Vellum & Razor Blades

Vellum & Razor Blades Helping AECO firms deliver predictably with confident coordination. V&RB's Razor Method cuts ambiguity from project starts so design intent

Founder Rick Aspin brings 30+ years across four countries studying what breaks in delivery and sharing what works.

01/09/2026

Are you ready?

01/09/2026

Nearly a billion dollars has gone into three construction robotics firms. Gravis, Bedrock and FieldAI. How much of it went into making sure the data those robots work from is true and correct?

I went looking for that number. Nobody publishes it.

On 19 August, Bedrock Robotics announced its excavators are digging with an empty cab on live jobsites. A water treatment facility in Nevada with Sundt Construction. A 1.2 million cubic yard sitework project with Zachry Construction. That followed a year of testing under human supervision.

The system perceives its environment and executes complex tasks autonomously after a site manager sets the initial plan.

The plan. That is the input.

Construction Dive named the constraint in the same week. The most prominent hurdle is the data crisis in construction, because builders have historically struggled to use the proprietary data sitting in their own files. It will ultimately be data that feeds the machines doing the work.

Vincent P**n, VDC department manager at Structure Tone, said it at New York Build in March 2026. Garbage in, garbage out.

The machine will stop for a person in its path. Bedrock built a 360 degree sensor field to make sure of it. It will not stop for a level that was wrong on the drawing.

So should design documents continue to be developed as a set of two dimensional drawings? Or should they be understood as a by-product of truthful three dimensional coordination?

They're here. Are you ready to hand them the right data? I would like to hear your answer, please put it in the comments.

References in the comments.

Who owns the roof?It sounds like a simple question but it's not.The slopes could sit in the slab or the joists. They cou...
21/08/2026

Who owns the roof?

It sounds like a simple question but it's not.

The slopes could sit in the slab or the joists. They could sit in the rigid insulation above. The membrane is applied to whatever slopes result from those decisions. The minimum waterproof upturn can be correctly installed or not. The model proves which side of the fence you are on. The parapet height can only be correctly set on the back of this work.

Here is one I have experienced myself. The structural model used 2D linework to set out drains. The architectural model had the slopes fully coordinated in 3D. The contractor laid out the slab from the structural 2D linework. That is where slab information lives but with no team clarity on ownership issues arise.

Two sources, one truthfully coordinated and one not, both looking authoritative. Nothing in the process made it possible to tell which was which.

That is a model ownership question. It is answerable at the start of a project rather than discovered in the middle of one.

Roofs are complex assemblies requiring input from multiple sources. The detail that fails is rarely inside a single scope. It is in the space between two or more of them.

So I am curious, and I would like your honest feedback.

How do you manage yours? Are you ahead of the game with a plan or reacting to the issues that arise late?

Let me know in the comments.

Lack of coordination was the largest design defect type in Josephson's 1998 study of seven live building projects. It ta...
20/08/2026

Lack of coordination was the largest design defect type in Josephson's 1998 study of seven live building projects. It tallied 28% of the design defect cost. Roofs carried the highest design-originated share of any element.

See four items below that are worth getting spatially coordinated early. Include tolerances in design to allow things to flex with change and construction.

1. Minimum slopes and installation details. Resolved together in 3D coordination and not only in 2D stand-alone details.

2. Spacing between drain locations. Set and coordinated early enough that the falls respect the reality of design.

3. Constructability. How should this roof be sequenced and installed. Does the model show that sequence working through all conditions.

4. Where interior spaces step out onto a roof, exit stairs being a common issue. Accurately modelled slopes mean you are not finding an extra tread and riser late in the game.

None of this is exotic. It is the difference between a roof that was drawn and a roof that was proven.

Josephson, P.E. (1998), Defects and Defect Costs in Construction, Chalmers University of Technology.

The full Project Start Checklist goes out to subscribers. Subscribe via the Visit my website button.

A roof membrane upturn detail, developed in 2D on a large retail build had issues. It was correct at the point it was cu...
19/08/2026

A roof membrane upturn detail, developed in 2D on a large retail build had issues. It was correct at the point it was cut, but impossible to install at the apex. When it was modelled in 3D and spatially coordinated the parapet had to be raised by two to three feet. This of course cost time and money and led to the design intent being adjusted late in the game.

If the model is spatially coordinated in 3D this same issue becomes a design decision. A design decision that the design team has more control over. When found in construction there are many more risks to the design intent.

Here is what I keep coming back to. Josephson's 9.1% for roofs excludes delay costs entirely. He judged them too difficult to attribute to a specific defect. It also excludes what he called loss-of-reputation cost.

Delays, losses and reputation are risks for project leads from this type of defect.

So the number is real, and it is still underestimated, Josephson said that himself.

Josephson, P.E. (1998), Defects and Defect Costs in Construction, Chalmers University of Technology.

Want to look at how your next roof package gets spatially coordinated before it is priced?

Take the Project Delivery Health Check via the Visit my website button. In the optional question 9, tell us about your own experience and we will follow up with some clear direction.

On a large retail build I worked on, the minimum roof membrane upturn was set by a 2D detail. It was correct in the loca...
18/08/2026

On a large retail build I worked on, the minimum roof membrane upturn was set by a 2D detail. It was correct in the location the 2D detail was cut. Following it around every three dimensional condition showed issues. At the apex of the roof there was not enough space for the installation to work.

This turned out to be a site issue with delay, added cost and adjustments to design intent. The means of installing the roof were developed in a 2D detail vacuum. The issue remained hidden behind 2D deliverables appearing to be correct. In reality it was impossible to install.

