OPTIMAL DESIGN AND OPERATION OF HELIUM REFRIGERATION SYSTEMS USING THE GANNI CYCLE

OPTIMAL DESIGN AND OPERATION OF HELIUM REFRIGERATION SYSTEMS USING THE GANNI CYCLE V. Ganni, P. Knudsen Thomas Jefferson National Accelerator Facility (TJNAF) 12000 Jefferson Ave. Newort News, VA 23606 USA ABSTRACT The constant ressure ratio rocess, as imlemented in the floating ressure - Ganni cycle, is a new variation to rior cryogenic refrigeration and liquefaction cycle designs that allows for otimal oeration and design of helium refrigeration systems. This cycle is based uon the traditional equiment used for helium refrigeration system designs, i.e., constant volume dislacement comression and critical flow exansion devices. It takes advantage of the fact that for a given load, the exander sets the comressor discharge ressure and the comressor sets its own suction ressure. This cycle not only rovides an essentially constant system Carnot efficiency over a wide load range, but invalidates the traditional hilosohy that the ( TS ) design condition is the otimal oerating condition for a given load using the as-built hardware. As such, the Floating Pressure- Ganni Cycle is a solution to reduce the energy consumtion while increasing the reliability, flexibility and stability of these systems over a wide oerating range and different oerating modes and is alicable to most of the existing lants. This aer exlains the basic theory behind this cycle oeration and contrasts it to the traditional oerational hilosohies resently used. KEYWORDS: helium, cycles, refrigerator, screw comressor, efficiency INTRODUCTION Traditional cryogenic helium refrigeration and liquefaction rocess cycles are designed at secified maximum caacity oerating oint(s). In ractice however the actual refrigeration and/or liquefaction loads often vary. In addition the comonents used in the system do not always erform exactly as envisioned in the cycle design cases, which are traditionally reresented by the TS design diagrams. As such, for design and off-design modes, it has been traditionally the ractice to force the lant to oerate at the design ressure and temerature levels established in the cycle design (referred to as the TS design conditions) by regulating the turbo exander inlet valves, thereby (resumably) keeing the sub-comonents close to their eak (design) efficiencies. Common methods

for lant caacity reduction are the use of ressure-throttling valves, adding a load using heaters and/or byassing the cold and/or warm helium gas. These methods in themselves introduce inefficiencies, resumably to maintain the TS design condition or close to it. So, for traditional rocess designs, the actual oerating utility requirements (electric ower, liquid nitrogen and cooling water requirements) er unit load (of refrigeration and/or liquefaction) significantly increases at reduced loads. Although these mechanisms reduce lant roduction, they have only a limited effect on reducing the required utilities to maintain high lant efficiency. These traditional methods are analogous to driving a car with a fully deressed gas edal while controlling the actual seed with a brake. Thus, the underlying assumtion for traditional rocess designs is that the TS design condition is considered the otimum oerating condition for the actual equiment and actual loads. The Floating Pressure Process Ganni cycle has no such bias and instead adots a non-interference control hilosohy using only a few key rocess arameters. It assumes that most controls are for rotecting the equiment; e.g., reventing exander temeratures from getting too cold (to maintain the required bearing caacity etc.), or reventing the 1 st stage comressor suction from going sub-atmosheric or some minimum ressure to ensure oil removal effectiveness etc. However, this can be easier said than done, since it is not uncommon for equiment manufacturers to rovide very narrow oerating limits to rotect the equiment from unknown or non-otimum conditions rather than truly rotecting the equiment from damage. Also, the Floating Pressure Process Ganni cycle only utilizes key rocess arameters that are the indeendent system rocess variables. This is contrary to many traditional rocess cycles that attemt to maniulate a sizeable number of variables resumably for rocess otimization and equiment rotection. In this aer, the authors have attemted to demonstrate that the Floating Pressure Process Ganni cycle invalidates the traditional hilosohy that the TS design condition is the otimal oerating condition for as-built hardware and actual loads. BASIC FLOATING PRESSURE CYCLE Consider the basic system consisting of one comressor working with a cold box containing a heat exchanger and a turbo exander as shown in FIGURE 1. This is a simlified arrangement for a tyical gas (shield) refrigerator used in many alications (e.g., 20-K systems). The control scheme for the Floating Pressure Process shown in FIGURE 1 oerates as follows: (a) Comressor byass (BYP) will resond to revent the comressor suction ( l,1 ) from going below the set (minimum) ressure (usually ~1.05 atm.). Basically, the sole function of the comressor byass is to revent the comressor suction from becoming sub-atmosheric. (b) Mass-in valve (MI) will resond by oening, charging the system with gas from the gas (or liquid) storage, if the comressor discharge ressure ( h,1 ) falls below the set oint.

