S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8 FA R M E R & K E N N E Y • 1 A TALE OF TWO STONES: EVALUATION OF A STONE PANEL CLADDING SYSTEM MATTHEW C. FARMER WISS, JANNEY, ELSTNER ASSOCIATES, INC., FAIRFAX, VA CLARISSA S. KENNEY WISS, JANNEY, ELSTNER ASSOCIATES, INC., FAIRFAX, VA ABSTRACT Looks are not always what they seem. The evaluation and analysis of problems associated with a natural stone cladding system involving two distinctly different types of natural stone will be presented. What began as a response to a single isolated panel failure of one of the stone types led to some surprising test results with respect to both stones. Our discussion will highlight the details of our investigative findings, the results of material strength testing, and the impact of each stone’s unique characteristics on our perceptions of these cladding materials, and the resulting challenges to developing appropriate recommendations for this unique set of project conditions. SPEAKER MATTHEW C. FARMER — WISS, JANNEY, ELSTNER ASSOCIATES, INC., FAIRFAX, VA MATTHEW C. FARMER joined the New Jersey office of Wiss, Janney, Elstner in 1986. Since then, he has been involved with numerous evaluations of concrete, steel, and timber structures, as well as those involving clay, concrete, stone, and cast-stone masonry. He has concentrated his practice in the area of building envelope cladding systems design, investigation, analysis, and repair, including numerous engagements as an expert witness. Mr. Farmer served as manager of the Fairfax office from 1994 until 2006, when he became a principal with WJE. Farmer is a graduate of the University of Colorado and Cornell University. He is licensed as a professional engineer in the District of Columbia, Virginia, and Maryland. COAUTHOR CLARISSA S. KENNEY — WISS, JANNEY, ELSTNER ASSOCIATES, INC., FAIRFAX, VA After completing an undergraduate dual degree program in both architecture and civil engineering at The Catholic University of America, CLARISSA KENNEY joined the Washington, DC, office of Wiss, Janney, Elstner in 2004. Since then, Ms. Kenney has been involved with the investigation and repair design of existing structures, including several landmark buildings listed on the National Register of Historic Places, which have required the development of repair drawings and specifications as well as on-site construction observation and field-testing services. Her investigative experience includes a wide range of building façade materials, including clay brick, concrete masonry, cast stone, concrete, metal panel systems, and numerous stone veneer panel systems. 2 • FA R M E R & K E N N E Y S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8 INTRODUCTION Looks can be deceiving; the presence of visible defects and deficiencies do not always represent the true character and durability of a material, nor does the absence of defects necessarily indicate a material with no flaws. After owner-implemented building expansions were conducted several times in the past 70 years, the subject building, located in the mid-Atlantic Coast region, underwent revitalization efforts during which a natural stone cladding system was applied in the 1970s to unify the appearance of the adjoining buildings of different vintages. Two distinctly different stone panels, serpentine and white marble, were used to architecturally unite the building as well as modernize its appearance. What began as an investigative response to a single isolated failure of an approximately 35-yearold serpentine cladding panel led to some surprising findings and test results as well as a realization that opinions of stone cladding quality after a visual assessment and initial assumptions may not be the same after a more in-depth evaluation. In this instance, the results of materialstrength testing resulted in contradictory perceptions of the stone used and caused modification of the approach taken to develop appropriate recommendations for this unique cladding system and its unusual project conditions. BACKGROUND History of Construction The building is five stories tall, but approximately 105 feet in height. The original building has a concrete structural frame with an exterior composed of limestone and brick masonry. Several expansions have occurred that both widened the building and increased its height. The expanded portions of the structure are constructed of an expressed concrete frame infilled with concrete masonry units. A façade comprised of 2-cm-thick stone panels was installed as a barrier system over the new combined structure (with the exception of a portion of the east elevation) approximately 35 years ago, in an effort to unify the appearance of the multiple buildings and additions. The stone cladding panels vary in size, but the most common panel dimensions are 27 inches wide by 54 inches tall. The cladding panels are made of two types of stone: a locally quarried green/black serpentine (see Figure 1), and a white marble quarried domestically (see Figure 2). Minimal documentation was retained over the years regarding building