Case Study 9.2 — The $60,000 Detail: Tracing a Sealant Failure from Value Engineering to a $2.1M Remediation

All people, companies, and projects in this book are illustrative composites. The building described here is a composite of the kinds of envelope failures that show up in warranty on mid-rise commercial buildings; the numbers are internally consistent and typical, not drawn from a specific real project.

Setup

Project: Alderwood Commons — a six-story, 168,000 SF speculative office building in Rivermont. Lump-sum contract, $38,200,000, completed by Kestrel Construction Group four years before Northgate started. Ray Alvarez was the project manager.

Facade: Architectural precast panels — 312 punched aluminum windows in a precast panel grid, with horizontal joints at each floor line and vertical joints between panels. Roughly 14,200 linear feet of exterior sealant joint.

Owner: a regional real estate investment partnership that leased the building to eleven tenants.

The job finished on time, under the contract sum, with a clean punch list and a client who sent a letter. Ray has a photograph of that building on his office wall. He keeps it there for a specific reason.

What Happens

Design, Month 4 — the value-engineering meeting

The project was over budget by about $900,000 at design development, and the team ran a value-engineering exercise. Eighty-one items went on the VE log. Item 47 read:

VE-47. Delete secondary (inner) seal at precast panel joints; provide single-stage exterior sealant joint only. Delete sill pan flashing at punched window openings; provide sealant perimeter seal at window frames. Estimated savings: $60,000.

It was accepted in about ninety seconds. It was one of the smallest items on the log. Nobody in the room was an envelope specialist and nobody asked the question that would have stopped it:

After this deletion, what is the second line of defense?

The answer was: nothing. The exterior sealant bead became the only thing standing between wind-driven rain and the inside of the building, at 14,200 linear feet of joint and 312 window perimeters. The building went from a two-stage, drained assembly to a face-sealed assembly — the design philosophy the industry abandoned for good reasons.

The same VE exercise deleted the specification's performance mockup requirement as a "testing cost" — item VE-63, $41,000. That deletion is the one that mattered even more, and it also passed without discussion.

Construction — four moments the building tried to tell somebody

Moment one: the shop drawings. The precast shop drawings showed the revised single-stage joint. There were no weeps and no internal drainage path, because there was nothing to drain to. The window shop drawings showed a sealant perimeter and no sill pan. Both packages were reviewed and stamped by the architect and forwarded by Kestrel with no comment.

Nobody traced the water layer with a finger. Everybody checked dimensions, anchorage, finish, and glass type. Those were all correct.

Moment two: the field mockup. A field mockup was built — an eight-foot panel section with one window. It was reviewed for finish, color, joint width, and sealant color. It was approved.

It was never water tested, because the performance mockup requirement had been deleted in VE and the field mockup's specification language said the mockup would be "reviewed for aesthetic conformance."

A garden hose and twenty minutes would have failed that mockup.

Moment three: the first ninety days of installation. The specification contained a line that appears in thousands of specifications: "Field testing of installed assemblies shall be performed as directed by the Architect."

The Architect never directed any.

That clause is worth studying. It is not a testing requirement. It is a permission to require testing, assigned to a party with no budget for it and no incentive to spend one. A clause with no owner, no frequency, and no funding source is a clause that will never be exercised.

Moment four: the eleven-month warranty walk. Two tenants reported staining at the base of exterior walls. The sealant subcontractor was sent back, cut out about forty feet of joint at the two reported locations, re-caulked, and the items were closed as "isolated sealant failures."

Nobody asked why two unrelated locations on two different elevations had failed at eleven months. Nobody opened a wall. Nobody ran a water test. The symptom was treated and the file was closed.

Year 4 — the letter

Forty-one months after substantial completion, the owner's property manager sent a letter reporting water intrusion at forty separate office suites across five floors and three elevations, plus visible corrosion staining bleeding from beneath several window frames.

An envelope consultant was hired. The investigation took eleven weeks and included water testing at forty-eight locations, six exploratory openings, and infrared thermography of the entire building on two cold nights.

The findings were consistent and unsurprising to anyone who had read the VE log:

1. Sealant adhesion loss at roughly 22% of the vertical panel joints, concentrated on the north and west elevations, with clear evidence of inadequate primer and application to cold, damp substrate during a January installation window.
2. No secondary seal anywhere, so every adhesion failure was a direct path to the interior.
3. No sill pans at any of the 312 windows, so any water reaching the rough opening ran into the wall and down the interior face of the panel.
4. Wet insulation and corroded window anchor clips at 61 locations, several with measurable section loss.

The consultant's conclusion was a single sentence Ray has quoted in every preconstruction meeting since: "The building has no capacity to tolerate the failure of a single material."

The Money

Cost of the remediation      $2,131,230
Savings from VE-47              $60,000
                             ----------
Ratio                              35.5 to 1

And that ratio understates it, because it ignores the value of VE-63 — the deleted $41,000 mockup test that would have caught the whole thing before a single production panel was set.