(c) Mass-out valve (MO) will resond by oening, discharging the system by sending gas out to gas storage, if the comressor discharge ressure ( h,1 ) rises above the set oint. (d) There is a fixed offset in set oints between the MI and MO valves (say ~0.2 to 0.3 atm.); with the MI valve set oint lower than the MO valve set oint. This offset is imortant for adjusting the charge in the system for a given load. (e) The comressor discharge set oint is a function of the load. For this alication, the discharge ressure set oint is the outut of a control loo that looks at the shield return temerature for its rocess variable. (f) During steady state oeration, the MO, MI and BYP valves are ALL CLOSED. FIGURE 1. General Arrangement for Floating Pressure Process Cycle (atent ending) The availability to the cold box (and to the load) is set by the comressor system and is roortional to both the system mass flow and the logarithm of the (high to low) ressure ratio. For the Floating Pressure Process, the gas charge (i.e., system gas mass in the cycle) is maniulated by the MO and MI valves. Since the exander and comressor are essentially constant volume flow devices, they each set their own resective inlet ressures for a given mass flow rate at their oerating temeratures. Since both devices have the same mass flow rate, this characteristic establishes a ressure ratio that is essentially invariant to the mass flow. So, for a given system gas mass charge, the discharge ressure is set by the exander flow coefficient and the suction ressure is set by the comressor dislacement. Although, in theory, either the discharge or suction ressure signals could be used as the key rocess variable used (to adjust) to match a given load, in ractice, using the discharge ressure rovides larger signal and thus leads to a very stable system. The (essentially) constant ressure ratio maintains a (nearly)

constant enthaly dro across the exander (assuming constant efficiency) which results in an (aroximately) invariant mass secific load enthaly difference. Since isothermal comressor efficiency is rimarily deendent on the ressure ratio [1], with an essentially constant ressure ratio, the mass secific comressor inut ower is nearly constant. Further, as the system temeratures vary a icayune amount (even under varying load conditions), the mass secific load exergy is nearly invariant. This results in an essentially constant system Carnot efficiency over a very wide load range. The following illustrates this mathematically. TABLE 1. Symbols and Subscrits Symbols: Symbols: C Secific heat at constant [J/g-K] φ = ( γ 1/ ) γ [non-dim.] ressure h Enthaly (mass secific) [J/g] γ Ratio of secific heats [non-dim.] Δ h Enthaly difference [J/g] η Efficiency [non-dim.] m& Mass flow rate [g/s] ρ Density [kg/m 3 ] or [l/s] N Units conversion constant [kpa/atm] τ 0 (=101.325) = Δ Thl / Tl [non-dim.] Ntu HX number of transfer [non-dim.] ξ Actual to design mass [non-dim.] units flow ratio Pressure [atm] r Pressure ratio [non-dim.] Subscrits: Δ Pressure difference [atm] C comressor q Heat transfer [W] D Design (or observed) Q Volumetric flow [l/s] h High ressure stream s Entroy (mass secific) [J/g-K] hl Difference between h & l streams Δ s Entroy difference [J/g-K] i isothermal T Temerature [K] l Low ressure stream Δ T Temerature difference [K] L (shield) load ( UA ) HX thermal rating [W/K] m motor W & Power [W] r ratio w Secific work [W/(g/s)] v volumetric Ε Exergy [W] x exander ε Physical exergy (mass [J/g] 0 Reference (state) secific) κ Flow coefficient (units defined by equation) 1,2,3 Temerature level From the secified design load conditions, the comressor dislacement caacity, exander flow coefficient and heat exchanger (HX) size are determined, considering the comonent limitations and best oerating ranges for overall otimum efficiency [2, 3]. Now, assuming all the flow is going to the load (i.e., there is no byass etc.), m& = m& x L

