DEVELOPMENT OF THE M ECHANICAL DESIGN FOR A FREEZE - OUT PURIFIER By Duncan Kroll A THESIS Submitted to Michigan State University in partial fulfillment of the r equirements for the d egree of Mechanical Engineering Master of Science 2020 A BSTRACT DEVELOPMENT OF THE MECHANICAL DESIGN FOR A FREEZE - OUT PURIFIER By Duncan Kroll Purification systems are necessary to support commissioning and operation of helium refrigeration and associated experimental systems. These systems are typically designed for a low level of impurity (i.e., in parts per million), since a 4.5 K or 2 K helium system will solidify, or freeze o ut , every other substance. The trace impurities can block and/or change the flow distribution in heat exchangers and potentially damage turbines or cryogenic compressors operating at high speed s . Experimental systems, such as superconducting magnets, requi re helium purification due to inherent characteristics in their construction. These are also used for the commissioning of sub - systems, like the compressors, and cold boxes. As known from experience, molecular sieves do not remove low - level moisture impuri ty sufficiently. Typical commercial freeze - out purifiers using molecular sieves have very short operating times between regeneration s and are inefficient , requiring substantial utilities like liquid nitrogen and high - pressure operation. Based upon proven e xperience from a freeze - out purifier design for Brookhaven National Lab (BNL) in 1983, a liquid nitrogen assisted freeze - out purifier has been designed. This design includes a multi - pass and multi - stream heat exchanger and an activated carbon bed. The heat exchanger design is expected to minimize the liquid nitrogen usage and extend the capacity and the operating pressure range, thereby the time interval between regeneration. The goal is to provide a simple , naturally balanced design procedure to develop an d opera te an efficient purifier system. iii ACKNOWLEDGEMENTS First, I would like to express my sincere gratitude to my primary advisor, Dr. Abraham Engeda, for his support of my graduate studies. He initially connected me with FRIB, where I was able to do interesting and meaningful research. He continues to encourag e me to aspire to greater achievements. Dr. Pete Knudsen, Dr. Rao Ganni, and Dr. Nusair Hasan have extensive experience in cryogenics, and I am lucky to be able to draw from their wealth of knowledge on this subject. Their constant support of my project an d availability to me when I have questions has been crucial to my success. Their ability to see the big picture when I had been focusing on the minutia of specific components was very helpful. Their work on the initial process study for this project was wh at allowed me to do this work in the first place. Dr. Hasan worked very closely with me throughout the design and writing of this thesis. I am very grateful for their guidance. I would also like to thank Dr. Rebecca Anthony, along with Dr. Engeda, Dr. Knud sen, and Dr. Hasan for agreeing to be a part of my thesis committee. Further thanks goes out to my coworkers. First, and building this helium purifier with me. I have learned a lot building the purifier and its prot otype with him. Mat Wright assisted in my initial understanding of the existing FRIB purification system. His previous work in helium purification provided a useful guide to base my work on. My fellow graduate students Jon Howard and Tasha Williams were he lpful in reviewing my work and providing a peer support system. Special thank s to Fabio Casagrande (Cryogen ic Department Manager, FRIB), Thomas Glasmacher (Laboratory Director, FRIB) , and the Accelerator Science and Engineering Traineeship program for prov iding me with the opportunity to work at FRIB. iv Finally, I would like to thank my family. My parents, Kevin and Suzanne, and my sister, Sydney , have been very supportive throughout my education and have always pushed me to be the best I can, believing I wi ll achieve it at every stage. This material is based upon work supported by the U.S. Department of Energy Office of Science under Cooperative Agreement DE - SC0000661, the State of Michigan and Michigan State University. Michigan State University designs and establishes FRIB as a DOE Office of Science National User Facility in support of the mission of the Office of Nuclear Physics. v TABLE OF CONTENTS LIST OF TABLES vi LIST OF FIGURES vii i CHAPTER 1: INTRODUCTION 1 1.1 Background 1 1.2 Motivation 2 1.3 Project Description 2 CHAPTER 2: PROCESS DESIGN 3 2.1 Heat Exchanger 4 2.2 Nitrogen Boiler 7 2.3 Carbon Bed 8 CHAPTER 3: MECHANICAL DESIGN 11 3.1 Design Considerations 12 3.2 Mechanical Design of Heat Exchanger 14 3.3 Mechanical Design of Nitrogen Boiler 19 3.4 Mechanical Design of Carbon Bed 21 3.5 Mechanical Design of Process Piping 24 3.6 Mechanical Design of Insulating Vacuum Jacket 25 3.7 Selection of Miscellaneous Components 26 CHAPTER 4: FABRICATION PROCESS 27 4.1 Fabrication Considerations 27 4.2 Fabrication Plan 27 4.3 Fabrication of Prototype 30 CHAPTER 5: Modes of Operation 33 5.1 Modes of Operation 33 5.2 Description of Operating and Maintenance Procedures 34 5.3 Valve Position Matrix 49 CHAPTER 6: SUMMARY AND CONCLUSION 50 APPENDICES 51 APPENDIX A: STRESS ANALYSIS 52 APPENDIX B: PROCESS CALCULATIONS 66 APPENDIX C: MECHANICAL CALCULATIONS 67 BIBLIOGRAPHY 71 vi LIST OF TABLES Table 2.1: Design requirements for the freeze - out purification system 4 Table 2.2: Constants for Equation 2.2 8 Table 3.1: Basic dimensions of the major components of the purifier 12 Table 3.2: Design parameters of the purifier 13 Table A.1: CAEPIPE B31.3 code compliance for nitrogen piping from nitrogen boiler to heat exchanger 53 Table A.2: CAEPIPE B31.3 code compliance for helium piping from supply to the heat exchanger 55 Table A.3: CAEPIPE B31.3 code compliance for helium piping from the heat exchanger to the nitrogen boiler 57 Table A.4: CAEPIPE B31.3 code compliance for helium piping from the nitrogen boiler to the carbon bed 59 Table A.5: CAEPIPE B31.3 code compliance for hel ium piping from the carbon bed to the heat exchanger 61 Table A.6: CAEPIPE B31.3 code compliance for helium piping from the heat exchanger to the recovery system 63 Table A.7 : Analysis of maximum stress results 65 Table B .1: Carbon bed sizing calculations 66 Table C .1: B31.3 piping pressure design 67 Table C .2: BPVC internal pressure design 67 Table C .3: BPVC external pressure design 67 Tabl e C .4: HX - 1 mandrel vertical rod supports 68 Table C .5: Reinforced nozzle opening in carbon bed top head 69 Table C .6: Carbon bed screen supports 70 Table C .7: Component weight 70 vii Table C .8: Component cool - down enthalpy 70 viii LIST