Fallingwater: Restoration and Structural Reinforcement

Tyler Meek

Abstract

Fallingwater, Frank Lloyd Wright's masterpice design, was a radical departure from the typical residence of the time period in which it was designed. The reinforced concrete construction with long multiple cantileverd elements pushed the limits of existing construction technology and capabilities. The structural design proved to be problematic as the main cantilevered beams were determined to be under-reinforced. This under-reinforcement combined with an ongoing creep condition in the concrete had caused deflections as high as 7 inches on a 15 foot cantilever. Under such conditions, the reinforcing steel was continuing to yield, placing Fallingwater in position of imminent collapse.


Keywords


Cantilever, Post-tensioning, Frank Lloyd Wright, Fallingwater, Restoration, Repair, Shoring, Deflection, PSU
Fallingwater_049.jpg
Figure 1: Wright's Falling Water (Meek Photo)



History

Frank Lloyd Wright was contracted by Edgar Kauffman Sr. to design Fallingwater as a weekend home for his family in 1934. Edgar Kauffman Sr., a successful business owner, was convinced by his son, Edgar Kauffman Jr., to hire Wright as the architect. Edgar Jr. had been exposed to Wright's mastery of bold, mature aesthetics while working in Taliesin, Wisconsin, home to one of Wright's architectural studios.


In December 1934, Wright visited the site the Kauffmans had selected and was immediately inspired by the presence of Bear Run Creek and the waterfall it presented. He later wrote, "The visit to the waterfall in the woods stayed with me, and a domicile has taken vague shape in my mind to the music of the stream." Although Wright was immediately inspired; he did not start transferring his ideas to paper until the late summer of 1935 when the Kauffmans scheduled a visit to his studio. On the morning of the Kauffmans' arrival, Wright started and completed the floor plans for Fallingwater. The elevations were finished later that night by apprentices while Wright was out to dinner with the Kauffmans. Between these impromptu drawings and the construction set, very little changed.

While the Kauffmans had expected the residence to be situated below the waterfall, Wright placed it directly on top of it. Wright believed that if the Kauffmans were provided a daily view of the waterfall, they would lose appreciation for its power and magnitude. To avoid this desensitization, Wright created the illusion that the stream had eroded the structure’s foundation over many years and therefore magnifying the power of water.

Construction began in 1936 and upon completion in 1937; Wright wrote "Fallingwater is a great blessing - one of the great blessings to be experienced here on earth." (Lemley, 2003) Fallingwater was donated to the Western Pennslyvania Conservancy in 1963 by Edgar Jr. and was opened to the public in 1964.

Additional information can be found on the Fallingwater website.


Structure

Fallingwater.2.jpg
Figure 2: View of Bolsters and Cantilever (fallingwater.org)


Fallingwater uses four main piers or bolsters for its foundation, three of which are reinforced concrete with the
fourth being stone masonry. In the image shown, the four bolsters can easily be seen below the first floor cantilever. Four cantilevered beams project 15’ from these piers suspending the first floor out over the stream. Concrete joists measuring 4” x 23" spaced at 4’ on center span between the beams and transfer the first floor loads to these cantilevers. To increase the strength of the cantilevered floor system, Wright designed a 4” concrete slab that was monolithically cast with the underside of the beams and joists. Moving the concrete slab from above the joists to below them, placed the majority of the concrete in the compression face of the cantilever thus creating a very efficient inverted T-beam system.


The cantilevered portion of the second floor, which served as the master bedroom terrace, extends 6’ beyond the floor below. It consists of two edge beams with joists and slab spanning between them. Before the structure was forensically investigated, it was believed that the second floor cantilever acted independently of the first floor. On the contrary, four steel T window mullions support the south end of the second floor cantilever and transfer its load to the edge of the first floor cantilever.





The Problem

Deflection issues and questions of sufficient strength have been associated with Fallingwater ever since the initial construction process. When the formwork for the first floor cantilever was removed, the construction workers recorded a downward movement of 1.75 inches. A small amount of deflection is normal when scaffolding is removed but in the case of 1.75 inches, the bending was extremely pronounced. A problem with the structural design became more apparent with the completion of the second floor cantilever supporting the master bedroom terrace. Soon after the formwork was stripped, two large cracks formed in the terrace’s parapet.