Most of the helium refrigeration/liquefaction cycles are constructed with constant volume dislacement comressors (e.g., screw, recirocating comressors) and centrifugal turbo exanders. Also, assuming that there is no comressor byass, m& = m& C x With, m& C = ηv QC ρl,1 (1) N0 l,1 and, ρ = l,1 φ C Tl,1 Since the exander is essentially a constant volume flow device, the flow through the exander is (essentially) choked (or critical) flow, h,2 m& x = κ x (for rx, 2) (2) Th,2 For this analysis, we will assume C to be constant and the same for both high and low ressure streams. Equating the comressor and exander mass flows, we have a characteristic ressure ratio that is essentially constant. h,2 ηv Q C 1 Th,2 r = Constant (3) l,1 κ x φ C Tl,1 Now, examining T h,2 and T l,1 for the HX shown in FIGURE 1, since there is balanced flow in the HX, ΔTmax Δ Thl,2 = and, Δ Thl,2 =Δ Thl,1 (1 + Ntu) with, Tl,3 = Tl,2, Δ Thl,1 = Th,1 Tl,1, Δ Thl,2 = Th,2 Tl,2, Δ Tmax = Th,1 Tl,3 and, ( UA) Ntu = mc & Kee in mind that T h,1 is set by the ambient temerature (or 80-K liquid nitrogen recooling) and Tl,2 ( = Tl,3 ) is assumed to be maintained by the control system to satisfy the load requirements; so, ΔTmax is constant. Now, assuming that the ( UA ) scales with the mass flow by aroximately, the relationshi between the ( UA) at a given m& and the design (or observed) ( UA ) at the design (or observed) n ( UA) m& = where, n 0.67 ( UA) D m& D letting, ξ = mm & & D, ( UA) ( UA) Ntu Ntu = = = m& m& C ξ m& C ξ D D D then, n (1 n) (1 n) (1 n) D D m& D is,. (4) This means that if the mass flow decreases to 35% of the design flow (ξ = 0.35), the HX Ntu s increase by ~40%. Further, recalling the equation for Δ T hl,2, for large Ntu s (20+), this will have a diminishing effect on Δ T hl,2 and Δ T hl,1 (and therefore, T h,2 and T l,1 ). Since T l,1 and Th,2 change very little (and T h,2 even less), even over wide variations in mass

flow, we find that the characteristic ressure ratio, r = h,2 l,1 is aroximately and ractically constant. The secific inut ower to the comressor is, W& Tl,1 φ C ln( r, C) C wc = = (5) m& C ηi ηm where the comressor ressure ratio is h,1 Δh rc, = = r+, with, Δ h = h,1 h,2 l,1 l,1 Note that the comressor isothermal efficiency η i is rimarily a function of the comressor ressure ratio rc,, which is essentially constant [1]. The secific load exergy (reversible work) is, Ε L =Δ ε L =Δ hl T0 Δ sl (6) m& L with, Th,2 φ 1 Trx, = = 1 ηx (1 rx, ) T h,3 q L Δ hl = = C ( Tl,3 Th,3 ) = C Tl,3 1 m& L ΔThl,2 τ 2 = T l,2 Δ l = h,3 l,1, Δ L = h,3 l,3 h,2 r rx, = = h,3 l l,1 ( 1 +Δ / ) Δ / = 1 l,3 L l,1 ( 1 +Δ / ) h,3 l l,1 ( τ + 1) 2 T r, x ( τ + 1) T h,3 h,3 2 ΔL / l,1 Δ sl = C ln φ ln = C ln + φ l n 1 T l,3 l,3 T r, x ( 1 +Δl / l,1 ) (8) since, r constant, and Tl,3 constant, Δl ΔL constant, constant, rx, constant, Trx, constant l,1 l,1 Further, since changes in τ 2 are small, the secific load exergy (Δε L ) remains aroximately constant. So for all ractical uroses, the Carnot efficiency is ΕL Δε L ηcarnot = = Constant (9) W& w C C (7)

