D+D MAGAZINE - APRIL 2014
Capitol Challenge
Conservators battle water intrusion at the Rhode Island Statehouse.
BY BARBARA THORNTON, AIA, BREWSTER THORNTON GROUP ARCHITECTS
AND
WILLIAM WATERSTON, AIA, RRC, WISS, JANNEY, ELSTNER ASSOCIATES
The Rhode Island Statehouse, and iconic 1904 McKim, Mead and White masterpiece, is a beautiful, albeit leaky, mass of white Georgia marble with brick backup, located in Providence, R.I.
Wiss, Janney, Elstner Associates (WJE) and Brewster Thornton Group Architects (BTGA) diagnosed and treated moisture migration at the base of the Statehouse dome, 2011-2013. It’s the fourth-largest self-supporting marble dome in the world.
The work revealed consequences of several well-intentioned but ultimately misguided energy retrofits, cleaning efforts and materials applications. WJE and BTGA are monitoring the repairs with humidity, temperature and moisture sensors to observe the hoped-for return to equilibrium in this complicated masonry system.
BUILDING DESCRIPTION
The upper roof of the Statehouse contains the main dome in the center and a “tourelle,” or open marble “turret,” at each corner. Each tourelle’s base comprises four faces, with two shorter walls facing the upper roof and the barrel below the main dome, and two longer, taller walls facing the main roof level, approximately 28 feet (8.5 meters) high. The tourelle bases form the corners of the main plinth supporting the central dome.
The main roof level is at the approximate level of the interior plaster ceiling of the smaller corner domes.
THE ISSUE
The third floor of the Statehouse was restored circa 2000. The ceiling received faux painting and gilt accents in the molded and pulled plaster corbelling. The state replaced the roof in 1999 with a liquid-applied roofing membrane. The southwest corner was repointed in 2010.
However, these efforts did not address all of the water-related issues.
Officials selected BTGA, Providence, partnering with WJE’s Boston office, in 2011 to investigate and repair the issues causing visible paint deterioration in the restored ceiling.
The project got increased attention when falling plaster struck the governor’s policy director a glancing blow as she exited a third-floor office in the building’s southwest corner in June 2012.
Officials authorized emergency sounding and plaster removal as the larger investigation moved forward.
INVESTIGATION
Though original drawings and historical documents revealed little structural information, the Rhode Island Department of Administration and the Rhode Island Historical Preservation and Heritage Commission offered valuable clues to past work in their repair records and correspondence.
The air circulation pumps in the basement, for instance, were decommissioned in the 1960s to conserve energy and many of the air shafts sealed. In-room vents still connect to the above-ceiling interstitial spaces, and significant airflow occurs above these areas. The only air conditioning is from isolated window units.
A low-pressure water-only wash was specified to clean the marble facing in 1976, but a contractor offered to sandblast the marble to improve “whiteness” and cut construction time. The approved sandblasting damaged the marble and increased its porosity.
Initial field observations in 2011 found damp and bubbling plaster in the northeast and southwest corner domes of the ceiling, and flaking paint on barrel vaults in all four quadrants.
Visual inspections throughout the area of study identified deteriorated exterior masonry conditions with potential avenues for moisture migration.
The Department of Administration, Facilities Division, authorized the BTGA and WJE investigative program for 2012.
The investigation included creating inspection openings on the interior; documenting exterior conditions of marble, mortar joints, windows, sealants and roofing; selectively removing stone to determine conditions of flashing and other materials; and observing conditions after rain events precipitating wetting of the plaster areas.
INTERIOR SURVEY AND INSPECTION
Openings — Inside, access ceiling openings allowed inspection of the domes from scaffolds and lifts, in the problem northeast and southwest corners and in all four barrel vaults. Investigators found saturated blown-in cellulose insulation, some of it dripping wet, above all ceiling openings, which helped explain the deteriorating plaster beneath.
During water-testing, the investigators viewed the interior surfaces of the exterior walls through the access openings.
Exterior Survey — Visual inspection of marble surfaces revealed voids, cracks and sealant improperly installed. Mortar in the upward facing joints was eroded. Flat capstones at tourelle corners sloped backwards, creating catchment areas draining into open joints below column bases. Sealant on the top edge of the wall flashing likely trapped water in the wall.
Water Testing — Spray racks water-tested all four tourelle side walls, from below the roof level moving upward. Four inches of standing water flooded the tourelle interior decks. Investigators sprayed third-floor roofs and side walls of tourelles and flashing with hose streams for about an hour from the base of the main dome roof level up the wall on the short walls.
A spray rack doused the larger walls for approximately three hours.
