with support from
D. A. Deuser, K. T. Slattery, and G. W. Wille of The Boeing Company
F. Arcella and B. Cleveland of AeroMet Corporation
* Work supported by U.S. Department of Energy through University of California, San Diego Purchase Order 10087872.
Abstract - Two fabrication processes are being developed that may significantly enhance fusion as a competitive electrical power source. Currently, the capital cost for experimental, developmental, and commercial fusion power cores is too high. Innovative ultra-low cost fabrication methods being developed by Boeing and its vendors may significantly lower the cost of normally-conducting toroidal field (TF) coils. Laser or plasma-arc forming is proposed for the large copper centerpost. This process uses a laser or plasma arc to "additively machine" the centerpost with all necessary features including integral coolant passages. The 2.7 x 106 kg aluminum TF return shell can be fabricated using a spray casting process that atomizes and sprays molten aluminum onto a preform structure and around embedded coolant tubes. A scoping cost study indicated the capital cost of the TF centerpost and return shell leg will be reduced by $330M (greater than an order of magnitude) which will reduce the projected cost of electricity (COE) by approximately 10%.
For years, the ARIES system code  has evaluated conceptual tokamak power plant designs with differing plasma configurations and power core systems and options to determine and optimize the capital and operating costs of the entire power plant. To efficiently examine the multidimensional performance and economic parameter, the level of detailed cost estimating is constrained by the level of a system or a component, e.g. Starfire  and ARIES-RS .
In the case of the spherical tokamak, the optimal configuration generally favors a low aspect ratio of 1.4 to 1.6 that causes the plasma to be very tall and very close to the centerpost. It is not feasible to have a power-generating or tritium-breeding blanket on the inboard area of the centerpost. The centerpost may or may not be shielded; hence the TF coils are normally-conducting. Although the blanket and shield system of the spherical tokamak is not as extensive and costly as the conventional tokamak, the cost of the TF coil system increases significantly as the height increases. Moreover, the TF centerpost must be replaced frequently.
The ARIES-ST team thought it would be worthwhile to conduct a more in-depth study of the evolving TF coil design and commissioned Boeing to review the design approach, develop a fabrication concept for the centerpost and return shell, and prepare a more detailed cost estimate for the dominant capital and operating cost item in the plant, the TF coil system.
The design basis for the ARIES-ST centerpost and TF return shell is shown in Fig. 1. A continuous shell was employed for the return TF coil system to obtain a low aspect ratio plasma, low TF field ripple, and moderate PF stored energy. This continuous shell could also function as the vacuum vessel.
The 30-m-long centerpost, shown in Fig. 1, is comprised of an upper tapered section to electrically mate with the outer shell, a straight cylindrical section that is in close proximity with the plasma, and a larger cylindrical section at the bottom to increase the conductive area to the maximum extent to reduce ohmic losses. The diameter of the central region is 1.6 m and the lower section is 3.35 m. The 30-meter, unitary centerpost is constructed of high-conductivity copper and cooled with low-temperature water in a single pass flow configuration. The centerpost is comprised of roughly 85% copper and 15% water. This part weighs approximately 0.851 x 106 kg.
The outer shell has three distinct parts, all water-cooled. An upper shell extends from the top of the centerpost to the midplane, where it is connected to one of the power supply busbar leads. The middle shell extends from midplane, where the other busbar connection is made, down to a maintenance break. The removable third shell extends from the maintenance break down to the lower connection to the centerpost. The outer shells could be made from either copper or aluminum, with thicknesses adjusted to achieve the same coil conductance.
The conventional method of constructing the copper centerpost assembly with internal water passages would be to fabricate a cylindrical wedge segment with grooves for the coolant and then weld the entire assembly in a fixture to minimize distortion. This difficult and costly technique would involve labor intensive welding and inspection. The unit cost to conventionally fabricate this part is well over $80/kg, thus this large assembly would likely cost $70M to $100M or more. Also, this component must be replaced every two to three years, and this will impact the COE as an operating expense.
The thicker cross-sections of the outer TF shells at the top and bottom complicate the fabrication approach. The TF shells would probably be cast in smaller pieces and welded together. If aluminum is chosen, separate stainless steel coolant tubes may be necessary for high reliability and long-term operation. The unit cost to fabricate such an aluminum component is over $100/kg. The total weight of the three aluminum shells is 2.69 x106 kg, which would suggest a total cost of $270M or more. If it were constructed of copper, the system would be significantly more costly.
Therefore, a conventional approach is not an attractive or cost-effective way to fabricate these components. Instead alternative, innovative fabrication technologies were investigated.