There are a few additional key observations from the above analysis: 1. Since Q= m& / ρ ~ m& / and m& ~ l and h ~ r l, the volume flow (and thus the velocity) at any oint in the system remains aroximately constant as the system ressure varies. This is readily aarent recalling that both the comressor and exander are constant volume flow devices. As such both the exander efficiency (i.e., one utilizing a variable brake) and the oil removal efficiency remain aroximately constant (recalling that the control scheme will not allow any comressor byass flow until the actual comressor suction ressure falls below the set oint ressure). So, the Floating Pressure Process does not ose any additional threat to the comressor system oil removal as long as the comressor byass is not used at reduced oerating ressures. 2. In the first order, the ressure loss Δ l reduces the refrigeration caacity (by reducing the comressor suction ressure and thus the mass flow rate) and Δ h increases the comressor inut ower by increasing the ressure ratio. 3. The minimum turn-down (load decrease) before throttling and/or load heaters would be necessary is determined by the minimum comressor suction ressure (i.e., the set oint, below which the comressor byass oens) and/or the exander oerational limit (equiment limitations such as bearing thrust or shaft natural frequency coincidence). In summary so far, the Floating Pressure Process allows the system ressures to adjust (as reviously described) at a nearly constant ressure ratio. In turn this rovides an essentially constant system Carnot efficiency over a wide load range (i.e., caacity turn-down from the design load). TS DIAGRAM DESCRIPTION FIGURE 2 deicts the TS diagram for a shield refrigerator (as in FIGURE 1) at the ideal design conditions. That is, assuming ideal gas behavior with constant secific heat (i.e., fluid ideality), neglecting stream/load ressure dros, heat leak and non-constant rotating machinery efficiencies (i.e., rocess idealities). From this diagram, there are several initial observations to be made: Y-axis is the natural logarithm of temerature Between any two arbitrary oints 1 and 2, the difference in s values between Δ s = ( s s ) = C ln( T / T) φ l n( / ), or these oints is, 2 1 { 2 1 2 1 } Δ s = C { n( T ) φ n( )} T T / T = / l r l r ; where r = 2 1 and r 2 1 So, at constant temerature (isotherms), Δ s = φ C l n( r), and, At constant ressure (isobars), Δ s = C l n( Tr) Sloe of isobars is equal to the secific heat at constant ressure ( C ).

If all comonents and loads were exactly as designed, the Floating Pressure Process would automatically adjust the system to oerate at the TS design condition. However, in ractice, no actual system oerates exactly as er the TS design conditions [4]. The main reasons are: 1. Given manufacturing and erformance tolerances, it is ractically imossible to exactly redict the erformance of the comonents (e.g., comressors, exanders, heat exchangers, ressure dros, heat leaks, etc.). 2. The system is designed for maximum caacity or for some rare oerating mode (like cool down) but is not required for normal oeration. 3. The margins allocated (e.g., load, system caacity) in the system design are in the actual oerating case either in excess of actual needs or not sufficient. 4. The load characteristics (e.g., refrigeration, liquefaction, actual load size) have changed or a different than as secified in the original design. FIGURE 2. TS Diagram of Shield Refrigerator CAPACITY MODULATION So, to effect caacity modulation while maintaining otimal efficiency (i.e., a roortional decrease in inut ower to a given reduction in load) at off-design rocess conditions, it is crucial to maintain a (nearly) constant entroy difference ( Δ s ) at each temerature throughout (from warm end to cold end) and to allow the mass flow ( m& ) to decrease roortionally with the load ( q L ).