OF FIGURES Figure 2.1: Simplified flow diagram of the freeze - out purification system 3 Figure 2.2: Solid - liquid (S - L) saturation temperature of moisture as function of the mole fraction at different stream (operating) pressures [7 ] 6 Figure 2.3: Heat exchanger cooling curves for HX - 1 (left) and HX - 2 (right) 6 Figure 2.4: Effect of stream operating pressure on purifier operating period at maximum design contamination (moisture) 7 Figure 2.5: Adsorption curve for nitrogen on PCB carbon in terms of liquid nitrogen 9 Figure 3.1: Sketches showing (a) complete purifier assembly, (b) nitrogen boiler, (c) freeze - out heat exchanger (without oute rmost shell) and (d) carbon bed 11 Figure 3.2: Example of u - bend to limit thermal stress 1 3 Figure 3.3: Finned tubes being measured on reception 14 Figure 3.4: Cross - section of the heat exchanger 15 Figure 3.5: Rods supporting HX - 1 mandrel for external pressure design 18 Figure 3.6 : Example head analyzed using Ansys 18 Figure 3.7 : FEA stress distribution of heads in heat exchanger 19 Figure 3.8 : Detailed cross - sectional view of the nitrogen boiler assembly 20 Figure 3.9 : Helium tubing coils in the nitrogen boiler 20 Figure 3.10 : Left shows top header, right shows bottom heade r of helium coil 20 Figure 3.11 : Detailed cross - sectional view of the carbon adsorber bed assembly 2 1 Figure 3.12 : Stainless steel screen and associated components 22 Figure 3.13 : 2 3 Figure 3.14 : Band heater model 2 3 Figure 3.15 : View of purifier piping 2 4 ix Figure 3.1 6 : Purifier head and connections 2 5 Figure 4.1: Fabrication plan step showing weld 28 Figure 4.2: Model of two sections of shell installed separately (arrows point to welds for this part) 29 Figure 4.3 : Coiled tubes carefully spaced out on the mandrel 30 Figure 4.4 : Tubes tied together with wires and cinched down with sheets and clamps 31 Figure 4.5 : Heat exchanger with shell installed 32 Figure 5.1: Regular Operation P&I 36 Figure 5.2: Helium Blo w Down P&I 38 Figure 5.3: LN2 Evaporation and Warm Up P&I 40 Figure 5.4: Heating P&I 42 Figure 5.5: Pump P&I 44 Figure 5.6: Backfill P&I 46 Figure 5.7: Cool Down and Purge P&I 48 Figure 5.8 : Valve position matrix 49 Figure A.1: CAEPIPE model of nitrogen piping from nitrogen boiler to heat exchanger 52 Figure A.2: Stress distribution for helium piping from supply to the heat exchanger 54 Figure A.3: Stress distribution for helium piping from the heat exchanger to the nitrogen boiler 56 Figure A.4: Stress distribution for helium piping from the nitrogen boiler to the carbon bed 58 Figure A.5: Stress distribution for helium piping from the carbon bed to the heat exchanger 60 Figure A.6: Stress distribution for helium piping from the heat exchanger to the recovery system 62 Figure A.7 : Stress distribution of heat exchanger headers/rings 64 1 CHAPTER 1: I NTRODUCTION 1.1 Background Helium has a wide range of applications in various scientific, space, medical and process industries. These industries take advantage of the very low (cry ogenic) boiling temperature and chemically inert nature of helium . The known helium reserves are deple ting, and this is reflected in the recent price escalations. Hence, not only from a technical aspect, but an economic one is there a need for helium purification and/or recirculation to minimize wast e. In 2015, more than one - third of the total helium consu mption in US was in the cryo genic refrigeration sector [1]. Cryogenic refrigerators which utilize helium as a refrigerant are necessary for systems using superconducting devices, such as magnetic resonance imaging and particle accelerators. These refrigera tion systems operate at 4.5 K (the just above normal boiling point of helium), down to 1.8 K (which requires helium with vapor pressure of 16 mbar). At these very low temperatures, the presence of any other substances (contaminants) except helium will resu lt in solidification. This can lead to damage to moving parts of the cryogenic system and/or affect the flow distribution in heat exchangers and flow blockage in valves. Obviously, these can have a deleterious effect on the refrigerator capacity an d operat ions. Although usually better than industrial Grade - A (also Grade 4.7) purity helium is used in these refrigerators, contaminants are inadvertently introduced to the system through residuals leftover from a clean - up, air in - leaks to systems operating below 4.5 K, and out - gassing from cooled devices (e.g., magnets). The constituents from the first two are of oxygen, nitrogen and moisture. After the initial clean - up, these constituents are present in relatively low concentrations, of the order of 10 ppm or le ss. Although, this seems small, it can (and does) build up over time and consequently pose threat to the reliable and efficient operation of the equipment. 2 1.2 Motivation From operational experience at Jefferson Lab and the Spallation Neutron Source [2], it w as found that the molecular sieve is unable to remove low level moisture sufficiently, despite reasonable regeneration practices. This was evident from the pressure build - up in the helium - helium - nitrogen heat exchanger used to cool the helium to liquid nit rogen temperatures. To address these issues, several different methods of low level impurity removal [3 - 5] have been investigated in the past, including freeze - out (or refrigeration) purification [4]. For the latter, a heat exchanger specifically designed to accommodate the solidified moisture from a contaminated helium stream is used, rather than molecular sieve. This is a very effective method for removing low level moisture contamination due to the very low saturation vapor pressures. However, it require s a heat exchanger design that is well suited for contaminate solidification distribution and minimal impact on flow distribution. Typical commercially available freeze - out purifiers have a much shorter operating time in between regenerations and are not o ptimized for low pressure operation or efficient LN usage [6] . As such, there is a need for fundamental improvements of this critical sub - system. 1.3 Project Description Th e development of a helium purification system utilizing freeze - out purifier heat exchanger is reported. The purification system is designed to remove low level impurities (mainly air), typically present in systems using superconducting devices at or below 4.5 K. The goal is to provide a simple design procedure to develop an energy/utility efficient helium purifier with a long operating interval between regenerations. This purifier will serve as the primary helium purification system for MSU - FRIB cryogenic r efrigerator and superconducting magnet testing facility. 