Two new fabrication processes are being developed that may be well suited to fabricate these components. The first fabrication process is laser or plasma arc forming, which melts powdered metals and deposits them as required to form the part. The second process is an in-place spray casting of molten metal.
Laser or plasma arc forming
The laser or plasma arc forming process is derived from existing stereolithography processes that construct solid plastic or cardboard models directly from CAD drawings. These solid, low-strength models are useful for quick fabrication of prototypes and models. This process has now been extended into metals and other more substantial materials for robust prototypes and limited production components.
The next evolutionary step is to extend this technology to the fabrication of larger and higher strength production parts. The Johns Hopkins Applied Physics Laboratory  in cooperation with the Advanced Research Laboratory at Penn State  and MTS Systems Corporation , under DARPA/ONR sponsorship, have developed a laser-based process to fabricate structural components of titanium alloys. In 1998 and 1999 Boeing, Northrup Grumman, and Lockheed-Martin awarded AeroMet Corporation , a subsidiary of MTS Systems Corporation, contracts to fabricate titanium aircraft parts. Results from these developers show progress in obtaining the desired material properties, namely fatigue strength. The results to date are between traditional HIPed investment castings and forgings used in aircraft structures for titanium. It is anticipated that other materials, including copper, would exhibit similar results.
So why would laser forming be an improvement over conventional techniques? Historically, to obtain cost effective, lightweight, high-strength components, there has been an evolution from riveted sheet metal, thin welded plate, and machined forgings to high speed machined plates and forgings.
The laser forming process, shown schematically in Fig. 2, is closer to the original stiffened skin process in that it starts with a thin panel or skin of the required thickness. Then a stream of powdered metal is directed toward the part and a laser melts the metal onto the substrate with little waste. Like a stereolithography process, the new material is applied only where required to form the 3-D part. The application rate of the material is limited only to the power of the laser and the compatibility with the material. Plasma arc sources seem to be better suited to copper at present. Surface finishes are typically 32 to 64µ min. To obtain better finishes, the application rates are reduced.
The laser forming process is ideally suited to construct the copper centerpost shown in Fig. 1. It is a simple geometry of cylinders and cones with multiple, continuous coolant passages from the top to the bottom of the part. An initial blank or preform will be used to start the process at the bottom. It is estimated that 20 kg/h of material could be deposited by a single forming head. A set of 10 laser forming heads would simultaneously deposit each layer of material. If a high surface finish is desired, dedicated laser heads could be assigned for that desired fabrication. At the combined rate of 200 kg/h and 24-hour operation, the 0.85-Mkg part could be completed in roughly eight months assuming a generous allowance for downtime.
The centerpost would be gradually grown in place from the initial preform. First, the lower 3.35-m-diameter cylinder would be constructed with the integral coolant passages. Continuing upward, the laser heads would form the tapered cylinder with the coolant passages being transitioned to those in the central 1.6-m-diameter cylinder and the top tapered section. Any errors detected during construction could be machined away and new material would be deposited.
Due to the overall size and weight, it is anticipated that the part would be fabricated on the reactor site. Since the lifetime of the centerpost is approximately two full-power years, the fabrication time is well matched to the component lifetime and replacement schedule.
Aluminum was chosen as the material for the outer return TF shell as it was lighter, easier to fabricate, and perhaps less expensive to fabricate. Conceivably, the outer shell could be constructed with the same laser forming process as mentioned above. But because the weight of a TF shell is more than three times as much as the centerpost, the time and cost to construct would be correspondingly higher. Therefore, spray casting of the shell was evaluated.
Spray casting of a molten metal involves holding the metal just above the melting temperature, atomizing it, and spraying it onto a preform structure. This process has fast deposition rates, up to 0.5 kg/s per head. Four heads are used to help speed the fabrication process, yielding 7200 kg/h at full production rate. Labor hours were calculated using an efficiency factor of 50% with an operator and an assistant or inspector in attendance. This would enable all three shells to be spray cast in less than six months with normal work shifts.
As a result of the faster deposition rate, the finish detail is reduced. If a smooth finish or precision dimensions are needed, some final machining will be required. On the TF return shells, this final finish will only be required locally at the flanges and joints with the interfacing hardware, such as the busbars and vacuum pumps.
This process requires a preform structure to initiate the process. This is a desirable feature for the TF shell since the preform will serve as the power core vacuum vessel. The interior preform structure is 0.5 in. (or 1.25 cm) thick. An inspection can verify vacuum integrity before initiating the spray casting process. This structure is fabricated in the conventional manner. After the vacuum inspection is concluded and the spray cast process starts, the vacuum vessel will be an integral part of the TF shell.