repair and maintenance; however, we found that the sealant at joints between stone panels was in poor condition at the time of our investigation, and supplementary anchors were installed along the lowest row of marble panels at each elevation. To support the new stone cladding over both the original 1930s limestone masonry structure and the newer walls with concrete masonry unit infill and expressed concrete floor slabs, a supplemental metal framing system consisting of steel “unistruts” was used (see Figure 3). The vertical members generally align with vertical stone panel joints and are attached to the concrete floor slabs or supplemental steel backup structure where required, using steel clips. The steel back-up is attached to the original façade using a combination of throughbolts, mechanical expansion anchors, and chemical fasteners. “Continuous” horizontal aluminum anchor rails span the unistruts and engage the stone cladding at kerfs (slots cut into the stone panel edges to receive an anchor) to transfer the lateral and gravity loads from the stone cladding into the unistruts using bolts and clamping washers. A TALE OF TWO STONES: EVALUATION OF A STONE PANEL CLADDING SYSTEM Figure 2 – Overall view of marble stone panels. S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8 FA R M E R & K E N N E Y • 3 Figure 1 – Overall view of serpentine stone panels. Note the extent of pronounced veining. Initiation of Investigation: Panel Failure After over 30 years of service, one of the serpentine (dark green, almost black) marble panels fractured. The upper portion of the panel slipped downward and was retained behind adjacent panels; the remaining lower portion of the panel fell to the ground. The fracture occurred along a horizontal vein partially filled with a brittle epoxylike material, likely applied as a surface treatment. The owners became concerned that other serpentine panels with similar horizontal veining might disengage, so they initiated an investigation to evaluate the cladding and anchorage system. POSTFAILURE OBSERVATIONS After addressing any immediate safety concerns by limiting pedestrian access, a visual survey of the serpentine stone was conducted from the ground in an attempt to assess if there were any additional panels that might be unstable. The marble panels were also surveyed from the ground for any visible damage. In addition, both types of stone cladding panels were observed at close range from a personnel lift, and selected panels were removed to examine the backup and cladding support conditions. Samples of the stone were also taken for material-strength testing in the laboratory. Stone Cladding Anchorage System Of some concern was the lack of any compressible fill in the stone panel kerfs that serve to evenly distribute lateral and gravity loads from the panel to the aluminum rail. Without any material in the kerfs, the loads can be unequally distributed throughout the kerf, causing uneven loading and stress concentrations at contact points between the anchor and the stone panel (see Figure 4). Small shims were observed between the anchor rail and the stone, presumably used to level the stone cladding panels and transfer the gravity load into the anchor rails. These shims were composed of a thin fiber board that was deteriorated, allowing direct contact between the stone cladding panels and the anchor rails (see Figure 5). The thickness of the shims varied, as well, Figure 3 – Unistrut support structure for stone cladding. Note horizontal anchor rail spanning between vertical unistruts. Figure 5 – Spotted serpentine panel (left), and veined serpentine panel (right). 4 • FA R M E R & K E N N E Y S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8 Figure 4 – Stone panel kerf engaging horizontal anchor rail (circle). Note lack of sealant and damaged shims (arrows). resulting in reduced engagement of the stone cladding panel kerf with the anchor rail and a subsequent reduction in load capacity. Serpentine Serpentine or serpentinite is part of the marble class of stone; however, ASTM distinguishes it from marble with respect to its physical properties. It is a metamorphic rock formed by the recrystallization of carbonate-based rock, but unlike typical marbles with calcite or dolomite mineral content, the stone is largely composed of the mineral serpentine, which gives the stone a greenish hue. The stone is typically known for its numerous crossing veins and rifts created by the entrapment of other minerals during its geologic formation. The serpentine panels varied greatly in color, texture, and veining. Two primary types of serpentine cladding were identified. The first was greenish-black serpentine with lighter (almost gray) spots evenly distributed on the surface. This stone material was found to be sound and generally free of defects. The second is a green-black serpentine with pronounced brown-red veins running throughout the panel. The minerals that comprise the veins weathered more rapidly than the base material, such that the fillers and surface treatments were no longer present. The