3 CHAPTER 2: PROCESS DESIGN The h elium purification process in the freeze - out purifier begins with the contaminated helium cooled to approximately 80 K in a counter - flow helium - helium - nitrogen heat ex changer (HX - 1 and HX - 2 in figure 2.1). Any moisture in the contaminated helium stream is solidified on the HX - 1 surface. The contaminated helium is t hen cooled to at or below 80 K in a liquid nitrogen (LN) boiler, after which it flows through an activated carbon bed (also maintained at 80 K) where the remaining contaminants (like oxygen and nitrogen) are removed. Pure helium leaves the carbon bed, and its enthalpy is recovered in the counter flow heat exchanger s (HX - 2, then HX - 1) , exiting near ambient condi tions from HX - 1. Design goals for the freeze - out purificat ion system are listed in table 2.1 . Figure 2.1: Simplified flow diagram of the freeze - out purification system 4 Table 2. 1: Design requirements for the freeze - out purification system Mass flow rate (helium) 30 g/s Operating pressure 6.0 to 16.0 bar (helium) Design pressure 18.0 bar (helium), 5 .0 bar (nitrogen) Design max. pressure drop 0.25 bar (tube side / shell side) Design. max. contamination 30 ppm v water, 30 ppm v nitrogen Minimum time between regenerations 14 days Design LN usage 0.05 m 3 /hr 2.1 Heat exchanger The heat exchanger is a major and critical component of the freeze - out purification system. Its effectiveness plays an important role in the purification capacity and LN consumption of the system. The type of heat exchanger is paramount to achieving the desired design requirements in a cost - effective m anner. For this application, a coiled fin - tube heat exchanger type was selected. They is somewhat similar to those used in the small - scale re frigerators, and also known as Collins heat exchanger s . The model for this heat exchanger was developed following the work reported by Yuksek [7 ] , studied for the Linde 1600 helium refrigerator. This type of heat exchanger is comprised of one or several tubes wrapped fin - to - fin, in a helix around a mandrel, and enclosed by an outer shell. There can be one or multiple passes that are arranged in one or multiple wraps supporting low pressure operation to reduce compressor power. However, these multiple passes increase the heat exchanger mechanical design and fabrication complexity. 5 The contaminated helium flows in the annular space in - between and over the finned - tubes in a locally cross - flow manner (although the heat exchanger is overall in a counter - flow configuration). This design inherently has t he characteristics for high contamination holding capacity with lower impact on the heat exchanger performance , like an increase in pressure drop or a reduction in effectiveness . The purified helium stream flows through the tubes which are wound about a ma ndrel and bounded by the outer shell. For this design six parallel passes of coiled fin - tubes (12.7 mm outside diameter tube, 4.8 mm fin height and 0.5 mm fin thickness) are used. For geometrical compactness and segregation of the trapped contamination (mo isture), the heat exchanger is physically split into two sections (HX - 1 and HX - 2 referring to figure 2. 1). HX - 1 is designed for freeze - out entrapment of the moisture from the contaminated stream. Figure 2. 2 shows the calculated solid - liquid (S - L) saturatio n temperature of moisture at the stated (total) pressure. The S - pressures and polynomial fits to measured saturation temperatures obtained from [8 ] . It is observed that the S - L saturation temperature varies between 200 K and 240 K over the range of operating pressures. As such, HX - 1 is designed to cool the contaminated stream from 300 K to 180 K. In this way, the trapped moisture stays in this section which facilitates the regeneration. An additional coiled fin - tubing is used in this heat exchanger section to recover the refrigeration from the nitrogen vapo r stream (exiting the nitrogen boiler). HX - 2 is designed to cool the contaminated stream from 180 K to 80 K, recovering the exergetically more valuable the refrigeration from the purified helium stream. The calculated cooling curves for both of these heat exchangers are shown in figure 2. 3. From the HX - 1 cooling curve, it can be estimated that approximately 25% of HX - 1 contaminated stream moisture content from the design maximum of 30 ppm to 0.3 ppm ( i.e. 1% 6 of the initial value). The heat transfer surface area corresponding to this length is approximately 6.5 m 2 . Figure 2.2: Solid - liquid (S - L) saturation temperature of moisture as function of the mole fracti on at different stream (operating) pressures [7 ] Figure 2.3: Heat exchanger cooling curves for HX - 1 (left) and HX - 2 (right) 7 Based on a design goal of maximum pressure drop of 0.25 bar and the surface area available to capture the moisture, it is estimated that up to 2.5 kg of moisture can be captured by HX - 1. A parametric study on the effect of the operating pressure on the pur ifier operating time was performed and the results are shown in figure 2. 4. From this figure, it is observed that with a moisture concentration of 30 ppm (at 30 g/s), the operating period of the purifier ( i.e. time before HX - 1 reaches a pressure drop of 0. 25 bar) is about 30 days or longer. Figure 2.4. Effect of stream operating pressure on purifier operating period at maximum design contamination (moisture) 2.2 Nitrogen Boile r The nitrogen boiler is the next major component of the purification system. The design of this component was performed following Wright, et al [9 ] . Based on an estimated LN consumption 8 of 0.05 m 3 /hr., a 0.17 m outside diameter (OD) vessel (approx. 0.05 m 3 vol ume) was selected for the nitrogen boiler. 