Now, as is the case for the Floating Pressure Process, if r is constant, then recalling the observations of the TS diagram descrition, it is straight-forward to recognize that Δ s =Δ sc = ( sl,1 sh,1) and Trx, = ( Th,2 / Th,3 ) are (essentially) constant. With these, the following additional observations can be made, again referring to FIGURE 2: The mass secific availability to the cold box is essentially constant, and is equal to the area under the rocess ath from (h,1) to (l,1); it is aroximately equal to the mass secific isothermal comressor ower, w& Ci, ΔsC φ C Tl,1 ln( rc, ) (10) Mass flow ( m& ) is roortional to the system gas charge (mass) which is roortional to the absolute system ressure levels; i.e., h or l The mass secific (shield) load ( ql m& ) is essentially constant and, neglecting load irreversibilities, is equal to the area under the rocess ath from (h,3) to (l,3); it is aroximately, 1 ( ql m& ) =Δhx Δhhl,1 C {( TSR +ΔThl,2 ) ( 1 Tr, x ) ΔThl,1} (11) This floating ressure rocess is also a variable (mass) charge system. Referring to FIGURES 1 and 2 and Case #1 in FIGURE 3, as the load ( q L ) decreases and the mass out (MO) valve resonds by releasing gas back to gas storage, the rocess cycle translates right without size change (i.e., floats ) from the black lined cycle to the dashed red lined cycle. The ressure ratio remains essentially the same, but the system has decreased its gas charge, thereby decreasing the mass flow rate ( m& ). Therefore, the width and height of the rocess cycle remains essentially unchanged as the rocess resonds to changes in the load. That is, the entroy difference between (h) and (l) streams at each temerature is nearly constant. It should be recalled that T h,1 is fixed (by the comressor cooling water or 80-K liquid nitrogen re-cooling) and the control system is maintaining T = T at the desired set-oint. This load caacity modulation is done without l,3 SR introducing rocess mechanisms that in themselves roduce exergy losses. Further, it is imlicit that the exander adiabatic efficiency does not decrease with a load reduction (i.e., which is a good aroximation for a variable brake exander and no flow throttling). TABLE 2. Methods to Control Shield Refrigerator Caacity Case # Load Adjustment Mechanism Constraint Zero Comressor Byass ( m& 1 Comressor Discharge Pressure ( h ) BYP ); i.e., r = constant 2 Load Heater ( q HTR ) Comressor Suction Pressure ( l ) 3 Exander Inlet Valve ( Δ x, i ) Comressor Suction Pressure ( l ) 4 Comressor Discharge Pressure ( h ) Comressor Suction Pressure ( l ) 5 Exander Inlet Valve ( Δ x, i ) Zero Comressor Byass ( m& BYP ) 6 Exander Byass ( m& x, BYP ) Comressor Suction Pressure ( l ) Cases #2 to #6 in TABLE 2 and FIGURES 3 to 5 are traditional methods to achieve a turn-down (i.e., load reduction) in lant caacity. Also, Cases #2 to #4 and #6 maintain the same total comressor mass flow uon a decreasing load. Interestingly, only Case #2 follows the TS ath uon load turn-down (though it will be later shown that it is not

necessarily more efficient to do so). Case #2 adds a heat load, so that the total heat load equals the design load. Cases #3 and #5 throttle the exander inlet valve to waste the availability generated by the comressor that is not required by the load. Case #3 maintains the design comressor suction ressure, but accumulates comressor byass uon a decreasing load. Case #5 maintains zero comressor byass, but this results in the comressor ressure ratio increasing uon a decreasing load (even though mass flow decreases). Case #6 reduces refrigeration roduced by the exander by byassing mass flow around the exander. Case #4 decreases discharge ressure uon a decreasing load, but maintains the design comressor suction ressure. Cases #4 and #6 can resent comressor system oil removal roblems due to decreasing discharge ressure, but no reduction in mass flow. In some instances, a combination of Cases #2 to #6 are also used for caacity modulation. In summary, excet the floating ressure case (#1), all other cases waste some of the secific exergy develoed by the comressor and do not allow the inut ower to decrease, in a significant way, uon a decreasing load. As such, only the Floating Pressure Process (Case #1) can offer an essentially constant efficiency uon a decreasing load. FIGURE 3. TS Diagram of Floating Pressure Process (Case #1) and for Case #2 FIGURE 4. TS Diagram of Cases #3 & #4

