Excavating beside the Dime Savings Bank, a landmark that had to stay exactly where it was
Excavating beside a landmark is a reading exercise before it is a digging one; what the instruments say between lifts decides the next lift, and the sequence is written to be interrupted.
A survey prism is a small glass target, the size of a coin, fixed to a wall so that a total station can find it from across the street and report its position to a fraction of a millimetre. Before any ground came out beside the Dime Savings Bank of Brooklyn, prisms like that were on the bank, and they were read before, during and after every day of digging. This piece is for geotechnical and structural engineers, for superintendents who have to excavate against a building that cannot be allowed to move, and for developers who want to know what that regime consists of before they pay for it.
The Brooklyn Tower needed a basement and foundations beside a landmark that was staying put, so the bank carried instruments the whole way down and each lift of ground was braced before the next came out. Those are the facts. What follows is the method that sits underneath them, in the order the method runs.
What is measured
Three things, and they are measured on the building rather than in the ground, because the building is what matters. Position, through prisms and survey points, so that any movement of the wall in any direction is known. Tilt, through instruments fixed to the structure that report whether a wall or a column has begun to lean. Cracks, through gauges across existing cracks so that a change in width is caught before a change in length. Vibration is added when the work involves driving or breaking, because a masonry building has a tolerance for vibration that is lower than a person's.
There are readings in the ground as well: water level in wells, movement of the soil behind the support. But those are early warnings. The building's own readings are the ones a decision is made on.
When it is read
Before anything happens, for long enough to know what the building does on its own. A masonry building moves with temperature and with the seasons, and a regime that has not measured that baseline will mistake a warm afternoon for a problem, or a problem for a warm afternoon. Then during the work, at an interval that tightens as the excavation gets deeper and closer. Then after, until the readings have stopped changing and the engineer is satisfied that the ground has finished adjusting to what was taken out of it.
The interval is a decision, and it is one of the places a developer sees the cost. Readings taken every day cost more than readings taken every week, and readings taken automatically and continuously cost more again. Beside a landmark the argument for continuous reading is strong, because the reading that matters is the one that arrives between two lifts of excavation and not the one that arrives on Friday.
What a reading means
A reading on its own means nothing. It means something against the baseline and against a set of thresholds agreed before the work began. The thresholds are set in tiers: a level at which the engineer is told, a level at which the work pauses and the sequence is reviewed, and a level at which the work stops. Every tier has a name and every person on the site knows which tier the current readings sit in. The thresholds are set for the specific building by the engineer responsible for it, and the numbers are theirs, so I am describing the shape of the system and leaving the values where they belong.
When a reading moves
This is the part of the regime a developer has to understand before agreeing to it, because it is where the schedule lives. When a reading crosses the first tier, the next lift does not proceed on the plan. It proceeds when the engineer has looked at the reading, decided what it means, and either released the lift or changed it. Changing it might mean a shallower lift, more support installed before the next one, or a pause while the ground is watched. The sequence is written to be interrupted, and a sequence that cannot be interrupted has no business beside a landmark.
Support as the ground comes out
The support of excavation is installed in step with the digging rather than after it. Each lift exposes a band of the retaining wall, and that band is braced before the next lift goes down, so that at no point is there more open face than the wall was designed to hold unbraced. The readings feed straight into this: if the building moves more than expected after a lift, the next band of support goes in sooner, or heavier, or both. The engineer for support of excavation and the engineer reading the instruments have to be talking every day, and on a site like this they were the same conversation.
Where I could be wrong
A geotechnical engineer reading this could argue that I have described a regime for a building that is being protected and not one that is being built beside, and that the more important controls were in the design of the support itself, which is meant to make the readings boring. That is a fair reading and I half hold it. A well designed support system should produce readings that never leave the first tier. But the readings are what let anyone know that the design is behaving, and a support system without a reading regime is a design being trusted rather than confirmed. The people who ran this were a structural engineer experienced in support of excavation, a geotechnical engineer, and a surveyor whose work was checked by someone other than the surveyor. Anyone excavating beside a designated building should retain all three, and should retain the first before the basement is drawn.
The compromise was pace. The excavation went at the speed the readings allowed, which was slower than the speed the schedule wanted, and on a site run this way there are days when nothing comes out of the ground because a number has moved and somebody has to decide what it means. The bank did not move, and that was the whole point.