13 ksi (90MPa). [11]
3.2 2008 NYC code
The current New York City Building Code, issued in 2008, is a local adaptation of the In-
ternational Building Code now used in nearly all jurisdictions in the United States. [12] The
provisions of this code were used as written with the exception of the seismic loading provi-
sions. Seismic loading was excluded because (1) it was not used at all at the time of construc-
tion and (2) it may be neglected for existing buildings in new York under current code as long
as there is no expansion of the building. If the Tower Building still stood, it would be exempt
from seismic analysis unless it was expanded, so an analysis of its original configuration is
exempt from seismic analysis. [13]
The dead load used for modern analysis is obviously the same as that used in the replicated
1887 analysis. The floor live load is the 50 psf (2.4kPa) used for offices, which was also used
for roof occupancy. Current snow loading is 23 psf (1.1kPa), using the ASCE 7 method speci-
fied in the New York code and therefore over-ridden by the roof occupancy loading. Wind
load, per ASCE 7 and the New York code, varies from the minimum allowable value of 20 psf
(1.0kPa) at grade to 24.5 psf (1.2kPa) at the roof elevation. Under current code, wind load is
applied at all floors, including this shadowed by neighboring buildings.
The most interesting aspect of the modern code loading is that it is significantly lower than
Trautwine's recommendation of 40 psf (1.9kPa). The total 2008 wind load is less than 60% of
the total 1887 Trautwine wind load.
Finally, the modern code does not address allowable stresses in wrought or cast iron. If
wrought iron is treated as a weaker form of steel, current code would allow tension of 0.6
times the yield stress, or roughly 15 ksi (103MPa). The best current method of evaluating
cast-iron columns as used in American practice provides provides a critical LRFD compres-
sive stress of 17 ksi (117MPa) for all columns with slenderness less than 108 (the cellar and
subcellar columns have slendernesses of 18), to be used with a strength reduction factor φ =
0.65 leading to a design stress of 11 ksi (76MPa) to be used with factored loads. [14]
3.3 Method of frame analysis
The west-wing building frame was extremely simple and lends itself to 2-dimensional
analysis. Wind load in the north-south direction was resisted by seven braced frames consist-
ing of the a cast-iron column at the north wall, a cast-iron column at the south wall, a
wrought-iron intermediate vertical truss chord between (located 13 feet north of the south
wall), a wrought-iron girder at each floor level, and wrought-iron diagonals (running from the
south column to the intermediate truss chord) that alternated direction at each floor. (See Fig-
ure 3 for the bracing layout.) The bracing greatly resembles Gilbert’s description of it as “an
iron bridge truss stood on end” with the actual bracing confined to the southern 13 feet (4.0m)
7
Donald Friedman and Gabriel Pardo Redondo
of the wing, where the diagonals were hidden within room partitions, and the diagonal-free
northern 8 feet (2.4m) including the public hallway. [15] Given the small size of the truss di-
agonals, which were all double or quadruple angles, with no individual angle larger than 6
inches (0.15m) by 6 inches, the typical steel-truss-analysis assumption – that the truss mem-
bers carry only axial loads, with any minor moments released by plastic bending near the con-
nections – is accurate. The bracing truss was therefore analyzed as pin-connected, even
though the braces were connected by multiple bolts to bent-plate connectors riveted to the
floor beams. [6]
The analysis of a pin-connected, single-diagonal Warren truss, even one that is cantilevered
rather than simply supported, is easily accomplished using the method of sections. This
method has the advantage that it is the method that would have been used by the designers of
the Tower Building and the frames of similar truss-braced buildings of that era. The gravity
load in the cast-iron columns was computed separately and added to the wind-induced forces.
3.4 Results
Analysis of the truss frames gives two sets of results: the stresses in the members and the
allowable stresses under the 1887 and 2008 codes. Both sets are inherently inaccurate due to
missing information. For the member stresses under load, we do not know what wind load
was used in analysis, whether Trautwine’s 40 psf (1.9kPa) or some other (probably lesser)
amount; we do not know all of the member sizes; we do not know the exact floor dead load;
and we do not know what floor live load was used. For the allowable stresses, we have con -
tradictory information for cast and wrought iron in the 1887 sources and none in the 2008
code.
The most heavily loaded members are the cellar and sub-cellar braces and columns. If we
use the worst case 1887 loading – 40 psf (1.9kPa) wind, 120 psf (5.7kPa) live load – the
stresses in the braces are between 14 ksi (97MPa) and 16 ksi (110MPa) and the maximum
(gravity plus wind) stress in the columns is 14 ksi (97MPa). The best case loading is that in
the current code – 25 psf (1.2kPa) maximum wind, 50 psf (2.4kPa) live load – leads to brace
stresses of 9 ksi (62MPa) to 12 ksi (83MPa) and the maximum stress in the column is 12 ksi
(83MPa). The best case could be improved by taking advantage of two analysis techniques
not in use in 1887: live load reduction, and allowable stress increase for combined gravity and
wind load. Both are based on reducing the effect of low-probability events, respectively the
presence of full live load on a large area of floor or on multiple floors at once and the pres -
ence of full live load simultaneously with full wind load. The amount of benefit from these
two techniques varies, but live load reduction alone could reduce the column compressive
stress by as much as 15 percent; the reduction of the combined live and wind load stress by 25
percent would have a similar effect. [12]
The total gravity load in the cast iron columns was roughly three times the maximum wind
force (tension or compression) in the columns and the dead load in the columns was more
than twice the wind force, meaning that the cast iron was never subjected to tension from
wind loading. (Using the modern code, the maximum column dead load was 785 kips
(3.5MN) while the maximum wind tension was 342 kips (1.5MN).)