surfaces of the veins were rougher and more eroded than the surrounding stone matrix (see Figure 6). This roughness and material loss made the veins appear to be cracked when observed from the ground, and stability was difficult to ascertain (see Figure 7). From a distance, it was nearly impossible to distinguish between veining and fully developed cracks. Upon close examination, several panels did exhibit cracks along the veins and were displaced along the cracks, indicating potential instability (see Figures 8 and 9). Several of the veins and cracks were filled S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8 FA R M E R & K E N N E Y • 5 Figure 6 – Example of eroded, intersecting veins in serpentine panel. Figure 7 – View of serpentine panels from the ground. Note veins can easily be mistaken for cracks. (likely at the time of fabrication) with an epoxylike substance to conceal the more eroded veins and create a smooth, polished surface (see Figure 10). The serpentine panel that most recently failed had fractured along a cross-panel vein that cracked previously or cracked recently as a result of wind loading against the panel section weakened at the vein. Our survey identified numerous panels that had a similar potential for failure due to crosspanel veins or preexisting cracks. Given the long vertical span of the panels relative to their width and thickness, these veins create opportunities for future fractures when the panel is loaded, particularly if the veins run across the width of the panel or if the veins are present in zones where panel flexural stresses are highest. The varying conditions of the veining and the potential for cracks to form along the veins also made individual panel performance and stability highly unpredictable without examination of each panel at close range. Marble Marble is a metamorphic rock formed from preexisting rocks (typically limestone) that have been exposed to increases in pressure and/or temperature. The recrystallization of the rocks (composed mostly of carbonate) and their impurities create marble with its intrinsic veining or “inclusions” and its crystalline structure. The purest white marbles have very few impurities, while colored marbles assume the color gradation of their mineral impurities. Observations of the marble cladding panels revealed that they were very uniform in color and overall texture, creating a homogeneous appearance with no significant physical distress evident (see Figure 11); however, nearly 100% of the marble panels were visibly bowed and generally curved inward (see Figure 12). The only imperfections observed were some small, short cracks oriented in a perpendicular direction from the panel edges. Sealant joints were removed at the corners of several panels to determine the effect of the curvature on the kerf and anchorage system, revealing that the interior legs adjacent to the kerfs of many of the panels were broken at their ends where the kerf anchor was in direct contact with the stone (see Figures 13 and 14). This critical damage that directly reduces the load capacity of the stone cladding and anchor system was not evident from the exterior and could have gone unnoticed until a more apparent failure occurred. When viewed under a microscope, this marble is composed of a series of tightly interlocked crystalline calcite grains that adhere to each other with a calcite binder. This complex arrangement of calcite grains Figure 8 – Crack that is coincident with a vein in a serpentine stone panel. Figure 9 – Example of a displaced serpentine stone panel with a crack along a vein. Figure 10 – Polymer surface filler present in serpentine vein. Note heavy concentration of veins. 6 • FA R M E R & K E N N E Y S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8 within the crystalline structure of the stone, although relatively uniform in its geometry, can be subject to the formation of intergranular fractures, or microcracking, within the calcite binder of the stone when the marble is exposed to changes in temperature. After these cracks have formed, the calcite grains in the marble tend to disassociate or come apart within the calcite binder of the stone. Once loosened, these grains are then free to move slightly and reorient themselves within the crystalline structure of the stone. They typically will not return to their original position within the calcite binder once this process has begun. This change in the material properties of the marble effectively disrupts the otherwise uniform and tightly bound crystalline structure of the stone, resulting in a “net,” or a permanent volumetric expansion of the marble itself due to the reorientation of the grains within the calcite binder that results in bowing. This process of net growth as a result of cyclic influences is known as “hysteresis.” A coarse-grained marble will typically experience a more pronounced level of deformation or bowing as compared to the finer-grained marbles available for the same application. As the marble structure increases in volume, it also decreases in strength, due to