Conference PaperPDF Available

Hidden Strength in Historic Buildings

Authors:
  • Old Structures Engineering

Abstract and Figures

p>Most historic buildings include structural materials and systems that are in some way obsolete. This can range from materials no longer available (first-growth >mber) to those no longer considered safe in modern codes for new use (cast-iron columns). There is a large class of buildings that contain systems that are no longer used but are safe in use; because these systems are not used in new buildings, they are not discussed in current code. This paper reviews three obsolete structural systems that have a history of good performance and that have more capacity than ordinary modern analysis would suggest. First, terra-coDa >le arch floors are known to be strong themselves for their expected (gravity) loading, but can also, through pseudo-composite ac>on, strengthen the wrought-iron or steel beams suppor>ng them. Second, draped-mesh (catenary) floors have a load capacity defined by their reinforcing, but have shown to be s>ffer than expected because of mul>ple load paths within the slabs. Finally, the heavy masonry curtain walls typical used with steel-frame buildings before 1920 provide alternate load paths, addi>onal s>ffness, and addi>onal capacity for lateral loads. The presence of these systems can mean that historic buildings are stronger than we think.</p
Content may be subject to copyright.
Hidden Strength in Historic Buildings
Mona Abdelfatah, P.E.
Old Structures Engineering, PC
90 Broad Street
New York, NY 10004
(212) 244-4546
MAbdelfatah@oldstructures.com
Donald Friedman, P.E., F. ASCE
Old Structures Engineering, PC
90 Broad Street
New York, NY 10004
(212) 244-4546
DFriedman@OldStructures.com
1. Abstract
Most historic buildings include structural materials and systems that are in some way
obsolete. This can range from materials no longer available (first-growth timber) to
those no longer considered safe in modern codes for new use (cast-iron columns). There
is a large class of buildings that contain systems that are no longer used but are safe in
use; because these systems are not used in new buildings, they are not discussed in cur-
rent code.
This paper reviews three obsolete structural systems that have a history of good perfor-
mance and that have more capacity than ordinary modern analysis would suggest.
First, terra-cotta tile arch floors are known to be strong themselves for their expected
(gravity) loading, but can also, through pseudo-composite action, strengthen the
wrought-iron or steel beams supporting them. Second, draped-mesh (catenary) floors
have a load capacity defined by their reinforcing, but have shown to be stiffer than ex-
pected because of multiple load paths within the slabs. Finally, the heavy masonry cur-
tain walls typical used with steel-frame buildings before 1920 provide alternate load
paths, additional stiffness, and additional capacity for lateral loads. The presence of
these systems can mean that historic buildings are stronger than we think.
1 of 8
2. Introduction
As designers and builders develop new forms of construction technology, old ones are
sometimes abandoned. The buildings containing those old technologies often continue
in use, because no owner can afford to replace every building to bring them all up to
modern code standards. Society as a whole cannot afford the environmental impact of
wholesale replacement just to address old structure. In any case, code requirements are
a moving target, so every code update would require replacement of buildings unless
engineers try to accurately assess the structure of the extent building stock.
When engineers expend more effort in the analysis of obsolete existing structure, the
result is, usually, less effort spent in construction, in replacement or unneeded strength-
ening. Looking at means of analyzing obsolete structure is therefore one of the first
steps in reducing unneeded work.
3. Obsolete Structural Systems
Structural technology can be current, can be archaic (meeting current requirements
but in a form no longer used in new construction) or can be obsolete (meeting previous
requirements but not current ones). Riveted connections in steel frames are archaic
construction, as they can be analyzed using current-day codes combined with extant
standards for rivets. No special analysis is needed for such connections. The three ex-
amples discussed here are obsolete: they were explicitly designed to meet past codes
and standards, but do not meet all of the requirements of current codes and therefore
could not be used in their current forms in new buildings. Unreinforced masonry, such
as tile-arch floors and old brick walls, lack any of the ductility and continuity that we ex-
pect in new designs. Draped-mesh floors have some degree of the properties, but far
less than we expect.
3.1 Tile-Arch Floors and Beam Capacity
Terra-cotta tile-arch floors were the most popular form of fire-resistant floor in the
United States from the 1870s until the first decade of the 1900s, and continued in use
into the 1920s. [1] They consisted of segmental or flat vaults of hollow terra-cotta blocks
spanning between wrought-iron or steel beams, and topped with non-structural fill. De-
signers knew that these floors, as a form of masonry arch, generally had greater capac-
ity for gravity load than the supporting beams. Their use ended largely because their
construction was more labor-intensive than the concrete floors that were later devel-
oped.
