Target Pit Structural Analysis

TRIUMF: Canada's Particle Accelerator Centre and home of the world's largest cyclotron is expanding with it's Advanced Rare IsotopE Laboratory (ARIEL). This expansion adds two new beamlines, one additional proton beamline from the main 500 MeV cyclotron, and one electron beamline from their new e-linac (electron linear accelerator); this means the addition of two new target pits (the area where the beam of particles collides with whatever the experiment requires). 

The target pit itself is on the second floor below ground level, and consists of concrete, with deep cutouts where the beamline will go. In the target pit the beamline is made up of sections, each supported from above by a steel shielding block about 2 meter tall.  These steel shielding blocks ("plugs") are made by several horizontal blocks bolted together for ease of manufacture and transport.

Each plug is supported at the top by flanges that sit on the concrete of the target pit. At the bottom - where the sections of beamline meet each-other - there is a pillow seal that must engage (since the beamlines must be kept under vacuum); this pillow seal will enact 10 kN of force onto the beamline connected to the shield plug. For most of the plugs this does not pose an issue since there are two opposing seals on either side of the plug, however the target plug (the plug that holds the hermetic target vessel) does not have this linear geometry, it has two seals on adjacent planes to each-other. This will create a resultant force of about 14 kN.

Since this is a high precision environment the deformation due to this force must be incredibly small. 

The project was broken down into 4 steps:

1. Create an initial bolt layout, as a starting point for simulation

• This was done by the senior engineer before my addition to the project. 

• To do this, one approach would be to treat the target plug as a single rigid body and calculate the internal stresses, use this to calculate a force and use that number to dictate the size and amount of bolts required; then space the bolts out as even as possible to start

2. Verify the solution with hand calculations

• Hand calculations were completed in Mathcad

• This was broken into multiple sections since the bolts were not in pure tensile or pure shear loading

• Bolts were first evaluated in pure shear according to Shigley's Mechanical Engineering Design, section 8-12: Bolted and Riveted Joints Loaded in Shear, specifically the section on eccentric loading

• Bolts were then evaluated in tension/compression (since the loads are applied below the bottom of the bolts and will result in a moment). Since the joint is very long, and very narrow in places, we took a cylindrical representation of the joint; we used the cylinder directly below the washer as seen in An Introduction to the Design and Behavior of Bolted Joints by John H Bickford, Chapter 5, Section II-B: Computing Joint Stiffness, if joint thickness is greater than 8 times the diameter of contact (between the head of bolt or washer and the joint), then the diameter of the representative joint must be less than or equal to the diameter of contact - in this case the diameter of the washer.

• Once the shear and tensile stresses on the bolts were found, they were combined using Mohr's circle into an equivalent maximum tensile stress and evaluated against yield criteria

• Our initial solution passed hand calculations

3. Verify the solution with simulation

• A simulation was set up in ANSYS using the structural tool

• The simulation was set up in three steps:

1) Applying the preload

2) Applying gravitational load (weight)

3) Applying the pillow seal load

• As per BC building codes, an extra case was run using seismic forces up to 1/3 of gravitational forces

  • For this case we applied the seismic loading in the worst case direction

• In the results from the first solution, there was a separation of layers between the blocks that was much greater than desired; this was due to geometry of the bolt placements rather than total required bolts.  The shield plug has a cutout down the length of it to run services, leaving a thin flake only a few inches wide, and several inches deep. It was in this flake that we saw deformation since there was originally only one bolt there.

• We redesigned the bolt pattern to have two bolts there, but had to make them smaller to fit in that space with enough wall thickness beside them and between them, then we returned to step 2 to hand calculate our new analysis 

• Upon successful calculations for the second solution, the simulations were completed again, using the new geometry and were successful

4. Write an engineering note summarizing the design and analysis

• The engineering note was written to describe the methods used in calculation and summarize the results clearly and concisely

• All Mathcad calculations and screenshots of the ANSYS setup and results were included in the appendices