Figure 11 – Consistent, uniform appearance of the marble stone panels. Figure 12 – Curvature of marble stone panels due to hysteresis and associated strength loss. Figure 14 – Rear kerf leg damaged due to marble curvature or hysteresis. Figure 13 – Concealed damage of interior kerf leg, due to curvature of marble stone panels. S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8 FA R M E R & K E N N E Y • 7 the breakdown of the original crystalline structure. Strength loss resulting from the effects of hysteresis is both progressive and irreversible, with no known treatment or cure. As bowing becomes more pronounced, moisture ingress into the stone increases as the cracks and microfractures at the exposed panel surfaces begin to open and become more directly exposed to weather. The calcite binder is dissolved more quickly, resulting in an increase in the rate of disaggregation of the stone. This disaggregation, commonly referred to as “sugaring” (due to the presence of loose, powderlike calcite grains on the exposed marble panel surfaces resulting from disaggregation), will then lead to edge cracking and similar distress in the body of each panel. This is indicative of strength loss in the stone, and permanent change in the material properties of the marble. MATERIAL STRENGTHS In both new construction and building repairs, stone testing using ASTM1 International Standards for dimension stone is recommended to establish the physical properties and typical characteristics of a stone, such as porosity, density, and strength (flexural and compressive strengths and modulus of rupture, etc.). Due to the presence of cracks in the serpentine and the evidence of the marble losing strength due to hysteresis, both the white marble and the serpentine stones were tested for flexural strength, or bending strength of the stone. This is the most critical physical characteristic for thin-stone cladding systems, since their design is typically controlled by flexural capacity and its ability to resist laterally applied loads, such as wind. ASTM C880, “Standard Test Method for Flexural Strength of Dimension Stone,” was performed to determine the flexural strength. The values obtained through testing allow us to determine the overall flexural capacity of the stone panels as well as the adequacy of the kerf anchoring system. Based on ASTM recommendations, each sample was tested in the vertical and horizontal directions of the bedding plane (or grain structure), allowing us to distinguish which spanning direction was weaker. This is particularly important in stones whose bedding planes may not be immediately evident. According to the standard, samples should be tested in both wet and dry conditions; however, in an effort to minimize costs, all of the samples were tested in a wet condition that typically governs for design and evaluation purposes. Table 1 summarizes the results of the C 880 Flexural Tests for the white marble, veined serpentine, and spotted serpentine. Based on the test results, both the veined and the spotted serpentine far exceeded the recommended flexural strength, indicating a high quality of stone; however, the coefficient of variation for the veined serpentine was extremely high, indicating a lack of consistency in the material due to the pronounced veining. Our visual observations of cracking along veins and variations in vein condition supported this finding. By comparison, the coefficient of variation for the spotted serpentine data was very low, implying that the absence of veining results in a more uniform cladding material. While the original strength values for the marble cladding were not available, our experience suggests that this material would normally have had a flexural strength on the order of 1,200 to 1,500 psi. The flexural strengths obtained from the marble cladding panels were well below this value, even below the minimum for a stone to be classified as a marble. Though outwardly appearing undamaged, the low-flexural test results confirmed strength loss consistent with the effect of hysteresis over a prolonged period, reducing the capacity of the marble cladding to resist applied lateral loads. STRUCTURAL ANALYSIS OF THE STONE CLADDING The ASTM minimum specification requirements for stone are typically used as guidelines for choosing a material and as a basis for design when stone-testing data may not be readily available. Numerous other factors such as the anchorage system, panel height-to-width ratio, span direction, loading, and safety factors must be considered when designing or evaluating a cladding system. As a result, a structural analysis of both the marble and the serpentine using the documented test strengths and project-specific conditions was necessary to determine the adequacy of the panel and anchorage system in the existing application. Serpentine Based on the flexural strength results obtained, the physical panel dimensions, and current industry-accepted safety factors, it was determined that the veined serpentine cladding panels were not adequate to resist anticipated lateral loads