$2,131,230 / ($60,000 + $41,000) = 21.1 to 1

Who paid

After eighteen months of claim and mediation, the allocation looked like this:

Kestrel's fee on Alderwood Commons had been about $1,530,000. The $480,000 contribution erased roughly 31% of the fee on a job that had closed four years earlier, profitably, with a thank-you letter.

Analysis

The mechanism, not the moral

It is tempting to read this as carelessness. It was not. Every individual decision was defensible in the room where it was made.

The value engineer's job was to find $900,000, and VE-47 was a small, technical-sounding item with a clear price. The architect's reviewer checked what reviewers check. Kestrel's project engineer forwarded a submittal that had been stamped by the design professional responsible for the design. The sealant crew caulked the joints they were given, in January, on a lump-sum subcontract, with a schedule behind them.

What failed was not a person. What failed was that the building's redundancy was removed by one decision and nobody's job description included noticing.

That is the general pattern of systemic envelope failures. They are not caused by bad work. They are caused by a design change that eliminates the second line of defense, followed by a series of ordinary reviews that check everything except whether a second line of defense still exists.

The four moments, and what would have caught it

Every one of those is cheaper than $2,131,230 by at least a factor of thirty.

The clause that did nothing

Go back and read this again: "Field testing of installed assemblies shall be performed as directed by the Architect."

Compare it to what a testing requirement looks like when somebody means it:

Perform field water testing of installed assemblies in accordance with the referenced test method. Test the first three installations of each of the following conditions: typical vertical panel joint, typical horizontal panel joint, typical punched window head/jamb/sill. Thereafter test one randomly selected location per floor per elevation, selected by the Architect. Testing agency engaged and paid by the Owner. Retesting following a failed test, and all costs of correction, shall be at the Contractor's expense. Pass/fail criteria and observation period as specified. Submit test reports within five days of each test.

The second version has a frequency, a selection method, a funding source, a consequence, and a deliverable. The first version has a hope.

When you review a specification during preconstruction — and you should, before you bid it — search it for the phrase "as directed by" and for every requirement that lacks a frequency, an owner, and a consequence. Those are the clauses that will not happen. Decide during buyout whether you want them to happen anyway, and if you do, put them in the schedule and in somebody's scope.

What Ray changed

Three permanent practices came out of Alderwood Commons, and every one of them appears somewhere in this book:

1. Any value-engineering item that touches the exterior envelope requires a written statement of what redundancy is being removed and who accepts the risk, signed by the owner. Not a veto — a signature. Most items still get accepted. A few do not.
2. The field mockup gets water tested, on every job, whether or not the specification requires it. It costs a few thousand dollars and it is the single highest-return quality expenditure available to a construction manager. On Northgate this became Margo's four-unit curtain-wall test — eleven thousand dollars that found a backward gasket at unit four instead of unit three hundred.
3. Repeat warranty items get investigated as patterns. Two leaks at two locations is not two problems; it is one problem with two symptoms until proven otherwise.

Discussion Questions

1. VE-47 saved $60,000 and cost $2,131,230. Does that mean value engineering is a bad idea? Make the strongest case for value engineering, then describe the specific procedural safeguard that would have let VE-47 be evaluated properly without killing the process.
2. Kestrel paid $480,000 — 31% of its fee on that job — four years after closeout. Rank the three Kestrel-side failures (forwarding the shop drawings, not testing the mockup, closing the warranty items) by how much of that share each one earned, and defend your ranking.
3. Rewrite the specification clause "Field testing of installed assemblies shall be performed as directed by the Architect" so that it will actually be performed. Name the five elements a testing requirement needs.
4. The consultant wrote: "The building has no capacity to tolerate the failure of a single material." Take an exterior wall assembly you know and identify one material whose failure the assembly cannot tolerate. Is there one?
5. The eleven-month warranty walk found two leaks and treated them as isolated. What would a written protocol for warranty-period leak reports look like, and at what threshold does an isolated item become an investigation? Who should own that protocol — the CM, the owner, or the architect?

Your Turn

You are Kestrel's project engineer on a new building with the same VE pressure. The design team has proposed the following item at design development:

VE-19. Delete through-wall flashing, end dams, and weeps at the brick-veneer shelf angle at each floor line; provide sealant at the horizontal joint. Estimated savings: $88,000.

Write a one-page VE risk statement for that item containing:

• (a) What the deletion removes, in terms of the four control layers.
• (b) What the second line of defense is after the deletion, stated plainly. If there is none, say so in those words.
• (c) The failure mode: where water goes, where it shows up inside, and on roughly what timeline.
• (d) An order-of-magnitude estimate of the remediation cost if it fails in warranty, using the Alderwood ratio as a sanity check.
• (e) Your recommendation — accept, accept with conditions, or reject — and the signature line for whoever accepts the risk.

Then answer one question honestly: if the owner accepts VE-19 over your written objection, and it leaks in Year Four, does your risk statement protect Kestrel? Under what circumstances does it not?