2.3 Carbon Bed Activated carbon at 80 K is proven to be and effective media for adsorbing oxygen and nitrogen the major species making up the contaminant. The carbon bed was sized based on the vo lume of carbon required. This was determined from the design parameters for mass flow rate of helium, desired break through time, and pressure drop. Two methods were used to find the specific adsorbent capacity of the carbon. The first calculate s the exces s adsorption energy, (in cal/mol) using the following equation: Here, is the specific gas constant ( nitrogen ), is the operating temperature, is the saturation pressure of nitrogen at 80 K , and is the partial pressure of nitrogen. Then equation 2.2 is used. It is derived from figure 2.5 to find the nitrogen adsorbed (in cm 3 liq /100g activated carbon) . Table 2.2 : Constants for Equation 2.2 Constant Value A - 2.1720x10 - 18 B 2.9064x10 - 14 C 1.2939x10 - 10 D 1.3893x10 - 7 E - 2.7232x10 - 4 F 1.6802 Equation 2.2 Equation 2.1 9 Figure 2.5 : Adsorption curve for nitrogen on PCB carbon in terms of liquid nitrogen The volume of carbon was calculated. This was based on a pre - determined bed diameter and a diameter to length ratio. A 12 NPS pipe was chosen for this because it would allow the purifier to be the desired size. A diameter to length ratio of 5 was chosen based on analysis done by Wright, et al. Once the volume was established, the breakthrough time, or time that the carbon takes to come to its adsorption capacity, was calculated using the following equation. V C is the volume of carbon, C is the density of carbon, and m N2 is the maximum mass flow rate of nitrogen. The maximum flow rate of nitrogen was calculated using the maximum nitrogen Equation 2.3 10 contamination of 30 ppm and the planned flow rate of helium of 30 g/s. The result was a break throu gh time of approximately 22 days. This fits the goal of at least 14 days. The pressure drop over the bed was also evaluated. The Ergun Equation (equation 2.3 ) was used. inter - particle diameter. A pressure drop of 0.0708 psi was calculated. This is le ss than previous literature values of 0.1 psi, and below the 3 psi allowable limit. The results of these process calculations are found in Appendix B. Equation 2.4 11 CHAPTER 3: MECHANICAL DESIGN The purifier design has three major pressure vessels a freeze - out heat exchanger, a nitrogen boiler, and an adsorber bed. All three pressure vessels operate at cryogenic temperatures and are enclosed in a vacuum insulating shell. The complete purifier assembly along with its major components are shown in figure 3.1 below. Figure 3.1 : Sketches showing (a) complete purifier assembly, (b) nitrogen boiler, (c) freeze - out heat exchanger (without oute rmost shell) and (d) carbon bed Mechanical design of the purifier piping and pressure vessels were performed following American So ciety of Mechanical Engineers (ASME) B31.3 Code and ASME Boiler and Pressure 12 Vessel Code (BPVC), respectively. Piping flexibility analysis for the cryogenic process piping was performed for design in accordance with the ASME B31.3. Pressure design of diffe rent novel components were carried out using finite element analysis and following ASME B31.3 and BPVC (as applicable). Basic dimensions of the purifier are listed in table 3.1 , and their design details are discussed in the following sub - sections. Table 3 .1 : Basic dimensions of the m ajor components of the purifier Components Outside Diameter ( m ) Shell Thickness ( mm ) Nominal Length (m) Insulating vacuum shell 0.91 7.92 3.05 HX - 1 Mandrel 0.27 6.35 2.15 HX - 1 Shell 0.33 3.97 2.11 HX - 2 Mandrel 0.36 7.92 1.83 HX - 2 Shell 0.41 4.78 1.95 Nitrogen Boiler Shell 0.17 3.40 2.54 Carbon Bed Shell 0.33 4.57 2.03 3.1 Design Considerations The materials used in the purifier were chosen based on their specific uses. ASTM A312 TP304L stainless steel was used for all the piping and vessels. This was chosen because it is an ind ustry standard that will economically meet the requirements. ASTM SB75 C122 Copper was used for the finned tubes. The fins are ASTM SB75 C102 Copper. Copper conducts heat very well, allowing for very good heat tr ansfer between process streams. 13 Table 3.2: Design p arameters of the p urifier Desi gn Pressure 18.0 bar (helium), 11.0 bar (nitrogen) Design Temperature Range 80 K 300 K Thermal contraction and expansion was considered in this design. The thermal expansion coefficient of stainless steel (from 300 K to 80 K) is 17.3 x10 - 6 m/m/K. There are several points w h ere it could have caused stress . Thermal stress calculations were done. The only areas that failed stress tests were in the piping outside t he process vessels. Piping loop were incorporated into the piping to reduce heat leak and stresses due to thermal contraction (at cryogenic temperatures), as well as for thermal stability . An example of this is shown in figure 3.2 . The stress analysis of piping sections is shown in Appendix A. Figure 3.2 : Example of u - bend to limit t hermal stress 3.2 Mechanical Design of Heat Exchanger The heat exchanger is a major and critical component of the freeze - out purification system. Its effectiveness plays an important role in the purification capacity and LN consumption of the 14 system. The type of heat exchanger is paramount to achieving the desired design requirements in a cost - effective manner. For this application, the coiled fin - tube heat exchanger type was selected, which is somewhat similar to those used in the small - scale refrigerators, also known as a Collins heat exchanger. The finned tubes allow for a very large heat transfer surface area, while keeping the volume low. Figure 3.3 is a picture of the finned tubing used in this purifier. Figure 3.3 : Finned tubes being m easured on reception The model for this heat exchanger was developed following the work reported by Yuksek [ 7 ] , studied for the Linde 1600 helium refrigerat or. This type of heat exchanger is comprised of one or several tubes wrapped fin - to - fin, in a helix around a mandrel, and enclosed by an outer Multiple passes allow for higher volume (mass) flow, at a lower pressure drop and thus supporting low pressure operation to reduce compressor power. However, these multiple passes increase the heat exchanger mechanical design and fabrication complexity. The heat exchange r is shown in figure 3.4 below. 15 Figure 3.4 : Cross - section of the heat exchanger The contaminated helium flows in the annular space in - between and over the finned - tubes in a locally cross - flow manner. The heat exchanger is overall in a counter - flow confi guration, as the flow globally goes up on one side and down on the other. This design inherently has the characteristics for high contamination holding capacity with lower impact on the heat exchanger performance. For geometrical compactness and segregatio n of the trapped contamination (moisture), the heat exchanger is physically split into two sections (HX - 1 and HX - 2 referring to figure 2 . 