An interesting observation was made by engineer Gunvald Aus in 1895, during load
tests to qualify segmental tile arches for the federal Appraiser’s Warehouse then under
construction in New York. [2] The arch floors used at the warehouse span up to 4.67m
(15 feet, 4 inches) between 610mm (24 inch) I-beams. (See Figure 1.) The fill above the
arch was concrete but its composition would be considered too weak and irregular for
2 of 8
structural use by modern standards and, as Aus said, it was “still green” (i.e., uncured)
at the time of the load tests. Aus’s report on the tests focused on the strength of the
arches (which were loaded to 57 kPa (1200 psf) without failure), but his most interest-
ing comment was at the end, where he notes that the supporting beams “showed no de-
flection” despite the extreme loading. He suggests that this was due to the concrete fill
which “in connection with the beams, forms combination girders, the concrete forming
the compression, the beam the tension chord.”
Figure 1: Terra cotta tile-arch floor as tested. [2]
This is an accurate statement of the concept of composite beams, made 41 years before
composite construction was first explicitly referenced in an American code. [3] As there
is no specific analysis method to combine a tile arch floor compositely with steel beams,
the idea was reviewed by (a) checking the capacity of the fill as a compression flange us-
ing current composite formulas, (b) checking the contribution of the solid area of the
tile arch as part of the compression flange, and (c) checking the stress-transfer mecha-
nism at the tile-arch/fill/steel interface. The first two steps are similar to modern analy-
sis. Since there are no shear connectors, the transfer depends on friction. The arch
thrust creates a high frictional resistance stress where the terra cotta meets the steel
(along the relatively small interface area of the arch contact with the steel), and the
lower frictional resistance stress along the fill/terra-cotta interface is at a much larger
surface area. In short, the arch thrust is high enough to ensure that the stress transfer
functions. The beams are considerably strengthened in both bending stress and deflec-
tion by the composite effect.
3.2 Draped-Mesh Floors and Slab Capacity
One of the successors to the tile-arch floor was the draped-mesh floor, a concrete slab
system reinforced with wire mesh that forms catenaries supported by the floor beams.
[1] Because the load is, in theory, entirely carried by the reinforcing catenaries, the con-
crete used was often lightweight material of poor composition: a popular variant used
coal cinders in the mix as the coarse aggregate. Cinder concrete is roughly 35 percent
lighter than stone concrete, but rarely has a compressive breaking strength higher than
3.4 MPa (500 psi).
If one assumes that the wires are arranged in perfect catenary curves, the expected de-
flection under load is quite small. An elongation on the order of 0.1 percent of the span
3 of 8
length leads to a downward deflection of 0.003 percent of the span, or 0.08 mm (.003
inches) for a 2.5 m (8 foot) span. However, the wires are in reality far from true cate-
nary curves. Because of the manner in which these slabs were built in the US during
the 1920s and 30s, the mesh typically does not hang in smooth curves. Construction
photos of the era show the method of construction: after the steel floor beams were in
place, wood forms were built for the slabs and the integral beam encasement. The beam
top flanges projected above the forms, so the drape was achieved by the laborers walk-
ing over the mesh after they rolled it onto the forms. (See Figure 2.) This led to a typi-
cal mesh configuration of a steep slope down from the beam tops to a long run of mesh
near the bottom of the slabs. In other words, the mesh forms a squared-off shallow “U”
shape rather than smooth curve. In theory, this curve would lead to higher observed de-
flections because the wires would elongate almost vertically at the ends when loaded.
Similar to tile-arch floors, cinder-concrete floors have been observed to provide more
strength and stiffness than simple analysis would suggest. During a 2010 load test of a
100 mm (4 inch) thick panel spanning 2.5 m (8 feet), with an expected slab deflection
6.4 mm (0.25 inches) using the realistic wire geometry, the senior author observed de-
flection of less than .5 mm (0.02 inches). This deflection was on the low end of what was
expected using the catenary-curve model even though the wire had been observed to
4 of 8
Figure 2: Mesh draped over beams (A). Note the sharp curve at the beam flange (B) and mostly flat curve
along bottom of slab (C). [4]
match the U shape. [5] This is an example of the fact that, despite being constructed
with poor-quality concrete, these floors have typically performed as well in service as
modern floors if protected from water so that the wire does not rust. The discrepancy
between analysis and observed performance seems to be related to discrepancies be-
tween the theoretical model and the actual built condition.