No cutouts or ports are assumed for this estimate. Vacuum vessel "orange-slices" will be bump formed into the approximate hemispherical elements and welded to form the upper hemisphere. Due to the size of the finished parts, welding and inspection will be on site.
The upper hemisphere would be spray cast in its final orientation and position in the power core. The weight of the vacuum vessel and spray cast shell is 1.56 x 106 kg, assuming 85% mass fraction (15% void for coolant). The two lower shell elements will be spray cast in an inverted position with added trunnions to assist in handling and inverting the large components. The weights are 0.584 x 106 kg and 0.548 x 106 kg for the middle and lower sections, respectively. A wastage allowance of 5% is allowed for the spray cast process.
It is not advisable for the coolant to be in direct contact with aluminum. Therefore, stainless steel tubes are embedded in the aluminum shell to contain and distribute the water. As the shell is being fabricated, the tubes can be placed in position. The aluminum is spray cast around the tubes, embedding them in the aluminum structure to cool the volumetric heating of the high-energy neutrons during operation.
Fig. 3 schematically illustrates the spray cast process. Large 1000 to 1200-lb. T-bars or sows of aluminum alloy are melted in the furnace. A 5000 series aluminum is used because it is a high conductivity alloy with low impurity content, which minimizes transmutation products. The melted aluminum is transferred to a holding furnace that precisely controls the liquid metal temperature. A low-pressure, liquid-metal pump removes the liquid aluminum from the holding furnace and transfers it to a distribution pump. This distribution pump sends the liquid metal through heated lines to the spray robots. High-pressure pumps atomize and spray the metal onto the preform surface. These robots are track mounted and can access the entire perimeter surface of the shells. A local cover-gas shield minimizes atmospheric impurities.
The cost estimates were obtained for the TF coil system using the new fabrication technologies. These processes have only been demonstrated on small components (compared to the scale of the parts referenced herein). The costs are developed based on a mature design and fabrication technology. The component costs represent tenth-of-a-kind articles with appropriate learning curves applied and are reported in 1998 dollars.
To assure the estimate is conservative, the process hardware costs associated with the laser or plasma arc forming and spray casting processes are included as direct capital costs. To account for possible material wastage, an allowance of 5% was added to the material quantities purchased. An efficiency factor of 50% was used in the spray cast rates due the lack of experience on parts of this scale. Allowances for inspection and rework were provided. Even the energy for the lasers and fuel for the furnaces were included. An overall contingency allowance of 20% was added to the total estimated cost, including material, labor, process energy, and process hardware. A prime contractor fee of 12% was also included.
Centerpost Cost Estimate
As compared to the conventional process to fabricate the centerpost, which was estimated to cost in the range of $70M to $100M, the laser-formed centerpost (0.85 Mkg) is estimated to cost $6.9M, or a fabricated unit cost of $8.09/kg. This value is only a few times the material cost of $2.86/kg. This significant cost reduction is possible because of the drastic reduction in labor in the fabrication process.
Vacuum Vessel Cost Estimate
The cost estimate for the conventional construction of the aluminum vacuum vessel (39,776 kg for three parts) is $3.4M, or a fabricated unit cost of $85.14/kg. This cost is typical for construction involving forming and welding a large structure.
TFShell Cost Estimate
As mentioned earlier, the conventional fabrication process for the aluminum shell was estimated to cost approximately $270M. The spray-cast TF shell (2.69 Mkg) is estimated to cost $13.1M, or a fabricated unit cost of $4.85/kg, excluding the vacuum vessel. This unit cost is only slightly greater than the material cost of $1.87/kg. This is possible because of the significant reduction in the fabrication labor. Two furnace manufacturers provided somewhat differing cost estimates for melting and holding furnaces, and the cost estimate quoted used the higher vendor quote.
Total Toroidal Field Coil Estimate
The cost of the TF coil system, if fabricated with these advanced, low-cost, and labor saving methods, would reduce the fabrication cost by greater than an order of magnitude (from $350M to $23M including the conventional fabrication of the vacuum vessel).
The innovative fabrication techniques of laser forming and spray casting offer an ultra-low cost approach for the centerpost and TF coil shell. These "additive" machining processes only add material where necessary and are extremely labor saving. These components are well suited to these techniques in that they are simple, continuous structures, have moderate stresses, and have distributed internal cooling passages. Due to the significant cost and time advantages that these ultra-low cost processes offer, it is recommended that these technologies be developed and applied to these and other appropriate structures.