In short, without access to accurate information about the original design that is apparently
not recorded, there is no way to be certain if the design met the standards of its day for stress
in both the wrought and cast iron. However, it can be stated that the apparent gap between ac -
tual stresses and allowable is relatively consistent, indicating that any design inadequacy rela-
tive to the 1887 code was based on the loads and allowable stresses used rather than an irra -
tional design or improper analysis of the bracing trusses.
8
New York’s Tower Building
4 CONCLUSIONS
The 1880s were still an era of empirical design in the United States. The New York City
Building Code of 1887 allowed bearing-wall buildings of any height to be constructed with-
out an engineering analysis of either gravity
or wind load, relying on a linear increase in
wall thickness to ensure adequate masonry
to resist the stress. None of the buildings that
are known to have met the provisions of this
code are known to have failed, but that is
simply because the rules were conservative,
and kept the masonry stresses low. Most
known failures of completed buildings of
that era concern design deviations from the
requirements or construction errors. The
code made little provision for the partial
skeleton frames already in construction and
none for the skeleton frame buildings that
followed the Tower, starting in 1890. This
situation was the logical outgrowth of the
mid-1800s, when no engineering design at
all was needed for the low-rise, masonry-
walled and wood-floored buildings that
made up nearly the entire country. In the few
cases where iron beams were used, the engi-
neering design was performed by the iron
contractors.
Structural design for American buildings
in the 1880s was, not surprisingly in those
circumstances, rather simple. The most im-
portant portion of the Tower Building’s
frame, famous at the time and in early histories of skyscrapers, was a single-diagonal Warren
truss, with cast-iron used for members that known to be in compression and wrought iron
used for those that would be subjected to tension. Double-diagonal trusses would have been
stiffer in the Tower Building, but would have been more difficult to build because of the
crossing connection. Since the truss was analyzed only for stress, and not for deflection, the
additional stiffness was not an observable benefit, while the additional cost of the more com-
plicated connection was certainly visible in a speculative venture such as this.
Within these constraints, the design of the frame appears to have made sense both by the
standards of its time and the standards of ours. The truss members appear to have been sized
to give a consistent maximum stress, which is both logical and efficient. The potential for cat-
astrophic failure of cast-iron stressed in tension, which was already known in 1889 and would
become a major debate in the American engineering community in the following decade, was
avoided through the top-heavy configuration of the exterior walls, which increased the dead-
load compression in the cast-iron columns. [16] While the building had a short life because it
became economically obsolete – it was replaced with a wider structure that had larger floors
that could be more easily rented for a higher price – its structure performed, which is all that
an engineering analysis demands of it. (See figure 5.)
9
Figure 5: Photograph of side girder to column connec-
tion during demolition. Note that one of the paired
spandrel beams on the left has been removed. [17]
Donald Friedman and Gabriel Pardo Redondo
REFERENCES
[1] n.a., Architecture and Building, v12 n9, March 1890.
[2] n.a., Plumbing of a large building. Carpentry and Building, v12 n6, 127-128, June 1890.
[3] n.a., Disputes of architects. The New York Times, August 19, 1899.
[4] n.a., New York’s first skyscraper and its architect. The Real Estate Record and Builders’
Guide, v88 n2275, 589, October 21, 1911.
[5] W. J. Fryer, ed., Laws relating to buildings in the city of New York... The Record and
Guide, §477, 1887.
[6] H. H. Quimby, Wind bracing in high buildings. Transactions of the American Society of
Civil Engineers, v27, 221-252, June 1892.
[7] W. J. Fryer, ed., Law limiting the height of dwelling-houses. The Record and Guide,
§477, 14-18, 1887.
[8] W. J. Fryer, ed., Laws relating to buildings in the city of New York... The Record and
Guide, §491, 38, 1887.
[9] W. J. Fryer, ed., Laws relating to buildings in the city of New York... The Record and
Guide, §483, 1892.
[10] J. C. Trautwine, The civil engineer’s pocket book, 15e. John Wiley & Sons, 216, 1891.
[11] J. C. Trautwine, The civil engineer’s pocket book, 15e. John Wiley & Sons, 443-446,
464, 1891.
[12] n.a., New York city building code, International Code Council, 2008.
[13] R. Visconti, Technical policy and procedure notice #4/1999, New York City Department
of Buildings, 1999.
[14] C. Paulson. R.H.R. Tide, and D. F. Manheit, Modern techniques for determining the ca-
pacity of cast iron columns, Standards for Preservation and Rehabilitation, STP-1258.
ASTM International, 1996.
[15] n.a., The birth of the new york skyscraper – a romance of architecture, The New York
Times, May 21, 1905.
[16] D. Friedman, The Darlington building collapse: modern engineering and obsolete sys-
tems, Proceedings of the Fourth Forensic Engineering Congress. American Society of
Civil Engineers, 2006.
[17] n.a., Condition of Tower Building, New York, after 25-yr service. Engineering News,
v71 n14, 748, April 2, 1914.
10