Structures Facilities Structures Research Structures Professors Structures Home Structures Facilities Structures Grad Program Kavanagh Lectures Structures Facilities Structures Grad Program Kavanagh Lectures Structures Facilities Structures Grad Program Structures Grad Program Structres Links Structures Seminars Structures Short Courses Structures Grad Program Structres Links Structures Seminars Structures Short Courses

 

 

 
 

click thumbnail for larger picture of Falling Water

 
     
Project Title: Fallingwater – Saving the Masterpiece: Field Monitoring and Analysis During the Rehabilitation of Frank Lloyd Wright’s Signature House – Fallingwater.
Investigator: Dr. Andrea Schokker
Research Assistants: Justin D. Brennan
   
Funding Agency: (No External Funds)
   
Project Duration: Mid-November 2001 – Mid-March 2002
References: Silman, Robert and John Matteo. Repair and Retrofit: Is Fallingwater Falling Down? Structure. July/August 2001. pp. 24-30.
  Hoffman, Donald. Frank Lloyd Wright’s Fallingwater: The House and Its History. Dover Publications. Second, Revised Edition. 1993
   

General Background:

To escape the hustle and bustle of city life in downtown Pittsburgh, Edgar and Lillian Kaufmann, owners of the Kaufmann department store chain, decided to build a summer retreat in rural area of Southwest Pennsylvania known as the Laurel Highlands. The grandiose setting of a waterfall on Bear Run was chosen as the site of the house. Due to the prestige of the Kaufmann family, and the beauty of the natural setting, Frank Lloyd Wright took interest and control of the designing of the house called Fallingwater. Construction began in 1936. In 1963, Edgar Kaufmann, Jr. donated the house to the Western Pennsylvania Conservancy.

Figure 2 Figure 3

The Problem:

In his design, Wright utilized multiple cantilevered terraces, to match the theme of the waterfalls. These terraces were designed of reinforced concrete. Wright had previously exploited the use of reinforced concrete, but had not undertaken such an extensive endeavor using this material as he would in Fallingwater.

Upon casting the lower terrace that cantilevered over the creek and the waterfall, large deflections were instantly discovered. Some of this initial deflection was due to the neglect to consider the wet weight of concrete by setting the forms level, instead of cambering them upward. The majority of the remaining deflection was caused by a lack of tensile reinforcement in the main inverted tee-beams that supported the lower terrace.

The lack of tensile reinforcement was not overlooked. In fact, the contractor added seven reinforcing bars to the beams after noticing the mistake. This infuriated Wright who would later say that the excessive weight of the added bars actually led to the excessive deflections. Regardless, even the addition of the extra rebar was not enough to control the nearly 8” of deflections of the end of the lower terrace.

Though the deflection is a great cause of concern, the cracks caused by the deflection also is of interest. There are a large number of cracks on the main beams of the lower terrace, as well as on the supporting concrete floor joists, and the upper terrace’s parapets, clearly evident in Figure 2.

Another interesting aspect of Fallingwater concerns the window mullions - the vertical steel tee-sections of the window frames that are embedded in the lower terrace parapet and extend into the upper terrace. These mullions were most likely considered as a non-structural element. However, finite element analysis has shown that the mullions act as structural elements and transfer loads from the upper terrace to the lower terrace. The mullions are shown in Figure 3.

Because of the structural deficiencies in Fallingwater, the Western Pennsylvania Conservancy decided to hire the engineering firm, Robert Silman and Associates of New York to investigate the house and to propose a solution to mitigate the problems. The house is now temporarily shored with steel to assure the safety of the structure. The steel shoring can be seen clearly in Figure 4.

Figure 4

The Solution:

Due to the historic nature of the house, many otherwise suitable methods of rehabilitation could not be considered due to the fact that the rehabilitation would alter the structure’s original appearance. In other words, the option that least changes the original nature of the house is the most optimal. Because of this, the beams are being externally post-tensioned. Post-tensioning is a construction/design method that utilizes high-strength steel cables or strands. The strands are typically bundled in a duct and anchored at one end of a span and pulled to a specified tension. Afterwards, the strands at the other end of the span are anchored. This pre-compresses the concrete, thus reducing tension. This is highly beneficial since concrete can handle compression very well and tension in concrete leads to cracking.

