Hammond Harwood House: 1990s Roof Work
Annapolis, MD:
By the mid-1970s, it was apparent that there were problems with the valleys of the main block. In the winter of 1978, an icicle formed coming from behind the wooden cornice on the west valley, reaching a length of 8 feet. Other signs of trouble included peeling paint stains appearing on the ceiling plaster of the upstairs bedroom. Standing on the street looking up at the cornice on the north and west valleys, it was possible to see that the cornice was actually coming loose from the building. This suggested that a leak had probably been ongoing for a long time, and had never been fully repaired.
The problem with this sort of roof damage is the unknowns: where are the leaks, which elements have been damaged, etc. It is hard to estimate what will be involved, how long it will take, and the cost of repair. To look at the problem you have to remove shingles and sheathing, but opening the roof and leaving it exposed to the elements while you investigate and renegotiate a contract is a less-than-ideal situation. Architect Charles Phillips and I proposed to Director Stephen Patrick that we spend a week in the attic fully documenting the structural system of the roof, attic floor, and, as best we could, the cornice framing, particularly the areas around the valley that are the most difficult to access because there is not enough clearance. Charles and his assistant David Clements produced a detailed set of drawings by studying the areas they could access.
We took this information and drew a section through a typical valley. Then, based on the symptoms we were seeing and drawing on our knowledge of 18th century contruction, made our best guess as to where the damage had started and subsequent damage. We then created a work plan to repair those areas, assuming worst case: everything that we could imagine could go wrong had gone wrong. We then gave a not-to-exceed figure for this work that we felt would cover any conditions that we might run into once we opened the roof. The idea was to erect a scaffolding and assemble a team of craftsmen to focus on the first valley area, and repair it and close it up as quickly as possible.
We billed the work at time and materials against this upset figure of $25,000. The idea was that from this first drop on the building we would discover the nature of the failure mechanisms, devise appropriate repair techniques, and determine how best to go on to the building's other valleys. Any money leftover from the $25,000 would be rolled into work on the next valley.
We were determined to understand the original structural design and make repairs that did not alter its function. We found the original system was superior in design and functioning well in every area except where neglect had allowed water in, leading to wood deterioration.
To protect the building and floors below from rain, we devised a simple system that allowed us to cover the roof within five minutes. It consisted of 5V-crimp metal roofing pieces that were kept stacked on the scaffolding. At the first sign of rain, they could be thrown across the rafters, and folds of EPDM roofing rubber could be unfurled and fastened over the tin with Bungee cords tied to the cornice and scaffolding.
Uphill, sheet metal flashing was slid under a course of slates, so the upper edge of the tarps could be secured to battens and slid under the sheet metal.
We discovered that the lower end of the valleys had been leaking water into the cornice. The valley sheathing and the rafter plate directly beneath the lower end of the copper valley had rotted and was almost completely missing. Water had worked its way down to the wall plate and into the cornice. Large sections of the rafter plate, several of the outriggers, and the large dovetailed joint where the diagonal dragon beam straddles the corner of the building and dovetails into the wall plate were also rotting. As a result, the outriggers and cornice sections they supported dropped had and rotated away from the building. This had apparently been going on a long time, as there were many painterly repairs: nails driven through portions of the cornice and lots of caulking.
When all the repairs were completed in about four weeks with a total cost of $19,700, the remaining $5,300 was rolled over to the gardenside valleys on the back of the house. We estimated the cost of repairing these two valleys at no more than $40,000, but they actually came in at $25,000. The remaining $15,000 was rolled over into extensive repair work on the rear cornice and cartouche.
We accomplished a great deal with this first drop to repair the valley and cornice of the main block of the Hammond Harwood House. We demonstrated to the owner that we were putting the building first. The whole process was created in order to respond to whatever was required to preserve the building.
We earned the trust of the client by working in incremental fashion, starting with our best assessment of what each portion would require. The not-to-exceed figure gave the client a sense of comfort to start into a complicated project. We had an open books policy, working time and materials, so the client could see exactly where every dollar was going. This is extremely important on a project of this type, once unexpected conditions inevitably come up that neither the assessment nor the budget took into account, and which may increase the scope and cost of work from original estimates
This first drop also allowed us to really understand the design of the original construction and the materials that were used, helping us to more accurately anticipate what would be involved on other sections of the building.