Planning the high volume production of Accelerator Cells for the International Linear Collider
Stanford University
Department of Engineering
IE225/ME217C
Professor J. Jucker
Authors:
Brent Cheldelin
Mats Cooper
Anne Robinson
Erik Teetzel
12/3/98
Table of Contents
1 Executive Summary 3
2 Introduction 4
3 Background 5
Cell Description 6
4 Robertson Precision Manufacturing Process 8
Existing Process 8
Process Definition 8
Process Times 11
5 The LMC Manufacturing Process 12
Company Profile 12
Proposed Process 12
Process Times 14
Process Considerations 15
6 Process Comparison 17
5.1 Setup for Robertson Process 17
5.2 Setup for outsourcing with LMC 21
5.3 Materials 23
5.4 Manufacturing with Robertson 27
5.5 Manufacturing with LMC 28
7 Cost comparison 35
Robertson Precision 35
LMC Process 35
8 Qualitative Analysis 38
Summary of SWOT Analysis 41
9 Recommendations 43
Appendix A: Equipment and Facility Drawings 44
Appendix B: Mori Seiki Machine Literature 45
Appendix C: LMC Adiabatic Process Manufacturing Technology 46
1 Executive Summary
The Stanford Linear Accelerator Center (SLAC) is currently developing the “Next Linear Collider.” Since the construction of the 1966 linear collider, LINAC, new technologies have emerged for the improved production of the accelerator structure cells, a critical component of the linear collider. The cells are each 2 ½” in diameter and 3/8” thick and are manufactured out of high purity copper. It is imperative that they hold a sub-micron tolerance on several major surfaces. At this time, SLAC is interested in determining if the aforementioned new technologies are a viable and feasible alternative compared with the current manufacturing process. This report will look at 2 alternatives for the rough machining of these cells and evaluate them based on feasability, cost, and scheduling issues.
The first plan is referred to as the Robertson Precision Machining plan as it is a modified version of the process developed by Robertson Precision for the previous linear collider. It involves purchasing 16 Mori Seiki machines, setting up a factory at SLAC, and manufacturing the ultra-high tolerance cells in-house. The major advantage of this method is the fact that SLAC will have a continuous-flow manufacturing system that is performed and managed by SLAC. The key disadvantages of the Robertson plan are that 1) it is very slow compared with the alternative process and 2) it is more costly.
The second plan evaluated is the outsourcing of the rough-machining process to LMC of Dekalb, IL. The major difference between LMC and most manufacturing methods is that LMC uses Adiabatic Manufacturing Process Technology to rapidly punch out copper disks from flat sheets of copper. The process then relies on coining the parts to tolerance. Obviously, if this unproven process is shown to work, it would result in a major reduction in the manufacturing costs and processing times of the disks. Shipping and the lack of internal quality control are the major disadvantages of this process.
From the data collected and the synthesis of the costs and schedules, the recommendation to SLAC is to pursue the LMC manufacturing technique. SLAC would realize about a $3 million cost savings and there is plenty of buffer inherent in this manufacturing process that would allow for any unforseen scheduling issues that are typical in a project of this stature.
2 Introduction
The Stanford Linear Accelerator Center (SLAC) is currently concerned with developing the “Next Linear Collider.” Since the construction of the 1966 linear collider, LINAC, new technologies have emerged for the improved production of the accelerator structure cells, a critical component of the linear collider. At this time, SLAC is interested in determining if these new technologies are a viable and feasible alternative compared with the current manufacturing process of these cells.
Currently, the manufacturing process is being redesigned for manufacturability. The difficulty with the manufacturing of the existing cell is that they are designed for a production size of one. Therefore, all costs associated with this process are extremely high. The main challenge posed to SLAC by the government was to develop a system that would make the manufacturing of the Next Linear Collider more robust and much cheaper. Thus, the purpose of this project is to analyze the two manufacturing processes for SLAC:
- the base-line process currently used
- the near-net shaping process[1]
Based on decision criteria identified by SLAC, the best option will be determined.