as originally designed. A reduction in the industryaccepted safety factor for flexural strength was considered in this case, due to the acceptable performance of the cladding in service; however, the reduced safety factor Table 1 ASTM C 880 STANDARD TEST METHOD FOR FLEXURAL STRENGTH OF DIMENSION STONE STONE TYPE VERTICAL HORIZONTAL Marble Min. (ASTM C503) 1,000 psi 1,000 psi Average 751 psi 724 psi Maximum 860 psi 868 psi Minimum 552 psi 521 psi Standard Deviation 117 psi 140 psi Coefficient of Variation 16 % 19 % Serpentine Min. (ASTM C1526) 1,000 psi 1,000 psi Veined Average 3,773 psi 3,504 psi Maximum 4,751 psi 4,628 psi Minimum 2,053 psi 1,598 psi Standard Deviation 1,026 psi 1,195 psi Coefficient of Variation 27 % 34 % Serpentine Min. (ASTM C1526) 1,000 psi 1,000 psi Spotted Average 3,430 psi 2,341 psi Maximum 3,811 psi 2,477 psi Minimum 3,118 psi 2,201 psi Standard Deviation 261 psi 102 psi Coefficient of Variation 8 % 4 % 8 • FA R M E R & K E N N E Y S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8 S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8 FA R M E R & K E N N E Y • 9 was rejected due to concerns with the high variability in panel conditions and the unpredictability of the impact of the veining on panel flexural capacities. Consideration of a reduced safety factor was deemed more appropriate for the spotted serpentine, due to its more uniform test results, but far fewer of these panels were installed on the building. Marble Upon review of the test data and panel geometry, the marble cladding panels were also found to be structurally inadequate, due to their lower flexural strengths, with no applied safety factor. The marble cladding panel flexural capacities were well below those required to sustain current code-mandated loading; however, no evidence of major structural distress (cracks, displacement) was observed or noted during the service life of the cladding. This may be partially explained by examination of weather data for the period during which the cladding was installed. The highest recorded wind speeds during this period were only two-thirds of currently accepted design wind speeds, resulting in actual loading well below the code-mandated minimums. The marble panels were also of structural concern, due to the loss of capacity at the kerfs resulting from bowing-induced kerf fractures. These were particularly worrisome because there was no outward evidence of the fractures, so compromised panels (or the extent of the damage) could not be readily detected. It was calculated that approximately 70 percent of the kerf length was required to transfer the codemandated design loads. Our observations suggested that no more than ten percent of the kerf lengths were damaged at the time of our examination, but this figure could increase as the panels continue to bow and lose strength in the future. SUMMARY Anchorage Conditions The original stone-panel installation is problematic overall because the kerfs were not filled with any material that allowed for the loads to be spread along their length; the panels were, in effect, loose between the kerf anchor rails, allowing movement and causing unpredictability with respect to load transfer from the panel to the backup support structure. Current practice includes filling the continuous kerfs with a flexible material such as elastomeric sealant so that the kerf anchor is not rigidly fixed in the kerf but also is not able to move freely. The vertical load of the stone panels (panel self-weight) was transferred into the continuous kerf anchors through shims installed below the panels, and these shims were deteriorated or missing. If the panels are not properly supported, then the panel can contact the kerf anchor in unanticipated ways that cause stress concentrations and eventual cracking. Future stone installations should include more durable setting blocks or shims in a length and size that are optimal for distributing the load to the backup structure and maintaining reliable, consistent kerf/anchor rail engagement. Stone Cladding Panels The appearance of the façade elements and stone cladding at first glance gave the impression that these materials were in good condition generally, despite using a stone panel thickness for both the marble and serpentine of less than the current industry recommended minimum of 3 cm. However, upon closer examination, the stone claddings (both serpentine and marble) had characteristics that undermined their stability, and compromised their structural integrity. The pronounced veining in the serpentine concealed cracks and represented potential planes of weakness that could fail over time. While strength of the serpentine was high relative to ASTM standards, it was not adequate if this system were to be installed under current code requirements. Even taking into account past “incident-free” performance of the cladding system, the inherent variability and unpredictability of the veined serpentine make it a poor choice for an exterior cladding application that carries an increased risk of performance-related problems. The