Roden Crater: A Lath and Plaster Work of Art

Vince Bailey | Photos By John Talbert

September 2005

On approaching what is destined to become "the world’s largest work of art,” you might expect to be confronted with something much more than what first meets the eye—some unforeseen effect or unknown experience lurking just beneath the surface. And when Scott Maki, president of MKB Construction, headquartered in Phoenix, penned the proposal assuming the pivotal role in the finishes of phase one on the Roden Crater project, the logistical challenges of working in an 840-foot long tunnel were immediately apparent. But the latent problems associated with providing the installation of ornamental metals, lath, plaster and paint for this once-in-a-lifetime project proved to be even more daunting for the longtime AWCI member.

Artist James Turrell spent more than two decades coaxing his "monument to light” through the conceptual stages; the actual construction of phase one of his magnum opus in installation art started in 1999 and is nearing completion at a cost of $10 million to date. The Roden Crater project is best described as a series of underground chambers and tunnels honeycombing an extinct volcano in a particularly remote part of northern Arizona. This complex system of subterranean rooms and culverts includes a number of precisely oriented skylights, designed to focus on certain celestial events. The "Alpha Tunnel,” the current centerpiece of the project, has been termed "the world’s longest naked-eye telescope.” Not your typical lath-and-plaster build-out.

Since the project’s first stages of emergence, Tom McGrath of Skystone Foundation, project manager and owner’s representative, has overseen the planning and activity. Initial site development and structural work, performed by The Ashton Company of Tucson, Ariz., began in early 1999 and entailed moving some 1.35 million cubic yards of dirt and cinders to sculpt the volcanic cone to the artist’s succinct specifications. Ashton then installed 660 tons of reinforcing steel, and poured 5,500 cubic yards of concrete to form the basic structure of the connected rooms and shafts. By early 2001, backfill and final grade was completed.

Turrell’s vision now had a skeletal substance, but the concepts for the finishes that would optimize his desired effect—the experience of light—were still in the developmental stages. Temporary finish work was minimally performed in a few areas, and the project went on hold for three years while Turrell contemplated the final product.

In the interim, Turrell’s other artistic endeavors led him to avail himself of the services of Dennis Hopper, artisan plaster contractor and principle of Hopper Handcrafted Specialty Finishes based in Phoenix. No stranger to unique or unusual projects, Hopper’s name has become synonymous with exotic plaster finishes throughout the country. While his involvement in the plastering business dates back some 46 years, he became fascinated with lime-based colored plasters during a 1987 vacation to Venice and has since developed a line of Venetian finishes that echoes the softly mottled look of old world frescoes.

Turrell and Hopper developed a mutual admiration, and Hopper’s Venetian plaster finishes were soon incorporated into Turrell’s plans for Roden Crater. One rub: Hopper’s scope of operation was limited to finishes only, and an additional contractor capable of a precisely installing a substrate of lath and base coat plaster would be needed. Hopper recalled prior collaborations with Maki, and recommended MKB Construction, a Phoenix-based lath, plaster, drywall and paint contractor, to provide the substrate.

By early 2004, Turrell’s final deliberations were translated into construction documents, and a project team was formed by Skystone Foundation in which Kiewit Western Company would serve as the general contractor and provide special installations and facilities, with MKB Construction providing lath, plaster and paint, and playing host contractor to Hopper Finishes and T.A. Caid of Tucson, Ariz., an ornamental metals fabricator.

While the scope of the project included provision of plaster and paint finishes in all areas, the lion’s share of work centered on the Alpha Tunnel, a keyhole-shaped, rising concrete passageway, averaging 12 feet in diameter and running the 840-foot long distance between two illuminated chambers. According to Turrell’s design, precision-cut, 2-inch by quarter-inch aluminum baffles would be mounted on the raw concrete interior walls at 4-foot intervals for the length of the tunnel. Between baffles, an installation of lath, base coat plaster and Venetian plaster would be applied over the concrete, to a depth at which 1 1/2-inch of the baffle surface would be left exposed. The aesthetic purpose of the baffle exposure is to fragment and mute the shaft of natural light cast on the tunnel walls from an oculus in the top chamber.