1). HX - 1 is designed for freeze - out entrapment of the moisture from the contaminated stream. The purified helium stream flows through the tubes which are wound about a mandrel and bounded by the outer shell. For this design six parallel passes of coiled fin - tubes are used for HX - 2, while seven are used for HX - 1. This amounts to six helium passes in HX - 2, with HX - 1 having f ive helium passes and two nitrogen passes. Annular flat heads at the ends of both heat exchangers serve as headers for tube and shell flows. At the top, the HX - 1 tubes go through the heads and into a mixing chamber. The helium then goes back through the h ead into the HX - 2 tubes. At the bottom the tubes go through the head, into the mixing chamber and exit through two inch pipes, or the reverse. The fin side flow comes in through two inch pipes which opens up to the fins. The helium flows up over the fins, then over 16 a ring connecting the mandrel of HX - 1 and the shell of HX - 2. It then goes out the same way it came in. The mechanical design of the heat exchanger was done using ASME code standards, including B31.3 and BPVC. Internal pressure of shells was calculated using BPVC Section VIII - 2, from the following equation: where P is design pressure, S is allowable stress for the material, E is the quality factor, t is vessel thickness, and R is inside radius. The results of t his and all following calculations can be found in Appendix C. External pressure of the shells was calculated using BPVC Section VIII - 2 as well. It uses the length to diameter and diameter to thickness ratios on the charts in appendix to fi nd the B value used in following equation: where P a is the maximum pressure the vessel can withstand, B is an intermediate factor based on vessel size, and D o is the outer diameter. For the HX - 1 mandrel, the external pressure required a thickness greater than desired for geometrical fit. Stiffening rings were considered to solve this. The allowable pressure when using stiffening rings was calculated using the same method as above, adjusting the effective length for the number of stiffen culated using the BPVC equation : Equation 3 .1 Equation 3.2 17 where I s is the moment of inertia, L s is the effective length, A s is the cross - sectional area of the stiffening ring s, and A is an intermediate factor based on vessel size. This was compared with the available moment of inertia for the ring cross - section. From this comparison, the necessary size of stiffening ring was chosen. Stiffening rings were found to be too bulky to fit in the small gap between the two heat exchangers, so another solution was pursued. This was using vertical tube supports at various points circumferentially between the shells. The calculation was done fol lowing for Stress and Strain , by Young and Budynas [10 ] . Table 15.2: Formulas for elastic stability of plates and shells details the calculation as shown below: w here r is the outer radius of the shell, n is the number of supports, l is the length of the shell, and v is the number of supports. Four supports were found to be necessary to support the external pressure acting on the shell. A segment of the shells showing t his design solution is in figure 3.5 . The two shells in the figure are the HX - 2 outer shell and the HX - 1, with the rods in the vacuum space between the two heat exchangers. Equation 3.3 Equation 3.4 18 Figure 3.5 : Rods supporting HX - 1 mandrel for external pressure design All of the heads (an example of wh ich is shown below in figure 3.6 ) were analyzed using finite element analysis ( FEA) in Ansys Workbench and compared to BPVC standards. These heads include boundaries between process streams, pressure bounda ries, and structural supports. Figure 3. 6 : Example head analyzed using Ansys The boundary conditions used were test pressure, vacuum side pressure (17 psia), and supports (in this case fixed and cylindrical). The reported values were equivalent von - mises stress, 19 as requested in BPVC. These values were compared to the maximum allowa ble stress values given in ASME BPVC Section II for the material being used. The results of the analyses come in the form of a maximum stress and a stress distribution, the latter of which is shown below in figure 3.7 for all four types of heads in the hea t exchanger. Figure 3.7 : FEA stress distribution of heads in heat exchanger 3.3 Mechanical Design of Nitrogen Boiler The nitrogen boiler consists of six parallel passes of stainless - steel tubing coiled inside a vessel, as shown in figure 3. 8 . Contaminated helium from the freeze - out heat exchanger (HX - 2) outlet flows through the coiled tubing submerged in the liquid nitrogen and is then fed to the 20 adsorber bed. The nitrogen boiler is nested inside the annular vacuum space of HX - 2 for compactness and minimizing radiation heat in - leak to the liquid nitrogen bath. Figure 3.8 : Detailed cross - sectional view of the nitrogen boiler assembly The coil consists of six tubes. This coil is shown in figure 3.9. They begin by coming out of the bottom of t he pipe that comes from the top. The flow recombines when the tubes go through a head at the bottom , as seen in figure 3.10 . Figure 3.9 : Helium tubing coils in the nitrogen boiler Figure 3.10 : Left shows top header, right shows bottom header of heliu m coil 21 The boiler was much simpler to design than the heat exchanger. There was one pressure vessel, designed using the equations above for internal and external pressure at 5 bar . The heads were designed using Ansys Workbench, the same way the heat exchanger heads were desig ned . 3.4 Mechanical Design of Carbon Bed The carbon bed is comprised of two pressure vessels, one nested inside the other. The outer vessel holds the adsorbent (activated carbon) in a fixed bed, while the inner vessel i s mounted at the center of the fixed bed, supported by the inlet and outlet nitrogen piping. Liquid nitrogen flows through the inner vessel keeping the adsorbent at a constant temperature (approximately 80 K). The adsorbent is held in place within the fixe d bed using layers of wire mesh screens and fiber - glass filter. In addition, sintered metal filters are used at the inlet and outlet nozzles to the adsorber bed to prevent any carry - over dust from the exiting pure helium. Band heaters are mounted to the ou ter vessel shell for the regeneration process. A detailed cross - sectional view of the adsorber bed assembly is shown in figure 3.11 . Figure 3.11 : Detailed cross - sectional view of the carbon adsorber bed assembly The two pressure vessels were designed as previously discussed. Wire mesh screens have been designed , with beam supports, to hold the we ight of the carbon in the bed. Equation 3.5 shows 22 t h e calculation for designing the beam supports , which was done following Stress and Strai n , table 15.1: w E is the modulus of elasticity, G is the shear modulus, l is the length of the beam, and a is half of the vertical depth of the beam. Stainless steel screens , as shown in figure 3.12 , were used to preliminarily contain the carbon and keep it packed. Affixed to the top screen is a pipe section with another screen on top. The purpose of this screen is to divert some of the flow to the outside of the bed for an even flow distribution , so it uses the full radius of the carbon bed to adsorb impurities. This prevents the need for premature