Of the areas tested, only the west face of the northeast tourelle leaked, as seen in the interstitial space.
Selective Stone Removal — An inspection opening at the reported leak during water testing showed that the original copper wall flashing was installed only 3 inches into the mortar joint. Workers removed the 6-inch-thick (15-centimeter-thick) marble face block from the backup brick. The mortar in the collar joint and the setting bed was friable and damp, likely the result of repeated wetting and freeze-thaw cycles, with cracked brick, split faces and brick fragments visible on the backup wall.
The contractor rebuilt the flashing base and installed a liquid-applied roofing membrane flashing the full depth of the marble, resetting it with stainless steel pins. Testing confirmed this work resolved the major leak area.
Materials Testing — Asbestos was absent from sealants and flashing, but present in bituminous damp-proofing material applied over sidewall flashings above the original roof.
Laboratory observations of the mortar between marble blocks showed a single layer of gray-white mortar, likely an adhesive mortar used in pointing. The original mortar was brown with high sand-aggregate content and low binder volume. Evidence of calcium carbonate pointed to the presence of a lime binder. Laboratory technicians found no Portland cement or slag particles in the sample.
ANALYSIS AND RECOMMENDATIONS
Although investigators found an active leak, it only partially explained the moisture in the northeast tourelle. They identified no active leaks or direct moisture paths on the southwest corner where the plaster fell.
The southwest tourelle had undergone an emergency repointing in 2010 before this investigation, but dampness remained. Efflorescence patterns reflect years of moisture migration from the tourelle bases, but investigation showed no rewetting, even during rain events. High levels of air movement suggest that condensation perhaps played a role in retaining interior-generated moisture.
Investigators recommended that the insulation, which retained and held moisture against the plaster, be vacuumed from the domes and adjacent arches experiencing issues — the northeast and southwest tourelles.
Investigators also recommended repointing rails and tourelle interior faces at all four corners, as well as sealant replacement and fastener corrections.
The combination of 1970s sandblasting, settlement, open joints, reduced interior drying and misplaced sealants blocking drainage contributed to saturation in the mass masonry wall.
The porous backup brick of the marble-faced tourelle bases, 3 feet (1 meter) thick in some locations, absorbs significant amounts of water. When in equilibrium, the wall expels the moisture before water migrates to the interior.
However, reduced air movement in the system, the insulation, the reduced exterior evaporation from repointing mortar and sealant blocking bulk water from exiting at the flashing, and the wall’s own physical properties overwhelmed the system.
The goal was to return the system to equilibrium.
CONSTRUCTION REPAIRS
The conservators presented their findings in July 2012 to the state’s Department of Administration. Officials approved a budget for repairs to be completed by December. Work began in early September with pointing and crack repair. The existing mortar proved hard and tenacious. Workers removed it carefully with sealant cutters, hand-chiseling the edges.
A repointing trial tested two putty mixes: one with pre-mixed lime putty and sand, and one using hydraulic lime and sand. Workers used brush-tamped finishing and created custom-made tools for placing mortar lifts in the 1/8-inch-wide (3-millimeter-wide) joints. The lime putty worked and finished well. The initial hydraulic lime putty mix from the Midwest caused brown-green stains and had to be replaced.
As work moved toward the Jan. 1 completion date, weather forced the use of the hydraulic lime mix and the budgeted enclosure of the southeast and northwest corners to heat the space and let the mortar cure.
The construction team began plaster restoration when exterior wet work ended. Removing the cellulose insulation and drying the base plaster was step one. Base plaster in most areas was more than 4 inches (10 cm) thick. The southwest areas dried well, permitting quick repainting. The northeast, where base plaster meets the outside wall, and where the heaviest damage occurred, dried slowly. Workers accomplished shaping, but finish paint was no-go as surface moisture, tested by pin-type meter, stayed above 25 percent at some locations.
The team applied painted panels over the trouble areas to allow continued drying during the upcoming legislative session before final repaint.
Since only two corners needed heated enclosures, the construction team used the budget savings to repoint all connecting walls of the dome base, including window restoration and sealant replacement. This outdoor-only work was accomplished during the session.
An interesting condition developed in the north roof access stairwell in early November 2012, at the start of the heating season.
Previously, the stair was chilly, but not uncomfortable, with the door at the top propped open for work access. Then on one visit, workers found the stairwell temperature suddenly topping 80°F (27°C), with water rivulets condensing on the walls and outside door.
The building’s steam heat had been activated. A steam line leaking below the state library on the second floor sent steam through perimeter vents in the library’s stained glass ceiling into the interstitial space connecting with the stair, and up the stairwell, boosting temperature and humidity.