marble cladding panels have suffered from hysteresis, resulting in a reduction of flexural strength below the industry minimum for marble and below values necessary to resist code-mandated lateral loads but with no readily observable distress that reveals the dramatic strength loss it has undergone. The bowing inherent with the hysteresis process also led to localized kerf damage that is concealed and therefore not easily monitored, further compromising the ability of the marble panels to resist lateral loads. Strength loss, as well as the associated bowing, kerf failure, and formation of additional cracks in the marble cladding panels, will continue to occur over time, further reducing their ability to resist in-service loading. RECOMMENDATIONS Recommendations for corrective action to address deficiencies with a cladding system are always difficult because past performance may not necessarily warrant drastic action. Often, deficiencies are discovered by accident, not through occurrence of a failure. For example, cladding systems may be underdesigned but perform well due to a lack of exposure to maximum design loading, conservative safety factors, or unanticipated structural redundancy. It is simple to arbitrarily dictate that such a cladding system must be removed because it does not meet current code, but if past performance has been satisfactory, then the decision to do away with the cladding may represent a substantial economic waste. In such a situation, other options should be explored to manage the risk associated with the deficient cladding system. This might include enhanced inspection or monitoring protocols, partial replacement, repairs, strengthening, even reevaluation of potentially overly conservative analytical assumptions or code requirements. Often, managing the risk associated with underdesigned or underperforming cladding systems can be more desirable and less disruptive than removing the risk altogether. In the case of this project, failure of a single cladding panel triggered an investigation that concluded that both the existing serpentine and marble cladding panels were theoretically underdesigned. Upon closer inspection, several other serpentine panels were indeed cracked or otherwise unstable; however, the majority of serpentine panels contained heavy veining but were performing with no evidence of structural distress. In addition, the marble panels appeared to be performing well under in-service loads, despite experiencing strength loss and kerf fractures. We presented several recommendation options to the client with a range of costs and associated risks. Because of the immediate need to replace damaged and unstable panels, we recommended a complete, 100 percent close-range visual examination to be performed in conjunction with joint sealant replacement to take advantage of available access and staging. Removal of the sealant also helps to eliminate any redundant support provided by the sealant itself and allows stability of the cladding panels to be evaluated directly. Any panels found to be unsound, unstable, displaced, cracked, or otherwise structurally compromised, were to be replaced with material that matched the original and complied with ASTM minimum material properties. In addition to replacement of damaged-panels, a program of frequent inspections was required to ensure that any future panel damage could be identified and addressed quickly on an ongoing basis as part of a maintenance program. This option carried the least nearterm cost but the most risk of future problems. A second option was presented that included the same close-range examination of the cladding and replacement of any damaged panels. In addition, we recommended modifying the anchorage system in a manner that reduced the flexural stresses below the code-required maximums by reducing the spans between anchor points. While more costly to implement, the risk of additional panel failure was reduced by modifying the conditions that could lead to a future failure, and the existing cladding panels that are sound could be reused. Continued inspection of the cladding was required but at less frequent intervals than the previous option. This option was more costly, but reduced the risk of future problems by addressing the stone cladding structural deficiencies directly. The third option presented for consideration was replacement of the stone cladding with alternative materials such as metal panels or an exterior insulation finish system (EIFS) and reusing the existing stone cladding support system. This was the most expensive option presented but also carried the least future risk of failure by removing the underdesigned and/or deficient cladding material altogether. Due to the least amount of risk and reduced long-term impact to the owners, this option was chosen over the first two; recladding design is currently underway. FOOTNOTE 1. Formerly, the American Society for Testing and Materials. 1 0 • FA R M E R & K E N N E Y S Y M P O S I U M O N B U I L D I N G E N V E L O P E T E C H N O L O G Y • OC T O B E R 2 0 0 8