We work in 3D models that have the potential for truthful spatial coordination. Traditional 2D approaches let issues sit unnoticed until they are pressure tested. This often occurs in procurement or construction. By that time the fix is reactive and it affects many more people than it should. Their work was based on hidden errors that will always need correcting.

This is what Level of Development is for. LOD is not a compliance exercise. It defines which elements carry enough information to prove the installation is possible. This should be done early and in the model, while it is still cheap to develop and to achieve a proven truth to stand by.

Josephson's 1998 study of seven live building projects found that 69% of roof defect cost originated in design. This was the highest design share of any element measured.

That figure is his. The mechanism I am describing is my reading of why it happens even today.

It doesn't have to be this way.

I write about this in my blog. Subscribe via the Visit my website button.

A Swedish study of seven live building projects published in 1998 shows insight. It measured what it cost to correct def...
17/08/2026

A Swedish study of seven live building projects published in 1998 shows insight. It measured what it cost to correct defects on each part of a building.

Roofs came out worst. 9.1% of roof cost went on correcting defects. Higher than any other element measured.

I've seen where that time and money goes even in the current day.

On a large retail build, the detail that set the roof membrane upturn was deficient. It used only 2D stand-alone elements voiding the 3D opportunity provided by a model. At the location the detail was cut, it was correct and worked no problem.

It passed every review because they all checked 2D documentation not 3D coordination. The 2D exports from a model maintain the traditional approaches. However, the skills to review a 3D model are critical to ensuring coordination is true.

The cost of this error just sat there waiting to be found in 3D. That either happens in the 3D model or the risk is it shows up as an emergency in construction.

Josephson, P.E. (1998), Defects and Defect Costs in Construction, Chalmers University of Technology.

It doesn't have to be this way.

The Project Delivery Health Check tells you whether your next project is starting in control or hoping for control. Find it at the link in my bio.

A free tip that costs nothing and takes about ten minutes.When I open a new model to check for quality assurance, the fi...
13/08/2026

A free tip that costs nothing and takes about ten minutes.

When I open a new model to check for quality assurance, the first place I look is the roof, to see whether the slopes are modelled.

Slopes can be drawn flat in 2D. Or modelled to the wrong gradient. Only when they are spatially coordinated in 3D is the risk actually removed.

Josephson's 1998 paper, Defects and Defect Costs in Construction - A study of seven building projects in Sweden, backs up the idea that roofs should be a prime focus for QA.

Roofs carried the highest defect cost of any element he measured. 9.1% of the roof's own production cost went on correcting defects, against a project average of 4.4%.

Of that roof defect cost, 69% originated in design. Workmanship accounted for 8%.

He does not break it down to the actual issues, so that part is my experience rather than his finding. But if design is where roof defects come from, and slopes are the thing most often drawn rather than modelled, I suggest you start checking your roofs in 3D first.

I'm interested to see what the figures show when we investigate more recent studies.

Open your roof. Check the slopes before you check anything else.

Three days in and I have mostly written about what defects cost. Here is the other side.The research group behind his pa...
12/08/2026

Three days in and I have mostly written about what defects cost. Here is the other side.

The research group behind his paper was not only academics. It included the seven companies whose projects were measured. People with their own skin in the game.

Their estimate: about half the defect cost could be eliminated by simple means. That is a group with the firms who ran the projects sitting in it, not researchers working alone.

You don't have to chase all of it either. Josephson sorted every defect by cost, most expensive first, then added them up as he went.

By the time he reached the worst 1%, he had accounted for 25% of the total cost. By 5%, more than half. By 20%, 79% of the total defect cost.

Twenty nine defects out of 2,879 carried a quarter of the total defect cost.

Design is where it's most reachable. 91% of design defects were judged findable earlier by the people involved.

Put a number on it. Defect costs ran at 4.4% of production cost. The Swedish industry profit margin at that time was about 2%.

If you could halve the defect cost then you have found more than your entire margin again. That is my arithmetic rather than his, to put it into context. Even small fractions change what a project returns.

That's the prize, we don't need perfection. If even a small number of costly issues surface early enough in design, the effect can be dramatic.

I will be reviewing and writing about many more studies such as this to build a picture over time. I am very interested to see just how much improvement has taken place over the decades.

If you want the articles as they land, see the link in my bio to subscribe.

What do you think? Does the opportunity still exist today? I would love to hear your personal opinion in the comments.

Per-Erik Josephson's study is the focus of this week's content. It has two categories of design defect that I had to rea...
11/08/2026

Per-Erik Josephson's study is the focus of this week's content. It has two categories of design defect that I had to read twice.

Faulty design: solutions that are impossible to build on site.

Unsuitable design: solutions that can be built, but obviously should not be.

Those are his definitions, not mine. Together they cost 31% of the design defect money. Nearly a third.

Here is the part that stands out the most. Unsuitable design cost more than faulty design. 18% against 13%.

Think about what that means on site. A design that cannot physically be built is obvious earlier and gets stopped. Someone picks up the task, reads the drawing, prepares for the work, and it fails in front of them. Still expensive, but caught and corrected through re-coordination and redesign.

It's likely that an unsuitable but buildable design creates deeper and more costly rework. It gets built first, then re-coordinated, redesigned and rebuilt.

The more sub-standard the design, the faster it gets stopped. The mediocre design has the potential to hide for longer.

That's my reading of it. Josephson gives the numbers but he does not provide the full explanation.

This study was published in 1998. Has our industry solved the issues behind faulty and unsuitable design?

I would love to hear your thoughts in the comments.

Address

P. O. Box 344
George Town
V8A5C2

Telephone

+13453246794

Website

,

Alerts

Be the first to know and let us send you an email when Vellum & Razor Blades posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Shortcuts

Share