In 1937, Kaufmann Sr. contracted Metzger-Richardson to conduct load tests on the structure and determined that stresses in the cantilever exceeded the appropriate margin of safety. Metzger-Richardson suggested placing props below the first floor cantilever to prevent the structure from collapsing into the stream. Wright convinced Kaufmann Sr. that his design was adequate and Kaufmann decided to not use any type of propping system and proceed with the building as planned. Although Kaufmann Sr. listened to Wright and did not utilize a propping system, he commissioned a surveyor to monitor and measure deflections on a regular basis. These measurements were performed from 1941 until Kaufmann Sr. died in 1955.

When Robert Silman Associates was hired to examine the houses structural issues in 1995, a total deflection of 7” was measured at the edge of the 15’ cantilever. During an initial probing investigation, a very large crack was found in the center of one of the cantilevered beams that effectively formed a plastic hinge at the face of the adjoining bolster. Pennsylvania State University was employed to monitor the deflections and gauge the size of the cracks over an 18 month period. By recovering the past deflection data and comparing the data from Penn State, it was determined that the deflections were worsening and the cracks continued to grow.

Robert Silman Associates created a structural model using the SAP 90 finite element program. Initial models used uncracked sections and showed a tendency for the load to be carried by the master terrace edge beams and parapets due to their higher relative stiffness. Using the properties observed in the field, a new model was created that produced values that corresponded well to the real deflections. This new model was used as a non-destructive method to analyze the structure. Forces in the structural elements were found in the computer model to be approaching failure. The concrete, which has an ultimate strength of 5,000 psi, was stressed up to 4,380 psi and the reinforcing steel was stressed up to 41,720 psi which is approaching its yield strength of 42,000 psi.

To further investigate the state of the cantilevered beams the floor system had to be removed. During this process, unexpected damage was discovered; the existing floor joists and supporting members were found to be severely deteriorated along the eastern side of the structure. Four of these joists had to be removed and replaced due to extreme cracking. It is believed that this damage was not a result of an insufficient structure but it was caused by an incident when a tree fell and hit the structure during a storm.


Causes


The main cause behind all the deflection issues was insufficient structural capacity of the cantilevered girders. As one of the first floor cantilevers was being constructed, the reinforced concrete contractor, who was also an engineer, noticed that there were only 8 reinforcing bars in the girder which is a very small amount of reinforcing steel for the particular girder and expressed his concern to Mr. Kaufmann. This contractor ran his own set of calculations and determined that the number of reinforcing bars should at least be doubled to 16. Mr. Kaufmann passed the concern on to Wright who took the correction as a personal attack. An infuriated Wright wrote back to Kaufmann Sr. saying, “I have put so much more into this house than you or any client has a right to expect that if I haven’t your confidence – to hell with the whole thing.” After Kaufmann “smoothed Wright’s ruffled feathers” (Lemley), trust was regained but the extra reinforcing was placed in the girder.

When the first floor cantilever deflected 1.75” immediately after the formwork was removed, suspicions were raised again. Mendel Glickman, the engineer who designed the cantilever, was notified by telephone of this deflection and after a brief check of his calculation; he is reported to have said, “Oh my God, I forgot the negative reinforcing!” (Feldman, 2005) This error clearly explains why the cantilever deflected so much upon removal of the formwork. Without sufficient reinforcing steel in the top of the girder, the existing steel most likely reached its yield limit and began to elongate. The increasing deflection of the cantilever proves that the steel had yielded and it was only a matter of time before it failed completely.

