Plaster Disaster to Digital Master

Reverse Engineering

—  Aft Pylon Fairing  —

     I chose this job as a case study because it has the full range of reasons, that are common in the industry, for a part to need reverse engineering. Some of these reasons are very universal to other industries as well. Some of the reasons include:

  1. The part was created by hand, by master tool-builders, as no CNC machines were used to cut the profiles.
  2. Engineering data for the profile was hand drafted on Mylar at 1:1 scale for visual inspection of the rib templates.
  3. Rib templates are placed on flat tooling plate and set to specific distances apart, as parallel as possible to each other, with a grid of thin supports between them.
  4. Plaster is poured into cavities between rib templates with a thin sheet of material stretched tightly over the assembly to help form and retain the plaster.
  5. After curing, the plaster is hand blended in between the rib templates to form the final contours.
  6. There is an incredible amount of man hours and talent invested in this procedure, and this plaster master engineering tool has made successful parts for many years. However, there is a fair amount of hand fitting required to get the individual components to fit to each other because of the tolerance build-up in each of the master tools and additional tolerance build up error in the parts themselves.
plas2

(Very damaged tool)

      Well here it is fresh out of the box, and oh isn’t it lovely.;–)  Don’t mind the missing pieces, we’ll digitally fill those in later. Notice all the different colors of plaster, indicating repaired, replaced and or modified sections in the tool. This tool was part of a joint effort program between the United States and France, so it’s probably seen more miles in transit back an forth than most Sky miles platinum members, and I’m afraid it’s beginning to show.
      The individual aluminum “loft plates” that basically control or create the shape, were first hand drafted onto Mylar sheets at a 1 to 1 scale. The aluminum pieces are then cut and trimmed to visually match these Mylar “loft curves”. The Mylar now becomes the actual inspection tool for the plates, and thereby transfers the engineering authority to the “loft plate”. This, of course, cannot be done completely accurately, so in this case we also want to inspect not only the plaster surfaces, but also the aluminum loft plates themselves, against the Mylar, this will show the error in the “loft” plates, which are the basis of the plaster model. To do this we must first digitize the Mylar sheets. With the sheets being 4′ x 16′ in length, using a traditional scanner would result in a great deal of error, so the Laser Tracker is a clean and accurate method of capturing this legacy data. Once the 2-D data is captured, we then model the data in 3-D CAD. This is now the most accurate representation of the original design intent of the part.
We then digitize the aluminum sections of the master tool and compare them with the nominal Mylar data.		loftdetail

(Close up of “loft” plates)

spline1

(CAD model of digitized Mylar loft data)

       As you can see in the picture below, the deviation error (shown in red +, andblue –) between the nominal Mylar data and the actual part is very inconsistent. This is as expected when you consider the method of production, and that the shaping and inspection of the aluminum “loft”  plates was done visually, by hand. These errors are now forever part of the master inspection tool and those errors  are inherited in the subsequent parts that this tool inspects.. This is a case where it is not desirable to simply just reproduce the actual surfaces of the plaster model, this would also reproduce all the errors and damage that the plaster mould contains, and make those errors a permanent part of every actual finished part that is made from here on out. Rather, we should attempt to capture the original intent of the designers, which in this case is represented on the hand drafted Mylar.
     If there are  large  errors between the plaster and the Mylar, we must add a third artifact to help determine which geometry is correct. This would need to be one of the following:

  1. An actual part (pylon fairing) that has been proven to fit well on the aircraft.
  2. A CAD model of the mating part(s) geometry, or again we could digitize the mating part(s) if no CAD data were available.
  3. Data from other hard tooling or jigs that actually were used to make the most recent successful part.

As no actual parts or mating part data was available for this project, we turn to the hard tooling that holds the actual part, and drills all the 5 axis mounting holes that secure this fairing to the mating pieces. This is called a drill or jig  fixture. Once again, the Laser Tracker  shows excellent capability as a portable co-ordinate measurement machine, or “portable CMM” to capture the data.

(Actual error between plates and Mylar)

 platedev

( Interface drilling fixture)driljig

      After careful evaluation of all the digitized data, we compile an accurate CAD model that accurately represents the original engineering intent and also incorporates all the updates and changes that have been made to the hard tooling and plaster master since the original Mylar were drawn. This is a far better solution than simply using any single one of the  original tools and calling that “right” or nominal, and now all of the storage for all of the associated inspection tools can be reduced to a simple CD, not to mention we can now modify, reproduce, and create any of the required geometry at will.

Finished CAD model 

 cad1

Top of page