regeneration (before all the carbon i s saturated) because the helium is only flowing over the center of the bed, saturating only the carbon in the center. Figure 3.12 : Stainless steel screen and associated components Equation 3.5 23 filters were used to make sure no carbon gets in the process stream outside of the adsorber bed. A model of the one used is shown below in f igure 3.13 . Figure 3.13 : filter assembly Band heaters , as shown in figure 3.14 , were wrapped around the bed to heat it above ambient temperature during regeneration. Six hea ters were used, spaced 12.5 inches apart to assure equal heating through the bed. Figure 3.14 : Band heater model 24 3.5 Mechanical Design of Process Piping Piping design was done using ASME B31.3 Code. The following equation was used to calculate the necessary pipe thickness: every pipe in the purifier, including helium and nitrogen lines, and the nitrogen tank in the carbon bed. A portion of the purifier piping is shown in figure 3 .15 . Figure 3.15 : View of purifier piping Equation 3.6 25 Flexibility analysis was done on the piping in Ansys . Thermal stresses were taken into account when this was done. The results of these calculations are shown in Appendix A. 3.6 Mechanical Design of Insulating V acuum Jacket The vacuum jacket for the purifier is a 36 NPS pipe with a standard ASME dished head. The vacuum shell was designed as a pressure vessel. All components inside the shell are mounted from this head. The insulating vacuum shell is attached to th e head using a flanged connection, allowing access to the inner cryogenic components without cutting the vacuum shell. Cryogenic valves, instrumentation and maintenance ports are mounted to the top and side of the dished he ad. Figure 3.16 is a view of the head with all the connections. Fig ure 3.16 : Purifier head and connections 26 3.7 Selection of Miscellaneous Components All cryogenic components are wrapped with multi - layer insulation (MLI) to minimize radiation heat in - leak to the process. In addition, there is an external valve and instrume nta tion panel. All of the vales in the purifier are controlled from this rack. The design of this valve rack was optimized for ease of use. The valves are organized into rows and grouped by their main functions. For example, when in normal operation, only the top row of valves are open. This same pattern is seen in the instrumentation p anel. This is further detailed in section 5.1: Modes of Operation. A recirculation blower and an evaporator will be used for warm up of the purifier from the 300 K end circulating helium in the tube side, nitrogen boiler, and carbon bed. This circulation i s further explained in section 5.1: Modes of Operation . Band heaters are mounted on the outside of the carbon bed for further warming above the ambient temperature . The nitrogen vessel in the carbon bed will assist the cool down following regeneration. 27 CHAPTER 4: FABRICATION PROCESS 4.1 Fabrication Considerations There were many considerations taken into account when planning the fabrication of this purifier. The largest consideration was the small spaces available for welding. The order in which parts were assembled was cr ucial in solving these problems. There were some welds that were initially planned to be done near more heat - sensitive parts (namely the copper fins on the tubes). The method of assembly was changed to avoid welding near these parts. ASME welding standards disallowed some of the initially intended welds, because of the proximity of the welds to thin parts or other welds. The design of those parts was changed as necessary to allow for ASME - approved welds. 4.2 Fabrication Plan A cross - sectional sketch of the heat exchanger was notated with the required welds in order of fabrication to show its viability. A sketch was drawn for each of the initial designs. The design that allowed for the easiest fabrication was chosen. A detailed, step - by - step fabrication plan was created from this sketch, showing a picture from the model of each of the 26 steps, including welds. This plan significantly helped the design process. It made some issues with the design more apparent. Some welds, like th e one sh own in figure 4.1 , are very difficult to complete , or cause other problems [1 1 ] . The problem with this particular step is that the heat - effected zone of the weld between the two sections of shell includes the ropes underneath. This would cause the ropes to burn. The solution to this was to keep the shell in one piece, despite the larger amount of friction caused by the longer shell. The fabrication plan was then changed accordingly. 28 Figure 4.1 : Fabrication plan step showing weld Anot her area that needed to be reconsidered was the connection between the finned tubes and the rings that combine the flow into a mixing chamber. The copper finned tubes needed to be brazed to stainless steel tubes, which were then welded to the ring. It was difficult to guarantee the position of the ends of the finned tubes, so the stainless steel tubes needed to be field - fitted to assure a good fit. The space between the end of the copper tubes and the holes in the ring were Step 20 29 measured. Then, stainless steel t ubes were cut and bent into shape. The stainless steel tubes were installed by pulling the copper tubes away from the mandrel and brazing them to the steel tubes, then feeding the steel end through the hole in the ring. The weld is the last step. This proc ess underwent several iterations, including welding the ring to the mandrel at different times in the process, before this order was decided on. Some small sections of shell need to be placed in between two rings that need (by ASME code) to be as wide as the shell. This means that the shell cannot be slid over the rings. Therefore, either one ring has to be installed after the section of shell, or the shell needs to be decided on to allow for easier fabrication of other parts of the heat exchanger. The bottom sections of shell, surrounding the mixing chambers between the rings, were chosen to be installed in two pieces, welded together in place. A model of this is shown in figure 4.2. Figure 4.2: Model of two sections of shell installed separately (arrows point to welds for this part) The rest of the purifier, including the carbon bed and nitrogen boiler, will be fabricated according to the design laid out in Wright, e t al [9] . The changes made to that design are not significant enough to change the fabrication process. 