TMS ONLINE | TMS PUBLICATIONS | SITE MAP An Article from the March 2003JOM: A Hypertext-Enhanced Article

VISIT THE JOM COVER GALLERY JOM MENU

Feature: Materials World

Analyzing and Characterizing the Steel Used at Frank Lloyd Wright’s Fallingwater

Louise Dean

---

Figure 1. Fallingwater, designed by Frank Lloyd Wright. Photograph by Robert P. Ruschak courtesy of the Western Pennsylvania Conservancy.

Author’s note: The units are both metric and U.S. customary, as the publications and the steel sections referred to used U.S. customary units.

INTRODUCTION

Fallingwater was designed by Frank Lloyd Wright in the mid-1930s as a country home for the wealthy Edgar Kaufmann family of Pittsburgh, Pennsylvania. The house is located in the forests of Mill Run, south of Pittsburgh. Spanning the Bear Run stream as a cantilever (Figure 1), Fallingwater, owned and operated by the Western Pennsylvania Conservancy, is recognized as one of the most significant works of American architecture.

In the past 68 years, restoration of the house has been necessary to provide structural stability and to accommodate tours of more than 145,000 people per year. In 2002, the most significant structural restoration work took place, predominantly to repair deflection of the largest terrace toward the stream and corrosion of the steel supporting the hatch steps. The goal of the restoration is to provide structural support without altering Wright’s use of materials. Fallingwater is an engineering marvel and second-guessing Wright’s designs and use of materials is not without risk.

The renovation has allowed samples of the original steel used for construction in the 1930s to be captured and analyzed. Two steels were sampled and tested; one represents the reinforcing bar used in the largest terrace and the other is from the hatch steps. The reinforcing bar was exposed when the new hatch steps were mounted in the concrete above it. Westmoreland Mechanical Testing & Research performed all the testing and analyses.

FALLINGWATER REBAR CHARACTERIZATION

A sample of the original 0.635 cm (0.25 in.) diameter rebar was taken for analysis. Both 1.58 cm (0.625 in.) and 0.635 cm (0.25 in.) rebar was used to reinforce the concrete. The reinforcing bar had not corroded or yielded. Restoration of the concrete itself was performed in a limited area exposing a spot where a sample of the smaller diameter rebar could be taken (Figure 2). The chemical composition was determined and would conform to the 1948 SAE No. 1018 or 1021 (Table I).

Table I. Chemical Analysis of Fallingwater Rebar Sections and Typical SAE No. Compositions
	C*	Mn	P	S*	Si	N*	Al
0.635 cm (0.25 in.) rebar	0.19	0.64	0.03	0.030	0.3	120 ppm	0.006
1018	0.1–0.20	0.60–0.90	0.040	0.030	—	—	—
1021	0.18–0.23	0.60–0.90	0.040	0.030	—	—	—
* Leco techniques used, Mn by OES and ICP for remaining elements for 0.635 cm (0.25 in.) rebar

These are nominal compositions of that alloy in 1948 per Reference 1 because, as in 1939, no SAE grades fit the actual composition. However, Bessemer Steel reinforcing bar did (Table II).

Table II. Bessemer Steel Compositions for Reinforcing Bar
Bessemer Steel	C	Mn	P	S
Reinforcing bars2	0.15-0.35	0.70 max.	0.11 max.	0.08 max.
Reinforcing bars3	0.15-0.35	0.70 max.	0.12 max.	0.08 max.
Reinforcing bars3	0.08-0.15	1.00 max.	0.12 max.	Normal or added

The 0.635 cm (0.25 in.) rebar was likely made by the Bessemer steelmaking process in a Pittsburgh area mill. An article published at around the time Fallingwater was built, 1935–37, shows the typical products of Bessemer Steel and includes reinforcing bars.² The composition shown in Table II matches that of the rebar in Table I (Reference 2 is from 1939, Reference 3 is from 1948 and shows two grades of Bessemer reinforcing bar had been developed by then). Reference 3 also contains a footnote on reinforcing bar compositions that indicates by that time, specifications governing reinforcing bars prescribed mechanical tests rather than chemical compositions. In 1935, over 2.7 million tonnes of Bessemer steel were produced in the United States.⁴ Open hearth was the predominant process, with over 30.8 million tonnes produced in the United States.⁴ Bessemer and openhearth furnaces are no longer used in the United States.