External post-tensioning generally specifies that the duct is not cast throughout its length inside the concrete member, but rather runs through anchor blocks and deviators attached to the sides of the beam. The deviators provide the necessary moment-arm to counteract some of the beams loads. External post-tensioning is favorable because after the stone floor is replaced, there will be little evidence that any work even took place. When the strand will be tensioned, the lower terrace will rise upward. Though it is very possible to tension the strands enough so as the lower terrace will rise the 8 inches required to be level again, it is highly unfavorable. This is because doing so will crack non-structural materials, like windows, since the non-structural materials have had many years to deflect along with the terrace. Therefore, the terrace will only be raised about an inch – enough to make the house safe.

Figure 5

Penn State's Contribution:

One of Penn State’s main objectives in the Fallingwater rehabilitation project is to successfully monitor the house through a variety of different instruments throughout the many different stages of the construction. One important consideration is not to permanently alter the house by installing the instruments. Because this was such an important consideration, all of Penn State’s instruments were chosen such that they would do little to no harm. Gauges are either clamped or bonded with epoxy. The epoxy marks can be sanded and repainted after the gauges are removed. Penn State can be shown instrumenting Fallingwater in Figure 5.

Penn State is monitoring multiple effects during the entire construction process. First the deflection of the terrace throughout the construction process is monitored with dial gauges. The window mullions (vertical components of the window frames), though they are non-structural items actually are taking some of the loads. Because of this, the strains in these mullions are being monitored through all stages of construction with use of vibrating wire gauges and Demec gauges. Strains are also being monitored with vibrating wire gauges on the three beams that support the lower cantilevered terrace during the post-tensioning process only. Lastly, the widths of the cracks in the cantilever beams are being monitored, as the cracks will close when the beams are compressed. It is also of interesting note that the company performing the post-tensioning operation is also taking deflection measurements during the tendon jacking through the use of the linear variable differential transducer (LVDT). Deflection results will then be compared.

 
Figure 6   Figure 7

Instrumentation Types used by Penn State:

Dial Gauges (Deflection) – See Figure 6
To monitor the deflection in the cantilever throughout the entire construction process, six dial gauges have been installed under the lower terrace slab, attached to the temporary shoring. Dial gauges work through the use of a spring coil attached to a piston. The piston is always in contact with the bottom of the cantilever slab as long as the reading is in range. This is due to the spring force in the gauge. As the slab rises or falls, the piston extends or compresses, respectively. The dial gauges in use at Fallingwater are sensitive to 0.001” and have a range of two inches.

Vibrating Wire Gauges (Strain) – See Figure 7
To monitor the strain (which is directly proportional to the change in length) in the window mullions, columns of the stairwell to Bear Run Creek, and the main cantilevered beams, vibrating wire gauges (VWG) are utilized. The vibrating wire gauge is an electrical resistance strain gauge that uses an electrical current to determine vibration frequency, and therefore is converted to strain. The gauge is epoxied to the mullions, columns, and cantilevered beams, although it can be welded to the steel mullions and columns as well. Readings are taken by attaching the VWG housing to a readout box, which compensates for temperature and gives strain directly. Four VWG’s are attached to the four south-facing window mullions and three are attached to each of the two columns that support the stairs to the Bear Run Creek. The cantilever beams have five vibrating wire gauges attached at the high moment area of the beam: one on the center of the top face of the beam, one on the center of the beam on the slab underneath, and the other three are attached to the side of the beam, evenly spaced. This will allow a strain distribution to be determined during the process of post-tensioning. This method is highly beneficial, since it will allow both the reader to be out of the way of the post-tensioning process to assure safety while still providing near-continuous strain data.

Demec Strain Gauges (Strain) – See Figure 7
To provide a check for the strain readings in the window mullions taken be the Vibrating Wire Gauges, which are electrical resistance strain gauges, a Demec strain gauge system is in use as well. The Demec strain gauge system is a mechanical strain gauge system, contrary to the VWG’s. The Demec system incorporates a reference bar that is used to set two initial Demec studs. After the studs are initially placed, subsequent readings are taken using a Demec gauge, which closely resembles a Vernier Caliper scale. The Demec gauge measures the change in length, taring the gauge to the reference bar every time. The change in length and temperature are then used to determine a strain reading. Four mullions are being monitored at this time.


       
     




Structures Home CEE Home Contact