The concrete used in Fallingwater may not have been as consistent as one would expect in today’s construction industry. Workers were forced to make thousands of small concrete batches by hand in the field because the remote site was unreachable for concrete trucks. The amount of skilled workers was also limited because at the time of construction, the Works Progress Administration had just formed and was recruiting the area’s top stonemasons for work on government jobs.
As mentioned in the History section, the drawings for Fallingwater were developed very quickly. Wright had visited the site only once before creating these drawings and the site visit had been months earlier. Because of this long stretch between inspiration at the site, and putting his imagination onto paper, there must have been issues that Wright overlooked or forgot in the rush to finish the drawings before the client’s arrival. Very few changes were made to the drawings from these preliminary drawings and the drawings that were used in the field during construction. Robert Silman, of Robert Silman Associates, suspects that the engineers were being rushed by the contractors to produce the structural drawings. (Feldman, 2005)

The four larger window mullions that support the second floor terrace and transfer load to the first floor cantilever are evidence of either poor communication with the structural engineers or a last minute adjustment to support an undersized cantilever.


Restoration Techniques


steel_truss_shoring.jpg
Figure 3: Steel truss used to support the sagging cantilever (STRUCTURE Magazine)

Based on the findings of the 1995 structural inspection, the Western Pennslyvania Conservancy choose to place temporary shoring beneath the main level terrace to carry the load and stop deflections until a permanent solution was completed.
The conservancy had many requirements for the permanent solution but the two main requirements were strength and aesthetics. The solution had to be strong enough to completely stop any and all deflections but at the same time be respectful to the historical building and be relatively inconspicuous to visitors.

Many options were considered by the owner and engineer. The options discussed are best presented in a bulleted list:
  • Simply leave the shoring in place permanently
  • Section enlargement
  • Carbon fiber strengthening
  • External post-tensioning

External post-tensioning was selected as the appropriate solution for Fallingwater because the system designed was “relatively lightweight and therefore did not add a significant amount of additional dead load to the already overstressed girders.” (Loper & Hughes, 2003) Post-tensioning was not possible in the second floor cantilever because there was not a sufficient back-span. Because the second floor cantilever could not be reinforced, the post-tensioning had to be strong enough to lift the entire cantilevered structure and monitored closely enough to avoid damaging the original concrete or the structural mullions on the second floor. Before the post-tensioning tendons were tightened, the structural mullions were reinforced by welding additional plates to them.

Post tensioning tendons consisted of ½” diameter, 7-wire low relaxation strands to allow the tension stress to be applied slowly to avoid causing more problems in the original materials. Thirteen-strand tendons were used on each side of two of the cantilevered girders while only a ten-strand tendon was used on the western girder. Eight monostrand tendons were used to span in the east-west direction. The ten-strand and thirteen-strand tendons were tensioned to an approximate jacking force of 300 and 390 kips, respectively. The eight monostrand tendons were tensioned to approximately 43 kips each.
PT_graphic.jpg
Figure 4: Graphic representing the moment imposed by post-tensioning: Permission Pending


Pennsylvania State University and the post-tensioning contractor installed a strain and deflection monitoring system during the stressing process. Strain gages were used on the structural window mullions and concrete girders. Penn State Civil Engineering Professor Andrea Schokker explained,“This system was extremely useful because it provided immediate and continuous feedback during stressing operations and helped us to cross reference estimated deflections from the engineering analysis. In fact, we were able to use strain gage data to make slight adjustments during the final stage of post-tensioning. This fine-tuning helped ensure relatively uniform stresses across the structure.” (Loper & Hughes)


When the post-tensioning strands were in place and tightened, the positive moment caused by the eccentric axial force essentially balanced the negative moment caused by the cantilever. The existing deflections could not be removed because restoring the structure back to its original form would cause more cracks to open due to the years of creep associated with the deflected shape. After the tendons were tightened, the cantilever raised about ¾” off the temporary shoring.

The structural elements that were used and remain on the structure are completely invisible to a typical visitor and the conditions are back to near original. When asked about the deflection that was left in the cantilever portion of Fallingwater’s structure, Silman replied, “that sag is part of the story of the building.”








Closing Remarks

Similar to the Leaning Tower of Pisa, Fallingwater is venerated for its structural flaws. Although structural repairs had to be made to Fallingwater, many people with Wright when he name Fallingwater one of the great blessings that can be experienced on earth.