In Bessemer steel, pig iron was charged into the vessel and air was blown through it, oxidizing the silicon, manganese, and carbon (in that order). Pig iron for the Bessemer furnace contained 1.00–1.80%Si, 0.085–0.100%P, 3.80–4.00%C with Mn under 0.60% and S between 0.03–0.08%. The manganese in the steel would have been restored by adding FeMn to the ladle. Aluminum is only present in the rebar at a residual level, indicating none was added to the heat. The high nitrogen content of the rebar, at 120 ppm, vividly points to the Bessemer as being the steelmaking process used, as air was used to blow the heat. Pig iron for the Bessemer furnace was required to contain very low phosphorus contents, which was possible from the use of low-phosphorus iron ores mined in the United States. In 1875 there was a Bessemer furnace at the Carnegie Steel Company’s Edgar Thompson Works in Braddock, Pennsylvania (now part of U.S. Steel) and at Aetna Standard Iron and Steel in Mingo Junction, Ohio (now part of Wheeling- Pittsburgh Steel). It is quite possible the rebar for Fallingwater was produced at one of these nearby mills.

Figure 2. A section above the hatch steps of Fallingwater, showing exposed 0.635 cm (0.25 in.) diameter rebar and hatch steel corrosion. Photograph by Louise Dean courtesy of the Western Pennsylvania Conservancy.

FALLINGWATER HATCH STEEL CHARACTERIZATION

The restoration of Fallingwater included removing the original hatch steps to replace the steel strap supports with more corrosion-resistant steel. The steel strap supports and all the visible steel used at Fallingwater was painted Cherokee red. Fallingwater’s preservation philosophy states that Wright chose the red color as he felt it was suggestive of both iron ore and the fiery method used to create steel. The hatch steps were connected to the house using these steel straps, with the steps going down to the stream itself. Approximately 68 years of exposure to the environment above the Bear Run stream had corroded this original carbon steel, particularly near the first floor terrace (Figure 2). The steps were made of rolled flat bars 5.87 cm × 0.79 cm (2.3125 in. × 0.3125 in.). The steel supports had been painted since Fallingwater was built.

Samples of steel from the hatch steps were tested for hardness, microstructure and micro-cleanliness, chemical composition, tensile strength, yield strength, reduction in area, and elongation. The microstructure and tensile properties were determined in both the longitudinal and transverse directions.

Table III shows the chemical analysis results for the hatch steel. The composition is typical of a 1939 SAE No. 1020 steel, with low levels of impurities for steel produced at that time (phosphorus and sulfur). The steel was deoxidized with aluminum and silicon and was likely produced by the basic open-hearth process and semi-killed ingot cast. Semi-killed products were made when the carbon was ~0.2%, the silicon was <0.04%, and the aluminum was <0.01%. Open-hearth steelmaking was the predominant method used in the United States from the 1930s through the 1960s. Both Jones & Laughlin Steel Corporation and the Carnegie Steel Company operated open-hearth furnaces in the Pittsburgh area in the 1930s.

The mechanical properties are shown in Table IV. The hatch steel exhibited a yield point, which is typical in hot-rolled steels (and not in cold-rolled steels). It is of interest to note that a paper written in 1936 by the metallurgist at Jones & Laughlin Steel Corporation, Pittsburgh,⁵ showed the same level of mechanical properties as found here for 1020 hot rolled bar: 407.5 MPa (59.1 ksi) ultimate tensile strength, 263.4 MPa (38.2 ksi) yield point, 37% elongation, and 62% reduction of area. This suggests that the steel used in the hatch steps was typical of steel production. Wright must not have specified a grade or condition that varied from common grades available at that time.

