Why Excavation Beside a Landmark Is Designed Backward

By Michael Stern, JDS Development Group

The building standing at the edge of my excavation has been there for a century and is not permitted to move. That one condition governs how deep the cut goes, the order in which soil leaves the site, and the shape of the foundation that eventually replaces it. This is written for developers, geotechnical and structural engineers, and construction managers who are about to open the ground beside something that is protected.

Most foundation work is designed forward. You know the building you want, you know what it will weigh, and you work down to something that will hold it. Beside a designated structure the sequence runs the other way, and the drawing set reads differently because of it.

What a Landmark Can Tolerate Sets the Design

The first quantity that matters on a site like this belongs to the old building rather than to the new one. Old masonry has a limited appetite for movement, and it has a different appetite for each kind. Vertical movement as the soil beside it is relieved, lateral movement as the ground behind a retaining element relaxes, and uneven movement between one length of old foundation and another are three separate problems with three separate limits.

So the work starts with a condition survey and a set of thresholds. The engineers agree what that structure can take before anything is dug, expressed as a limit on vertical drop, a limit on tilt, and a limit on how far an existing crack may open.

Those thresholds come from the condition of the old fabric and from what its construction will tolerate. They do not come from the ambitions of the tower. I am deliberately not putting figures here, because the figures belong to a specific building and are established by the geotechnical engineers and the structural engineer of record reading that building, usually alongside a preservation consultant who knows what the fabric is.

Two old buildings of the same age will not answer with the same limits either. Softer historic mortar lets a wall redistribute load slowly and absorb small movement that a harder later mortar would send straight into a crack, and a building that has already been repointed in stiffer material can behave more brittly than its age suggests. Reading which of those you have standing beside the site is part of the survey and not a detail to be settled on the day the machines arrive.

How Underpinning and Bracing Follow the Tolerance

Once the thresholds are fixed, the support of excavation becomes a problem with a known answer rather than a matter of preference. Every element below ground is selected against a movement limit that was written down before anyone priced it.

Underpinning extends the old foundation down to the level of the new cut so that the existing building is bearing at a depth the excavation will not undercut. The work proceeds in short lengths because a long open run under a wall is exactly the condition the thresholds forbid, and the length is chosen from the tolerance rather than from the crew size. That is a decision the thresholds make for you.

The retaining element works the same way. A stiffer wall moves less, and the stiffness is selected from the movement limit on the old building rather than from what the soil alone would require.

Bracing then follows the dig level by level, each strut or tieback installed before the soil beneath it is taken out, because the wall is never allowed to cantilever further than the limit permits. Deep foundation practice has a long body of work on how those systems behave, and the choice among them is an engineering judgment made on the specific ground rather than a preference carried in from the last site.

111 West 57th Street rose in the courtyard of the landmarked Steinway Hall. A site described that way tells you the protected structure was not on one side of the work but around it, which means there is no elevation of the excavation where the question of movement goes away.

Why Instruments on the Old Walls Pace the Excavation

The thresholds only mean something if somebody is reading the building against them while the dig is running. So the old walls get instrumented before the first machine arrives: survey targets and optical prisms for position, tiltmeters for rotation, gauges across existing cracks, and monitoring points on the ground between the old foundation and the cut.

Readings are taken on a schedule and compared against two levels rather than one. The lower level triggers a review, which means the engineer looks at the trend and at what was excavated that day. The upper level stops the work.

How often those readings are taken is itself a design decision. Manual survey rounds at a fixed interval are adequate where the work is predictable, and automated instruments reporting continuously are worth their complexity where a single shift can change the condition under a wall.

A reading taken off the old wall can stop the dig, and nobody on the construction side has the standing to argue with it.

That authority is the reason the numbers are agreed at the beginning. A threshold written in advance is written by people who have no schedule pressure on them yet, and it holds later precisely because it was fixed before anyone wanted it moved.

The readings do a second job as well. They test the model the support of excavation was designed on, and a building that moves less than predicted tells the engineer something useful about the ground, just as a building that moves more tells them to look again at the sequence before it reaches the upper level.

What This Approach Costs and Who Should Decide It

A competent engineer can read all of this and conclude I have the order wrong. The argument against me is that designing from the old building outward oversizes everything below ground, that an observational approach starting with a lighter scheme and stiffening it only if the readings move is the better engineering, and that I am buying protection against movement that would never have occurred.

That argument is a good one and it wins on ordinary sites. My answer is narrower than a general defense of conservatism. A designated building has no recovery path, so the asymmetry between being too careful and not careful enough is severe enough that I am willing to pay for the difference on this one class of site.

The compromise I accept in exchange is time. Underpinning in short lengths is slow by construction, monitoring can hold the dig for days while a trend is understood, and the schedule carries that from the start rather than absorbing it as a surprise.

The people who should decide any of this are the geotechnical engineer, the structural engineer of record, and a preservation consultant who reads the old fabric, with the monitoring program specified by those engineers rather than by the contractor whose work it paces. Where the protected status of the neighbor governs what may be touched, that is a question for the agency with jurisdiction, which in New York City is the Landmarks Preservation Commission, and the protection of adjoining property during construction sits with the Department of Buildings. A recorded right of support along the line between the two buildings is a separate question again, and I have written elsewhere about how a party wall fixes the edge of a new building.

The old building is the first client on a site like this, and everything below it is designed to leave it where it already stands.

Michael Stern is the founder and chief executive officer of JDS Development Group.