The realistic geometry of the mesh interacts with another element not present in the
theoretical model of the slabs: the beam encasement. Catenary theory does not require
a shear transfer at the ends of the slabs, but the encasement was required as a means
of providing fire-protection for the beams. In general, the encasement has the effect of
shortening the free span of the catenaries. If the rigid zone of the encasement is carried
up through the slab at a 45º angle, the span is effectively shortened further. A shorter
span is significantly stronger, since the carrying capacity is inversely related to the
square of the span. If the 2.5m (8 foot) span observed in 2010 is recalculated using a
2.25 m (7 foot, 4 inch) span to account for the encasement at each end, the allowable
load increases by roughly 20 percent. The encasement also accounts for an increase in
apparent stiffness by reducing the scope of possible movement of the vertical (or sloped
and nearly vertical) ends of the wire spans.
3.3 Masonry Curtain Walls and Frame Capacity
Steel skeleton framing, with metal frames explicitly used to support non-structural cur-
tain walls was first used in the US in the 1890s. The technology developed rapidly and
generally had replaced the use of bearing walls in most building types by 1905. How-
ever, the curtain walls in use through 1920 generally were still solidly built, brick or
brick faced with stone or other masonry veneer, and at least 300 mm (12 inches) thick.
(See Figure 3.) Walls were reduced further in mass in the 1920s, with the inner wythe
often built of hollow terra cotta blocks rather than brick, but generally remained the
same thickness. The use of expansion joints in masonry curtain walls did not become
standard practice in the United States until after 1960, so the masonry in these build-
ings is effectively continuous from top to bottom of the curtain walls. In short, there is
little physical difference between curtain walls of the 1890 to 1930 era and the bearing
walls that might have been used if the buildings were not constructed with steel
frames.
Obviously the strength and stiffness of the curtain walls may not be sufficient to per-
form structurally for tall and slender buildings. It is difficult to make a blanket state-
ment because of the widely varying geometry of height, slenderness, percentage of win-
dow area in the walls, and interior structure, but it is clear that the walls are suffi-
ciently strong to support load in low-rise buildings of this type. In a previous analysis of
this issue, the senior author found that the 300mm (12 inch) walls were sufficiently
strong to carry a portion of gravity and lateral load. [6] More importantly, the walls are
stiffer than the frame when loaded in plane, which means that the walls will be loaded
simultaneously with the frame and will carry the majority of the load until and unless
5 of 8
they fail. The walls failing under load is technically a structural failure but has no ill ef-
fect other than the appearance of cracks in the masonry, since the failure simply throws
the load back into the steel frame that was designed for it.
Without disassembly, there may be no way to determine whether the masonry is carry-
ing some or most of the load in a low- or medium-rise building of this type. Unless the
masonry is overstressed, the unplanned structural action is an overall benefit.
4. Discussion
All three examples described above are unplanned structural effects that exist in addi-
tion to the intended structural action. None of the these forms of structure is explicitly
6 of 8
Figure 3: Section of solid brick and terra-cotta
curtain wall at the Central Bank Building, New
York. The double I-beam is the spandrel beam;
the dashed rectangles are the spandrel column lo-
cation. [7]
allowed under current US codes, but they are present in hundreds of thousands of
buildings across the country. The existing buildings are grandfathered - allowed to re-
main in use as structures legalized under old codes or regulations - under a provision
that has been present in US codes since they were first written. [8] However, under-
standing the design of the existing forms of structure, including specifically their unin-
tended actions, is a useful tool in determining whether existing buildings are generally
safe rather than whether they are code-compliant.
The first two examples, generally suggest that the two most common forms of floor
used in fire-rated buildings for the 1870s to the 1920s are stronger than they were in-
tended to be, because their reality is more complex than the models used to define
them. The non-structural fill of a tile arch floor contributes by increasing the friction be-
tween the tile and the beam; the poor mesh geometry of a draped-mesh floor helps by
moving the portion of wire most likely to contribute to downward deflection to the rigid
encasement zone; the non-structural curtain walls contribute gravity and lateral resis-
tance through their inherent strength. Similar systems may have similar effects, such
as tile arches topped with loose fill rather than concrete, where the tile blocks them-
selves serve as the compression flange.