To achieve the intended effect, the layout of the baffles would have to be uniform—precisely aligned to the center of the barrel portion of the keyway, or "image eye”—at a tolerance of plus or minus one-sixteenth of an inch in 840 feet. As a practical matter, the baffles would serve as plaster grounds, and their exact placement would become the means to correct any imperfections or serpentine undulations in the structural alignment of the tunnel. If MKB could meet Turrell’s expectations, they would, in effect, be building an 840-foot segmented gun barrel—more of an instrument than an installation.

During the pre-construction stages, a three-dimensional laser scan of the tunnel showed that variances in the concrete substrate intruded upon the parameters of the theoretical image eye. This required Kiewit to perform substantial bush hammering of the concrete to bring the substrate into conformance. The process of bush hammering revealed the first of a number of concealed issues. The tightly quilted embedment of reinforcing steel lay a mere inch beneath the finished surface of the concrete. The immediate problem this posed was apparent in that corrective bushing would be limited to a depth of one inch. The imminent problems associated with attachment of the baffles and lath were now becoming evident as well.

Baffling Problems
In September 2004, while the shops of T.A. Caid were still occupied with laser cutting and machining the 210 Alpha Tunnel baffles, Zeke Delgadillo, MKB framing and lathing foreman, arrived on site and immediately set to the task of resolving issues. He began by testing fasteners for the attachment of metal lath, baffles and furring. The installation of the metal lath alone would require in excess of 45,000 fastening points. In addition to the shallow embedment of the rebar, the concrete substrate—batched for the 3500 psi/4-inch slump design specifications—had aged for well over three years. This was a recipe for a surface hardness that spit most powder-actuated fasteners back at the installer, and ate hammer-drill bits for breakfast. After ponderous trials, newly developed products allowed the best production time with a minimum of attachment failures.

Next, Delgadillo planned the layout for the installation of the baffles. Turrell’s instructions for this were explicit: Each keyhole-shaped assembly was to be center-oriented horizontally, vertically and rotationally to within plus or minus one-sixteenth of an inch of the theoretical image eye. Each baffle would be oriented perpendicular to the rising gradient (8 degrees) and square with the centerline of the pathway. According to Caid’s shop drawings, each baffle assembly would consist of five components—one semicircle radius piece, two radius-to-straight transition pieces, and two straight legs—forming the keyhole configuration. Components would be fixed together at overlapping joints by a screw-attached clip that would serve to mount the baffle to the substrate as well. Aluminum horseshoe shims of various thicknesses would be provided for mounting the clips and fine-tuning the alignment. Delgadillo foresaw that a template should be fabricated to hold the circular baffle parts together and in precise alignment during installation. Word was sent to Caid’s project manager, Ron Rice, and a collaboration between Rice and Delgadillo generated a mounting template that would do just that.

A preliminary survey was performed by Kiewit to provide vertical center marks on the floor and elevation center marks on the sidewalls every 60 feet, thus establishing the theoretical center of the image eye. Corresponding marks were made on the template and the baffle components for reference, and, after much head scratching, a layout plan was initiated.

The process of accurately laying out and installing the baffles was critical and painstaking, as location determined the placement of subsequent work, and orientation was required in multiple dimensions. The 4-foot spacing was determined by aligning the baffle placement with segment joints of a lighting curb, already installed by Kiewit. Elevation (horizontal center of the image eye) was established by transfer of the surveyed elevation points via laser level; the self-leveling feature was overridden due to the gradient, and the beam was adjusted to align the two nearest survey marks on each side of the tunnel—four total. Mid-point marks on the template were brought into this alignment. By aligning opposing mid-point marks on each side of the radius template, correct elevation and correct rotational position were determined; the template was positioned accordingly and held in place by screw-jacks.

Vertical orientation of the template to the vertical center of the image eye, was established by transferring center survey marks on the tunnel floor to the top dead center on the template via another laser level in plumb mode. The template was positioned and held fast by template set-screws.

Perpendicular orientation to the 8-degree gradient of the image eye line was determined by using a "6-8-10” squaring formula at mid-point of each side of the template along the lasered elevation line (horizontal center of the image eye). Continuity of this perpendicular orientation was achieved by aligning a rotational laser light along the plane of the baffle.

