Problem
Carbon fiber shells, measuring almost 3m long and 1m in diameter, were to be manufactured at LBNL for the Large Hadron Collider at CERN, but the current equipment was inadequate. The shells were the largest to be produced at LBNL yet, and the autoclave was not set up to make them.
Design
The shells were almost as large as the autoclave that they are cured in, so three modular configurations of the tooling were designed. The process was broken down into three stages: rolling the mandrel to layup carbon fiber, rigidly supported while in the autoclave, and cantilevered to removed the cured part. The CAD program used was Creo Parametric with Windchill.
Careful measurement and modeling of the autoclaves track, thermocouple, and vacuum connections were required to understand the physical constraints of the new tooling. Even after a new cart and thinner mandrel supports were designed, the tooling only had three spare inches in the autoclave.
First, a new cart was designed to support the different configurations. The cart was designed as long and as low as possible to fully utilize the space in the autoclave. Next, a rolling configuration was designed to rotate the aluminum mandrel during layup. Single setup CNC machining and off the shelf parts were used to reduce the cost of manufacturing. Where parts needed additional rigidity and could not be bolted, welding was utilized. Next, a cantilever assembly was designed to bolted on the end of the cart, so one side can be freed to remove the shell.
FEA Results
Ansys was used to perform finite element analysis simulations to obtain stress, strain, and modal properties. Results from the analyses were used to determine thicknesses of the aluminum mandrel and steel supporting bracket. To achieve more accurate results from Ansys, submodels were used with added details and had strains imported from the full assembly simulation.
Drawings and Manufacturability
OFFICIAL DRAWINGS RELEASED WITH PERMISSION FROM LBNL
Drawings were produced to communicate with both LBNL and contractor machinists. GD&T was used where appropriate to specify cylindricity, perpendicularity, surface finish etc. Many parts were chosen from readily available sites to minimize the amount of parts produced by machinists and reduce the cost of the project.
Here is the poster that I presented at the end of the summer.