3 Background
The Stanford Linear Accelerator Center (SLAC) is the world’s largest linear accelerator research facility. SLAC is located in Palo Alto, California, on Stanford University grounds northwest of the main campus. At SLAC, physicists explore properties of matter by colliding electrons and positrons at extremely high speeds. Achieving this result requires a linear collider, or similar tool.
Currently, SLAC is using the “LINAC”, a two-mile long linear collider completed in 1966, to assist their research. However, their current theory and understanding have surpassed the capabilities of the LINAC, prompting SLAC to develop and build a larger and more powerful tool. The proposed solution is titled the Next Linear Collider (NLC) (Figure 2-1), a 25 km long particle accelerator that is approximately ten orders of magnitude greater in power than the LINAC. Although the conceptual NLC in design is far more effective, it is significantly more difficult to manufacture.
Figure 2-1: The Proposed NLC Layout
The proposed collider is a 25 km tool that is composed of 9,056 1.8 m long structures (Figure 2-2) aligned end to end. Each structure is comprised of 207 precision turned cells, all of which are held to sub micron (0.0005 mm) tolerances. A total of 1.9 million ultra-high precision cells are required to construct the proposed NLC.
Figure 2-2: An Accelerator Structure
Cell Description
The cell itself is circular in shape with varying complex internal features that are critical for efficient particle acceleration (Figure 2-3). These internal features include an iris and four milled Higher Order Mode (HOM) slots, all of which are held to ultra high precision tolerances. Because the cells are diffusion bonded together, the two surfaces need to be both flat and parallel to within two microns.
This report considers only the roughing operations. The next step is final machining and is not part of this project. Appendix A, Figure 1 shows the part prior to final machining; this will be the geometry that each of the competing processes are required to deliver. Because of the ultra high precision demanded for the final cell, the dimensions on the roughed part must be held to a tolerance of ±0.003”.
Figure 2-3: An Accelerator Structure Cell
As stated in the introduction, there are two distinct processes, which this paper will be comparing. The current process, developed by Robertson Precision, utilizes conventional milling and turning operations to produce the roughed part. A divergent alternative from this traditional method is detailed in this report as the LMC near net shaping process. The technology developed at LMC, adiabatic forming, possesses the potential to make a near net shaping process possible for the cell manufacturing. This report will compare and contrast the two aforementioned processes, in the interest of recommending the most cost-effective solution. To assist in the costs of developing the process at small manufacturing firms, SBIR (Small Business Innovation Research) grants are given by the federal government in the interest of stimulating technological advancement. SBIR grants awarded to firms of this nature usually dedicate $750,000 to the development of unproven technologies. It is SLAC’s hope that both Robertson Precision (section 3) and LMC (section 4) are able to utilize this monetary channel to further develop the cell manufacturing process.
4 Robertson Precision Manufacturing Process
Existing Process
Robertson Precision in Redwood City, CA is redeveloping the existing process for the cell’s rough machining. Currently, the roughing operations are being performed on two separate machines- a lathe for the turning operations and a mill for the cutting of the HOM holes. However, the process defined in this report assumes a unified twin spindle, CNC lathe and milling center. Robertson Precision will procure this machine, the Mori Seiki DL-25MC, during the Phase II of the SBIR process, starting in April of 1999. (For more information on Mori Seiki machines, please see appendix B). Although the DL-25 will greatly reduce the set-up effort, the chip cutting times recorded on the existing lathes and milling machines are comparable to the Mori Seiki’s predicted performance.
In an effort to help establish current machine performance, Robertson Precision manufactured a batch of test cells. These cells were analogous to the current design in both size and tolerance, however they lacked a complicated iris profile as well as the milled HOM slots and holes. The times recorded for these test cells are listed in Table 3-1 below.
Table 3-1 Test Cell Rough Machining Times
Note: The times recorded by Robertson Precision are strictly cutting times, they do not include part or tooling set-up times.