Finally, when precise template placement was achieved, the radius and transition components of the baffle were clamped to the template, aligning the corresponding marks. Clips and shims were installed, the baffle assembly was fastened to the concrete substrate, and the template was released and removed for reuse on the next baffle in sequence. After a ponderous learning curve, it was determined that a two-man crew could erect eight baffles in a ten-hour day, a rate that would keep the operation within the estimated labor budget.

Due to the logistical restrictions of working in a 12-by-840-foot tunnel, Delgadillo requested a skeleton crew for the installation of baffles and lath. He and one helper would focus on erecting the baffle assemblies, while three other journeymen lathers would follow behind, installing furring, lath and trims. While the plans and specs called for 3.4 diamond mesh metal lath to be directly fastened to the concrete substrate, variances in the alignment of the structure often required furring of the tunnel wall to keep base coat plaster thickness within the specified 1 1/4-inch limit. Where these circumstances prevailed (nearly half of the 28,000 square-foot tunnel wall area), 1/2-inch or 7/8-inch, 20-gauge furring channels were shot to the structure at 12 inches on center to provide an underpinning for the lath. Where metal lath could be directly applied to the concrete, "dimpled” or "self-furring” lath was used, as its configuration holds the plaster key slightly away from the substrate surface, allowing the scratch coat of plaster to key completely into the mesh. A "2-A” expanded metal corner bead was then wire-tied to the lath and/or shot-fastened to the concrete at the radius-to-vertical transition, forming a corner at the transition, while maintaining the 1-1/2-inch exposure on the baffle. A 3/8-inch-deep "number 66” casing bead was added at the top of the lighting curb as a termination ground for the plaster.

At Delgadillo’s direction, MKB’s baffle and lath crew started at the topmost termination of the Alpha tunnel at the East Portal and began to work downhill at a rate of 32 feet per day. It would take a full month to reach the tunnel’s mid-point, and the snail’s pace would take its toll on the lathing foreman’s nerves. "After getting past the learning curve, I expected better progress because of the repetition, but the constant fine-tuning for the close tolerances on the baffles and trims, and all the time-consuming furring to compensate for the tunnel deviations—all this ate us up,” he said.

Plaster and Scaffolding
Rick Howell, plaster superintendent for MKB, understood the imminent challenges associated with his scope of work in the Alpha Tunnel from the outset. He began making regular visits to the remote site to assess his approach to the work even before the contracts were signed. Howell was to provide a scratch-and-brown base coat of plaster as a foundation for Hopper’s Venetian finish. Maintaining a uniform depth and a precise baffle exposure would be critical. Hopper would demand a "baby’s butt” smooth substrate for his work. Skilled personnel would be vital and hard to attract to such a remote site.

Scaffolding the barrel to maintain good working reach would be tricky. Logical sequence would have to be strictly observed due to the spatial restrictions, curing times between coats, and extent of scaffold. Proper temperature and ventilation would have to be maintained for curing throughout the operation. And all this was just for starters.

Howell first tried to determine how delivery of the plaster product from the mixer to the installation area would be accomplished. For ease of delivery and the high production levels associated with spray application, his first choice was to use a plaster pump and hose to convey the mud. If he set up two mixing stations, one at the top chamber called The East Portal, and one at the bottom chamber, called The Sun and Moon Room, the length of hose needed would be just over 400 feet at the furthest point—still within acceptable limits. Even so, Howell was concerned about potential hose blowouts and hours of daily cleanout and disposal at such a distance.

Drawbacks notwithstanding, Howell submitted his plan to incorporate pumping and spraying the scratch and brown coats to the general contractor and the owner’s rep. Unfortunately, due to concerns over protection of the existing stonework and the solid bronze stair in the East Portal, Kiewit and Skystone requested that MKB eliminate that area from consideration as a mixing station. They would have to submit an alternate plaster delivery plan.

Howell was stumped. Pumping plaster through 800 feet of hose uphill (the rise in elevation from the Sun and Moon Room to the East Portal is 150 feet) was not feasible. Visions of an easy delivery and high productions began to dissolve. Hand wheeling the mud uphill for such interminable distances at such a grade was equally unacceptable. That’s when the notion of using a motorized Georgia buggy came to him.