30 4.3 Fabrication of Prototype A prototype of HX - 2 was built and tested to find and solve the issues that ca me up with the fabrication. This was done duri ng the design process, so many of the parts were not their final size. Th e largest difficulty encountered during the fabrication of the prototype was tightening the coils of finned tubes together and to the mandrel. The coils needed to be tightened so ther e was no space between them. This was to prevent any gas bypassing the fins. They had significant friction between each other because of the fins, making this difficult. The coils had to be pulled apart slightly in order for them to be moved. This spreadin g of the some of the coils is shown in figure 4.3 . The first three coiled tubes are spread apart, so the next three can be placed in the gap. Here, small metal sheets are being used to keep the fins from interlocking and stopping the movement of the tubes. In the figure, the second three tubes are partially screwed into place. Figure 4.3 : Coiled tubes carefully spaced out on the mandrel The coils were kept together by twisting steel wires around pairs of them at many points around the circumference , as shown in figure 4.4 . Once they were tight and tied together, they were unlikely to move out of place because of the friction the fins provide. 31 Figure 4.4 : Tubes tied together with wires and cinched down with sheets and clamps Another significant difficulty was sliding the shell over the finned tubes once they were in place. There is minimal clearance between the mandrel, tubes, and shell. Again, this is to prevent bypass. The tubes were tightened to the mandrel using thin metal sheets and hose cla mps, as shown in figure 4. 4 , above. This was done along the length of the heat exchanger. The initial method for sliding the shell on was attaching flanges to the top end of the mandrel (the shell was being slid on from the bottom) and the top end of the shell. All - thread rods were used as bolts. The nuts on the flange on the shell were tightened on two sides simultaneously to pull the shell over the tubes. This process was done a few threads at a time, stopping to make sure the ropes were not getting caug ht and were staying in place under the shell. The metal sheets were left on the tubes as long as possible to keep them tight to the mandrel. The sheets were 32 removed, one by one, as the flange approached them. The heat exchanger after the shell was installe d is shown in figure 4.5 . Figure 4.5 : Heat exchanger with shell installed As the shell was pushed on, some of the fins stuck out further than the inside diameter of the shell, so they were bent so they would fit inside. This is clearly not ideal. The d esign was since changed, including larger ropes and a slightly larger shell diameter. This allows more clearance between the fins and the shell , while still blocking bypass . The entire apparatus was mounted equipment that allows it to be rotated. This allo ws much easier access to the fins and ropes during fabrication. 33 CHAPTER 5: O PERATIONAL AND MAINTENANCE PROCEDURES 5.1 Modes of Operation The purifier has several modes of operation for regenerating the adsorbent and heat exchanger. This must be done to avoid surpassing the capacity of these components. A detailed o perational scheme is laid out here . This includes P&Is an d valve settings for each mode. The first mode is Regular Operation. This is the mode that the purifier will be in most of the time, wh ile it is purifying helium. The second mode is Helium Blow Down. This step begins the regeneration process. The helium inlet and outlet valves are closed, and a vent is opened to release the potentially dirty helium in the system. The pressure is reduced to approximately 20 psig. This leaves the system at positive pressure, helping with warm - up. The third mode is LN2 Evaporation and Warm Up . The fin side of the heat exchanger is isolated . The liq uid nitrogen inlet is turned off . Moderate heat is added to the carbon bed, warming it up to approximately 200 K. This boils off the LN2. The rest of the helium in blown down, reducing the system pressure to approximately 1 - 2 psig. The fourth mode is Heating. Heaters are turned on around the ca rbon bed to heat it up further, to approximately 350 K . Helium is circulated at approximately 0.5 g/s and warmed to approximately ambient temperature by opening it to a vaporizer and a blower circuit. 34 The fifth mode is Pump . The helium circulation and heaters are stopped. T he shell side of the heat exchanger and the carbon bed are separately pumped down to a vacuum, removing any moisture left in the heat exchanger and regenerating the carbon . The sixth mode is Backfill . The shell side of the heat exchanger and the rest of the purifier are separately backfilled with helium to approximately 1 - 2 psig . This is the first mode of operation when starting up a new purifier, or one that has been unused for some time. Repeat modes 5 an d 6 in sequence 3 times. The seventh mode is Cool Down and Purge . The purifier circulates helium as if in normal operation, except that it is fed with clean helium. The liquid nitrogen inlet is turned back on. This continues until desired temperatures are reached, then the purifier can be put into regular operation mode. 5.2 Description of Operating and Maintenance Procedure s This section will give a detailed description of purifier operation and maintenance , including P&I s showing flow paths and valve settings. A green - highlighted valve is open (or in operation, for control valves), while a red - highlighted valve is closed. The green streams are nitrogen. The blue streams are helium that has not yet or being purified . The red stream s are clean helium , after purification. 35 5.2.1 Regular Operation a) Close the cooldown valves (MV75171 and MV75172) b) Verify the blower circuit is closed (MV75154 and MV75156) c) Verify the helium vent is closed (MV75153) d) Open the helium inlet and outlet valves (MV751 11 and MV75119) 36 Figure 5.1: Regular Operation P&I 37 5.2.2 Helium Blow Down a) Make sure the compressor discharge is aligned to the other purifier. b) Shut the helium inlet valve (MV75111) c) Shut the helium outlet valve (MV 75119) d) Open the blow down valve (MV75153) e) Verify LN2 is on (PV75132) f) Verify the inlet to the blower is closed (MV75154) 38 Figure 5.2: Helium Blow Down P&I 39 5.2.3 LN2 Evaporation and Warm Up a) Isolate the fin side of the heat exchanger, close PV75113 b) Close the LN2 inlet control valve (PV75132) c) Open the nitrogen boiler vent (MV75135) d) Close the heat exchanger nitrogen vent (MV75134) e) Slowly open the GN2 as needed to evaporate LN2 (MV75136) f) Begin blower cycle a. Open MV75154 and MV75156 b. Turn on blower c. Turn on heater so the carbon bed reaches 200 K 40 Figure 5.3: LN2 Evaporation and Warm Up P&I 41 5.2.4 Heating a) Verify that the GN2 inlet is closed (MV75136) b) Turn heater up so the