None of these examples is particularly difficult in concept, but all are obscured by the
ordinary assumptions made in modeling. Gunvald Aus was one of the most talented en-
gineers of his era: he was a bridge designer before working for the US Department of
the Treasury when he observed the floor tests, and went on to design the frame of the
Woolworth Building. In his comments on the tile-arch floors, he followed his observa-
tions rather than the normal model, and thus anticipated a now-common structural
form.
5. Conclusions
Modern engineering education and research is largely concerned with the design of
new structures. It is well known that the majority of extant buildings do not meet cur-
rent code requirements; every time codes are modified, more buildings become techni-
cally obsolete. However, a lack of code compliance does not automatically mean that the
old structural systems are weaker. The details of the old systems, specifically the lack of
ductility in unreinforced masonry, may make them less suitable for seismic loading, but
many of these building exist in areas where wind load is the dominant lateral load. In
short, the capacity of obsolete structure needs to be examined on a case-by-case basis
rather than on assumptions based on code-compliance, the status of the technology
used, or inaccurate structural models.
6. References
(1) D. Friedman, “Analysis of Archaic Fireproof Floor Systems,” in Structural
Analysis of Historical Constructions: Preserving Safety and Significance; pro-
7 of 8
ceedings of the Sixth International Seminar on Structural Analysis of Historical
Constructions, 2-4 July 2008, Bath, United Kingdom, D. D’ayala and E. Fodde,
eds. Boca raton: CRC Press, 2008. pp. 129-136.
(2) G. Aus, “Tests of Long-Span Terra-Cotta Floor Arches,” Engineering News, vol.
34, no. 19, p 314, 1895.
(3) n.a., “Specification for the Design, Fabrication, and Erection of Structural Steel
for Buildings,” New York: American Institute of Steel Construction, 1936.
(4) n.a., “Wire Fabric Reinforcing in Buildings,” New York: American Steel & Wire
Company, 1925.
(5) D. Friedman, observations on site at a 1950 government-owned building in New
York. The details of the project are covered by a confidentiality agreement.
(6) D. Friedman, “Analysis of Steel-Structure/Masonry-Wall Interaction in Historic
Buildings,” in Structural Analysis of Historical Constructions: Possibilities of
Numerical and Experimental Techniques; proceedings of the Fourth Interna-
tional Seminar on Structural Analysis of Historical Constructions, 10-13 Novem-
ber 2004, Padova, Italy, C. Modena, P. Lourenço & P. Roca, eds. Leiden:
Balkema, 2004. pp. 103-10.
(7) W. Birkmire, The Planning and Construction of High Office-Buildings, 4th edi-
tion. New York: John Wiley & Sons, 1906, page 113.
(8) n.a. 2018 International Building Code. Washington DC: International Code
Council, 2018, section 102.6.
8 of 8
ResearchGate has not been able to resolve any citations for this publication.
Conference Paper
Full-text available
As engineers and builders developed modern steel framing in the late nineteenth century, the existing options for floors to span between steel beams were forms of masonry vaults. Many possible alter- nate floors were developed in the United States, but few had rational bases for design. Testing programs put in place by building officials in New York City promoted the use of certain systems in New York and, by providing a rationale for those systems, nationwide. Ten systems are described and analyzed.
Specification for the Design, Fabrication, and Erection of Structural Steel for Buildings
  • G Aus
G. Aus, "Tests of Long-Span Terra-Cotta Floor Arches," Engineering News, vol. 34, no. 19, p 314, 1895. (3) n.a., "Specification for the Design, Fabrication, and Erection of Structural Steel for Buildings," New York: American Institute of Steel Construction, 1936. (4) n.a., "Wire Fabric Reinforcing in Buildings," New York: American Steel & Wire Company, 1925.
observations on site at a 1950 government-owned building in New York. The details of the project are covered by a confidentiality agreement
  • D Friedman
D. Friedman, observations on site at a 1950 government-owned building in New York. The details of the project are covered by a confidentiality agreement.
The Planning and Construction of High Office-Buildings
  • W Birkmire
W. Birkmire, The Planning and Construction of High Office-Buildings, 4th edition. New York: John Wiley & Sons, 1906, page 113.
Tests of Long-Span Terra-Cotta Floor Arches
  • G Aus
G. Aus, "Tests of Long-Span Terra-Cotta Floor Arches," Engineering News, vol. 34, no. 19, p 314, 1895.
Analysis of Steel- Structure/Masonry-Wall Interaction inHistoric Buildings,” in Structural Analysis of Historical Constructions: Possibilities of Numerical and Experimental Techniques; proceedings of the Fourth International Seminar on Structural Analysis of Historical Constructions
  • D Friedman