"I’d used them before, but never in such a restricted space as a tunnel,” he observed. The biggest drawback to this approach would be access through the scaffolding. Howell selected a suitable piece of equipment, a 130-horsepower gas-driven Ingersoll Rand buggy, then measured the wheelbase, its widest point. It was 42 inches—just 6 inches under the inside width of most of MKB’s 5-foot walk-through scaffold frames. With such limited clearance, the length of each scaffold setup would best be limited to 200 feet or so to minimize the odds of a collision.

The mixing station could now be located at the staging area outside the entrance to the lower Sun and Moon Room. It was the solution that Howell would settle for, but it meant hand application of all of the 28,000 square feet of base coat plaster. Schedule and budget would now loom larger as potential issues.

Howell and Delgadillo had originally agreed that scaffold setup and plastering should commence when the lath and trims were complete on the top half of the tunnel. But now the added factor of the buggy running up and down the tunnel every half hour or so would conflict with the setup of the baffle template, which effectively blocked the tunnel at each setup. Plaster would have to be held off until all of the baffles were erected. Fortunately, Tom McGrath of Skystone and Scott Davis, project superintendent for Kiewit, recognized that their concerns over protecting the existing work had created a scheduling hardship for MKB; they worked up a revised schedule that accommodated the logistical delay. In order to further minimize the impact of the conflict, Delgadillo began to erect only the radius portions of the baffle assembly to hurry the duration of template use along.

Like many EIFS and stucco contractors, MKB runs several scaffold erection crews in-house, and maintains a yard stacked to the heavens with every kind of frame, brace, plank, rail and connector known to man. Howell determined that stationary scaffold would be used for the plasterers’ scope, and, to facilitate a good working reach in the barrel portion of the tunnel, a stepped, tri-level working platform would be erected. At the centerline of the tunnel, 5-foot wide by 6-foot tall walk-through frames were placed on screw jacks and adjusted to a platform height of 6.5 feet. These frames were erected gravity-plumb against the 8-degree slope of the tunnel floor, requiring the addition of swivel bases and tube-and-clamp side braces in lieu of the standard X-braces to conform the setup to the ramping fall of the tunnel.

Cleats were nailed to the bottom of the planks to prevent downhill drift. Outriggers were added at either side to form a step-down level 18 inches lower than the main platform. From this configuration, all portions of the tunnel were placed within comfortable working reach of the plasterers’ trowels and rods, while easy access through the setup could be maintained for workers and for the buggy. And though minimal length of the setup would be preferable, sequence of work would ultimately decide the length factor.

Early in the pre-activity stages, it was determined that MKB would need a minimum run of 200 feet of scaffold set at a time. "With a cure time of 48 hours between scratch and brown coats,” Howell says, "half our eight-man crew could drop back after scratching 25 panels or so and begin to brown the first panels scratched.”

As a matter of courtesy, however, MKB’s managers had agreed to leave the scaffold erected for Hopper Finishes’ plasterers to complete their work as well. This would require at least another 100 feet of run to accommodate Hopper’s seven-man crew of Navajo artisans. In a perfect world, laborers would tear down behind the finishing crew and set the next 100 feet of platform, keeping all phases of plastering going simultaneously on 300 feet of scaffold. However, a five-day cure time requirement between the brown and finish coats created a time lag that would necessitate a total of 400 feet of scaffold standing in the first run in order to keep everyone working. This, of course, required twice the expected amount of equipment (walk-throughs that would accommodate the buggy were getting scarce) and would double the obstruction factor in the tunnel. Remarkably, there would be no incidence of collision while negotiating the motorized Georgia buggy through hundreds of feet of limited access for dozens of trips through the obstacle course per day.

Howell’s solution to providing a uniform depth and a precise 1 1/2-inch baffle exposure was consistent with this unique project’s constant demand for improvisation. He ordered several 6-foot magnesium "feather edge” rods, cut them into 50-inch lengths, and notched the ends 1-9/16 inches deep from the leading edge. This simple but resourceful modification allowed the rod man to ride the notched portion of the rod along the baffle edge, using it as a ground and screeding the surface of the brown coat at a plane recessed 1 9/16 inches from the baffle face. Howell thought to add an extra 1/16-inch to the depth of the rod to allow for the negligible thickness of Hopper’s finish.