Richard Cleary, a Wright Scholar and architectural historian at the University of Texas at Austin, is one of those people. He believes, “The house is a real presence in the American psyche. [The] view of it over the waterfall is a sort of perfect American fantas. It showed we culd have both technology and the natural world, and make them work beautifully together.” (Lemley, 2003)

Multiple architects have come to the defense of Wright’s design and the impending structural problems, claiming that the structural flaws do not reduce Fallingwater’s value. Michael Sorkin, the director of the graduate urban design program at City College and Robert A. M. Stern, dean of the Yale School of Architecture, are two such architects.

Mr. Sorkin says, “After 70 years, a little structural problem on the best house ever designed doesn’t strike me as something that raises any kind of revisionist issues in terms of Wright’s career. He was working on some kind of edge, and a certain amount of risk is entailed.” (Wald, 2001)

Dr. Stern takes a similar stance on the subject by concluding, “When you’re involved with an experiment, you’re often ahead of the curve.” (Wald, 2001)

Both Dr. Stern’s and Mr. Sorkin’s views on the matter revolve around the same idea. Wright’s gravity defying design for Fallingwater was ahead of it’s time. Had it been built using today’s techniques and materials, there would have been no structural issues. Part of what makes Fallingwater so fantastic is the era that it was built in. Wright took what materials he had at his disposal and created the perfect blend of nature and technology by designing something that no one else would have.




References

  • Brown, J. (February 2002). "Fallingwater Restoration Uncovers More Damage." Civil Engineering, p. 24
      • Brown discusses the unexpected spalling and cracking discovered during the restoration process, as well as what may be done to correct these issues.

  • Dean, Louise (March 2003). "Analyizing and Characterizing the Steel Used at Frank Lloyd Wright's Fallingwater." Materials World (Retrieved 9.27.2010)
      • As the name implies, this article produces an in depth, detailed analysis of the reinforcement used in the original design of Fallingwater and compares its characteristics and behavior to that of modern concrete reinforcement.

  • Feldman, Gerard (September 2005). "Fallingwater is No Longer Falling." STRUCTURE magazine, pp. 46-50
      • An article from STRUCTURE magazine discusses the placement of reinforcement both before and after the repairs of the cantilevered beams supporting the floor slabs.

  • Gonchar, J. (March 2002). " 'Wrighting' a Fragile Landmark Sagging for Nearly 65 Years." Engineering News Record, pp. 84-90
      • This article provides a general background to the overall rehabilitation process and the main goals of the process.

  • Silman, Robert (September 2000). "The Plan to Save Fallingwater." Scientific American.
      • Written by Robert Silman, president of Robert Silman Associates, this document is a firsthand account of the structural repairs to Fallingwater.

  • Silman, Robert and Matteo, John (September 2001). "Repair and Retrofit: Is Fallingwater Falling Down." STRUCTURE magazine (Originally Retrieved 9.20.2007) Viewed 9.27.2010
      • The precursor to "Fallingwater is No Longer Falling," this article discusses the other options that were discussed but not put into action. The modeling and predesign is also discussed here.

  • Lemley, B. (January-February 2003) "Saving Fallingwater." This Old House, pp. 84-90
      • A great appreciation for Fallingwater shines through in Lemley's article which gives a wonderful back-story to the design of the residence and a detailed description of the prescribed and carried out repair process.

  • Loper, James and Hughes, Jason (April 2003). "Post-Tensioned Retrofitting Maintains Landmark's Aesthetics." Concrete International, (Retrieved 9.27.2010)
      • A few less known topics are discussed here including: inspection and evaluation methods, repair system selection and a few repair process challenges.

  • Wald, Matthew (September 2001). "Rescueing a World-Famous but Fragile House." New York Times 2, (Retrieved 9.27.2010)
http://www.nytimes.com/2001/09/02/us/rescuing-a-world-famous-but-fragile-house.html?scp=1&sq=&st=nyt&pagewanted=2
      • This article gives a simplified introduction and history to the structural issues that Fallingwater faces along with the proposed solution.