Occupants knew about the leak, the librarian said, but library temperatures were bearable. Since the roof-access stair was seldom used, conditions went unnoticed.
That steam line was disconnected, and the library became more comfortable. However, escaping steam may have pumped water into the base of northeast tourelle for several years.
MONITORING SYSTEM
WJE installed 21 wireless electronic sensors in early December to measure air temperature, relative humidity and the moisture content of the material to which they attached. They automatically gather and transmit data at approximately five-minute intervals.
Conservators located the sensors in the corner domes on the third floor at the exterior brick masonry walls above the plaster ceiling; in the air space above the ceiling; and in the plaster on the attic side; and on the interior sides of the corner domes. They installed sensors at the arches by the corner domes also in the exterior wall, and in the plaster on the attic side.
Two sensors on the interior of the brick masonry measure surface temperature and condensation potential on the north and south walls.
Previously cut openings accessed the exterior wall above the dome and arch ceilings on the third floor, where workers attached an adhesively applied hook to the brick masonry. They hung a sensor from the hook with its probes contacting the brick.
Plaster-mounted monitors fastened directly into the plaster with bright metal screws. Attic air temperature sensors hang in the space on fishing line.
The team also installed two data collectors, called gateways. The gateways collect and transmit the sensor data via cellular modem to a remote site, from which it becomes available to view or download for monitoring.
The gateways are mounted in closets off the northeast and southwest third-floor corridors, and plugged into standard electrical outlets. Power outages will interrupt data collection and transmission, but they automatically resume when power returns. Monitoring software sends an alert email in case of power loss, sensor failure or other inactivity.
RESULTS
After one year, the exterior walls are slowly drying as hoped. Inside, the plaster is slowly drying, and no deterioration has been observed. Technicians repainted the repaired areas in fall 2013 except for two sides of the northeast tourelle, where painted plaster adhered directly to brick was not drying fast enough.
In January when the surfaces still weren’t dry, workers installed floating surface on furring, vented top and bottom, over the plaster. With ventilation behind, the wall can continue to dry, and the painting won’t deteriorate.
Investigators haven’t observed condensation, although conditions for it have been favorable several times.
No interior leaks have been observed except for a persistent, though much-reduced, leak in the corner of the northeast tourelle, still being investigated. Additional fastener replacement in fall 2013 helped greatly.
The humidity levels in the southwest tourelle exterior masonry wall still spikes, possibly from another unidentified steam leak or some other aspect of building operation. The wall’s 2010 repointing mortar mix contains Portland cement and may also slow exterior evaporation.
The team is considering replacing that repointing for consistency with the final repair work on the other three corners. The speed with which the older, emergency repointing responds to rain events, even though water testing failed to replicate the condition, implies that hairline cracks may still remain, or bond separation has occurred in the 2010 mortar work.
Although the tourelle bases are stabilized, observed deterioration of the exterior tourelle components and roof continues. The vaulted brick roof structure parge coat shows accelerated spalling. Spread cracks in the marble cornice ring were also noted. This deterioration is outside the area studied are above the level of water infiltration. Investigators recommend further study and stabilization of these free-standing elements for the next phase of work, and monitoring continues.
While insulation was removed from the two problem corners, the remaining material is still considered a potential liability. It reduces temperatures above the ceilings and exterior walls, and heightens relative humidity and potential for condensation.
Plans are in place for a new HVAC system within the building, to be completed by 2016. The system would use the interstitial spaces for duct distribution and seal the remaining vent openings from individual spaces.
That plan is recommended for re-evaluation in light of findings from this study. Cooler, drier conditioned air may help wick away moisture from high summer humidity levels above the ceiling. On the other hand, if the inside face of the plaster gets colder, moisture could condense on the back.
The authors have recommended that the state extend the monitoring through completion of the HVAC project.
This article is based on a paper presented by the authors at the 29th International RCI Convention & Trade Show.
ABOUT THE AUTHORS
Williams Waterston is an associate principal at Wiss, Janney, Elstner Associates in Boston. His work includes the investigation, evaluation and design of roofing and waterproofing systems. He is also experienced in construction document preparation and specification writing. He is registered as an architect and roofing consultant to help solve waterproofing and roofing challenges for new and existing buildings and has authored several articles on roofing material choices and roofing practices.
Barbara Thornton has been a partner in Brewster Thornton Group Architects LLP in Providence, R.I., for 14 years. Licensed in Rhode Island, Massachusetts and Vermont, she has developed a specialty in shell restoration and public projects during her 27 years of professional practice. Thornton holds four degrees from the Massachusetts Institute of Technology and is a commissioner on Rhode Island’s Fire Board.