Two problems emerged here: The soft magnesium would wear at the notches from continual rubbing on the aluminum screeds, thus allowing the rods to cut deeper. More rods would have to be ordered, cut and notched to regularly replace the worn ones. Also, a number of the baffle clips, which were located a scant 1/16-inch below the surface of the plaster, became exposed in the screeding process. Grinders were brought in where this occurred, and the protruding portions were ground off to ensure that the thin finish coat would sufficiently cover them.

Problems large and small seemed to come in droves. Even material procurement would become an issue. Having run work in the Flagstaff area before, Howell knew that the local sand would not be acceptable for the base coat plaster of the Alpha Tunnel.

"There’s two kinds of Flagstaff sand,” says Howell, "concrete sand, which is too coarse an aggregate for plaster brown coat, and mortar sand, which is too fine—it creates shrinkage cracks.” Consequently, MKB ended up hauling 150 tons of plaster sand 200 miles from Phoenix, 30 tons at a time. To Howell, going to such lengths was less a matter of preference than a matter of standard. "I can’t say we put all 150 tons in the tunnel, though,” the plaster superintendent laughs when asked to calculate his material use. "I think at least a third of that sand is blowing around up on the streets of Tuba City by now,” referring to the 60 mile-an-hour winds that are frequent visitors to the Crater. He often speaks of one fated delivery made during one of these windstorms that could never be accounted for.

But in spite of all efforts to keep the sand pile covered on the off hours, the persistence of the driving wind worked another hardship on MKB that was no laughing matter. During working hours, and even as the tender shoveled the sand into the mixer, gusts would intermittently blow the fines out of the sand, leaving it coarse in some batches, while less exposed mixes yielded more acceptable batches.

The inconsistent scouring effect on the surface of the brown coat from rodding the coarse aggregate would become a substrate issue to be taken up with Dennis Hopper later. In addition, the absence of fines in some of the panels made it nearly impossible to hold the depth and baffle exposure to the prescribed tolerance, as much of the coarse aggregate was comprised of grains that exceeded 1/16-inch diameter. Hosing down the sand pile regularly during blustery days helped to control the problem, but intermittent scouring of the brown panels would continue to plague the plaster crew throughout the tunnel.

It’s All in the Grit
Hopper was visibly troubled on his first visit to the tunnel following completion of the first several panels of base coat plaster. The care that MKB had given to the even surface of the brown coat and the consistency of the depth was apparent, and with this he was quite pleased.

But some of the deep scratches left by the feather-edge would be difficult to conceal with his finish work. Hopper consulted with MKB’s Howell, and both agreed that changing to a finer sand on the brown coat would cure the scouring problem, but would also result in shrinkage cracks—a condition to be avoided at all costs.

Both also agreed that Hopper’s usual finish process—a leveling coat of primus, two coats of Venetian and a sealer—would not provide enough buildup to cover the blemishes. After much discussion and several trial panels, Hopper and Howell arrived at a solution: Hopper would change his leveling coat formula to include 20-grit silica sand, rather than the usual 60-grit component. This approach would lend the needed buildup to conceal the scratches, but care would be needed not to exceed the 1/16-inch depth allowed on the baffle exposure for Hopper’s finish.

Hopper was also concerned that the new formula for the leveling coat might cause the character of his Venetian finish to deviate from the approved samples, but inspection of the finish on the first few completed panels by Turrell and McGrath soon relieved him of these concerns.

All involved parties agree that the collaboration between MKB Construction and Hopper Finishes was key to the success of phase one of the project. The finished product met with James Turrell’s stringent expectations, and the effects are striking. In fact, McGrath hopes to reunite the team members soon for phases two and three of Roden Crater, which consist of the addition of still more subterranean tunnels and chambers.

Says McGrath: "I started out thinking that these guys were privileged to be working on such a unique project, but I came to realize that we were just as privileged to have such a rare convergence of gifted craftsmen here.”

About the Author
Vince Bailey is a construction manager and free-lance writer based in Phoenix.

About the Photos
Photos by John Talbert of MKB Construction appear courtesy of James Turrell and Skystone Foundation.