FIGURE 5. TS Diagram of Cases #5 & #6 By relieving over-determinate rocess constraints, the Floating Pressure Process entreats the question of whether the TS design conditions are otimal for as-built hardware. That is, should the TS design conditions be coerced on as-built hardware to achieve otimal efficiency for the design load? The answer to this question can be found in examining off-design conditions by introducing a small variation in one of the equiment arameters. TABLE 3. Effect of Variations in Equiment Parameters Selected Equiment Parameter Less Consequence at Same Load Case # Than Design Value Pressure Ratio ( ) Mass Flow ( m& ) A HX Size ( Ntu ) Increase Increase B Exander Efficiency ( η x ) Increase Increase C Exander Flow Coefficient ( κ x ) Increase Decrease D Comressor Vol. Efficiency ( η v ) Decrease Increase Referring to TABLE 3 and FIGURES 6 and 7 (Cases A to D), the black lined cycle is the (intended) TS design condition and the dashed red lined cycle is how the actual cycle would oerate under the design load using the Floating Pressure Process for a small decrease from the design value for in the selected equiment arameter. If, instead of using the Floating Pressure Process (as discussed in Case #1), one of the load adjustment mechanisms in Cases #2 to #6 were imlemented in attemting to bring the off-design condition back to the TS design condition (i.e., the black line in FIGURES 6 and 7) one of two results would occur: 1. For the selected equiment arameter which is less than the design value, the shield load cannot be met and system Carnot efficiency is reduced. 2. For the selected equiment arameter which is greater than the design, the shield load can be met (matched) but at a system Carnot efficiency less than is ossible. r

FIGURE 6. TS Diagram of Cases A & B FIGURE 7. TS Diagram of Cases C & D It is imortant to notice that the Floating Pressure Process is not contingent on recise instrumentation for successful oeration, and in fact, will oerate quite contently regardless of the calibration or accuracy of the instrumentation. This is due to decouling secific values of rocess variables from resumed system load caacities. THE GANNI CYCLE The Floating Pressure Process is alicable not only for a shield refrigerator but also an isothermal refrigerator and/or a liquefier. FIGURE 8 deicts the suerosition of several shield refrigerators [3, 4], each oerating at a rogressively colder temerature level. This suerosition of shield refrigerators is the recycle flow or exander loo(s) in the traditional Claude cycle, and comrises 60 ercent or more (u to 90 ercent) of the total comressor flow. Of course, the shield load for each exander in the Claude cycle is the cooling required for the liquefaction load, heat exchanger losses, heat leak and any shield loads that may be resent.

FIGURE 8. Suerosition of Shield Refrigerators The majority of the helium refrigeration and liquefaction system exergy losses (u to aroximately 2/3 of the total loss [5]) are a result of comressor system inefficiencies. As such, it is imortant to integrate the comressor efficiency characteristics in the cycle design. As shown in [1], the otimum (maximum) isothermal (and volumetric) comressor efficiency is rimarily deendent on the ressure ratio, and it is around 3 to 4 for screw comressors. Now, consider if the exanders recycle flow is allowed to oerate using the Floating Pressure Process at the otimum ressure ratio, and the refrigeration load return is segregated from the exander recycle return, so as to maintain the lowest ossible constant refrigeration load temerature. Such an arrangement is the Ganni cycle. FIGURE 9 deicts a ossible multi-stage comressor arrangement for maximizing the exergy suly to the cold box and achieving good overall system efficiency within ractical ressure limits. Although it is not necessary (nor erhas ractical) to comletely segregate the exander recycle flow from the refrigeration load return flow, it is caable of achieving greater efficiency and stability. The guidelines for otimal arrangement of exanders in the cold box using the Carnot ste are given [2, 3]. Some more detailed alications to helium cycles are given in the various arrangements for Ganni Helium Process Cycle (US atents 7,278,280 & 7,409,834 and the atent ending for the Floating Pressure Process).