Table III. Chemical Analysis of the Fallingwater Hatch Steel
	C*	Mn	Si	Al	P	S*	Cu	Cr	Mo	Ni
Hatch steel	0.217	0.374	0.027	0.016	0.007	0.018	0.011	0.006	0.024	0.006
1020	0.15–0.25	0.3–0.6	—	—	0.045	0.055	—	—	—	—
* Leco techniques used, remaining elements were determined by OES

Table IV. Mechanical Properties of Fallingwater Hatch Steel
Tensile Sample	Ultimate Tensile Strength (MPa) (ksi)	Yield Point (MPa) (ksi)	0.2% Yield Strength (MPa) (ksi)	Elongation (%)	Reduction of Area (%)	Modulus (MPa) (msi)
Longitudinal	424.7 (61.6)	279.9 (40.6)	230.3 (33.4)	40	67	174,437 (25.3)
Transverse	446.8 (64.8)	237.9 (34.5)	231.7 (33.6)	33	60	213,738 (31.0)
Hardness	65.59 Rockwell B

Figure 3. Inclusions in the hatch steel. The microcleanliness per ASTM E-45 was found to be 1.0 thin D oxide inclusions and 1.5 thin A sulfide inclusions.

Figure 4. The pearlite and ferrite microstructure of the hatch steel, etched with 1% Nital.

The microcleanliness of the hatch steel is shown in Figure 3 and Table V. The sample was mounted, polished, and viewed under the microscope at 100X to determine the micro-cleanliness in both the longitudinal and transverse directions. It was found that the steel was clean for steel produced during that era, with few inclusions in the longitudinal direction and a rating according to ASTM E-45 of 1.0 thin D oxide inclusions and 1.5 thin A sulfide inclusions. In the transverse direction, the cleanliness was found to be the same. ASTM E-45 is not used to rate cleanliness in the transverse direction; however, the rating is provided here for general information. The microstructure consisted predominantly of ferrite with a lesser amount of pearlite (Figure 4).

Table V. Micro-cleanliness Results to ASTM E-45
Sample	Type D Oxide	Type A Sulfide
Longitudinal	1.0 Thin	1.5 Thin
Transverse	1.0 Thin	1.5 Thin

The steel was found to conform to hot-rolled 1020 grade with a higher cleanliness than typical of steel produced in the 1930s. As with the rebar, it would have been produced prior to the use of the basic oxygen process or continuous casting. It was likely made using an open-hearth furnace, cast in ingot molds, and rolled to the bar shape using handmills. During this time little use was made of desulfurization, leading oftentimes to larger inclusions. In this case, the steel cleanliness was higher than expected. The largest areas of corrosion occurred near the terrace and were observed to be on the order of 3.8 cm (1.5 in.) of steel corroded from the original bar size. That translates to a corrosion rate of approximately 0.06 cm/y (3.8 cm/68 y) [0.02 in./y (1.5 in./68 y)]. These flat steel bars had been painted since installation and severe corrosion had occurred only at the concrete-bar interface.

CONCLUSIONS

When Frank Lloyd Wright designed Fallingwater, he must not have specified a steel grade or condition that varied from common production available. Yet, it is notable that the Bessemer steel reinforcing bar had not deteriorated during service in the concrete terrace. Also, the open-hearth hatch steel performed well, as severe corrosion occurred only at the concrete-bar interface while the remainder of the flat bar, also exposed above the Bear Run stream for 68 years, showed only pitting corrosion. Fallingwater has timeless interest as an architectural achievement, and some of the steel used to produce this landmark was made using techniques whose time has passed.

References

1. Metals Handbook (Metals Park, OH: ASM, 1948), p. 307.
2. Metals Handbook (Metals Park, OH: ASM, 1939), pp. 778–781.
3. Metals Handbook (Metals Park, OH: ASM, 1948), pp. 320–322.
4. Making, Shaping and Treating, 8th edition (Pittsburgh, PA: United States Steel Corp., 1964), p. 439.
5. J.E. Beck, “Cold Forming Processes-Drawing Rods and Bar,” The Working of Metals, (Metals Park, OH: ASM, 1937).

For more information, contact Louise Dean, Westmoreland Mechanical Testing & Research, PO Box 388, Youngstown, PA. 15696-0388; (724) 537-3131; e-mail louise@stargate.net.