Clawfoot Reproduction

I had recently picked up an antique clawfoot bathtub to refinish and install in my home, and it came equipped with 3 matching claw feet and one oddball. Seeing an opportunity, I decided to use one of the existing feet to produce a new mold pattern for casting, with the intent of producing a small volume for future sale. This would give me the opportunity to work out the reverse engineering process using the tools I had developed, and provide a real, marketable product to sell and help pay that nagging mortgage that doesn’t seem to be going away any time soon.

 The first step in the process was to identify the appropriate part for reproduction. The original castings were scaled with rust. Some had chips, and the lockscrew that holds the casting onto the tub had seized on all of them. They all looked pretty nasty.

I chose the least ugly casting of the lot (no two were exactly the same) and prepared it for scanning by coating it with a white primer material for better registration by the laser digitizer. For more recent scans, I prefer to use a chalky material known as developer, which can easily be removed with soap and water. The white primer I used is somewhat permanent, and would need to be chemically stripped or sandblasted to be remove, but it smooths over surface defects and stands up to constant handling.

The initial scans were done at medium resolution, and rescaled to roughly 500,000 vertex points to create the initial surface. I had to reduce the surface detail of the scan before I could bring it into Solidworks, so boths scans are shown below.

 

 

 

 

 

 

 

 

 

 

 

During the course of this project, I downloaded the free Solidworks 2009 training package, to see if it was capable of reproducing the complex organic shapes that often appear in vintage castings. I am a big fan of Solidworks, and have used it professionally for 5 years, but I found it struggled with large triangulated surface files at this resolution, and found it to be a poor fit for the work I intended to do. The simplified surface that I finally got to open in Solidworks was lacking some detail, and appeared blocky. Even at this scale, it still slowed the computer to a near halt, and made progress slow and tedious.

After a couple of days of work, a string of predictable but nonetheless frustrating software crashes, and one completely irretrievable corrupted file that crashed when I was nearly done, I had completely rebuilt the part parametrically in Solidworks. The realtime rendering looks beautiful, and is a close fit for the nickel plated finish I intended for these parts.

 

 

 

 I chose to use a parametric modeling package because it allows me to control unique parameters that define a part (parametric modeling). That means I can selectively build relationships between features on the part, such as the overall height, and vary those parameters to fit different applications.

In the case of the clawfoot casting, there are two distinct parts required for a bathtub; a short one for the side where the water drains out, and a taller one for the opposite side. The parametric model allowed me to quickly adjust for the 1/4″ height difference by modifying the height while maintaining the same basic shape. The short and tall feet are shown overlaid below. The green casting is the taller of the two.

 

 

 

I can also quickly vary the shape of the mounting tab on the rear of the casting (to accommodate different tub mountings) as well as the shape of the profile (to accommodate different tub shapes). The work I did to create a single casting can now be expanded to create a whole family of related castings that all share the same basic look. The variations can later be documented in a tabular format (using a program such as Excel) and generated automatically by the computer. Probably beyond the scope of this project, but a possibility just the same.

 

 

Once I had defined the shape of the part I wanted to build, it was time to create the mold patterns for the sand casting. This is where the electronic data manipulation really saves time.

I created a single block of material in Solidworks to represent the top and bottom of a finished sand mold. I positioned my part according to the parting line, and split the virtual block along the curved parting line and created a mold cavity in the blocks. The mold cavity was scaled up to account for shrinkage in the iron, and the two resulting mold halves now represented the finished sand mold.

I then used the virtual molds to create a set of foundry patterns, by filling the cavity of the virtual mold. I could have gone straight from the part to the pattern, but I liked to see all of the pieces as I progress to better prepare for possible problems.

I generated machine toolpaths and ran a test part by cutting polyurethane foam on the CNC. The foam was a lot cheaper than ABS or aluminum, and brought to light a few issues with my machining that weren’t apparent in the simulations, so I could make corrections for the final pattern.

 

 

I took my test patterns to a local foundry to work out the final pattern design. I determined a good match plate size, and designed a pattern to accommodate all four feet in a single casting, with the proper gating system to faciltate a good iron casting.

 

I then created the top and bottom match plates to create the cavity for all four castings and the gate, runner, and risers system within a 16″x12″ box.

 

The pattern was now ready for the final tooling production on the CNC.