carbon bed reaches 350 K 42 Figure 5.4: Heating P&I 43 5.2.5 Pump a) Turn off the heat, and blower b) Close MV75153, MV75154, and MV75156 c) Close backfill valves if open (MV75151 and MV75114) d) Verify that all valves that need to be closed are, so there is no unwanted gas in the system a. Especially MV75113 (it separates the fin side of the heat exchanger) e) Verify that the cold trap is clean and in place f) Verify that the vac u um pump(s) is attached to both port s properly g) Slowly open the vacuum valve s (MV75151 and MV75114 ) a. Close when pressure stops reducing 44 Figure 5.5: Pump P&I 45 5.2.6 Backfill a) Close the vacuum valves (MV75151 and MV75114) b) Slowly open the purifier backfill valve s (MV75152 and MV75171 ) a. Close when pressure gets to the designated pressure (1 - 2 psig) c) Repeat steps Pump ( 5.2.5 ) and Backfill ( 5.2.6 ) , in sequence, three times or until the baseline pressure (pressure after pumping) stops reducing between repetitions 46 Figu re 5.6: Backfill P&I 47 5.2.7 Cooldown a) Close heat exchanger shell side backfill valve if open (MV75152) b) Open the heat exchanger nitrogen vent (MV75134) c) Close the nitrogen boiler vent (MV75135) d) Turn on the LN2 inlet control valve (PV75132) e) Open the heat exchanger shell side to the rest of the purifier (MV75113) f) Open the cooldown return valve (MV75172) g) Open the cooldown supply valve (MV75171) h) When all the temperatures are where they need to be, transition into purification mode 48 Figure 5.7: Cool Down and Purge P&I 49 5.3 Valve Position Matrix The matrix in figure 5.8 shows the physical positions of the valves as they will be on the valve rack. It also shows the progression of the valves during operation and regeneration. Figure 5.8 : Valve position matrix 50 CHAPTER 6: SUMM ARY AND CONCLUSION The design of a helium purification system utilizing a freeze - out heat exchanger for application in systems requiring helium refrigeration is reported. The purification system is designed to remove low level air impurities . This is done using freeze - out purification to remove moisture and adsorption purification to remove air. Key features of the process design, mechanical design , fabrication, and operation procedures are discussed is this paper. The most critical tasks of the design were the pressure design (especially of shells) and physical design for facilitating simple fabrication. The fabrication of a prototype of the heat exchanger greatly assisted this. It showed the limitations of the design and solutions were found to overcome th ose limitations, namely larger clearance between the finned tubes and the shell, larger ropes, and a rotatable fixture for holding the apparatus. The process design included heat exchanger design, and component selection and sizing. The mechanical design i ncluded design and stress analysis of vessels, heads, and piping. Detailed analysis of the purification system demonstrates an effective and efficient design for supporting the 6 - 16 bar operation, with operating period of at least 2 2 days at a design conta mination level of 30 ppm in 30 g/s of helium and an LN consumption of approx. 0.05 m 3 /hr. at full capacity. Thi s design and analysis has shown that this purifier can be a good tool to serve as the primary helium purification system for MSU - FRIB cryogenic r efrigerator and superconducting magnet test facility. 51 APPENDI CES 52 A PPENDIX A : STRESS ANALYSIS Figure A. 1 : CAEPIPE model of nitrogen piping from nitrogen boiler to heat exchanger 53 Table A.1: CAEPIPE B31.3 code compliance for nitrogen piping from ni trogen boiler to heat exchanger 54 Figure A.2 : Stress distribution for helium piping from supply to the heat exchanger 55 Table A.2 : CAEPIPE B31.3 code compliance for helium piping from supply to the heat exchanger 56 Figure A.3 : Stress distribution for helium piping from the heat exchanger to the nitrogen boiler 57 Table A.3 : CAEPIPE B31.3 code compliance for helium piping from the h eat exchanger to the nitrogen boiler 58 Figure A.4 : Stress distribution for helium piping from the nitrogen boiler to the carbon bed 59 Table A.4 : CAEPIPE B31.3 code compliance for helium piping from the nitrogen boiler to the carbon bed 60 Figure A.5 : Stress distribution for helium piping from the carbon bed to the heat exchanger 61 Table A.5 : CAEPIPE B31.3 code compliance for helium piping from the carbon bed to the heat exchanger 62 Figure A.6 : Stress distribution for helium piping from the heat exchanger to the recovery system 63 Table A.6 : CAEPIPE B31.3 code compliance for helium piping from the heat exchanger to the recovery system 64 Figure A.7 : Stress distribution of heat exchanger heade rs/rings 65 Table A.7 : Analysis of maximum stress results 66 APPENDIX B: PROCESS CALCULATIONS Table B .1 : Carbon b ed sizing calculations 67 APPENDIX C: MECHANICAL CALCULATIONS Table C .1 : B31.3 piping pressure d esign Table C .2 : BPVC i nternal p ressure d esign Table C .3 : BPVC e xternal p ressure d esign 68 Table C .4: HX - 1 m andrel v ertical r od s upports 69 Table C .5: Reinforced n ozzle o pening in c arbon b ed t op h ead 70 Table C .6: Carbon b ed s creen s upports Table C . 7 : Component w eight Table C . 8 : Component cool - d own e nthalpy 71 BIBLIOGRAPHY 72 BIBLIOGRAPHY [1] Hamak J E 2015 Minerals Handbook. United States Geological Survey) p 9 [2] Casagrande F, Campisi I, Gurd P, Hatfield D, Howell M, Stout D, Strong H, Arenius D, Creel J, Dixon K, Ganni V and Knudsen P 2005 Status of the Cryogenic System Commissioning at SNS. In: Proceedings of the 2005 Particle Accelerator Conference, pp 970 - 2 [3] Thomas E R and Denton R D 1988 Conceptual studies for CO2/natural gas separation using the controlled freeze zone (CFZ) process Gas Separation & Purification 2 84 - 9 [4] Dauvergne J P, Delikaris D, Haug F and Knoops S 1994 A helium freeze - out cleaner operating at atmospheric pressure Cryogenics 34 135 - 8 [5] Scholes C A and Ghosh U K 2017 Review of Membranes for Helium Separation and Purification Membranes 7 9 [6] Perso nal Conversation with V. Ganni, Director for MSU Cryogenics Initiative [7 ] Yuksek E 2009 Capacity Improvement of a Small - Scale Cryogenic System Using Exergy Analysis. In: Mechanical Engineering : Old Dominion University) p 79 [8 ] Flatau P J, Walko R L and Cotton W R 1992 Polynomial Fits to Saturation Vapor Pressure Journal of Applied Meteorology 31 1507 - 13 [9 ] Wright M 2009 Design and Development of a Helium Purifier. In: Mechanical Engineering, (Virginia, USA: Old Dominion University) p 87 [10 ] Young, W arren C, and Richard G Budynas. Roark's Formulas for Stress and Strain . 7th ed., McGraw - Hill, 2002. [11] Personal Conversation with M. Wright, Cryogenic Engineer, FRIB