Process Definition
As previously described, all of the cell’s rough machining will be performed on a twin synchronous spindle Mori Seiki manufacturing center. This machine will be able to accept bar stock from a feeder, and produce a roughed part without any operator intervention. The current process is as follows:
Figure 3-1 Robertson Precision Cell Manufacturing Flow
Insert Bar Stock
Currently, the process is baselined with the bar stock being loaded manually into the feeder. This requires the effort of an operator, but would only be necessary once every 25 hours (assuming the current manufacturing rate) so it may be possible for one operator to feed all of the parallel manufacturing centers.
Rough Turn Side 1
The bar stock is feed through the center of the spindle, and immediately faced with a turning tool. The tool path will likewise continue to the side of the bar stock and turn down the outer diameter.
Drill HOM Holes
After the first face is turned, the HOM holes will be drilled to the desired dimension.
Rough Mill Side 1
After the drilling operation, the machine head rotates to the next tool, a small diameter end mill. Using the new tool, the HOM slots will be roughed on side 1.
Part from Bar Stock
Once the milling operation has finished, the two independent spindles of the Mori Seiki are engaged to the same RPM. The tool changer again switches cutters, from a mill bit to a parting bar. Spinning at an identical angular rate, the two spindles extend towards each other, and the turned part is grasped by the empty chuck. While being constrained by both spindles, the bar stock is parted by the cutting tool. Both spindles then return to their original position.
Rough Turn Side 2
With the part now constrained in the second spindle, a cutting tool turns the second face and outer diameter of the cell.
Rough Mill Side 2
The second turret now rotates to get a milling bit in position. Once the proper tool is in place the HOM slots are cut. Once this operation is complete, the part is then removed from the machine and a second cell is ready to start the manufacturing process.
Figure 3-2: The Robertson Manufacturing Process
Process Times
In practice, this entire process shall be free from human interaction, thus reducing the variability in cycle times to an insignificant level. From Mori Seiki’s literature, we find that a common tooling change, with the rapid movement of the cutting head will be around 3 seconds. Further, it is estimated that the synchronous switch from one turret head to another will be performed in 30 seconds, with the inclusion of the parting operation. Table 3-2 is the current estimated times for the roughing operations of the cell manufacturing process.
Table 3-2: Current Operation Times
The values for the Turning Operations were estimated by multiplying the values generated by Robertson Precision by 1.5. This scaling factor is due to the increased complexity of the features from the test cell to the current cell. Although the process is numerically controlled, form factor does in fact influence cutting times.
As seen from Table 3-2, the current manufacturing process takes almost nine minutes for each cell. This is the baseline cycle time that this report shall use to estimate the time and cost of manufacturing all 1.87 million cells.
5 The LMC Manufacturing Process
*The following section will describe the potential outsourced process for the manufacture of these cells.
Company Profile
Lindell Manufacturing Company (LMC), a small company based in DeKalb, Illinois, has expertise in adiabatic manufacturing. Like traditional stamping operations, LMC has developed machinery that creates parts through a shearing motion. LMC's uniqueness, however, arises from the speed at which the material is sheared. Unlike traditional stamping, the LMC manufacturing process is so fast that localized annealing occurs, leaving a high precision part with very little or no strain hardening around the parted surface. Because the machine fires at a high speed, there is little need for contaminating lubricants or cutting fluid. The process is fast, clean, and repeatable, making it a viable candidate for a near net shaping process of the Next Linear Collider cells.
Proposed Process
For LMC to properly meet the tolerance and volume demands of the NLC cells there will be a significant cost in the development of tooling and machinery. Funding will start through the submission of an SBIR request, and capital equipment shall be procured from this grant.
The proposed process flow for the adiabatic forming of the NLC cells came from a meeting between Jeff Rifkin of SLAC and Lennart Lindell of LMC. The low part thickness to part diameter ratio coupled with the high malleability of copper, makes it impossible to part off blank disks from copper bar stock using LMC's current Excalibur[2] process. However, Lennart is confident that the same adiabatic processing technologies will work for stamping out disks from copper plate stock, giving parts with high accuracy with very little strain hardening. The plate stock has a cross section of 2.5" by 0.375", as seen in Figure 4-1.