Although the Floating Pressure cycle can be alied to systems roduced by most manufacturers, there are some systems to which it can be alied over a wider range than the others. As such, there are two notes of caution in alying the Floating Pressure cycle: 1. When the comressor byass valve is used for reasons based on other control needs, the velocity through the oil removal system ceases to remain constant. In these cases the functionality of the oil removal system should be carefully checked. 2. For systems with at least a modest liquefaction load, an efficient design requires the exander mass flows to be relatively close to each other [2, 3]. In systems where this is not the case, careful attention should be given in balancing between trying to achieve more otimal Carnot stes (exander temerature level sacing) and a turbine s safe oerating range. FIGURE 9. Simlified Ganni Helium Process Cycle APPLICATIONS TO DATE Fundamental asects of the Floating Pressure Process were originally alied to the cryogenic system for the Suerconducting Suer Collider Laboratory (SSCL) string test lant (known as ASST-A) in 1992 to allow the refrigerator to resond efficiently to various modes of oeration, including magnet string quench recovery. In 1994-95, the Floating Pressure Process was alied to all four major cryogenic lants at Jefferson Lab (JLab), which are manufactured by different vendors. Later it was alied to Michigan State University (MSU) [6], the Sallation Neutron Source (SNS) [7], Brookhaven National Laboratory (BNL) [8] and for NASA at the Johnson Sace Center (JSC) [4]. In all cases, it has resulted in substantial imrovements in the system s efficiency, caacity, reliability and stability. Presently JLab licensed the Ganni Cycle Floating Pressure Process technology to Cryogenic Plants and Services a Division of Linde BOC Process Plants, LLC for world-wide commercialization.

SUMMARY & CONCLUSIONS In summary, the Ganni cycle Floating Pressure Process: 1. Provides a basis for an otimal design at maximum load, turn-down cases and mixed modes, addressing the comressor system as the major inut ower loss contributor 2. Provides a solution to imlement on as-built systems (existing or new) to imrove system efficiency, reliability, availability and load stability under actual loads and hel to imrove the exerimental envelo 3. Invalidates the hilosohy that oerating as-built systems at the TS design conditions is otimal by roerly identifying the fundamental rocess system arameters for control 4. Is a constant ressure ratio rocess cycle (as the Sterling Cycle is a constant volume rocess and the Claude Cycle is a constant ressure level rocess) and maintains the comressor efficiency for varying loads 5. Is a variable gas charge system, whose gas charge is automatically adjusted to satisfy the given load 6. Maintains a constant volume flow (and thus the velocity) at any oint in the system and reserves the exander efficiency and the oil removal effectiveness during the turn-down cases 7. Has been licensed by JLab to Linde Cryogenics, Division of Linde Process Plants, Inc. and Linde Kryotechnik AG for world wide commercialization ACKNOWLEDGEMENTS The authors would like to exress their areciation and thanks to the TJNAF management for their suort. This work was suorted by the U.S. Deartment of Energy under contract no. DE-AC05-06OR23177. REFERENCES 1. Ganni, V., et al, Screw Comressor Characteristics for Helium Refrigeration Systems, Advances Cryogenic Engineering 53A, 2008,. 309-315. 2. Knudsen, P., Ganni, V., Simlified Helium Refrigerator Cycle Analysis Using the Carnot Ste, Advances Cryogenic Engineering 51B, 2006,. 1977-1986. 3. Ganni, V., Design of Otimal Helium Refrigeration and Liquefaction Systems, CEC-ICMC 2009 - CSA Short Course Symosium. 4. Homan, J., et al., Floating Pressure Conversion and Equiment Ugrades of Two 3.5kW 20-K Helium Refrigerators, Advances Cryogenic Engineering 55, 2010, to be ublished. 5. Ziegler, B.O., Second Law Analysis of the Helium Refrigerators for the HERA Proton Magnet Ring, Advances Cryogenic Engineering 31, 1985,. 693-698. 6. McCartney, et al., Cryogenic System Ugrade for the National Suerconducting Cyclotron Laboratory, Advances Cryogenic Engineering 47A, 2002,. 207-212. 7. Arenius, D., et al., Cryogenic System for the Sallation Neutron Source, Advances Cryogenic Engineering 49A, 2004,, 200-207. 8. Than, R., et al., The RHIC Cryogenic System at BNL: Review of the modifications and ugrades since 2002 and lanned imrovements, Advances Cryogenic Engineering 53, 2008,. 578-587.